ANTIRETROVIRAL (ART) AND PREVENTION OF MOTHER TO - CHILD TRANSMISSION (PMTCT) THERAPIES ADHERENCE IN GREATER GABORONE, BOTSWANA: A MIXED METHODS HEALTH GEOGRAPHIC STUDY By Matlhogonolo Kelepile A DISSERTATION Submitted to Michigan State Universit y in partial fulfillment of the requirements for the degree of Geography Doctor of Philosophy 2019 ABSTRACT ANTIRETROVIRAL (ART) AND PREVENTION OF MOTHER TO - CHILD TRANSMISSION (PMTCT) THERAPIES ADHERENCE IN GREATER GABORONE, BOTSWANA: A MIXED ME THODS HEALTH GEOGRAPHIC STUDY By Matlhogonolo Kelepile Botswana is among the highest HIV/AIDS epidemic burden bearing countries in the world. While treatment for HIV was introduced in Botswana in 2002, the uptake and adherence of the antiretroviral (AR T ) a nd Prevention of Mother - to - Child Transmission (PMTCT) therapies are still relatively low. The goal of this research is to provide recommendations on how to improve adherence to AR T and PMTCT treatment regimens for adult men and women, and pregnant women an d new mothers living with HIV in Botswana. With improved ART adherence, there is high potential for a reduction in new HIV infections as a result of suppressed viral load. This reduction will contribute to the global goal of the UNAIDS (United Nations AIDS to eliminate new HIV infections by 2030. AR T a dherence is measured in two ways (a) biological ( serum CD4 count and viral load) and (b) using a composite score of biological and behavior, including reporting missed dosages . With this i n mind, this study will examine three dimensions of AR T non - adherence in a sampled population of HIV infected adult men (n=239) and women (n=419), including pregnant women or new mothers (n=67) attending 21 ART clinics in Greater Gaborone . First, ART adher ence was investigated at the individual - level to assess and evaluate differences by gender, age and geography. Second, PMTCT was investigated at the individual - level to identify challenges to adherence during pregnancy and early motherhood. Third, the clim ate of Botswana is hot, arid and dry and this study therefore, investigate the impacts of weather extremes on adherence behavior. The findings from this study show that men have lower adherence to ART than women; and men 30+years are at the highest risk of transmitting HIV infection . Lower adherence in men 30 - 40 years may in part, be explained by taking other drugs for co - conditions in addition to ARVs. With increasing age, lower adherence among men 50+ years is largely explained by single marital status an d the need to seek other companions for support later in life. Second, a mong pregnant women and mothers, a majority cohabit with HIV in fected men (n=45) and many (n=22 - partum, many women do not disclose their o as trust is evolving in the relationship. Not knowing a nificantly lead to poor adherence and increase d the risk of HIV transmission, however, declining trends in CD4 counts are observed, suggestive that there are difficulties adhering to ART for both partner and mother post - partum . Third , high ambient temperat ure is an important barrier to monthly clinic visits for both men and women living in urban, peri - urban and rural areas; and water rationing is a significant risk factor for non - adherence in rural areas. Being turned away by the clinic because of no water availability, is another important barrier for women including, pregnant and post - partum mothers and men living with HIV in Greater Gaborone. The findings from these three studies recommend (a) targeting men 30+ years living with HIV for medical and social support to ensure high ART adherence, to reduce HIV transmission; (b) provid e additional social services for women and their partners, during pregnancy and post - partum to enhance trust associated with HIV disclosure, to ensure high ART adherence; and (c) increase public health messaging about how to overcome environmental barriers to ART adherence, especially on extremely hot days and days of water rationing, while also ensuring water availability in all ART clinics. These recommendations may be extended b eyond Botswana, to other sub - Saharan African countries to optimize AR T treatment success and the elimination of HIV transmission over the next decade . iv This dissertation is dedicated to the memories of professors Florence Margai (1962 - 2015) and Michael Bernard Kwesi Darkoh (1940 - 2016). Although you were my inspiration to pursue my doctoral degree, you were unable to see my graduation. This is for you. v ACKNOWLEDGEMENTS First, I would like to give my outmost thank you to my advisor and chair, Dr. Sue Grady for the time and effort she generously spent in shaping these manuscripts. Her support, comments and guidance throughout this research, I am forever grateful. My committee members, Drs. Winkler, Zulu and Hughes, I appreciate the time you dedicated in helping me to dev elop these studies. Your mentorship and high standards helped me to develop and see to fruition my research ideas. The comments, guidance and reviews you provided has made me to see an idea turn into a full dissertation. I thank you all! My home departmen t of Geography, Environment and Spatial Sciences and Center for Gender and Environment (GENCEN) here at Michigan State University (MSU), I am grateful for the resources you provided for the data collection and dissertation completion. Ministry of Health a nd Wellness (MoHW) in Botswana, in no particular order, K. Motlhanka, L. Leu, T. Nong, Dr. Simoonga, Dr. Pone and others not mentioned by name here, thank you for the secondary data provided and IRB permissions. I would also like to thank all the District Health Management Teams (DHMTs) of Gaborone, Kgatleng, Kweneng East and South East districts for the permission to enter their respective clinics. My other appreciations go to Department of Surveys and Mapping, Botswana for the provision of shapefiles as w ell as A. Lefaphane and Ms. Ntshwarisang from Department of Meteorological Services, Botswana for the climatic data provided. Phase II clinic matron Mrs. V. Thuku and Mrs. L. Goitsemang from Tlokweng Main clinic, I thank you for always being there to answe r my endless questions on the ARV regimes. Mrs. M. Mmelesi from UN Botswana, thank you for the insightful interview and documents provided. I would make a mistake if I do not thank the patients and key informants from the 21 study clinics in Greater Gaboro ne who agreed to take part in this study. Your participation in the study was of great value. I would like to give a big thank you to my friends and research assistants: Loraine, Neo, Bame and Oarabile for the assistance in interviewing the participants. I would not have managed without you. Finally, I would like to thank my vi husband, Kabelo Kelepile who has been supportive throughout this journey. He endured to be alone a thousand miles away in Botswana while I was in MSU studying. He was always there when I needed him. He also followed up with any of the resources that I could not reach by myself in Botswana. I cannot thank you enough and am grateful that finally, I will be reunited with you once again ! vii TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ................................ ....... xi LIST OF FIGURES ................................ ................................ ................................ ................................ .... xiii CHAPTER 1: ANTIRETROVIRAL (ART) AND PREVENTION OF MOTHER - TO - CHILD TRANSMISSIO N (PMTCT) THERAPIES ADHERENCE IN GREATER GABORONE, BOTSWANA: A MIXED METHODS HEALTH GEOGRAPHIC STUDY ................................ ................................ ................................ ............................... 1 1.0.INTRODUCTION ................................ ................................ ................................ ................................ . 1 1.1. SOCIO - D EMOGRAPHIC DRIVERS OF HIV IN SOUTHERN AFRICA ................................ ....................... 3 1.1.1. Culture ................................ ................................ ................................ .............................. 3 1.1.2. Behaviors ................................ ................................ ................................ .......................... 5 1.1.3. Poverty ................................ ................................ ................................ ............................. 6 1.2. WEATHER AND CLIMATE IN BOTSWANA ................................ ................................ .......................... 7 1.3. PURPOSE OF STUDY ................................ ................................ ................................ ......................... 9 1.4. BACKGROUND ................................ ................................ ................................ ................................ 10 1.4.1. Biology of HIV ................................ ................................ ................................ ................. 10 1.4.2. Stages of HIV Disease ................................ ................................ ................................ ..... 11 1.4.3. Diagnostic Tests for HIV ................................ ................................ ................................ .. 11 1.4.4. Treatment for HIV in Botswana ................................ ................................ ...................... 12 1.4.5. PMTCT Treatment ................................ ................................ ................................ .......... 13 1.4.6. ARV Uptake and Adherence in Botswana ................................ ................................ ....... 13 1.4.7. PMTCT Uptake and Adherence in Botswana ................................ ................................ .. 16 1.4.8. Gaps in the Literature ................................ ................................ ................................ .... 18 1.5. STUDY GOAL ................................ ................................ ................................ ................................ .. 20 1.5.1. Objective 1 ................................ ................................ ................................ ..................... 20 1.5.1.1. Research Questions ................................ ................................ .......................... 20 1.5.2. Objective 2 ................................ ................................ ................................ ..................... 20 1.5.2.1. Research Questions ................................ ................................ .......................... 21 1.5.3. Objective 3 ................................ ................................ ................................ ..................... 21 1.5.3.1. Research Questions ................................ ................................ .......................... 21 CHAPTER 2: Antiretroviral Therapy (ART) Adherence among HIV Infected Adults: A Health Geographic Study in Greater Gaborone, Botswana ................................ ................................ ................................ . 22 2.0. I NTRODUCTION ................................ ................................ ................................ .............................. 22 2.1. GOAL OF THE STUDY ................................ ................................ ................................ ...................... 24 2.2. BACKGROUND ................................ ................................ ................................ ................................ 24 2.2.1. HIV Life - Course ................................ ................................ ................................ ............... 24 2.2.2. ART Adherence ................................ ................................ ................................ ............... 25 2.2.3. Barriers to ARV Adherence ................................ ................................ ............................. 29 2.3. THEORETICAL PERSPECTIVE ................................ ................................ ................................ ........... 30 2.4. STUDY DESIG N ................................ ................................ ................................ ............................... 32 2.5. DATA AND METHODS ................................ ................................ ................................ ..................... 32 2.5.1. Study Area ................................ ................................ ................................ ...................... 32 2.5.2. Primary Data Co llection ................................ ................................ ................................ .. 33 viii 2.5.3. Sampling Methodology ................................ ................................ ................................ .. 34 2.5.3.1. Districts selection (n=4) ................................ ................................ .................... 34 2.5.3.2. Clinics selection (n=21) ................................ ................................ ..................... 34 2.5.3.3. Reliability and Validity of Data Collected ................................ .......................... 38 2.5.4. Non - Adherence Definition ................................ ................................ .............................. 38 2.5.5. Descriptive Methods ................................ ................................ ................................ ...... 39 2.5.6. Analytical Methods ................................ ................................ ................................ ......... 40 2.6. RESULTS ................................ ................................ ................................ ................................ ......... 41 2.6.1. Life Course of People Living with HIV and ART - Initiation Post - HIV Diagnosis ................. 41 2.6.1.1. Women ................................ ................................ ................................ ............ 41 2.6.1.2. Men ................................ ................................ ................................ ................. 44 2.6.2. Non - Adherence Levels ................................ ................................ ................................ .... 51 2.6.3. Risk Populations and Risk Factors for Non - Adhe rence ................................ ................... 54 2.6.4. CD4 Count as a Proxy of Non - Adherence ................................ ................................ ........ 58 2.7. DISCUSSION ................................ ................................ ................................ ................................ ... 62 2.7.1. Limitations ................................ ................................ ................................ ...................... 64 2.8. CONCLUSIONS AND RECOMMENDATIONS ................................ ................................ ..................... 65 CHAPTER 3: Prevention of Mother - to - Child Transmission (PMTCT) Adhere nce and HIV status disclosure in Pregnant Women and New Mothers in Greater Gaborone ................................ ............................... 67 3.0. INTRODUCTION ................................ ................................ ................................ .............................. 67 3.1. PURPOSE OF STUDY ................................ ................................ ................................ ....................... 68 3.2. BACKGROUND ................................ ................................ ................................ ................................ 68 3.2.1. Maternal and Child Mortality in Botswana ................................ ................................ ..... 68 3.2. 2. ART and PMTCT Programme in Botswana ................................ ................................ ...... 69 3.2.3. Challenges to PMTCT in Sub Saharan Africa and Botswana ................................ ............ 70 3.2.4. Reoccurring Barrier to PMTCT: HIV Status Disclosure ................................ ..................... 76 3.2.4.1. HIV Status Disclosure ................................ ................................ ....................... 76 3.2.5. Constraints and Promoters of HIV Status Disclosure of Wome n ................................ ..... 77 3.3. GOAL OF THE STUDY ................................ ................................ ................................ ...................... 80 3.4. THEORETICAL AND CONCEPTUAL FRAMEWORK ................................ ................................ ............ 81 3.4.1. Health Belief Model ................................ ................................ ................................ ........ 81 3.5. STUDY DESIGN ................................ ................................ ................................ ............................... 82 3.6. DATA AND METHODS ................................ ................................ ................................ ..................... 83 3.6.1. Study Area ................................ ................................ ................................ ...................... 83 3.6.2. Primary Data Collection ................................ ................................ ................................ .. 84 3.6.3. Sampling Methodology ................................ ................................ ................................ .. 86 3.6.3.1. Reliability and Validity ................................ ................................ ...................... 86 3.6.3.2. Ethical Review ................................ ................................ ................................ .. 87 3.6.4. Descriptive Methods ................................ ................................ ................................ ...... 87 3.6.5. Thematic Analysis ................................ ................................ ................................ ........... 88 3.7. RESULTS ................................ ................................ ................................ ................................ ......... 89 3.7.1. Non - Adherence Levels ................................ ................................ ................................ .... 89 3.7.2. HIV Status Disclosure ................................ ................................ ................................ ...... 91 3.7.3. CD4 Count and Year of HIV diagnosis ................................ ................................ ............. 93 3.7.4. Themati c Analysis: Qualitative analysis ................................ ................................ .......... 96 3.7.4.1. Theme 1: Conflict in Autonomous Decision - Making ................................ .......... 96 3.7.4.2. Theme 2: Adherence and Compli ance ................................ ............................... 99 ix 3.8. DISCUSSION ................................ ................................ ................................ ................................ . 101 3.8.1. Limitations ................................ ................................ ................................ .................... 105 3.9. CONCLUSIONS AND R ECOMMENDATIONS ................................ ................................ ................... 105 CHAPTER 4: Impacts of High Ambient Temperatures and Water Rationing on Antiretroviral (ART) and Prevention of Mother - to - Child Transmission (PMTCT) Therapies Adherence in Great er Gaborone, Botswana ................................ ................................ ................................ ................................ ............ 107 4.0. INTRODUCTION ................................ ................................ ................................ ............................ 107 4.1. PURPOSE OF THE STUDY ................................ ................................ ................................ .............. 108 4.2. BACKGROUND ................................ ................................ ................................ .............................. 108 4.2.1. Water, Sanitation and Hygiene (WASH) Barriers ................................ .......................... 110 4.2.1.1. Physical Infrastructure Barriers ................................ ................................ ...... 112 4.2.1.2. Gender Differences in HIV ................................ ................................ .............. 114 4.2.1.3. Gender and Water ................................ ................................ ......................... 115 4.2.2. Bot swana ................................ ................................ ................................ ...................... 117 4.2.2.1. Water Rationing ................................ ................................ ............................. 120 4.2.2.2. Power Infrastructure and Water Extraction in Botswana ................................ 121 4.2.3. Gaps in the Literature ................................ ................................ ................................ ... 122 4.3. GOAL OF THE STUDY ................................ ................................ ................................ .................... 122 4.4. DATA AND METHODS ................................ ................................ ................................ ................... 123 4.4.1. Study Area ................................ ................................ ................................ .................... 123 4.4.1.1. Greater Gaborone ................................ ................................ .......................... 125 4.4.1.2. Study clinics (n=21) ................................ ................................ ........................ 126 4.5. STUDY DESIGN ................................ ................................ ................................ ............................. 127 4.5.1. Primary Data Collection ................................ ................................ ................................ 127 4.5.2. Sampling M ethodology ................................ ................................ ................................ 129 4.5.2.1. Reliability and Validity of Data ................................ ................................ ....... 131 4.5.2.2. Ethics Overview ................................ ................................ .............................. 131 4.5.3. Non - Adherence Definition ................................ ................................ ............................ 131 4.5.4. Descriptive Methods ................................ ................................ ................................ .... 132 4.5.5. Analytical Methods ................................ ................................ ................................ ....... 133 4.5.6. Thematic Analysis (Qualitative) ................................ ................................ .................... 133 4.6. RESULTS ................................ ................................ ................................ ................................ ....... 134 4.6.1. Descriptive statistics ................................ ................................ ................................ ..... 139 4.6.2. Health Workers Analysis on Water issues, High Temperatures and Non - adherence: Qualitative Results ................................ ................................ ................................ ................. 140 4.6.2.1. Clinic Operatio ns and Water Issues ................................ ................................ 140 4.6.2.2. Clinic Attendance and the Impact of Heat ................................ ....................... 141 4.7. DISCUSSION ................................ ................................ ................................ ................................ . 142 4.7.1. Limitations ................................ ................................ ................................ .................... 145 4.8. CONCLUSIONS AND RECOMMENDATIONS ................................ ................................ ................... 145 CHAPTER 5: Discussions and Conclusions ................................ ................................ ............................ 147 5.0. DISCUSSION ................................ ................................ ................................ ................................ . 147 5.1. CONCLUSION ................................ ................................ ................................ ................................ 149 5.2. RECOMMENDATIONS ................................ ................................ ................................ ................... 151 5.3. FUTURE RESEARCH ................................ ................................ ................................ ....................... 151 x APPENDICES ................................ ................................ ................................ ................................ ........ 153 APPENDIX A: Adults on ART Survey ................................ ................................ ........................ 153 rvey) ................................ ................................ ................. 156 ................................ ................................ .................... 159 REFERENCES ................................ ................................ ................................ ................................ ........ 162 xi LIST OF TABLES Table 1. HIV Diagnos es Across the Life Course by Current Age and Low Estimate 1 of Years ARV Initiated Post - HIV Diagnosis Sample of Women (n=419) Living with HIV in Greater Gaborone, Botswana, 2016 Table 2. HIV Diagno ses Across the Life Course by Current Age and High Estimate 1 of Years ARV Initiated Post - HIV Diagnosis Sample of Women (n=419) Living with HIV in Greater Gaborone, Botswana, 2016 Table 3. HIV Diag noses Across the Life Course by Current Age and Low Estimate 1 of Years ARV Initiated Post - HIV Diagnosis Sample of Men (n=239) Living with HIV in Greater Gaborone, Botswana, 2016 Table 4. HIV Diagn oses Across the Life Course by Current Age and High Estimate 1 of Years ARV Initiated Post - HIV Diagnosis Sample of Men (n=239) Living with HIV in Greater Gaborone, Botswana, 2016 Table 5 . Demograph ic Characteristics of Sampled Women (n=419) and Men (n=239) Living with HIV in Greater Gaborone, Botswana 2016 . 52 Table 6 . Behavioral Characteristics of Sampled Women (n=419) and Men (n=239) Living with HIV in Gr eate r Gaborone, Botswana, 2016 Table 7 . Health Care Facility Characteristics for Sampled Women (n=4 19) and Men (n=239) Living with HIV Table 8 . Unadjusted Odds Ratio Estimates of Non - Adherence for Men (n=239) vs. Women 1 (n=352) Living with HIV, Greater Gaborone, Botswana, 2016 5 5 Table 9. Odds Ratio Estimates of Non - Adherence for Men (n=239) vs. Women 1 (n=352) Liv ing with HIV, controlling for the Number of Years ARV Initiated Post - HIV Diagnosis, Greater Gaborone, Botswana, 2016 . .5 5 Table 10. Unadjusted Odds Ratio Estimates of Non - Adherence for Men (n=239) vs. Women (n=352) by Age Group . 5 6 Table 11. Adjusted 1 Odds Ratio Estimates of Non - Adherence for Men (n=239) vs. Women 2 (n=352) by Age Group 5 7 Table 12. D escriptive Statistics of CD4 Count for Sample of Women 1 (n=352) and Men (n=239), Greater Gaborone, Botswana, 2016 xii Table 13. Linear Regression Adjusted Estimates of CD4 Count for a Sample of Men 30 - 39 Years (n= 60) Living with HIV, Greater Gaborone, Botswana, 2016 6 0 Table 14. Health Profile of Men 30 - 39 Years (n= . 6 1 Table 15. Linear Regression Adjusted Estimates of CD4 Count for a Sample of Men 40 - 49 Years (n=124) Living with HIV in Greater Gaborone, Botswana, 2016 .6 1 Table 16. Linear Regression Adjusted Estimates of CD4 Count for a Sample of Men 50+ Years (n=56) Living with HIV in Greater Gaborone, Botswana, 2016 .. 6 1 Table 1 7. Demographic Characteristics of Mothers (n=67) vs. Non - Mothers (n=377) Living with HIV in Greater Gaborone, Botswana 2016 9 0 Table 18 . Maternal Characteristics by Adherence vs. N on - Adherence for a Sample of Women (n=444) in Greater Gaborone, 2016 .9 0 Table 19 1 Status, Greater Gaborone, Bots wana, 2016 . 91 Table 20 1 for Mothers Enrolled in PMTCT by Year of HIV Diagnosis and Number of Children . 92 Table 21 ges by Marital Status, Greater Gaborone, Botswana, 2016 93 Table 22 : HIV S tatus Di sclosure of Mothers and Family Member L iving with HIV 93 Tabl e 23 . Environmental Characteristics and ARV Adherence vs. Non - Adherence 1 in a Sample of Wom en (n=419) and Men (n=239) in Greater Gaborone, Botswana, 2016 134 Table 2 4 . Environmental Characteristics and ARV Non - Adherence Stratified by Location in Gr eater Gaborone, Botswana, 2016.. . ........................................... ........................................................................ . 1 35 Table 25 . Unadjusted Estimated Odds of Non - Adherence if Missed an Appointment due to Heat by Travel Model, Greater Gaborone, Botswana, 2016 1 37 Table 26 . Multilevel Model Estimating the Effects of Environmental Variables on Adherence among Adults Living with HIV attending Clinics in Greater Gaborone, 2016 38 Table 27. Multilevel Model Estimating the Effects of Environmen tal Variables on Adherence among Adults Living with HIV attending Rural Clinics in Greater Gaborone, 2016 . . ... 1 39 Table 28 . Environmental Variables Impacting Mothers (n=67), Children and Infants, Greater Gaborone, Botswana, 2016 1 39 xiii LIST OF FIGURES Figure 1: Map of Study Area Showing ART Health Facilities and D istricts in Greater Gaborone, Botswana 36 Figure 2 : Women and Men Ages 20 - 29 Y 47 Figu re 3: Women and Men Ages 30 - 39 Y 48 Figu re 4: Women and Men Ages 40 - 49 Y . 49 Figure 5: Women and Men Ages 50+ Years 50 Figure 6 : Histogram of CD4 Count for a Sample of Women (n=352) Living with HIV in Greater Gaborone, 2016 58 Figure 7 : Histogram of CD4 Count for a Sample of Men (n=239) L iving with HIV in Greater Gaborone, 2016 59 Figure 8 : Fitted Plot of CD4 Counts for a Sample of Men 50+ Years (n=56) Marital Status Single vs. Cohabiting, Greater Gaborone, Botswana, 2016 62 Figure 9 : CD4 Count by Year of Diagnosis, Sample of Mothers (n=67) Living with HIV in Greater Gaborone, Botswana, 2016 94 Figure 10 : CD4 Count by Year of HIV Diagnosis for a Sample Status, Greater Gaborone, Botswana, 2016 95 Figure 1 1 : CD4 Count by Year of HIV Diagnosis for a Sample of Mothers (n=67) and HIV Disclosure Status to Partner, Greater Gaborone , 2016 95 Figure 1 2 : Average M o nthly Temperature and Rainfall T rends in Africa from 1991 - 2016 1 09 Figure 13 . Average Monthly Temperature T rends in Botswana from 1991 - 2016 1 19 Figure 14 : Maximum Monthly Temperature T rends in Gaborone, Botswana from 1985/86 - 2015/16 12 0 Fig ure 15 : Map of Botswana S h owing Wet (Okavango Delta) and D ry (National Parks and Kal ahari Desert) Areas and Mining P laces 124 Figure 16 : The Responses of Patient s Attending 21 Clinics in Greater Gaborone on w hether Water Rationing Affected their Health at Home 1 3 6 1 CHAPTER 1 : ANTIRETROVIRAL (ART) AND PREVENTION OF MOTHER - TO - CHILD TRANSMISSION (PMTCT) THERAPIES ADHERENCE IN GREATER GABORONE, BOTSWANA: A MIXED METHODS HEALTH GEOGRAPHIC STUDY 1.0. INTRODUCTION In the early 1970s, the first cases of human immunodeficiency virus [HIV] and Acquired immunodeficiency syndrome [AIDS] were reported in Kinshasa, Democratic Republic of Congo (DRC). The speculation was that HIV was brought upon by a traveler from Cameroon via river transport into the DRC [Joint United Nations Programme on HIV and AIDS] (UN AIDS), 2012). A decade later, the incidence of HIV skyrocketed factors underlying the early diffusion of HIV were widespread rural - to - rural and rur al - to - urban labor migration, the low social - status of women, the high - ratio of men to women in the cities and the availability of sex workers (Pepin, 2011; UNAIDS 2013 ). In the mid - 1980s in East Africa, sex workers testing positive for HIV reached 36.7%; a nd disturbingly, when seroprevalence studies were conducted among pregnant women , HIV prevalence was 8.6% (Maticka - Tyndale, 2012). HIV epidemics however, were not observed in West African countries, such as Cameroon and Gabon where HIV was known to origina te, probably because of the long distances and difficult travel between cities, which limited sexual networks such as those that fueled the spread of HIV in East Africa (UNAIDS, 2013). In the late - 1980s, there was a third shift in HIV diffusion, this tim e to countries in Southern Africa, including Botswana, Malawi, Zambia and Zimbabwe , where HIV epidemics were emerging (World Health Organization [WHO], 2011). Factors contributing to the southern diffusion of HIV included the in - delivery goods into and across the region, high levels of unemployment and labor migration, such as miners to South Africa and other migrants, as well as soldiers (Renton, 2006; Reiners & Watkins, 2010) and finally, the culture of polygynous relationships and multiple concurrent partners increased local transmission (Levinsohn, 2011; Hanson, 2008; Mathers et al . , 2008) . In Southern 2 Africa, the sexual networks initiated by men were generally accepted, and even encouraged in some cultures. Women were more vu lnerable to infection if sex was unprotected due to potential of tear and blood entry into her body. While condom use was the most effective mode of HIV transmission prevention, women were not able to insist on condom use by their male partners because of the subservient role of women , when making decisions in culture. Men dominated the sexual relationship a difficult cultural practice in terms of limitin g HIV transmission, that continues to be observed today. The HIV epicenter of adult HIV prevalence in the world today is in Southern Africa, specifically the four neighboring countries: Eswatini (27.4 per 100 adults), Lesotho (23.8), Botswana (22.8) and S outh Africa (18.8) (UNAIDS, 2018). In 2015 UNAIDS established the 90 - 90 - people living with HIV will know their HIV status, 90% of all people with diagnosed HIV infection will receive antiretr oviral therapy (ART) and 90% of all people receiving ART will experience viral suppression (UNAIDS, 2017) thereby, eliminating HIV transmission. Of the four countries with the highest HIV prevalence, Eswatini and Botswana have met their 90 - 90 - 90 programmat ic goals (UNAIDS, 2019) ; however, both countries continue to have a substantial number of new infections and AIDS - related deaths . In 2018, Eswatini reported n= 7,800 (incidence rate, 8.62 per 100 population) and Botswana n= 8,500 (incidence rate, 4.36) new H IV infections. To assess if these new HIV cases are contributing to growing or shrinking HIV trends, t he Incidence Prevalence Ratio (IPR) is used , with an IPR < 0.03 indicati ng the HIV epidemic is expected to decline (UNAIDS, 2019) . In 2018 Eswatini, the IPR = 0.037 and Botswana, IPR = 0.022 suggestive that both countries can expect a decline in their HIV epidemics, with Botswana expected to decline slightly faster than Eswatini. Another indicator , the Incidence Mortality Ratio (IMR) identifies a point at which HIV related health care costs diminish , with an IMR < 1.0 ( in a scenario of low mortality ) indicating a decline in the HIV population , thereby, the ability to reduce the costs of ART 3 services. In Eswatini , the IMR = 3.25 (case - fatality rate = 11.4 deaths per 1,000 HIV infected individuals ) and Botswana, IMR = 1.77 (case - fatality rate = 12.9 7 ). The case - fatality rates for Eswatini and Botswana are higher than that of South Africa (9.22) but lower than that of Lesotho (17.9) , suggestive of the need t into the near future . Excluding South Africa because of its high population, Botswana has the highest number of people living with HIV (n=370,000) , which may be a reflection of surveillance an d reaching its 90 - 90 - 90 targets . However, Botswana continues to have a high number of new infections and a relatively high number of AIDS related death s . For these reasons, Botswana is an extremely important country to study HIV beyond success; specifically, the population(s) that continue to transmit a nd acquire new infections, and the reasons behind this transmission most importantly, AR T treatment adherence . The findings from these investigations will help to further re duce HIV transmission and AIDS - related deaths , while also promoting longer life expectancy and quality of life in Botswana. First, more discussion is provided on the structural and physical drivers of the HIV epidemic in Southern Africa , while also providing a more in - depth understanding of those structures and physical drivers in Botswana . 1.1. SOCIO - DEMOGRAPHIC DRIVERS OF HIV IN SOUTHERN AFRICA 1.1.1. Culture It has been documented widely in the literature that int ergenerational sex and cultural roots are major drivers of the persistence of HIV particularly when Southern Africa became the AIDS belt (endemic) (Leclerc - Madlala 2008; MacPhail and Campbell 2001; Hunter 2007; Auerbach et al . , 2011 and Yeboah 2007). Lecle rc Madlala (2008) divides this concept of intergenerational sex in two ways: age disparate i.e., men who are five years older than their female partners and intergenerational i.e., men ten years older than their female partners. Throughout Southern Afric a and Botswana, high HIV incidence have 4 been found among older men and young women. This is explained by the practices of intergenerational sex, where men with higher disposal incomes buy sex from younger women. In most cases these acts are not considered prostitution, because sometimes these relationships last months or longer. In the Botswana context, older men tend to have higher disposal incomes because they have established job experience, hence higher incomes and some men have investments such as catt le (described below), which are also a sign of wealth in this region. In these relationships, condom negotiation for young women is very rare because older men demand respect from women of younger age (Malema, 2012). Leclerc - Madlala (2008) argues that unli ke in the literature that proposes women are driven into intergenerational sex because of poverty (Dow and Essex, 2010; Muula, 2008; Keetile, 2014; UNFPA, 2017; Pascoe et al . , 2015). Leclerc - hings such as cellphones needs (Nkosana and Rosenthal, 2007). The young women, because of the cultural attitudes, beliefs and practices embedded in them i.e., that the man is a provider, tend to be either in school or unemployed because they have reliable sources of income (Smith , 2002). There is also the tendency to blame the government for everything in Botswana (Malema, 2012). When graduates cannot fi nd jobs, the where the people are the receivers of the government, in the same way as women are receivers in intergenerational sex (Cullen, 2006). Othe r studies from Zimbabwe which have marriage patterns comparable with many other parts of the region, found young married women with partners , who were 16 or more years older , were at three times greater risk of HIV infection than those with partners 0 15 y ears older than themselves (Schaefer et al . , 2017). Culture embedded in terms of social roles and wealth place a significant role in shaping the geographic patterns of HIV prevalence (Kandala et al . , 2012). Mining towns and military camps are the landscap es of HIV in Botswana. It is in these places where HIV prevalence is as high as 43 .0 % ( National AIDS Coordinating 5 Agency [ NACA ] , 2015). In these mining towns, because of their remoteness, men tend to live by themselves, perhaps only checking on their famil ies once a month. These men are also paid higher salaries because they have extra benefits such as housing allowance, and scarce skill and remote area allowance, and with this income and far from their families are willing to participate in sex with young women who are willing to travel to these places. As the case of Botswana, Corno and Walque (2012) found that in Lesotho and Eswatini , women married to miners we re more likely to engage in extra - marital sex during absence. Additionally, mine rs we re less likely to use condoms, especially during extra - marital relationships (Corno and Walque, 2012). 1.1.2. Behaviors Alcohol abuse is another driving factor for HIV in Botswana because there is no age drinking limit (Keetile, 2014). Young women who could be younger than 18 years, have access to bars where they can meet older men who will buy them drinks and later lead to unsafe sex because of the impaired ability to negotiate safe measures. The legal age for purchase and consumption of alcohol in Es watini is 18 years (WHO, 2014). Even though the legal age limit is 18, young adults (less than legal age) still have access to alcohol and substance abuse in Eswatini . Use of alcohol and drugs appears most common in high school students and out - of - school y outh. The latter who reported drinking alcohol, 58.9% do so daily or weekly (Swaziland Ministry of Sports, Culture, and Youth Affairs Report, 2015). As mentioned above in the case of Botswana, . Status disclosure is a huge problem among HIV - infected pregnant women (as will be investigated in - depth in Chapter 3 of this dissertation) . There are various reasons why many pregnant and new mothers do not disclose to their partners and families. In Sou th Africa, 21.3% (n=80) of healthcare workers thought that HIV - infected individuals should not have children; 53.8% thought HIV - infected individuals were having too many children; and 46.3% thought that social grants were an incentive for HIV - infected wome n to have children (Mnyani and McIntyre, 2013). Some women were afraid of telling partners because they we re accused of bringing the virus home (Visser 6 et al . , 2008; Makin, 2010). In Botswana, most mothers were scared to disclose to families because they w ere afraid to be questioned about their lack of breastfeeding (Clark, 2006). 1.1.3. Poverty It is also evident however, that one cannot fight HIV without also addressing poverty, socioeconomic inequality and the broader issues of economic development (Aval os et al . , 2016). Poverty interplays a major role in sexual - risk vulnerability, and reluctance of people to seek protection from HIV as well as medical attention. Young women are often unemployed in Botswana, particularly after they finish their college ed (IMF, 2016). However, unemployment for the youth aged between 18 - 35 years is estimated at 25.1% and among university degree holders at 11.4% (Statistics Botswana, 2016). As regards to gender comparison, the female unemployment rate is higher (19.1%) than that of their male counterparts (16.3%) (Statistics Botswana, 2016). The Botswana Mu lt i Topic Household Survey of 2015/16 survey estimated youth aged between 15 an compared to 23.6% for males. Unemployment can drive young women to risky behaviors of transactional women who are less than 25 years and never married would account for the 55 .0 % of all the new HIV infections, while men of the same age ranges, as well as women older than 25 years would account for about a quarter of HIV incidence (Botswana Pilot Inciden ce Report, 2015). Unemployment in neighboring countries is very common particularly in Lesotho , where not only women are involved but young men who turn to being oung people in Lesotho. Poverty drives these young people to take up herding livestock as a full - time occupation, meaning many are deprived of formal education and lack access to health services, including HIV prevention, treatment and care (Lesotho Minist ry of Health, 2015). 7 1.2. WEATHER AND CLIMATE IN BOTSWANA Botswana is located in the subtropical high - pressure belt of the Southern Hemisphere in the interior of Southern Africa and away from oceanic influences (Kenabatho, 2016). There are 17 synoptic sta tions in Botswana, distributed in strategic places across the country, which record weather parameters (temperature, wind, humidity, pressure and rainfall). In addition, there are about 500 rainfall stations in the country that measure and record rainfall only. All the stations recorded warmer maximum temperatures for the period of 2015 , than in the previous year 2014 (Statistics Botswana, 2016). The highest mean monthly maximum temperature of the period (April to September) was 34.9C for Shakawe (North) in September 2015, while the lowest was 21.2 C for Tsabong (Kgalagadi District) in June 2015 (Statistics Botswana, 2016). Precipitation patterns are highly variable, temperatures are consistently high and water resources are very scarce (Kenabatho, 2016). Dr ought is therefore, a recurring element of Botswana; one in four years being a drought year. As climate change increases the chances of droughts and dry spel ls in S outhern Africa, the complex relationship between food security and HIV/AIDS can make matters worse for disaster management in two ways. First, the nutritional requirements of people living with HIV/AIDS are higher: up to 15 .0 % greater for protein an d 50 .0 % greater for energy (Piwoz and Preble , 2000). Secondly, HIV/AIDS amplifies the effect of drought on nutrition (Mason et al . , 2005). This creates a vicious circle: inability to provide adequate nutrition in times of drought weakens the immune system and increases susceptibility to opportunistic infections , such as tuberculosis, which in turn undermine the overall nutritional status of people experiencing this depravity in Botswana . In Botswana, it is believed that culture and climate /weather play impo rtant roles in HIV transmission and this research will further investigate these roles in Chapter 4 of this dissertation . The s patial variation of water availability in Botswana can increase the chances of HIV transmission among the north ern (wet and rainy ) and south ern (hot and dry) regions 8 of the country. This mismatch can provide for a circular migration where those in the south who have limited water resources move to the northern part of the country , increasing the risk of HIV transmission . Once peopl e have HIV it is important that they adhere to treatment regimes. The relationship between water scarcity, scorching temperatures and low treatment adherence is complex and may operate in several ways, including missing clinic appointments because of embar rassment by not being able to bathe , longer walk - times for water collection and missed appointments , scorching temperatures and unable to walk to the clinic, especially if the person is severely immune - compromised or pregnant or a new mother , conditions o f which are highly vulnerable because walking becomes impossible in scorching temperatures. In poor remote areas, where air conditioning in health facilities is a luxury becomes an inhibiting factor of adherence among the vulnerable poor. Since the adoptio n of Millennium Development Goals [MDGs], especially the right to human health, there has been a steady increase in the provision of antiretroviral (ARV) therapy in Botswana and other affected regions around the world. Unfortunately, despite intense public ity campaigns to expand the global use of ARVs , a gap remains between the governmental roll out efforts (coverage) and those who actually have access or receive the medication s . In particular, there are concerns with loss to follow - up rates, which have st eadily increased among adults (NACA, 2015) . There are also high treatment failure rates for adults (an increase in first line adult failure rate from less than 6 .0 % recorded in 2012 , to over 10 .0 % in 2013) due to the challenges of adherence to counselling and monitoring (NACA , 2015). The ART coverage is also compromised by the low testing rate of 63.7% (according to Botswana AIDS Impact Survey [BAIS] IV of 2013) of the population aged 15 - 49 years , who in the 12 months preceding the survey had a HIV test an d received result s (NACA, 2015). Adherence to ART is important for sustained HIV suppression, improved health and longevity and importantly from a public health perspective, reduced risk of drug resistance ( Chesney, 2006; WHO, 2003). No research to my kno wledge, has focused on how climate/weather extremes in Botswana , impacts ART and Prevention of Mother - to - Child Transmission [PMTCT] treatment adherence. 9 1.3. PURPOSE OF STUDY The purpose of this research is to assess ART adherence levels in population(s) at risk of HIV transmission and to provide recommendations on how to improve adherence to ART and PMTCT treatment regimens for adult men and women, and pregnant women and new mothers living with HIV in Botswana. ART is a lifelong commitment where as , the p erson should adhere completely in order to achieve viral suppression. A reduced viral load decreases significantly the risk to transmit new HIV infections. With this in mind, this study will examine three dimensions of ART and PMTCT non - adherence for peop le living with HIV in urban , peri - urban and rural areas of Botswana: To investigate the relationships between ART and PMTCT non - adherence and immunodeficiency at the individual - level, and socioeconomic, cultural and environmental factors, including health care services at the contextual - level; and to investigate the role of high ambient temperatures and water scarcity and rationing on ART and PMTCT adherence, for people living with HIV in Botswana. The format of this dissertation will begin with a descripti on of the biology of HIV, the states of HIV disease, routine diagnostic tests and ART and PMTCT treatment protocols. I will then discuss the status of and concerns around adherence to ART and PMTCT adherence to justify the need for this study. Following this Background Section, I will present in more detail the objectives that will be implemented to meet the goal of this research. Each objective will be presented in the format of a peer - reviewed paper , that will include the theory and conceptual framewor k, data and methods proposed for these studies and the results and conclusions. The final chapter of this dissertation will be a summary of the overall findings from these three investigations, and their policy and programmatic impacts. Finally, recommend ations for future research to monitor ART and PMTCT adherence Botswana, is provided. 10 1.4. BACKGROUND 1.4.1 . Biology of HIV Diseases are interruptions or disorders in body functions, systems or organs (Will iams and Williams, 2005; Merrill, 2011). They can either be infectious or non infectious, and the infectious diseases can be further classified as emerging and re - emerging diseases (ERIDs) (Margai, 2010; Merrill, 2011). Emerging infectious diseases are ill nesses that have newly appeared among people or that have existed but are increasing in incidence and geographical space (Morse , 1995; Margai , 2010). HIV was appropriately classified as an emerging infectious disease that evolved into a pandemic epidemic f ollowing its discovery in the early 1980s. Today HIV is considered endemic due to its persistence in populations throughout the world, and most prominently in Southern Africa. HIV is a retrovirus that occurs as two types: HIV - 1 and HIV - 2 (described below) . Both types are transmitted through direct contact with HIV - infected body fluids, including blood transfusion, sexual contact, mother - to - child transmission during pregnancy , delivery and breast milk, and sharing needles contaminated with HIV. When HIV ent ers the body, the virus attaches itself to dendritic cells, usually found near the mouth, upper gastrointestinal tract, vagina and rectum (Centers for Disease Control and Prevention (CDC), 2015). The virus can infect many cells in the body, including brain cells but it mainly targets the T - lymphocytes or cluster of differentiation 4 [CD4] cells also referred to as helper cells. Unlike other viruses for which there are successful vaccines, HIV quickly integrates itself into the deoxyribonucleic acid [DNA] of the host cell, where, in some cells, it remains latent and essentially invisible to the immune system (Chun et al . , 1998). Because latency is established very early - i.e. within days to weeks after infection, the window period in which HIV remains vulnera ble to eradication through immune responses is very short (Chun et al . , 1998). Once latency is established, it has not yet been possible to eradicate the virus, even in patients receiving highly active antiretroviral therapy (HA A RT) for extended periods of time (Chun et al . , 1998). Instead treatment will be ongoing to reduce the viral load of a person over their life span with HIV. 11 HIV - 1 is more common worldwide and in this study, the focus is on HIV - 1, thus HIV herein refers to HIV - 1. HIV - 2 is found mostl y in West Africa. HIV - 2 is less easily transmitted than is HIV - 1 (Pepin, 2011). The lower infectivity of HIV - 2 is likely related to lower ribonucleic acid [RNA] levels. In addition, HIV - 2 infection is characterized by higher CD4 cell counts and lower viral load. While HIV - 2 can be transmitted from an infected mother to her child, the rate of transmission is lower than with HIV - 1. Overall, without intervention 0 - 4% of breastfed infants become infected with HIV - 2 if the mother is infected compared to more tha n 25 .0% of transmission if the mother is infected with HIV - 1. Furthermore, HIV - 2 infection generally takes longer to progress to symptomatic and has a lower mortality rate than HIV - 1 infection. There are rare instances where a person may be infected with b oth HIV - 1 and HIV - 2 (CDC, 2008). 1.4.2 . Stages of HIV Disease There are three stages in HIV infection. The acute HIV Infection is the first stage where within two to four - cause the virus is replicating at high amounts (UNAIDS, 2015). During this acute stage, an infected person is at risk of transmitting the virus because of the high viral load (Dow and Essex, 2010). With no AVR treatment, the infection moves to the second s tage called , clinical latency (CDC, 2015). At this, stage, the virus is being produced at low levels and the person may appear asymptomatic for an average of 10 years. At this time, the CD4 count will begin to decline, while the viral load increases to the final stage known as AIDS ( Acquired Immunodeficiency Syndrome ) . During this final stage, the immune system is damaged and the body becomes vulnerable to an array of opportunistic infections, most importantly but not limited to pneumonia and cancer s (WHO, 2015). 1.4.3 . Diagnostic Tests for HIV The CD4 count is the number of CD4 T - lymphocyte cells in the blood, while the CD4 cells are the type of white blood cells that assist the immune system to fight infection (CDC, 2015). Viral load refers to the number of copies of HIV - 1 RNA per milliliter of plasma in the blood and is measured using polymerase 12 chain reaction (PCR) testing (Mellors et al . , 2007; CDC, 2008). This test can also be used to check the thy immune system ranges from 500 to 1600 cells/mm3). When HIV actively multiplies, it infects and kills CD4 cells falling below 200 cells/mm3 with AIDS (WHO, 2015). Generally, as the CD4 count decreases, the viral load increases, shortly after the person becomes infected with HIV. Although HIV ultimately resides within cells, the plasma measurement is an accurate reflection of the burden of infection and the magnitude of viral replication (Schacker et al . , 1998). It is used to assess the risk of disease p rogression and can help guide initiation of therapy. The range of detectable virus differs somewhat with each test, but the lowest level of detection generally is 20 - 75 copies/ml (Egger et al . , 2000). A viral load below this "undetectable" level indicates the inability of the test to detect HIV in the plasma, but does that there is an absence of HIV from the body (Hogg et al . , 2001). The highest levels of detection of the viral load assays typically are between 500,000 copies/mL and 750,000 copies/mL (Palel la et al . , 2003). Viral loads higher than these levels are reported, for example, as >500,000 copies/ml. A high viral load leads to a higher transmission risk and can be a sign of more severe disease progression. People in this stage who are not on ARVs c an survive an average of one to three years (UNAIDS , 2015). In Botswana, to my knowledge, the laboratory diagnostic limit is 400 copies/ml meaning that anything below 400 is considered suppressed viral load. 1.4.4. Treatment for HIV in Botswana First and second line ART regimens for new non - pregnant adult/adolescent patients included taking about three pills a day that were: tenofovir (TDF) + femtricitabine (FTC) (or 3TC) + efavirenz (EFV) or nevirapine ( NVP ) l, while the second line included: zidovudine ( AZT ) + 3TC + LPV/r (MoH, 2008). This was an era where the CD4 count threshold was <250 copies/ml. Prior to Aft er 2012, a fixed dose combination of Atripla combination of tenofovir (TDF) TDF/emtricitabine (FTC) 13 or 3TC/ efavirenz (EFV) was implemented for all patients regardless of their CD4 count and/or reproductive status (MoHW Botswana, 2015). Atripla is one pil l that contains these three medications, given daily with a low toxicity level. 1.4.5. PMTCT Treatment HIV positive women on ART who are considering to become pregnant minimize the risk of a negative pregnancy outcome, and it is recommended that the fol lowing conditions are met prior to becoming pregnant (i) viral load suppression of less than 400 copies/mL (ii) CD4 count above 200 cell/µL and (iii) stable on ART for no less than 2 years. In 2008, Botswana treatment guidelines recommended that tenofovir (TDF) should be avoided in pregnancy, and zidovudine (AZT) - based ART used instead. If AZT cannot be used, for instance due to anemia, then stavudine (d4T) should be used instead, but after delivery, d4T should be changed to TDF (MoH, 2008). In 2012, If her CD4 was less than 350 she will be initiated on ART for her own health benefit. But if the CD4 is above 350 she will be taking ART for Vision 2016 Policy all pregnant women living with HIV (MoHW, 2015). In 2016, recommendations were that all HIV positive ART naïve pregnant women be placed on ART for themselves and thei possible. In addition, it is recommended that during labor, all HIV positive women regardless of their ART regimens, should receive supplemental AZT (zidovudine) 300 mg every three hours not to exceed 1,500 mg (MoHW, 201 6). 1.4.6 . ARV Uptake and Adherence in Botswana Botswana was one of the first countries to take advantage of the M DG therapy to all infected citizens. Importantly, Botswana has a second higher percentage of HIV infected people taking ART (84.6%), (Free ART launched in 2002) compared to Eswatini (86 .0 %) (2003), Lesotho (74 .0 %) (2004) and South Africa (61 .0 %) (2004) (UNAIDS, 2018). However, the country continues to be 14 challenged by the epidemic despite an intensive and aggr essive HIV/AIDS prevention campaign (NACA, 2012). In 2000, Botswana went into a public private partnership with the African Comprehensive HIV/AIDS Partnerships (ACHAP), a non - governmental organization formed by the Bill & Melinda Gates Foundation and The M erck Company Foundation/Merck & Co., Inc. to make ARV therapy free and universal (MOH 2005; Bene and Darkoh, 2014; Farahani et al . , 2013). As part of this partnership, both Bill and Melinda Gates and the Merck Company Foundation contributed US$50 million e ach, while Merck & Co., Inc., pledged to donate two types of ARV medicines free of charge for the duration of the partnership (MOH , Darkoh, 2014). The name of the nat build a future for the nation' (Government of Botswana , 2006:3). In order to sus tain this programme, there are HIV policies in place that spearhead the financial allocation towards HIV prevention in Botswana. The latest financial year of 2016, saw only 10 .0 % of the financial resources going towards HIV prevention while 73 .0 % went to P MTCT (Avalos et al . , 2016). The government of Botswana is the major sponsor for all HIV/AIDS programs today in the country at 64.4% (this wa s 2.4% of the total government spending in 2015) (MoHW, 2016). Bilateral and multilateral donor s we re second at 31 .0 % while the private sector contribute d 4.6% towards funding HIV/AIDS programs in Botswana (Avalos et al . , 2016). In June 2016, Botswana launched a Treat All Strategy where all people living with HIV regardless of their CD4 count we re enrolle d. While loca l programs were underway, the CD4 350 wa s still in place. For a 15 - year period of 2015 - 2030, adopting a comprehensive Treat All Strategy, Botswana hopes to become one of the first countries in the world to achieve HIV epidemiological control by 2020 (MoH W, 2016). To track patients people living with HIV for ART and monitor the progress as well as the Emergency Plan for AIDS Relief (PEPFAR) established a m onitoring and evaluation unit within the national 15 ART programme in the Department of HIV/AIDS Prevention and Care in the Ministry of Health (Farahani et al . , 2013). The electronic patient tracking and outcome monitoring captures individual level patient d ata and the system is able to generate facility level reports that assist in both clinic management and patients care (Farahani et al . , 2013). Although the ART program has done well since inception, there are concerns with loss to follow - up rates which ha ve steadily increased among adults. Among the challenges encountered with growing numbers has been less time for one - on - one counselling of patients and intensive follow up through community channels for those who fail to honor clinic appointments (M oHW, 20 17). There we re also high treatment failure rates for adults (an increase in first line adult failure rate from less than 6 .0 % recorded in 2012 to over 10 .0 % in 2013) due to the challenges of adherence to counselling and monitoring (NACA, 2015). Adherence on ART is important for sustained HIV suppression, improved health and reduced risk of drug resistance (Chesney, 2006; WHO, 2003). Patients on ARVs are able to maintain consistent adherence levels that can result in viral suppression, increased CD4 counts and improved quality of life and survival. On the other hand, poor adherence from the initial and periodic lapses in adherence may lead to emergence of drug resistance and loss of future treatment options (Thompson et al . , 2012). Adherence drug resistance is not the only cause of treatment failure. The natural history of HIV infection is very unpredictable and . , 2000; WHO, 2006). If people are r immune system becomes compromised - ion and clinical condition (Schneider et al . , 2004). It is also vital that each patient receives and fully understands information about HIV disease and the goals of the therapy. The aim of ART involves achieving and maintaining viral suppression; decreasi ng HIV related mortality and preventing sexual transmission. However, education alone is not enough to assure high 16 levels of adherence because non - adherence is often a consequence of behavioral, structural and psychosocial barriers (Halkitis et al ., 2005) . Patients must also be positively motivated when initiated to maintain therapy. 1.4.7 . PMTCT Uptake and Adherence in Botswana Despite the availability of effectiv e interventions, most children we re infected through vertical transmission in Southern Afric a (Drobac et al . , 2008). Without any intervention, about 30% of babies born to HIV infected mothers will be infected (Drobac et al . , 2008). In 2007, a third of pregnant women were living with HIV and almost all of them were receiving antenatal care and del ivery in a health facility in Botswana (MOH, 2008). PMTCT services were introduced in Botswana in 1999, but by 2003, the program uptake was still only at 30 .0 %. There was an evident dramatic increase of 55 .0 % in 2007 and mother to child transmission droppe d to under 4 .0 % (Drobac et al . provided to all pregnant women living with HIV (MoHW, 2015). Due to this, mother to child transmission rates are expected to fall below 1 .0 % every year (MoHW, 2015). The risk of mother - to - child transmission can be significantly reduced by providing antiretroviral medicine (as lifelong therapy (ARV) or as prophylaxis) for the mot her during pregnancy, labor and delivery, with antiretroviral prophylaxis for the infant, and antiretroviral drug to the mother or child during breastfeeding (if breastfeeding), and use of safe delivery practices and safer infant feeding (Drobac et al . , 20 08). By the year 2014, the PMTCT programme was available in all the 634 health facilities that provide maternal child h ealth services in Botswana (MoHW, 2015). The percentage of pregnant women who were tested for HIV and received their results (during pre gnancy, during labor and delivery, and during the post - partum period, including those with previously known HIV status) was 94.3% in 2014, a decrease of 5 .0 % for 2013. However, male involvement remained low at 18 .0 % in 2014 a slight increase from 11 .0 % in 2011. In 2016, the number increased to 97 .0 % of new antenatal care (ANC) attendees who underwent an HIV 17 test. Of these, 26 .0 % of women who delivered during this 2 .0 % (n=51) also took both ZDV and NVP, while 869 (30 .0 %) were on Triple ART Prophylaxis (TAP), giving an uptake of 95 .0 % in 2016. In 2017, women testing remained at 97 .0 %, HIV positive number slightly decreased to 25 .0 % (n=3,171 of 12, 512), male testing increased to 21 .0 %, PMTCT uptake remained at 95 .0 % and 96 .0 % of infants born to HIV positive wo men were started on nevirapine [NVP] (MoHW, 2017). Although, the PMTCT program has done very well in Botswana, there are some challenges that need to be addressed in order to eliminate the vertical HIV infections. Challenges include low testing (46 .0 %) fo r children from six weeks to two months as reported in BAIS IV of 2013 due to poor turnaround times at the level of the laboratory (BAIS IV, 2013) as well as low male partner testing (21 .0 %) in 2017 (MoHW, 2017). PMTCT is a comprehensive therapy which mean s that other than the drugs provided for the woman, there is prophylaxis for HIV exposed children, periodic HIV testing to monitor the baby as well as partner testing and counselling. While the percentage of these babies who will ultimately receive their r esults improves over time, the purpose of early infant diagnosis and subsequently early treatment is lost for those who do not receive their results in time. Most of the health departments in Botswana also tend to work separately to achieve the same goal o f HIV elimination in newborns. This is indicated by the poor integration of services such as routine HIV testing, ART, Tuberculosis (TB) and Sexual Reproductive Health (SRH) (NACA, 2015). Finally, in relation to ARV non - adherence to expand on the discussio n above , people living with HIV w e re more likely to miss their hospital due to high temperatures and rare heavy rainfalls particularly in poorer ones who rely on walking. Rare heavy rains (and the fact that they are usually accompanied by lightning) in Bot swana usually impact the network services as well as the availability of electricity. Health workers rely on electricity to store patients results on computers and fridges that store blood results. If there is no electricity available, usually the work of the day stops, patients are sent home without their results and this can discourage them even them further. There are few public health facilities that have reliable power 18 back up. High temperatures as much as heavy rainfall, impact ART access and adherenc e negatively. People living with HIV particularly with low CD4 counts, pregnant women and new mothers are highly vulnerable because walking makes it impossible in scotching temperatures. In poorer areas such as rural areas, where air conditioning in health facilities is a luxury becomes even a promoting factor to non - adherence among the vulnerable poor. 1.4.8 . Gaps in the Literature Unlike many African countries, where government programs have not been aggressive, the government of Botswana is working ext remely hard to combat HIV/AIDS. HIV prevention methods and messages are everywhere, and news about the epidemic are the daily bread of newspapers, radio and television (Epstein and Ashburn, 2004). Free condoms are found in every corner of the country in remote clinics, bars and schools. Botswana was the first African country to offer ARVs to its citizens in 2002. Despite these efforts, the HIV epidemic in Botswana continues to be high. According to Green (2004), Botswana like many African governments reli es on western donors who spend a lot of money on promoting condoms, but put little emphasis on advising people to be faithful to their sexual partners (Green, 2004). Botswana is known historically as a culture of seasonal migrants where both men and women often spent time away from their homes and may be involved in long term relationships with different people in different places. Cattle in Botswana are valuable as they are the traditional form of wealth and are kept in remote cattle posts. Only boys and m en are responsible for taking care of them, this means that from time to time, they leave their partners behind to check on them. It is common in the country that most people do not use condoms regularly, but only with commercial sex workers and casual par later abandon them as a gesture of trust. These are the risky relationships because one partner can be committed in another relation with someone else. Hesselberg and Iversen (2011) term the basic cause of the HIV epidemic as multilocal nature of families and partners. In recent years, urban labor migration has 19 become so common in the country where people move away from rural areas in search of jobs. This rural - urban circulation particularly of young men play a significant role of widespread sexual connectivity which is a major risk of HIV (Hesselberg and Iversen, 2011). The relationship between HIV and climate is a complex one because vulnerability is involved. Vulnerability (Drimie and Casale, 2009; Chambers, 1989; Chambers, 1992) has two aspects which are external (stresses that people are subjected to) and internal (their capacity to cope with those stressors). While previo us studies have begun to document the drivers of HIV in Botswana (Bene and Darkoh, 2012 and 2014; Bene , 2014), including the increase in uptake of ART in peri - villages, however, the variation in HIV prevalence, ART/PMTCT uptake and adherence at the local l evel in Botswana is unknown. Examining the rate of adherence (which entails whether the patients are following the medical instructions) is a key piece of this study. Boonstra and colleagues (2002) categorizes adherence into two: full or acceptable adheren ce. The former means that the clients are taking their pills/drugs without any failure, while the latter means patients take their drugs, but they sometimes miss the doses for a variety of reasons and circumstances. How well patients adhere to the ARVs is influenced by many factors, including the doctor - patient relationship, acceptability, family support, religion, climate parameters such as ambient temperatures and water scarcity as well as the availability of food. The goals of the therapy are to increase CD4 count and reduce the viral load, therefore adherence is important. Studies have shown that ART/PMTCT can play an important role in preventing HIV transmission (Beaten et al . , 2011; Sheth et al . , 2009, Graham et al . , 2007). Lower levels and decreased c genital secretions that are on ART therapy (CDC, 2015). Research have illustrated that increased use of ARVs also lead to reduced rates of new HIV diagnoses; HIV transmission is now considered low when an ind adherence can assist health providers in addressing potential barriers to adherence . Little is known on the relationship between environmental factors such as high temperatures, water rationing and ART in Africa 20 as well as Botswana. The study results will fill an existing gap in the literature on climate and ART/PMTCT uptake. Also, in an era where treatment is used as prevention, more investigation is importa nt to understand HIV disease progression as there are cases of patients who enrolled later in the ART program. This study will contribute to this important epidemiological transition of HIV life course. The Ministry of Health and Wellness [MoHW] in Botswan a will benefit from these results and recommendations on how to address the disparities and adherence in the ART/PMTCT uptake. The study also serves as a useful reference for other regions as well as contributes to the emerging literature on climate and AR T/PMTCT adherence, especially in the African context. 1.5. STUDY GOAL The goal of this research is to provide recommendations on how to improve adherence to ART and PMTCT treatment regimens for adult men and women, and pregnant women and new mothers living with HIV in Botswana. With this in mind, this study will examine three dimensions of ART and PMTCT non - adherence for adults living with HIV in urban, peri - villages and rural areas of Botswana: 1.5.1. Objective 1 To investigate the life - course of HIV infe cted women and men from HIV diagnosis to newly - initiated ART and subsequent barriers to AR T adherence in Greater Gaborone , Botswana . 1.5.1 .1. Research Questions RQ 1 : In terms of life course, what is the length of time between HIV diagnosis and ARV initi ation? RQ 2 : What are the levels of ART non - adherence for adult women and men living with HIV and do these levels vary by age, and other socio - demographic characteristics? RQ3: How does length of time between HIV diagnosis and ARV initiation impact ART n on - adherence? 1.5.2. Objective 2 To investigate barriers to PMTCT adherence and their potential impact(s) on immunodeficiency (as a reflection on HIV transmission) for pregnant women and new mothers in Greater Gaborone, Botswana. 21 1.5.2.1. Research Question s RQ1: What is the level of ART non - adherence for pregnant and post - partum women (mothers) and all other women? RQ2: H ow does marital status vary for partners with HIV and/or mothers who do not disclose their own HIV status? RQ3: From their perspectives, w hat are the reasons for non - disclosure among pregnant and post - partum women (mothers)? 1.5.3. Objective 3 To investigate the role of environmental factors (high temperatures and water scarcity and rationing) on ART and PMTCT non - adherence, for adults liv ing with HIV in Greater Gaborone, Botswana. 1.5.3.1. Research Questions RQ1: Do environmental factors such as heat, water scarcity and rationing impact ART adherence for women and men living with HIV? RQ2: Do environmental variables such as heat, water sca rcity and rationing impact ART clinic attendance? RQ3: Are environmental factors barrier(s) to PMTCT adherence for mothers in particular? The first study (Objective 1) is outlined in Chapter 2; followed by the second and third studies (Objectives 2 and 3) in Chapters 3 and 4. 22 CHAPTER 2: Antiretroviral Therapy (ART) Adherence among HIV Infected Adults: A Health Geographic Study in Greater Gaborone, Botswana 2.0. INTRODUCTION In 2018, an estimated 36.4 million (range, 31.4 to 42.3) people worldwide were infec ted with the Human Immunodeficiency Virus (HIV), prevalence rate = 0.8 per 100 population (United Nations Joint Programme on AIDS (UNAIDS), 2019). Of these, approximately 1.7 million (range 1.4 to 2.3) were new HIV infections, incidence rate = 0.39 per 1, 000 population (UNAIDS 2019). Worldwide, a majority of new infections 1,000 population) (UNAIDS, 2019). In 2015, the World Health Organization (WHO) im plemented the 90 90 target treatment strategy for 2030, whereby 90% of people living with HIV know their HIV status, 90% of people who know their HIV - positive status receive antiretroviral treatment ( ART) (also referred to as ART uptake) and 90% of people on ART achieve a suppressed or non - detectable HIV viral load (UNAIDS, 2019). The premise of the program is, as ART uptake increases, the viral load of people living with HIV will decrease, thereby reducing person - to - person transm ission and the elimination of new HIV infections. In East and Southern Africa, there are approximately 20.6 million people living with HIV. Of these, n= 13,764,070 (68.0%) people receive ART, and 58.0% of those on ART have a suppressed viral load (UNAIDS, 2019). These findings demonstrate substantial progress toward warranted. Within the East and Southern Africa region, are four countries with the hig hest adult (15 - 49 years) HIV prevalence in the world (2018) Eswatini, n=210,000 (rate, 27.3 per 100), Lesotho, n=340,000 (23.6), Botswana, n=370,000 (20.3) and South Africa, n=7.7 million (20.4) (UNAIDS, 2019). Of these, South Africa has the greatest numbe r of people living with HIV followed by Botswana . Eswatini introduced free antiretroviral drugs (ARVs) in 2003, thereafter South Africa and Lesotho initiated treatment in 2004 (WHO, 23 2005). Botswana was the first African country to provide free ARVs in 20 02, yet remains one of the countries most affected by HIV in the world (UNAIDS, 2019); and is therefore, the focus of this study. HIV originated in Botswana in 1980. Although ARVs were free after 2002, employees of the mining industry and their family memb ers, and people with other positions of higher income were most likely to receive HIV treatment. In 2010, the WHO introduced the CD4 blood count to ascertain among those HIV infected, who should receive ART. It was recommended that HIV infected persons wi th a CD4 count of 350cells/mm 3 or less receive access to ART (WHO, 2013). In 2013, those guidelines were increased to CD4 count of 500 cells/mm 3 or less to ensure a greater number of HIV infected people receive treatment (WHO, 2013; 2016). In 2012, Botswa na adopted the 2010 recommendation of CD4 count of 350 cells/mm 3 or less as a guide by which to offer ART but did not adopt the subsequent 2013 guidelines (BIAS IV, 2013). Instead access to ART regardless of the CD4 count (UNAIDS, 2017). In June 2016, this program was officially launched and while some ART facilities are still using the CD4 count 350 criteria, the roll - out of universal access to ART continues. Recently the viral lo ad was added in conjunction with the CD4 count to evaluate treatment effectiveness. estimated 91.0% of people with HIV who know their HIV status, 92.0% of those who know their HIV positive status, receive ARVs and of those taking ARVs, over 95.0% have achieved viral suppression (UNAIDS, 2019). Despite this success, in 2018 there were n=8,500 new HIV infections in Botswana. Importantly, these new infecti ons varied by gender and age with females disproportionally affected (n=4,500) compared to men (n=3,700). Adolescent and young adult females were twice as likely as males to acquire a new HIV infection. In contrast, adult men (25 - 49 years) were 1.5 times more likely than adult women to be newly diagnosed with HIV. Furthermore, approximately, 25.0% of new infections were late diagnoses (CD4 count < 200 cells/mm 2 ) (UNAIDS, 2019). There is a need therefore, to further understand 24 treatment and the life - cour se of HIV infected women and men in Botswana. Finally, in 2017, only n=849 adults 15+ years were newly - initiated on ART (UNAIDS, 2019), which may account for less than 10.0% of new HIV infections that year (UNAIDS, 2019). To reduce viral load and ultimat ely eliminate HIV transmission among newly infected adolescents, young adults and adult women and men, and people currently living with HIV, it will be important to ensure ART access and optimize ARV adherence. 2.1. GOAL OF THE STUDY The goal of this res earch is to investigate the life - course of HIV infected women and men from HIV diagnosis to newly - initiated ART and subsequent barriers to AR T adherence. A survey was conducted during the winter in Botswana (May to August, 2016) among a sample of adult (n =667) women (n=428) and men (n=239) living with HIV, and also enrolled in ART treatment at 21 government clinics, in urban, peri - in Botswana. 2.2. BACKGROUND 2.2 .1. HIV Life - Course There are three stages of HIV infection (CDC, 2019) Stage 1 or acute HIV infection occurs within two to four weeks following exposure. People may experience flu - like symptoms as their viral load increases , and at this stage they are very contagious. Stage 2 or clinical latency stage, is when HIV infection becomes chronic, reproducing at lower levels. People are asymptomatic for up to a decade, during which time HIV transmission can still occur. If ART is initiated early in Stage 2 people living with HIV may lead productive lives, and often die of common chronic diseases. However, if people are not started on ART or poorly adhere to ARV treatment, the progression of disease may occur more rapidly as evi dence by dropping CD4 count or increase in viral load. Under these conditions, people living with HIV may acquire opportunistic infections, and move into Stage 3 also referred to as AIDS (Acquired Immunodeficiency Syndrome). Without ARVs, people living w ith AIDS may survive about three years. An AIDS diagnosis is made when 25 their CD4 count drops below 200 cells/mm 2 or if they develop opportunistic infections including cancers, ion of ARVs or late HIV diagnosis occurs when the CD4 count drops below 200 cells/mm 2 (UNAIDS, 2019), generally during the later period of Stage 2 or Stage 3 of HIV infection. Adherence to ARVs across the life - course of people living with HIV, is therefor the elimination of HIV transmission a public health priority. 2.2 .2. ART Adherence ( Haynes et al., 2005 The WHO adapted these definitions of long - taking medication, following a diet, and/or executing lifestyle changes, hence, there is no gold standard for adherence measurement. ARV adherence in a population is thus, generally estimated using indirect measures (DiMatteo, 2004) including pill counts, pharmacy refill questionnaires and visual analogue scales and biomonitoring measurements ( Golin et al., 2002; Paterson et al., 2000; Bangsberg et al., 2000; WHO, 2006). These indirect methods of monitoring adherence are briefly described below followed by studies demonstrating their successes and failures. Pill Coun t : is a measure of adherence calculated by counting the remaining doses, assuming the remaining pills in excess represent missed doses (Golin et al., 2002; Bangsberg et al., 2000; NcNabb et al., 2001). Pill 26 counting is commonly conducted in clinics or unan nounced home visits and is relatively easy to implement with cooperative patients. Pharmacy Refill Records : not obtained, it is assumed that the patient is not taking t heir medication (Turner, 2002; Steiner and Prochazka, 1997; Maher et al., 1999). Studies ( Hogg et al., 2002; Low - Beer et al., 2000; Wood et al., 2002) have shown that pharmacy refill measures that can be used in the evaluation of viral suppression, ARV re sistance, and progression to AIDS and death). When n=291 young adults on ARVs were studied in Gaborone, and the main outcome measure was treatment failure (defined as two consecutive HIV viral - months of follow - up) (Genn et al . , 2018). Pharm acy refill non - adherence was defined as two consecutive refill adherence measurements <95.0% during the same period. Authors found that 53 (18.0%) experienced treatment failure and 128 (44.0%) had refill non - adherence (Genn et al . , 2018). Pill counts and pharmacy refill measures are often used together to measure ART adherence. A study in Gaborone (Ndubuka and Ehlers, 2011) sought to understand whether pill counts and pharmacy refill records could predict immunological and viral responses, through increas ed CD4 cell count and suppressed viral load, for patients taking ARVs. Pill counts were calculated by dividing the number of pills actually taken by the total number of prescribed ARV pills expected to be taken over a 30 - day period and expressed as a perce ntage (Ndubuka and Ehlers, 2011) . The remaining pills in excess of what was expected were considered as missed doses on a monthly basis. To calculate pharmacy refill, exact dates when ARVs were collected from the pharmacy were compared with the expected re fill dates and the exact number of days of being late for pharmacy refills was recorded (Ndubuka and Ehlers, 2011) . Adherence was defined as a composite adherence rate equal to or higher than 95% while levels of adherence 94.9% and lower were considered no n - adherent. The mean adherence rates were 95. 9% for the pharmacy refill and 27 96.4% for the pill count, providing a mean composite adherence rate of 96.2% (Ndubuka and Ehlers, 2011) . Patients who were ART adherent, as measured by consistent pharmacy refills for greater than four months, were significantly more likely to achieve viral load suppression and benefit immunologically, than were less - adherent patients (Ndubuka and Ehlers, 2011). Pharmacy refills and pill counts adherence measurements were shown to b e reliable methodologies (Ndubuka and Ehlers, 2011) by which to monitor ARV adherence. Electronic Drug Monitors : The Medication Event Monitoring System (MEMS) uses a computer chip embedded in a specially designed pill - bottle cap to record the time and dura tion of each bottle opening. MEMS adherence is calculated by dividing the number of observed bottle openings, by the number of expected doses over the study period (NcNabb et al., 2001; Arnsten et al., 2001). Wisepill, a wireless pill container transmits a cellular signal when opened. In their study, Haberer and colleagues (2010) found that when adherence levels were measured by Wisepill, compared with unannounced pill counts, self - reports, and HIV RNA (Haberer et al . , 2010), real - time adherence monitoring was found to be feasible for resource - limited settings and may detect sub - optimal adherence prior to viral rebound (Haberer et al., 2010). Self - Reporting Tools : to monitor the viral load (Ferradini et al., 2006). Four - day recall adherence measures the number of tablets taken correctly as a proportion of those prescribed (Ferradini et al., 2006). Another self - reporting tool is the visual analogue scale, - i.e., the percentage of patie nts taking at least 95% of their tablets correctly. Both methods have been shown to aide both program and public health evaluation (Ferradini et al., 2006). Ferradini et al. (2006) found the average population adherence rate to be 85 .0 % but only 60 .0 % ac hieved the optimal the adherence rate (Ferradini et al., 2006). The authors concluded that having both self - reporting tools are ideal to verify the results (Ferradini et al., 2006). Nance and colleagues (2016) validated 28 what authors in Malawi proposed of using both measures in a study for reliability (Ferradini et al., 2006; Nance et al . , 2016). Other studies (Arnsten et al . ,2001; Paterson et al . , 2000; Golin et al . , 2002)) have found that a single low, untimed antiretroviral drug level was sensitive in id entifying those with very poor adherence (<60 .0 %), but therapeutic drug levels were not necessarily indicative of good adherence (Liechty et al . , 2004). Plasma concentrations are limited by their ability to detect only recent adherence behavior (Machtinger and Bangsberg, 2006.) Biological Markers: Biological markers include plasma concentrations of ARVs (Machtinger and Bangsberg, 2006). Plasma concentrations of a protease inhibitor is a class of ARVs that block protease (an HIV enzyme) to prevent new (imma ture) HIV from becoming a mature virus, that can infect other CD4 cells ( Dandache et al . , 2007). Biological markers are significantly associated with adherence behavior. Furthermore, low concentrations of antiretroviral drugs also may be caused by factor s other than adherence, such as malabsorption, drug interactions, and individual metabolic differences (Machtinger and Bangsberg, 2006.) Other studies have been conducted where antiretroviral concentrations in hair samples were evaluated as a marker for an tiretroviral exposure and, less directly, long - term adherence (Bernard et al . , 2002; Gandhi and Greenblatt, 2002). An et al., 2000; Rabound et al., 2011). Poor adherence to ARVs will not fully suppress viral replication (Rabound et al . , 2011) which in turn may mutate making the drugs no longer effective (Center for Disease Control [CDC], 2017). If resistance occurs, different medications are used ones to ta rget the virus (CDC, 2017). Resistance is especially difficult to manage in resource - limited settings, where there are fewer choices for ARV medicines (CDC, 2017). This is particularly true for African countries, already burdened with high HIV prevalence, low staff personnel, budget cuts and insufficient of ARV supply. ARV adherence is thus, critical to reduce viral load and the elimination of HIV transmission. People living with HIV who 29 have a low viral load also have a stronger immune system, fewer illne sses, and can lead a healthier life, thereby, improving overall population health. 2.2 .3. Barriers to ARV Adherence Studies that investigate barriers to ART adherence generally fall into three areas: patient - centered barriers, social - structural barriers, a nd service - centered barriers. Patient - Centered Barriers include age greater than 35 years and single (Peltzer et al . , 2011) and divorced (Birbeck et al., 2011) marital status, low education (Charurat et al . , 2010; Peltzer et al . , 2011; Birbeck et al . , 2011), low health literacy, inadequate knowledge about HIV, its treatment and how to interpret laboratory results, effects to the ARVs (Birbeck et al., 2011) , alcohol abuse, depression or other illness, and the use of other medicines (Kip et al . , 2009) and lack of interest or desire ). S tudies have indicated that patients with chronic co - morbidities, such as diabetes and hypertension take their ARVs only about 70.0% of the time (Hashmi, 2007 (77% adherence); Martin, 2005 (69.8%); Brown, 2011 (75 %); Charurat et al . , 2010 (70.4%)). Generally, ARV adherence is better in urban compared to rural areas (Peltzer et al . , 2011). Social - Structural Barriers include non - disclosure of HIV status and stigmas (Weiser et al., 2003; Birbeck et al., 2011), low le vels of social support (Yang et al., 2018), high - unemployment (Charurat et al . , 2010), food insecurity, religious rituals such as fasting and poverty (Bezabhe et al . , 2014). Stigmas surrounding HIV are particularly important. Weiser et al. (2003) found i n Gaborone and Francistown clinics (n=3), Botswana that 69.0% of patients kept their HIV status a secret from families and 94.0% from their community; and 32.0% of patients feared that they could lose their job if their HIV status was known, and 15.0% repo rted that stigma interfered with the ability to adhere to medication. Another study by Thupayagale - Tshweneagae (2010) in urban Botswana found that young adults would keep silent about 30 their HIV status, but sought treatment for other conditions (such as fl u) to avoid being stigmatized. Furthermore, young adults were concerned with begging forced to reveal their HIV status when pregnant, divorced or assaulted to further avoid being stigmatized. Service - Centered Barriers to non - adherence, include lack of t ransport and travelling costs, longer travel to the clinic, health workers' inability to conduct home visits and to contact defaulters, limited clinic operational hours, waiting times at clinics, nurses' attitudes and knowledge, poor labo ra tory services in cluding, delays in getting HIV testing results and insufficient supply of medications (Bezabhe et al., 2014; Birbeck et al., 2011; Kip et al., 2009; Halkitis et al., 2005; Stirratt et al., 2006; Carr and Gramling, 2004). In contrast, ARV adherence increas ed if people living with HIV lived < 2 hours from the clinic (Charurat et al., 2010), when provider - patient spoke the same language, when patient education and counselling were provided and dose - frequency was simplified (Orrell et al., 2003; Bezabhe et al. , 2014). While patient - related and service - centered barriers are more tangible to address, there are virtually no offered solutions other than education to social - structural barriers to improve ART adherence and will therefore, be a focus in this study. 2.3. THEORETICAL PERSPECTIVE This study adopts the Theory of Planned Behavior (TPB) within a framework of Intergenerational there is a 10 - year or more ag - Madlala, 2008). Intergenerational relationships on its own is not a theory, but gender ideology particularly in Africa. It is well suited under the framework of the Theory of Planned Behavior (TPB). TPB is a psychol ogical conceptual framework that explains human action (Ajzen, 2001). The theory was derived from the theory of reasoned action (Fishbein and Ajzen, 1975) that assumed that most human social behavior can be predicted from intentions alone (Ajzen, 2001). Th e new addition here is the component of perceived behavioral control, 31 2002). Theory of planned behavior explains that human behavior is guided by (a ) beliefs about likely consequences and attributes of behavior (behavioral beliefs), (b) normative expectation of other people (normative beliefs) and the (c) presence of factors that may further or inhibit performance of the behavior (control beliefs) (Aj a behavior that is, they illustrate how hard people are willi ng to try, and how much of an effort they are show the degree to which a person has favorable or unfavorable evaluation of behavior at the moment while the subjective norm refers to the perceived pressure of the performance (or not) of the behavior at hand (Ajzen, 1991 ). Attitudes can be internalized (self - stigmatization) or external. In a qualitative study (n=32) conducted in Gaborone, Botswana, partic ipants who had accepted their HIV status described These individuals were able to have better ARV adherence compared to those who were in denial (Nam e t al . , 2008). Subjective norms are a product of the normative expectations of other meaningful people towards a particular behavior and the motivation to comply with the opinions of others (Jewkes, Levin and Penn - Kekana, 2003). Stigma surrounding HIV and A RT is still prevalent in Botswana where HIV is thought to be contracted through promiscuous behavior. HIV/AIDS has many Setswana derogatory names such as ( unacc eptable moral behavior, which gives reason for judgement of an individual by family or community (Nam et al . , 2008; Heald, 2005). to their control ov er the behavior and the perceived power the individual feels to remove these barriers (Ajzen, 1991). Cross generation relationships align with the second arm of TPB (subjective norms), in the sense that the age - disparity between partners is a constraining factor the younger partner to negotiate 32 safer sex (Leclerc - Madlala, 2008; Jewkes, Levin and Penn - Kekana, 2003). This means that young women might be obliged to have unprotected sex with older men because it is a norm/culture to do so in their society or i voice might be dim when it comes to safe sexual encounters. 2.4. STUDY DESIGN This study utilized an observational, cross - sectional, population - based survey study design. Face - to - face interviews were conducted among n=667 adults living with HIV and attending 21 HIV clinics in the Greater Gaborone area. The questionnaire design consisted of closed and open questions, reflecting on retrospective recall of events, as well as current knowledge, attitudes and beliefs and behavioral practices. The sampling methodology to include participants is further described below. 2.5. DATA AND METHODS 2.5.1. Study Area This study took place in the Greater Gaborone area, which in cludes the capital city of Gaborone city (as its own district) and three surrounding districts, South East, Kgatleng and Kweneng East within which urban, peri - urban and rural areas were available for study. According to the last national census in 2011, t he city of Gaborone had a population of n=231,592, while the larger metropolitan area had about one - half million (n=421,907) people (Botswana Census, 2011). The prevalence of HIV in Gaborone in 2013, was 17.0 per 100 population (estimated HIV n=49,016); S outh East, 16.6 (n=35,443); Kgatleng, 19.9 (n=30,858) and Kweneng East, 21.5 (n=61,384). Using these HIV est imates to calculate ART uptake i n 2017, Gaborone and South East districts had the highest uptake, 68.0% (n=33,332 people receiving ARVs) compared t o Kgatleng, 35.7% (n=11,028) and Kweneng East, 43.4% (n=26,587) (Kelepile and Grady, 2019 in review). While ART uptake will not be a focus of this study, further understanding AR T adherence among people living with HIV is needed in this high HIV prevalenc e area of Botswana. 33 2.5.2. Primary Data Collection The data for this study were generated primarily from a survey (quantitative) administered to adult women and men who are living with HIV and enrolled on ART (face to face interview), at the study area. This survey was conducted during winter in Botswana of 2016 (May to August). A sample of adults (n=667) men (n=239) and women (n=428) living with HIV, who were enrolled in ART treatment at 21 government clinics, in urban, peri - urban and rural areas (in G reater Gaborone) were interviewed. These patients were required to attend the clinic every month, and it is at this visits where, they collect their ARV drug and have doctor consultation. Biological data such as CD4 count, viral load, date initiated on ART , weight and each time they visit their ART facility. Only the latest readings were taken as the patients take new readings at least three times a ye ar. The inclusion criteria were age 18 years and older and HIV positive, attending the clinic for ART. The questionnaires were administered to participants in Setswana (national) and English (official) languages. The use of Setswana language (local) was useful in this study because patients and staff understood it better than English. All participants were asked to sign an informed consent before partaking in the study to obtain their permission to participate. The principal researcher and four research a ssistants administered the questionnaires. The principal researcher took a training with the Michigan State University [MSU] IRB, while the principal researcher with materials trained the research assistants who were all Botswana locals, which were directl y obtained from the MSU IRB (#x15 - 1269e). All the assistants were fluent in both the languages. The complete survey is provided in Appendix A. Ethics Overview: The survey took place in the urban, peri - urban and rural areas of Botswana from May until Augu st 2016. Given the sensitivity of this research, an approval from the Institutional Review Board (IRB) was secured from both Michigan State University, and the Ministry of Health and Wellness in Botswana (see Appendices A and B) prior to administering the survey. After satisfying the requirements 34 of the MoHW IRB requirements, to obtain access to the different clinics, separate IRBs were sought from the District Health Management Teams [DHMTs] in Botswana. MOHW provides primary health care services through D HMTs, which are responsible for running a network of health facilities; hospitals, clinics, health posts and mobile stops as well as community based preventative and promoting services (MoHW, 2018). 2 . 5.3 . Sampling Methodology 2.5.3.1. Districts selection (n=4) The selected study area has a high HIV prevalence, high number of HIV positive patients as well as high number of ART clinics attending to patients receiving ARVs. Purposive sampling as conducted i.e., of non - al researcher deliberately selects specific elements or subjects for inclusion in a study in order to ensure that the elements will have certain characteristics relevant to the study (Palinkas et al . , 2016). The study covered Gaborone (urban), South East ( peri - urban), Kgatleng and residence; and had all the identified characteristics based on geography and HIV prevalence. To calculate the number of clinics needed, the national prevalence (22.0%) was used since the study did not consider districts as rigid boundaries. 2.5.3.2. Clinics selection (n=21) There were 84 ART health facilities in the study area, 22 in Gaborone (17.0 HIV prevalence), 6 in South Eas t (16.6% HIV prevalence), 26 in Kgatleng (19.9 HIV prevalence) and 30 in Kweneng East (21.5 HIV prevalence). Using the national prevalence (0.22*84 clinics), 18 clinics were recommended. To account for over and under sampling of facilities, the number was increased to 21 clinics. The clinics were then From the urban area of Gaborone, five clinics [ Broadhurst Traditional, Bontleng, Phase 2, Old Naledi and Village] were selec ted . Among these urban clinics, three clinics (Broadhurst, Bontleng and Old Naledi) are located in low - income 35 neighborhoods. Old Naledi is mainly a squatter settlement in the outskirts of Gaborone characterized by small businesses and relatively high popul ation density. This neighborhood is also close to Gaborone dam which is the city is major source of water, so minor fishing is evident as well as less dry landscapes. Five peri - urban [ Nkoyaphiri, Mafitlhakgosi, Mmopane, Mogoditshane and Tlokweng Main] ar e located in the peri - urban villages close to Gaborone. Due to the growing population of Gaborone, most people commute and reside in these peri villages. Due to high densities of people, clinics in these areas, are usually located around residential neighb roads close to the peri - urban clinics. The eleven rural clinics [Makakatlela, Siga, Phuthadikobo, Bokaa, Kopong, Gabane, Morwa, Mochudi1, Kgosing, Lentsweletau and Taung] usuall y had limited tarred roads and people use dirt roads to reach to these facilities. 36 Figure 1: Map of Study Area Showing ART Health Facilities and D istricts in Greater Gaborone, Botswana . 37 The second stage involved selection of adult men and women in the c hosen districts (Equation 1.1). The sample size for the participants was calculated using the formula below: The formula was adopted from a national stigma associated with HIV/AIDS survey conducted in 2014 (Botswana Stigma Index Survey Report, 2014). N= (t 2 pq)/ (d 2 ) [Equation 1.1] Where; n = minimum sample size required t = z value of confidence level desired (95% confidence level) p = estimated prevalence in target population (the national adult prevalence rate of 22%). d = precision. The higher t he precision desired the larger the sample size will be. 0.05 (5%) is the most common d value. q = 1 - p. Assuming 95% confidence and a precision of 0.05 Formula: Equation 1.1 N = (1.96 2 )*0.22*(1 - 0.22))/ (0.05 2 ) = 263.6. Design effect = 263.6 x 2 = 527.2 = 52 7 would be the minimum sample size required for adult men and women living with HIV and receiving ARVs in the study area, and to strengthen results and stable rates, this number was increased to 667 based on the extra availability of the patients at each c linic after reaching the targeted n=30 per clinic. To give the number of patients per clinic, the power to reject the null hypothesis was assessed in Open Epi Software (Dean, Sullivan, and Soe, 2013). The power of a statistical test is the probability that the test will correctly identify a false null hypothesis (Kar and Ramalingam, 2013). The power was set at 80% meaning that the study will be able to detect a true effect of the non - adherence 80% of times assuming that false negative results will occur onl y 20% of times. Failure to reject the null hypothesis when it is false is called type II error, so the higher the power of the study, the greater will be the required sample size (Kar and Ramalingam, 2013 ). The districts (n=4) where people came from in urb an, peri - urban and rural areas) 38 were known with all the 21 clinics. Thus the minimum required population of n=527 patients was used, and input 80% reject power required for each clinic, result came as 30. Even though 30 was the minimum, some clinics were o versampled. The clinics were geocoded in Google Earth to produce a KMZ file, which was later converted to a geographic shape file in ArcGIS 10.5 (ESRI, 2018). Following the selection of clinics and participants, systematic random sampling was employed to s elect the patients at their respective clinics. Every second patient to enter the door was selected until the minimum required sample of n=30 patients per clinic was reached. Participants were visited daily at their respective ART facilities in the study a rea. The health personnel and the researchers briefed the ART patients on the purpose of the study during morning health talks, before the interviews took place in interview rooms. 2.5.3.3. Reliability and Validity of Data Collected For reliability purpos es, all the patients were visited at their respective clinics at the same time (from 7:30 am until 2:00pm). In addition to the principal researcher, four experienced research assistants assisted with the data collection. These assistants were locals who we re trained on IRB requirements and fluent in Setswana language (as described above). The principal researcher and assistants attempted to make the patients comfortable in order to open - up about their treatment adherence behaviors, without the fear of judgm ent. The data were coded by the principal researcher to remove the bias which could be incurred if analysis is important because it allows readers 2. 5.4. Non - Adherence Definition [CD4_REC], (2) taking the pills as instructed by the doctor (100% dosage) [ARVPrescribe] and (3) having missed a (one) dosage sometimes [MissDose] (95% or at least missed 1 pill a month). If all of these 39 criteria were not - T treatment - The explanatory (independent) variables to predict Non - Adherence, were coded as sex (female=1, male=0); age groups ( base, 30 - 39 years=0, and 18 - 24=1, 25 - 29=1, 35 - 39=1, 45 - 49=1, 50+=1; Body Mass Index (BMI) height and weight from which to calculate their body mass index (BMI) as an indicator of overall nutritional status. Th ey were also asked the year diagnosed with HIV (HIV_Year) and the year ARVs were initiated (ARV_Cat). These variables were studied using continuous annual measures, and categories as follows: The first patient reported being diagnosed in Botswana was in 19 85 [Category 1: 1982 - 1985]; the universal free ARV drugs were introduced in Botswana in 2002 [Category 2: 2002 - 2016]. Any illness/condition [llness], this variable was coded as 1, and 0 for no illness. To understand the side effects of ARVs, the question on how they felt after taking ARVs [ARVFeel], was asked, and if one felt no side effects after a dose, the response was coded as a 0. Anyinfo (any information provided - open ended). This question provided for the richness of the discussion points that they researcher may have overlooked. All the continuous variables were assessed to check for distribution and normality through histograms and - skewed with many outliers, so patients with high numbers (<7000 copies m/L) were removed. Further the - which the number of partners that one had over the past year, had a couple of extreme values such as 16 partners in 12 months, so two patients w ere removed, leaving the n=658 in the analysis. 2. 5. 5 . Descriptive Methods The dependent and independent variables were examined for their frequencies and cross tabulations. Histograms were evaluated for the continuous variables. These descriptive analyse s, were also used to check if all the assumptions of regression analyses were being met. To assess the HIV life - course of 40 individuals, the stage of life at time of HIV diagnosis was estimated by taking their age in 2016 (age at interview) (Y axis) minus th eir year of reported HIV diagnosis (X axis). The range in numbers of years post HIV diagnosis, that ARVs were initiated, are provided in each x, y cell using low and high estimates for each age group. 2. 5. 6 . Analytical Methods Individual - level regression analyses were estimated in this study. However, imbalanced data was a limitation for many analyzes, thus population - level analyses were implement as described herein. Imbalanced datasets can significantly affect the performance of the classifiers as the w orking model will be biased towards the majority class to minimize the overall error term (Estabrooks, 2000; Dubey et al., 2014). The imbalanced data problem occurs when the explanatory variables are unevenly distributed among different classes. In the cas e of binary classification, the number of observations in one class is significantly greater than that of the other (Zheng et al., 2004). In this study, the data were frequently imbalanced . For instance, there were almost twice the number of females (n=419 ) sampled as there were men (n=239). Another example was the classes observed within the variable age, with 30 - 39 years for women having the highest frequency (n=148) compared to the same class for men (n=60). Another example of imbalance was seen with the unemployment variable, where 172 women were unemployed compared to (n=47) of men. To address imbalanced data, this study utilized the Least Squares (LS) - means procedure to estimate the marginal totals of population parameters in SAS 9.4 (SAS Institute In c., 2002 - 2012, Cary, NC, USA). The linear functions were defined by population marginal means of the corresponding means for balanced data (Cai, 2014). When performing an LS - means procedure with logistic regression in SAS, it produces a diffogram, and odds ratios with 95% confidence intervals. In a diffogram, each line segment corresponded to one pairwise difference between LS - means. A line segment centered at the LS - means in a pair and had both a vertical and a horizontal line that indicated values and 41 lev els that corresponded to the pair. The length of the line segment reflected the projected width of a confidence interval for the difference (Cai, 2014). 2.6. RESULTS 2. 6.1. Life Course of People Living with HIV and ART - Initiation Post - HIV Diagnosis The lif e - me between HIV diagnosis and ART initiation. Although it is unclear if these same patients received late HIV diagnoses, a summary of findings stratified by women and men is provided belo w. 2.6.1.1. Women Using the low age - gro up estimates (Table 1 ), there was no woman born with HIV. However, there were women diagnosed with HIV across the life - course, including the life stages of infancy (n=1), childhood (n=47), young adult (n=127), adu lt (n=120), mid - early adult (n=71), mid - late adult (n=38), older adulthood (n=14) and oldest adult (n=1). Using the h igh age - group estimates (Table 2 HIV diagnosis shifted to slightly higher age groups as expected. There were no wome n bor n with HIV, nor diagnosed in infancy. HIV diagnoses occurred i n childhood (n=8), young adult (n=47), adult (n=103), mid - early adult (n=139), mid - late adult (n=73), older adult (n=37) and oldest adult (n=14). In general, women diagnosed with HIV across the life - course began ARV treatment during Stage 2 of the disease, with a few individual women with late diagnoses i.e., late Stage 2 or possibly Stage 3. 42 Table 1 . HIV Diagnoses Across the Life Course by Current Age and Low Estimate 1 of Years ARV Initiated Post - HIV Diagnosis Sample of Women (n=419) Living with HIV in Greater Gaborone, Botswana, 2016. 1 Estimates based on low - age range. 43 Table 2 . HIV Diagnoses Across the Life Course by Current Age and High Estimate 1 of Years ARV Initiated Post - H IV Diagnosis Sample of Women (n=419) Living with HIV in Greater Gaborone, Botswana, 2016. 1 Estimates based on high - age range. 44 2.6.1.2. Men Using the low age - group estimates (Table 3 ), there was one man born with HIV and two men diagnosed during infancy. T here were men d iagnosed with HIV across the life - course, including the life stages of childhood (n=19 ), young adult ( n=35 ), adult (n= 56 ), mid - early adult (n=7 8 ), mid - late adult (n= 30 ) and older adulthood (n=1 8 ). Using the high age - group estimates (Table 4 ) men age of HIV diagnosis shifted to slightly higher age groups as expected with the same man diagnosed at birth, fewer diagnosed in childhood (n=7), young adult (n= 12 ), adult (n= 27 ), mid - early adult (n= 70 ), mid - late adult (n= 77 ), older adult (n= 32 ) and oldest adult (n=13 ). The majority of men diagnosed as mid - early adults (31 - 39 years) and mid - late adults (40 - 49 years) are currently in these same age groups, thereby receiving ARV treatment early in Stage 2. The re are also 21 men diagnosed during their mid - early adult years, who are now 40 - 49 years who initiated ARVs a mean of 10 years post - diagnosis (range, 3 - 15 years). Likewise, 11 men were diagnosed during their mid - late adult years, who are now 50 - 59 years w ho initiated ARVs a mean of 9 years post - diagnosis (range, 0 - 14 years). Finally, there are 6 men who were diagnosed during the older adult years 50 - 59 years, who are now in their 60+ years who initiated ARV treatment a mean of 9 years post - diagnosis (rang e, 4 - 11 years). In general, men were diagnosed with HIV later in their life - course compared to women who were diagnosed at younger ages. More older men also initiated ARV treatment later in Stage 2 and Stage 3 of their HIV disease, compared to women diag nosed at these same ages with a few outlying women initiating ARV treatment in Stage 3. 45 Table 3. HIV Diagnoses Across the Life Course by Current Age and Low Estimate 1 of Years ARV Initiated Post - HIV Diagnosis Sample of Men (n=239) Living with HIV i n Greater Gaborone, Botswana, 2016. 1 Estimates based on low - age range. 46 Table 4 . HIV Diagnoses Across the Life Course by Current Age and High Estimate 1 of Years ARV Initiated Post - HIV Diagnosis Sample of Men (n=239) Living with HIV in Greater Gaborone, B otswana, 2016. 1 Estimates based on high - age range. 47 When the number of years ARV is initiated post - HIV diagnosis for women and men in their current age group is plotted against their current CD4 count ( Figures 2 - 5 ) there are striking differences. Figure 2 shows young adult women and men (ages 20 - 29 years) with a linear upward trend in CD4 count with increasing length of time between HIV diagnosis and ARV initiation (range, 0 to 12.5 years). All but two women and one man initiated ARV treatment immediately following diagnosis through Stage 2. Women consistently have a higher CD4 count (trend > 500 cells/ml ) compa red to men (trend < 500 cells/ml ). Figure 2 : Women and Men Ages 20 - 29 Years. Figure 3 s hows women and men in their mid - early adult years (30 - 39 years) with a cross - over effect of ARV initiation around 7.5 years post - HIV diagnosis (range, 0 to 15.0 years). Women at HIV diagnosi s show CD4 counts > 500 cells/ml compared to men (CD4 count < 500 cells/ml ); however, with increased delay in 48 At this age - group, there are more women receiving delayed ARV treatment compared to men, which may influence these results. A substantial number of women and men are diagnosed late in the disease as evidenc ed by a CD4 count < 200 cells/ml . Figure 3 : Women and Men Ages 30 - 39 Years. Figure 4 shows women and men in their mid - late adult years (40 - 49 years) runni ng relatively parallel aro und CD4 count 500 cells/ml with women across the delay in ARV initation having a slightly higher CD4 count compared to men. The range in the number of years ARV initiated post - HIV diagnosis is greatest in this age group (range, 0 to 20+ years). 49 Figure 4 : Women and Men Ages 40 - 49 Years. Finally, Figure 5 shows women and men of the older and oldest - adult age groups (50+ years) with women and men experiencing a negative linear trend in CD4 count across the delay in ARV treatment (range, 0 to 12.5 years). In general, older men have lower CD4 counts compared to older women with the gap widening with delay in ARV treatment beyond 7.5 years. 50 Figure 5 : Women and Men Ages 50+ Years. In summary , these graphs show that in general, w omen experience higher CD4 counts compared to men at all life - course stages, except those women in their mid - early adult years (30 - 39 years) with extreme delays in ARV treatment. Women and men in their young adult and adult years of life generally respond to ARV treatment when initiated as evidence by a rising CD4 count. In contrast, women and men in their older and oldest stages of life respond less well to ARV treatment with increasing delays as also evident by a lower of CD4 counts. These population - b ased graphs are informative because they show general trends in potential ARV treatment success following differing periods of ARV initiation post - HIV diagnosis. With this background understanding of gender differences in year of HIV diagnosis, number of years ARVs initiated post - HIV diagnosis and current age - groups referenced CD4 counts, the population surveyed is further investigated for levels of ARV adherence and non - adherence. 51 2. 6.2. Non - Adherence Levels Non ART adherence in Greater Gaborone was 29 .1% for women and 46 .0 % for me n (Table 5 ). Across age groups, non - adherence was differe nt among women and men. For young adults and adults ( 18 - 29 years ) non - adherence was double that for women (21.7%) compared to men (12.6%) . This trend cross - over for th e older and oldest - adult age groups (50+) where women were substantially less non - adherent (2.9%) compared to men ( 12.9% ). By marital status, both women and men were mostly either single or cohabiting. Single men tend ed to be more non - adherent (21.3%) th an single women (15.9%). There were fewer women 5.5% (1.2% non - adherent) who did not have education compared to 15.9% (7.9% non - adherent) men. There were also high number of women with university education (10.02%) compared to 5.4% of the men with both ha ving about the same non - adherence levels (about 3% each). A m ajority of women were unemployed (41.1%) compared to men ( 19.7% ) with unemployed women more non - adherent (10.7%) compared to men ( 8.8% ). For religion, women were more of the Christian denominati on (87.4%) while the men were having different beliefs. Only 2.6% of the women practiced African beliefs of which 1 .0 % were non - adherent , compared to 11.3% men of which 5.4% were non - adheren t (Table 5 ). 52 Table 5 . Demographic Characteristics of Sampled Wo men (n=419) and Men (n=239) Living with HIV in Greater Gaborone, Botswana 2016. Demographic Characteristics Women Non - Adherence Men Non - Adherence No. (%) (%) No. (%) (%) Age (Years) 18 - 29 91 21.7 6.2 30 12.6 4.6 30 - 39 148 35.3 11.5 60 25.1 11.7 40 - 49 124 29.6 8.6 90 37.7 16.7 50+ 56 13.4 2.9 59 24.7 12.9 Marital Status Married 60 14.3 4.3 39 16.3 7.5 Single 223 53.2 15.9 105 43.9 21.3 Cohabit 118 28.2 7.9 88 36.8 15.5 Widowed 17 4.1 0.9 7 2.9 1.7 Divorced 1 0.2 0.0 0 0.0 0.0 Educati on No. Formal 23 5.5 1.2 38 15.9 7.9 Primary 89 21.2 6.2 73 30.5 14.6 Secondary 249 59.4 17.4 100 41.8 19.7 Vocational 16 3.8 1.2 15 6.3 0.8 University 42 10.0 3.1 13 5.4 2.9 Occupation Unemployed 172 41.1 10.7 47 19.7 8.8 Self - Employed 70 16.7 4.3 57 23.9 11.3 Farmer 14 3.3 1.9 14 5.9 2.9 Private 99 23.6 8.6 82 34.3 15.5 Government 64 15.3 3.6 39 16.3 7.5 Religion Christianity 366 87.4 25.1 156 65.3 30.9 African Beliefs 11 2.6 0.9 27 11.3 5.4 Other 42 10.0 3.1 56 23.4 9.6 Total 419 100.0 29 .1 239 100.0 46.0 Of the behavioral characteristics (Table 6), s exual behaviors showed that only 5.5% of the women had more than 2 partners within the past year and of these, 2.4% were non - adherent compared to 18.5% of the men who had more than 2 partner s, of which 10.5% were non - adherent. A m ajority of both women and men (72.5%) did not use a condom each time with a sexual partner and of those who did not use a condom, 21.7% of women were non - adheren t compared to 34.7% of men . A m ajority of the women 53 ( 72.6%) and men (74.5%) obtained condoms free from health facilities ; however, of these a bout 20 .3 % of women, were non - adherent compared to 33.9% of men. Table 6 . Behavioral Characteristics of Sampled Women (n=419) and Men (n=239) Living with HIV in Greater Gaborone, Botswana, 2016. Behavioral Characteristics Women Non - Adherence Men Non - Adherence No. (%) (%) No. (%) (%) No. Sex Partners 0 54 12.9 5.0 27 11.3 5.9 1 342 81.6 21.7 167 70.2 29.8 2+ 23 5.5 2.4 45 18.5 10.5 Condom Use Yes 115 27.5 7 .4 65 27.2 11.3 No 304 72.5 21.7 174 72.8 34.7 Obtain Condoms Workplace 2 0.5 0.2 5 2.1 1.3 Stores 113 26.9 8.6 56 23.4 10.9 Health Facility 304 72.6 20.3 178 74.5 33.9 Total 419 100.0 29.1 239 100.0 46.0 Of the health care facility characteristics (Table 7), c linic utilization showed that majority of both women (57%) and men (55.6%) attended rural clinics. Of these, 14.8% of women were non - adherent compared to 24.3% men. Non - adherence was double for men who lived in peri - urban areas (13.8%) compare d to 6.4% of women in the same location. A m ajority of women (44.4%) used public transit to reach the ART clinics compared to 28.1% of men who had about the same non - adherence levels (14.6 and 15.1%). Men who waited more than one - hour for a service at the clinic were more likely to be non - adherent (35.2%) compared to 18.9% of the women. 54 Table 7 . Health Care Facility Characteristics for Sampled Women (n=419) and Men (n=239) Living with HIV in Greater Gaborone, Botswana, 2016. Health Facility Characteris tics Women Non - Adherence Men Non - Adherence No. (%) (%) No. (%) (%) Clinic Location Urban 82 19.6 7.9 44 18.4 7.9 Peri - urban 98 23.4 6.4 62 25.9 13.8 Rural 239 57 .0 14.8 133 55.6 24.3 Transit to Facility Walk 211 50.4 13.1 138 57.7 23.4 Drive 22 5.3 1.4 34 14.2 7.5 Public Transport 186 44.4 14.6 67 28.1 15.1 Distance < 30 minutes 329 78.5 21.5 186 77.8 34.7 >= 30 minutes 90 21.5 7.6 53 22.2 11.3 Waiting Time < 1 Hour 122 29.1 10.3 57 23.9 10.9 >= 1 Hour 297 70.9 18.9 182 76.2 35.2 Total 419 100 29.1 239 100 46 Across all the demographics and socio - economic characteristics, men consistently showed high trends of non - adherence compared to women (Tables 5 - 7 ). 2.6.3. Risk Populations and Risk Factors for Non - Adherence Men of all ages showe d a high er odds of non - adherence compared to women. Older and oldest - aged m en aged 50+ versus women of the same age , had the highest odds of being non - adherent [OR 4.06 , 1.79 - 9.12] followed by those mid - late adult men ages 40 - 49 years [OR 1.88 , 1.06 - 3.56] (Table 8) . When controlling for the num ber of years ART initiated post - HIV diagnosis, men in their mid - early ages ( 30 - 39 years) , were also at increased od ds of non - adherence [ OR 2.05, 1.06 - 3.96] (Table 9 ). 55 Table 8 . Unadjusted Odds Ratio Estimates of Non - Adherence for Men (n=239) vs. Women 1 (n=352) Living with HIV, Greater Gaborone, Botswana, 2016. Sex by Age Odds Ratio 95% LCI 95% UCI Men vs. Women 18 - 29 Years 1.30 0.52 3.23 Men vs. Women 30 - 39 Years 1.89 0.99 3.58 Men vs. Women 40 - 49 Years 1.88 1.06 3.56 Men vs. Women 50+ Years 4.06 1.79 9.12 1 Excludes women on PMTCT (n=67). Table 9 . Odds Ratio Estimates of Non - Adherence for Men (n=239) vs. Women 1 (n=352) Living with HIV, controlling for the Number of Years ARV Initiated Post - HIV Diagnosis, Greater Gaborone, Botswana, 2016. Sex by Age Odds Ratio 95% LCI 95% UCI Men vs. Women 18 - 29 Years 1.24 0.50 3.11 Men vs. Women 30 - 39 Years 2.05 1.06 3.96 Men vs. Women 40 - 49 Years 1.91 1.06 3.45 Men vs. Women 50+ Years 3.95 1.73 8.97 1 Excludes women on PMTCT (n=67). Unadjusted odds of non - adherence showed that old and older men have the highest odds of being non - adherent compared to women of all ages. oldest - aged m en aged 50+ versus women of the same age , had the highest odds of being non - adherent [OR 4.06 , 1 .79 - 9.19 ] followed by those mid - adult men ages 30 - 3 9 years [OR 3.21, 1. 42 - 7.25 ] (Table 10) and those in their mid - late adulthood [OR 2.93, 1.37 - 6.28]. W hen adjusted, highest odds were still evident for oldest men aged 50+ compared to the women of the same age [OR 3.95, 1.74 - 8.97] followed by those aged 30 - 39 versus oldest women [OR 3.09, 1.36 - 7.03] and mid - late adult men [OR 2.86, 1.32 - 6.16] compared to women aged 50+ (Table 11). 56 Table 10 . Unadjusted Odds Ratio Estimates of Non - Adherence for Men (n=239) vs. Women (n=352) by Age Group. 57 Table 11 . A djusted Odds Ratio Estimates of Non - Adherence for Men (n=239) vs. Women (n=352) by Age Group. 1 Controlling for men vs. men and women vs. women by age group and the number of years ARV init iated post - HIV diagnosis. 2 Excludes women on PMTCT (n=67). 58 2. 6. 4 . CD4 Count as a Proxy of Non - Adherence In the following analy sis, only CD4 count was assessed. The decision to do so was based on the fact that after evaluating gender differences on non - adherence levels, low CD4 count accounted for the majority of the non - adherence levels. A continuous variable CD4 count showed tha t the range of 50 - 1,450 copies/ml (Figure 6 ). Women showed high levels of CD4 counts compared to men across the board. The mean CD4 count for wome n was 548.93 copies/ml (Figure 6 ) compare d to 447.23 of the men (Figure 6 ). The range CD4 count for men was 50 - 1,350 copies/ml, slightly lower than the maximum for the women (Figure 7 ). Figure 6 : Histogram of CD4 Count for a Sample of Women (n=352) Living with HIV in Greater Gaborone, 2016. 59 Figure 7 : Histogram of CD4 Count for a Sample of Men (n=239) Livin g with HIV in Greater Gaborone, 2016. By age, older women had the highest mean CD4 count (566.27 copies/ml ) followed by younger women (18 - 29) with 555.09 copies/ml. T he women mode (reoccur ring CD4 count was 400 copies/ml across all ages except for those ag ed 30 - 39 years . For m en, the age group 30 - 39 had the lowest mean CD4 count 428.68 followed by those aged 50+ at 429. 81 copies/ml (Table 12). Unlike the women, men had the lowest mode of 200 copies/ml which was followed by 300 copies/ml, still lower than t Younger males (20 - 29) had the highest mean CD4 count and mode compared to males of all ages (Table 12). 60 Table 12. Descriptive Statistics of CD4 Count for Sample of Women 1 (n=352) and Men (n=239), Greater Gaborone, Botswana, 2016. Sex by Age N Mean Median Mode Range Standard Deviation Women 352 548.93 522.00 400.00 1405 225.21 Ages 18 - 29 65 555.09 532.00 400.00 1251 249.72 Ages 30 - 39 117 542.59 523.00 391.00 1204 232.72 Ages 40 - 49 114 543.40 523.00 400.00 1278 208.62 Ag es 50+ 56 566.27 501.50 400.00 9325 216.78 Men 239 447.23 415.00 400.00 1276 199.49 Ages 20 - 29 30 470.33 447.00 400.00 690 149.88 Ages 30 - 39 60 428.68 399.00 200.00 1276 198.75 Ages 40 - 49 90 463.33 418.00 400.00 1156 195.51 Ages 50+ 59 429.81 380.00 300.00 1141 227.82 1 Excludes women on PMTCT (n=67). Men aged 30 - 39 (n=60) had low CD4 count due to being sick in the last 2 months (b - 99.39 p 0.0092) and taking other drugs (b - 98.74 p 0.04). Their CD4 count increased only when they had a healthy B MI and being initiated on ART before 2012 (Table 13). Table 13. Linear Regression Adjusted Estimates of CD4 Count for a Sample of Men 30 - 39 Years (n=60) Living with HIV, Greater Gaborone, Botswana, 201 6. CD4 Count Estimate Standard Error t - value Pr > [ t] Intercept 584.19 35.35 16.52 <0.0001 Taking Other Drugs - 98.74 47.74 - 2.07 0.0404 Sick in Last 2 Months - 99.39 37.65 - 2.64 0.0092 Mean BMI 9.28 2.76 3.35 0.0010 ARVs Initiated < 2012 89.33 36.24 2.46 0.0149 Viral Load - 0.03 0.04 - 0.73 0.4639 M ode l: df (5), F value=7.35, Pr>F <0.001 R - Square = 0.205, Coefficient Variation = 39.21, Root MSE = 219.10 (CD4 Mean=558.83). 61 Table 14. Health Profile of Men 30 - 39 Years (n=60). Illness Taking Medications Recent Sick Yes No Yes Hypertension 8 3 0 Tub erculosis 3 0 3 Anemia 4 3 1 Body Pains 1 9 1 Other 12 13 5 Total 28 28 10 For males aged 40 - 49 (n=124), their low CD4 count was due to high viral load (b - 0.11 p 0.0032) (Table 15). Finally, for males aged 50+ (n=56) was mainly to their single status (Table 16 ; Figure 8 ). Table 15. Linear Regression Adjusted Estimates of CD4 Count for a Sample of Men 40 - 49 Years (n=124) Living with HIV in Greater Gaborone, Botswana, 2016. CD4 Count Estimate Standard Error t - value Pr>[t] Intercept 547.14 35.51 15 .41 < 0.0001 Mean BMI 4.71 2.57 1.83 0.0698 ARVs Initiated < 2012 85.53 37.51 2.28 0.0244 Viral Load - 0.11 0.04 - 3.00 0.0032 Model: df (3), F value=6.62, Pr>F < 0.0004 R - Square = 0.141, Coefficient Variation = 37.03 , Root MSE = 203.29 (CD4 Mean=548.91) . Table 16. Linear Regression Adjusted Estimates of CD4 Count for a Sample of Men 50+ Years (n=56) Living with HIV in Greater Gaborone, Botswana, 2016. CD4 Count Estimate Standard Error t - value Pr>[t] Intercept 833.71 76.39 10.91 < 0.0001 Single vs. Co habiting - 303.21 84.34 - 3.60 0.0009 Model: df (1), F value=19.92, Pr>F 0.0009 R - Square = 0.259, Coefficient Variation = 34.55, Root MSE = 202.14 (CD4 Mean=584.92). 62 Figure 8 : Fitted Plot of CD4 Counts for a Sample of Men 50+ Years (n=56) Marital Stat us Single vs. Cohabiting, Greater Gaborone, Botswana, 2016. 2. 7. DISCUSSION By life course of how many years a patient can survive without ART post HIV diagnosis, results showed that men turned to survive a longer time than women. For both the women and m en, majority were diagnosed with HIV during their reproductive years. There were no women who were born with HIV or diagnosed as infants. Using CD4 count to assess the disease progression, women experience higher CD4 counts compared to men at all life - cour se stages, except those women aged (30 - 39 years) with delayed ART initiation. Women and men in their young adult and adult years of life generally respond to ARV treatment when initiated as evidence by a rising CD4 count. In contrast, women and men in th eir older and oldest stages of life respond less well to ARV treatment with increasing delays as also evident by a lower of CD4 counts. This implies that women are at more risk of acquiring HIV later in their life. 63 This study also uncovered that non ART adherence was (34%) among adult women (n=122) and men (n=110; 99 being men aged 30 and above) in Greater Gaborone. Men were further found out to have higher non - adherence rates (46%) compared to women (29.1%). In other studies, female gender has been show n to be associated with lower rates of ART adherence across several sub - groups (Tapp et al,2011; Berg et al . , 2004; Krawczy et al . , 2006; Anderson and Mitchell, 2000). Age was found to be an inhibiting factor when it came to adherence. Men aged 30+ showed to have the highest risk of non - adherence to ART when compared to women of all ages. Men aged 50+ had the highest odds of being non - adherent. Women across all ages were at a low risk of being non - adherent particularly women aged 50+. In Botswana, the age group 30 - 39, is the one with the highest HIV prevalence (BAIS IV, 2013). The same was observed in this study, and they were the same cohort that had low adherence levels. This age cohort is the working labor as well as the reproductive group, so if they ar e too sick and cannot work and care for their family, it is detrimental to the economy of Botswana. They are also at high risk of infecting their new babies (women) if they do not take the medications as instructed. Sexual behaviors showed that there were still both women and men who had more than two sexual partners in a year and condom use was minimal even though they obtained the condoms free from p ublic health facilities (Table 6 ). This shows that a lot must be done particularly with the men who usually tend to seek treatment late as they view health issues as women issues. Self - testing and assisted partner notification remain important but under - utilized methods to increase HIV diagnoses among men (UNAIDS, 2018). Women and men who lived in rural areas and peri - urban had high levels of non - adherence compared to t heir urban counterparts (Table 7 ). Locational and other demographics and socio - economic factors were associated with non - adherence in this study. Majority of the women were unemployed (41.1%) and about a quarter was non - adherent. Female unemployment rate in Botswana is higher (19.1%) than that of their male counterparts (16.3%) (Statistics Botswana, 2016). The Botswana Multi Topic Household Survey of 2015/16 survey estimated youth aged between 15 64 unemployment higher at 26.9% compared to 23.6% for males (Botswana Multi Topic Household Survey, 2016). Unemployment can drive young women to risky behaviors of transactional sex. Other factors such as bein g single (marital status) showed to be a constraining factor for older men aged 40 - 49. The lack of social support could be contributing to non - adherence. These findings were corroborated by other studies where social structural factors included low levels of social support (Yang et al., 2018), high - unemployment (Charurat et al . , 2010) and poverty. In Bezabhe and colleagues study (2014), patients cited unemployment, food insecurity, religious rituals such as fasting as their main concerns related to non - adhe rence (Bezabhe et al . , 2014). This study also found that biological factors played an important role in ART adherence. When a patient was healthy (normal weight BMI) and had low viral load, ART initiated before 2012, high levels of adherence were observe d. As previously mentioned, at the time of this study, (2016), CD4 count of below 350 copies/ml was used. To achieve viral suppression then becomes a difficulty because at this vulnerable stage, the patient is prone to opportunistic infections such as Tube rculosis (TB). Males aged 30 - 39 in this study showed that being sick of different illnesses and taking other drugs decreased CD4 count significantly. Having a high viral load was a risk factor for males aged 40 - 49. Several authors corroborated this study finding that taking medication every day is a challenge for anyone with any additional chronic illness such as diabetes and hypertension as they took their ARVs only about 70% of the time (Hashmi, 2007; Martin, 2005; Brown, 2011; Charurat et al . , 2010). 2 .7.1. Limitations The participants were geographically located in the southern part of Botswana , which contains urban, peri - urban and rural areas; however, since Gaborone is the capital city and circular migration happens from Gaborone, the study area may not be representative of Botswana as a whole. Nevertheless, this study utilized an adherence measure to evaluate the levels of non - adherence among adult men and 65 women that was sensitive in capturing non - adherence behavior and biological confirmation . In t erms of survey responses, r esponse bias could be a potential limitation because the participants could have provided answers they thought the principal researcher would like to hear. There could also be issues of recall bias because patients were asked to self - report how they took their medication. Another potential limitation is that the study interviewed only patients on ART and did not include patients lost to follow up/discontinued ART. It is possible that those patients were non - adherent and not attend ing their monthly appointments. This limitation would underestimate non - adherence levels in the study area. 2. 8. CONCLUSIONS AND RECOMMENDATIONS The current study demonstrated that male gender presents an additional barrier for adherence to ART among patie nts in Greater Gaborone, independent of other socio - behavioral, structural and clinical characteristics. This is evidence that even within a context where medical care and ART are provided free to HIV - positive individuals, there remain important healthcare access differences between male and female. Other socio - structural factors are also critically important in affecting adherence, and men often face barriers derived from their comorbidities and lack of social support. In order to improve adherence to ART among men and women, gender - specific interventions should be developed which recognize these unique barriers. Low adherence rates were mainly due to CD4 count particularly among the men. CD4 count is an important factor in monitoring the healthy immune sys tem of an HIV patient and if its low it can be an indicator that the medication is not working well. There is a need to educate men about the importance of taking their medications as prescribed because it will improve their viral suppression which in turn will minimize the new infections. Life course in HIV diagnosis and the length of time survival with ART showed that men in the study area tended to survive longer compared to the women. Furthermore, both genders were diagnosed with HIV in their reproducti ve ages implying the increase in new HIV infections in Botswana. The government of Botswana has made good progress in providing ART, condoms and safe male circumcision free of charge to people living with HIV. All these efforts are commendable 66 but without high levels of ART adherence, HIV/AIDS will never be kept in check because viral suppression will not be reached. The consequences of this effect are that new infections will grow rapidly as the treatment as prevention strategy will be failing. Education ( following medication instructions and awareness) is the most important aspect in the HIV/AIDS /ART matrix because if people hear and understand particularly older men, change can be guaranteed, and the 90 - 90 - 90 goals will be successfully reached in Botswan a by 2020 as the UNAIDS stipulates. Future directions will be to understand the matrix of gender and age in non - adherence to ART. Results showed that older males had highest rates of non - adherence when compared to females of all ages. This shows that more needs to be done to reduce the non - adherence rates because in the era where treatment is used as prevention, viral suppression should be reached. The older males have the potential to pass the virus to the females in the intergenerational sexual relations . When females of reproductive ages are infected, an extra burden is incurred because they risk passing the disease in cases of pregnancy and breastfeeding. This further emphasize the need to study the HIV life course to under stand the disease progression. There is a need to study the adherence levels of PMTCT to eliminate new infections in children. 67 CHAPTER 3: Prevention of Mother - t o - Child Transmission (PMTCT) Adherence and HIV status disclosure in Pregnant Women and New Mothers in Greater Gaborone 3.0. INT RODUCTION Comprehensive programmes for prevention of mother - to - child transmission of HIV (PMTCT) including ARV therapy and prophylaxis are current measures used to significantly reduce the number of infants who are living with HIV and promote better health for their mothers and families (Government of Botswana, 2015). PMTCT remains central to global HIV initiatives because international organizations such as the World Health Organization [WHO] and Center for Disease Control [CDC] as well as national governm ents are working together to build PMTCT capacities and achieve the global goal of reducing the HIV incidence rates among infants. National PMTCT programmes provide an important foundation for HIV prevention and treatment programmes. The Ministry of Health in Botswana has found out that the main reason for low rates of infant HIV testing is that early infant diagnosis is too centralized, with the only center in Gaborone (National AIDS Coordinating Agency [NACA], 2015). As of 2015, there were plans of openin g a second testing center in Francistown (second city of Botswana) and to roll out to four additional sites. A Sexual Reproductive Health (SRH) - HIV linkage project that was piloted in three districts and will be rolled out to all facilities nationwide to i mprove family planning for women, within the context of HIV in the near future (NACA, 2015). Benefits of this strategy is that it will increase the capacity of health facilities to cater for the SRH rights of HIV positive women and also specifically target s males to become more involved in SRH issues including partner testing and antenatal care. In addition, the Ministry of Health and Wellness is currently pursuing the treatment of all HIV+ pregnant women with triple antiretroviral therapy (ART) , regardless of cluster of differentiation [CD4] count, hence adoption of Option B+ (Ministry of Health and Wellness, Botswana [MoHW], 2015). According to WHO transmission approach for expectant mothers living with HIV in which women are immediately offered Even though, Option B+ is a great step towards eliminating vertical transmissions, Ebuy and Colleagues (2015) caution that with Option B+ 68 PMTCT, as the curr ent alternative, due to prolonged exposure to ART, high levels of non - adherence may arise for HIV - infected pregnant women (Ebuy, Yebyo and Alemayehu , 2015). 3.1. PURPOSE OF STUDY To investigate barriers to PMTCT adherence and their potential impact(s) on immunodeficiency (as a reflection on HIV transmission) for pregnant women and new mothers in Greater Gaborone, Botswana. The participants in this study included a sample of (n=67 pregnant and new mothers enrolled on PMTCT) and 84 health workers at 21 gove rnment clinics, in urban, peri - urban and rural areas of Greater Gaborone. To compare non - adherence levels among mothers and non - mothers, a sample of all women (n=419) was also used. This women sample is explained in detail how it was derived elsewhere (Ch apter 2). The findings from this study are used to create awareness of the WHO guidelines that recommend couple HIV counseling and testing which provides an opportunity for partners to test together thus enabling provider - facilitated disclosure and linkage to care if either partner is HIV - infected . 3.2. BACKGROUND 3.2.1. Maternal and Child Mortality in Botswana I n 2017, there were 52,358 live births, of which 52,242 were hospital births, representing 99.8 % of all births in Botswana. M aternal mortality rat io has declined from 156.6 deaths per 100,000 live births in 2016 to 143.2 deaths per 100,000 live births in 2017. T here were 75 maternal deaths in 2017 in Botswana classified into direct and indirect causes (Botswana - Maternal Mortality Ratio , 2017 ) . Over all, 76.0 % of all maternal deaths were due to direct causes, while the remainder were due to indirect causes. The leading direct c ause of maternal mortality was genital tract and pelvic infection following abortion and ectopic and molar pregnancy which ac counted for 20 .0% of all maternal deaths, followed by diseases of the circulatory system complicating pregnancy , childbirth and the puerperium at 10.7%, severe pre - eclampsia with 8.0% and rapture of uterus during labor at 8.0% ( Botswana - Maternal Mortality Ratio , 2017 ) . On the other hand, diseases of the circulatory system complicating pregnancy, childbirth and the puerperium 69 were the leading indirect cause of maternal deaths ( Botswana - Maternal Mortality Ratio , 2017 ) . T he highest maternal deaths were 24 (3 2 .0%) reported among age groups 35 - 39, followed by ages 30 - 34 (20.0% ), 20 - 24 with 16%, 25 - 29 with 14.7% and lastly age - groups 15 - 19 and 40 - 44 with 8.0 % respectively ( Botswana - Maternal Mortality Ratio , 2017 ) . No case of maternal death was reported for age s 45 and over in both direct and indirect causes. Diarrhea, pneumonia, s epticaemia, and s evere protein - energy malnutrition are the major causes of deaths among infants in Botswana . Latest statistics of 2016 show that under - 5 mortality in Botswana was 28.1, child mortality rate at 11 and infa nt mortality per 1000 live births was 27 (Statistics Botswana - Selected Statistical Indicators 1966 - 2016) . 3. 2 .2. ART and PMTCT Programme in Botswana sitive blood will be drawn to check her CD4 count. If her CD4 is less than 350 she will be initiated on ART for her own health benefit. But if the CD4 is above 350 she will be taking HAART for prophylaxis reason to transmission risks to the infant. Her par tner will also receive HIV counselling concerning their HIV status, available medical care for themselves and their families. HIV exposed babies will be given prophylactic treatment to reduce the transmission rate. And their mothers will be counselled on t wo breastfeeding methods, including exclusive breastfeeding and exclusive formula feeding. If she chooses exclusive formula feeding she will be provided with free infant formula until the baby is a year old. The baby will be tested for HIV at six weeks us ing dry blood spots (DBS) for Polymacros Chain Reagent (PCR). The baby will also be given crotrimazole preventive therapy until 12 months or when tested negative for HIV. Confirmatory test will be done immediately when the child is positive. If negative, r epeat testing will be done at 18 months using rapid test. All HIV positive infants whether symptomatic or not will be referre d to the ARV programme for ART W , 2011). Prompt and accurate HIV diagnosis in pregnant wome n is essential for referral for Triple ARV Prophylaxis 70 (MoHW, 2012). All pregnant women in Botswana who test HIV positive must immediately have blood collected for CD4 count to determine ART eligibility status. Regardless of CD4 count, all pregnant HIV - infected women qualify for Triple ART Prophylaxis during their pregnancy, therefore, all women found to be HIV - infected should be immediately referred t o local PMTCT services (MoHW, 2012). According to the Ministry of Health and Wellness in Botswana , the absolute number of women who newly tested HIV positive has declined from 14,058 in 2011 to 11, 845 in 2014 (MoHW, 2015). Re - testing at the third trimeste r of the pregnancy (36 weeks) among pregnant women remained below 2% in the reporting year of 2014 (MoHW, 2015). There was a repeated decrease in the number of pregnant women who received ARVs in 2014 at 90.8% (11,845) of HIV - positive pregnant women recei ved ARV medicine to reduce the risk of mother - to - child transmission compared to 95.9% in 2013 (MoHW, 2015). In the years of 2015 to 2017, the uptake remained unchanged at 95%. 3. 2 .3. Challenges to PMTCT in Sub Saharan Africa and Botswana Studies that inve stigate PMTCT adherence levels and challenges generally fall into three areas: patient - centered barriers, social - structural barriers, and service - centered barriers. Patient - Centered Barriers include Ochigbo (2013) who conducted a prospective cross sectio nal study using semi - structured questionnaire to determine the level of adherence among HIV infected pregnant women on PMTCT antiretroviral therapy in Area W Clinic, Francistown Botswana (Ochigbo, 2013). A total of 61 pregnant women participated in the stu had been enrolled in the PMTCT program at least more than one month previously. Assessing their demographics, a majority (75%) of the sample were between 26 and 42 years, 90% were single, 81% had attained secondary school education, and 60% were unemployed. Adherence levels were considered optimum if greater than or equal to 95%. Ochigbo (2013) found that the reported optimum adherence levels were 84% by virtual analogue assessment, 82% by pill cou nt and 98% indicated that they did not miss any dose during the last three days before the interview (Ochigbo, 2013). The factors influencing 71 non - adherence from the study were pregnancy - related illnesses, medication side effects, and month of pregnancy of the patient as participants tended to adhere less as they got closer to delivery (Ochigbo, 2013). The author recommended care - givers should carefully monitor patients for these factors, and to carry out continuous adherence counselling with special attenti on given to those approaching delivery in order to improve overall adherence to PMTCT therapy. He further stated that even though adherence levels to PMTCT therapy among the population sampled were high, more could be with interventions designed to cover a nd improve the highlighted areas in the implementation of the preventive therapy (Ochigbo, 2013). It is evident that there are a number of individual and structural barriers to PMTCT adherence in Botswana. HIV testing is only offered on week days, and this means that there are high chances of missed opportunities to test pregnant women especially in labor and delivery over the weekend because health care auxiliary personnel is off - duty in the evenings and during weekends (MoHW, 2017). Due to the disintegrat ion of SRH and ART services, almost 99% of health facilities still initiate ART at infectious disease care clinics (IDCC) and not at the sexual and reproductive health (SRH) or antenatal (ANC) sections, contributing to delays to treatment (MoHW, 2017). Of fering PMTCT late to pregnant women is detrimental because it exposes the unborn baby to HIV infection. The other concern is that there are some hospitals that do not keep ARVs in labor and delivery sections, hence late or non - bookers receive. There are in adequate skills on pediatric treatment initiation by medical officers in Botswana due to the complex systems/services involved in PMTCT therapy. There is still high stigma attached to infant formula feeding where by women still breastfeed their infants eve (Botswana Stigma Index Survey, 2013). There is evidence of low rates of testing among partners of pregnant women enrolled in PMTCT at 19% in 2015, 18% in 2016 and further decline to 16% in 2017 and i nadequate involvement of communities and civil society organizations is a challenge. This is mainly in part due to lack of HIV status disclosure of HIV pregnant woman and new mothers to their sexual partners 72 and family members. Disclosure of HIV status by women living with HIV to their sexual partners is important for the success of PMTCT as a comprehensive program (Rujumba et al . , 2012). Social Structural Barriers for women living with HIV in Sub - Saharan Africa face significant challenges including stigma and disclosure, hunger, cultural factors, lack of accurate health information, lack of social support, medication side effects and overcrowded health systems when seeking HIV care and adhering to antiretroviral therapy (Mepham et al . , 2011). Although ther e is evidence of these noted challenges, studies on challenges facing pregnant women when taking ART for prophylactic purposes in Sub - Saharan Africa is limited. Maintaining adherence to ART over time is a challenge in many settings also in Sub - Saharan Af rica where the average 24 - month retention rate in ART programs 2007 2009 was found out to be 70% (Fox and Rosen, 2013). Some studies in the region have identified social, cultural and economic barriers to antenatal and postnatal adherence to ART over time (Kirsten et al . , 2011; Nachega et al . , 2012; Mepham et al . , 2001; Nassali et al . , 2009; Duff et al . , 2010). Low levels of adherence have both individual and structural disadvantages because poor adherence to ART leads to virologic failure, increased risk of MTCT and a high risk of drug resistance that may require a change to more expensive treatment regimens (El - Khatib et al . , 2011; Ramadhani et al . ,2007; Bennett et al . , 2012; Hamers et al . , 2013). A sample of 501 pregnant HIV - 1 - infected women in Dar es Sa laam, Tanzania participated in a study called Mitra Plus (Ngarina et al . , 2013). It was an open - label, non - randomized, prospective PMTCT study where participants were initiated on triple ARV drugs from 34 weeks of pregnancy until six months post - delivery ( the breastfeeding period). Women enrolled with m l were put on ART for life and received significant adherence counseling and support in addition to transport incentives, free medical services including management of opportunistic infections , as well as formula or food supplements for their children after stopping breastfeeding if needed (Ngarina et al . , 2013). At the end of the study in 2009, viral - reported adherence, a high proportion of the women enrolled on ART for life had high viral loads at one and two years after delivery (Ngarina et al . , 73 2013). A semi - structured open - ended interviews to explore women own perceived barriers to adherence to ART post - delivery and after the cessation o f breastfeeding was conducted to investigate factors associated with better drug adherence among women in need of ART. Results showed that lack of motivation to continue ART after weaning the child, poverty and stigma were factors inhibiting uptake (Ngarin a et al . , 2013). The authors recommended that projects that simultaneously address stigma, their full potential (Ngarina et al . , 2013). Elwell (2016) conducted a qualitative study using interviews and focus groups to explore factors influencing women's adherence within PMTCT programs in southern Malawi (Elwell, 2016). Participants included current PMTCT patients, healthcare providers, community leaders, and patien ts who had that barriers to PMTCT uptake included fears of HIV disclosure to husbands, community - based HIV/AIDS stigma, and poor interactions with some health workers (Elwell, 2016). Facilitators of PMTCT adherence included the improved survival of PMTCT patients in recent years and the desire to remain healthy to care for one's children (Elwell, 2016). The author recommended that to improve access to med ical care for PMTCT patients, there should be integrated services to increase attention to confidentiality and minimize stigma, shared HIV testing and counseling for couples, and peer - led support groups to provide social support from other women with the s hared experience of an HIV - positive sero - status (Elwell, 2016). - Saharan Africa evaluating the PMTCT efficacy of triple therapy until cessation of breast feeding compared to short course zidovudine monotherapy in a predominantly breast feeding population (Mepham et al . , 2011). To further explain variations during objective adherence assessments, a sub - study was conducted at one site (Kwa Zulu Natal, South Africa) to examine the underlying adherence i ssues in 100 enrolled Zulu women. Information was supplemented 74 by unstructured, free - ranging interviews conducted by trained adherence counsellors on 43 consecutive women attending the trial clinic over a two - week period (Mepham et al . , 2011). Authors defi ned adherence as good at (>95% adherence), or poor (<95% adherence). Participants indicated that their poor adherence was due to therapy misconceptions/misunderstandings, ARV use by relatives, domestic violence, poverty and issues relating to disclosure an d stigma (Mepham et al . , 2011). About 61 .0 % (57/94) of antenatal women had good adherence with their PMTCT prophylaxis, with no significant difference shown between those taking the long and short course (Mepham et al . , 2011). Simo ni et al . ( 1995 ) and Kali chman ( 2003 ) have shown that women who disclose their HIV status is in part to reduce stress and increase access to social support (personal reasons) and to maintain honesty in relationships as well as protecting others from infection (interpersonal level) (Visser et al . , 2008). In a study conducted in Johannesburg, South Africa on HIV pregnant women, the main reason for HIV status disclosure was for the infant well - being such as breastfeeding avoidance (Varga et al . , 2006). Other studies done in Africa hav e shown that lack of HIV status disclosure to partners among women living with HIV inhibited the success of PMTCT program. In southern Malawi, Elwell (2016) conducted a qualitative study using interviews and focus groups to explore factors influencing wome n's adherence within PMTCT programs (Elwell, 2016). Participants included current PMTCT patients, healthcare providers, community content analysis and results showed that barriers to PMTCT uptake included fears of HIV disclosure to husbands, community - based HIV/AIDS stigma, and poor interactions with some health workers (Elwell, 2016). Facilitators of PMTCT adherence included the improved survival of PMTCT pati ents in recent years and the desire to remain healthy to care for one's children (Elwell, 2016). The author recommended that to improve access to medical care for PMTCT patients, there should be integrated services to increase attention to confidentiality and minimize stigma, shared HIV testing and counseling for couples, and peer - led support groups to provide social support from other women with the shared experience of an HIV - 75 positive sero - status (Elwell, 2016). Other studies have documented the fear of s tigma and abandonment (Mokhoka, 2000; Medley et al . , 2004; Awiti et al . , 2011; Varga et al . , 2006; Visser et al . , 2008; Rujumba et al . , 2012), loss of economic and emotional dependency (Varga et al . , 2006; Maman et al . , 2003; Larsson et al . , 2011; Visser e t al . , 2008; Rujumba et al . al . , 2006). Kebaabetswe (2007) argued that the existing belief that affordability of antiretroviral drugs offered to developing countries, does not translate to utiliza tion for the Prevention of Mother - to - Child Transmission of HIV Program in Botswana (Kebaabetswe, 2007). Botswana government offers free counseling, testing, antiretroviral drugs and infant formula for babies born to infected mothers and yet there are some women who not utilizing the program to protect their infants. She conducted a qualitative study on factors that facilitated and prohibited pregnant women from participating in the PMTCT program in Gaborone, Botswana. Forty subjects (ten HIV - positive pregna nt women who accepted the program, eleven who rejected the program, nine PMTCT health workers and ten key informants) participated in in - depth interviews (Kebaabetswe, 2007). Using thematic content analysis found uptake barriers to the uptake included the fear of knowing one's own HIV status, infant feeding distribution stigma, lack of male antiretroviral drugs and free infant formula (Kebaabetswe, 2007). Clinical Services - R elated B arriers include non - adherence to guidelines and recommendations made by PMTCT programs is a problem faced by both healthcare providers and HIV - infected women (Mofenson, 2010; du Plessis et al . , 2014; Gamell et al . , 2013; S pangler et al . , 2014). Other problems were late presentation for health care delivery service during pregnancy, inadequate ante - natal visits, poor opportunistic infection screening, non - disclosure of HIV - positive status to partners, and inadequate family p lanning counseling (du Plessis et al . , 2014; Bucagu et al . ,2013; Mnyani et al . , 2014). Low HIV testing rates during pregnancy, suboptimal healthcare coverage, and poor patient retention are among the issues 76 experienced by a lot of health institutions in Su b Saharan Africa (Prendergast et al . , 2015; Adetokunboh and Oluwasanu, 2016). The accessibility to continuous care for HIV - infected mothers during the post - natal period and Early Infant Diagnosis (EID) for HIV - exposed infants are both low (Mugasha et al . , 2014; Adetokunboh and Oluwasanu, 2016). Due to this loss of continuity in health care, there are high cases of delayed testing of HIV - exposed infants, which delays initiation of ART, and results in missed perinatal care opportunities (Lilian et al . , 2013; Adetokunboh and Oluwasanu, 2016). Cultural practices were associated with a risk of mixed infant feeding as a result of feeding preferences and poor maternal adherence to recommended infant feeding guidelines (Lawani et al . , 2014; Adetokunboh and Oluwasan u, 2016). In a South African hospital, Feucht and colleagues (2014) found that access to HIV services prior to conception, family planning, tuberculosis screening, HIV disclosure, psychosocial support, and post - natal care were not integrated within the PMT CT services and there has been an overall deterioration in consistent infant feeding messages conveyed to HIV - positive women (Feucht et al . , 2014). Plans to involve men in PMTCT services were limited to mostly HIV testing and counseling with minimal provi sion for men treatment, care and support and all these tended to alienate them rather than sustaining their involvement (Sherr and Croome, 2012; Adetokunboh and Oluwasanu, 2016). 3. 2 .4. Reoccurring Barrier to PMTCT: HIV Status Disclosure 3.2.4.1. HIV Statu s Disclosure Varga and colleagues (2006) identified four disclosure forms. The first two disclosure - status through a decision made and undertaken w ithout coercion. The individual is in control of when, how and to whom - status is revealed without the ssion or intent. The second two re address mode: an 77 thout it being openly articulated . , 2006 pg. 952 - 953) . In their study, Visser and colleagues (2008), explained that disclosure of HIV status involves a series of steps of decision making. These include (1) coming to terms with the HIV diagnos is, decision on the appropriateness of disclosing to a particular person, weighing the anticipated impacts and then choosing the appropriate situation for disclosure (Visser et al . , 2008) . 3. 2 .5. Constraints and Promoters of HIV Status Disclosure of Women HIV status disclosure serves as an important prevention strategy in prevention of mother to child transmission ( PMTCT ) and enhances adherence to key PMTCT interventions (Madiba and Letsoalo, 2013) . Disclosure also leads to increased utilization of prevent ive strategies and allows an individual to get partner or family support for preventive actions they may decide to undertake (Mucheto et al., 2011; Stirratt et al., 2006). It has been documented that women are more likely to adhere to PMTCT interventions w hen they have the support of their sexual partners (Medley et al., 2004). However, in order to receive support, the women must disclose their HIV - positive status (Stirratt et al., 2006 ; Madiba and Alaetsoalo, 2013 ). Women who hide their HIV status are less likely to adhere to PMTCT interventions and might put their children at risk beca use they may be unable to take antiretroviral t herapy (ART) until delivery (Madiba and Letsoalo, 2013), fail to give n evirapine syrup to the infant for recommended duration, and fail to adhere to the exclusive infant feeding option they selected (Kebaabetswe, 2007; Olagbuji et al., 2011; Rujumba et al., 2012). The importance of disclosure in the prevention of vertical transmission of HIV makes it a fundamental topic for inclus ion in counseling HIV pregnant women in the PMTCT program (Baek et al., 2009). P artner disclosure is a central concept in PMTCT pr ograms (Njunga and Blystad, 2010) and the success of members of her f amily (Igwegbe and Ugboaja, 2010). However, in South ern Africa n countries such as Botswana and South Africa , the HIV positive pregnant woman is often considered to be the primary target of PMTCT interventions 78 (Madiba and Letsoalo, 2013) . F or PMTCT to be fruitful, it is necessary to provide adequate information on PMTCT therapy to the general population, the family, the in - laws, and male partners of HIV - positive pregnant women (Madiba and Letsoalo, 2013) . There is a need for transparency abo ut HIV positive status within households to shape community norms on medication adherence, infant feeding , and couple testing. In their qualitative study, Madiba and Letsoalo (2013) interviewed 25 HIV positive women enrolled in a prevention of mother to ch ild transmission (PMTCT) program in Tshwane, South Africa. Four focus group interviews were used with each having at least seven participants. The women were included in the study if they were HIV positive and enrolled on PMTCT and their babies aged betwee n 6 weeks and 6 months (Madiba and Letsoalo, 2013). Data was analyzed using thematic analysis through Nvivo8. Findings showed that 18 women were single and 7 married. Out of the 25, 10 women have disclosed HIV status to partners only, 8 to partners and clo se family members, 2 to close members and 5 have not disclosed to their family (Madiba and Letsoalo, 2013). Women cited their reasons for disclosure as a way to gain support from family members to adhere to PMTCT. Moreover, women also disclosed to family members and friends they trusted. Women who have disclosed experienced various reactions from their partners. Findings showed that women received no support a nd others were abandoned by their partners (Madiba and Letsoalo, 2013). Other common reasons for non - disclosure were stigma, fear of rejection, protecting feelings of others and protecting then HIV status. For those women who disclosed to partner, they rep orted that adherence top PMTCT was easy. In conclusion, authors emphasized the need to develop interventions that address social and cultural contexts of HIV positive mothers to encourage HIV status disclosure in PMTCT programs (Madiba and Letsoalo, 2013). In the longitudinal study conducted among pregnant women (n=390) attending rural antenatal clinics in 79 PMTCT and maternal health services (Spangler et a l . , 2014). Baseline questionnaires were administered counselling and rapid testing followed by post counselling and PMTCT services for those who tested positive. HIV positive status disclosure was defined as having disclosed to anyone; a male sexual partner; immediate family member and to other persons including friends, distant relatives and the community (Spangler et al . , 2014). Chi - square test was perf ormed to examine differences between women who reported using the services and those who did not. Bivariate logistic regression using STATA12 statistical software package) then followed to investigate associations of different HIV positive status disclosur e categories with the use of antiretroviral drugs for PMTCT and birth in a health facility. Results showed that giving birth at a health facility was the highest among HIV positive women who had disclosed to anyone (49 .0 %). Those who have not disclosed had a lower rate at 21% (p<0.001). For the use of ARVs for PMTCT, HIV positive women (88 .0 %) who disclosed reported utilizing the service compared to 56 .0 % of those who had not disclosed. With regard to HIV status disclosure, n=72 (45 .0 %) of women disclosed t o anyone, 44 (28 .0 %) nonexclusively to a male partner, 32 (20 .0 %) to a family member and 19 (12 .0 %) to another person (Spangler et al . , 2014). When comparing women who had disclosed to anyone versus those who did not, they were 5 times more likely to use A RVs during pregnancy and 4 times likely to give birth in a health facility. Nonexclusive disclosure to a male partner was significantly associated with ARV use and birth in a health facility. In conclusion, authors highlighted that disclosure of HIV positi ve status may become critical social, economic and medical needs (Spangler et al . , 2014). In another Kenyan study, a case - control design was used among 180 mot her - baby pairs with HIV - exposed infants. Thirty - six pairs with HIV positive babies (cases) were compared to 144 pairs with HIV negative babies (controls) to investigate whether the failure to disclosure an HIV positive status to a male partner was associat ed with increased risk of infant HIV acquisition and whether part of the association was 80 explained by the exclusion of male partner in PMTCT programs (Nyandat and Van Rensburg, 2017). Non - disclosure of HIV status to partner was defined as failure by HIV po sitive mother to tell her partner prior Student t - test and chi - square were used for data analysis as well as conditional logistic regression in SPSS, STATA and MPLUS statistical software. Results indicated that 17 .0 % (n=30) of the mothers in the mother - baby pairs had not disclosed their HIV positive status to their partners (Nyandat and Van Rensburg, 2017). A higher proportion of the cases had no t disclosed their HIV status (52.8% vs 7.6%). Another finding was that low male partner involvement (18.3%) was partly responsible for the association between non - disclosure of HIV status on mother to child transmission. In conclusion the authors suggested that stakeholders need to re - examine the policies on disclosure and male partner involvement in an attempt to address the two behavioral contributors to mother to child transmission of HIV (Nyandat and Van Rensburg, 2017). 3.3. GOAL OF THE STUDY The aim of this paper is to understand the constraining factors associated with PMTCT adherence among pregnant women and new mothers in Greater Gaborone, Botswana. The investigation is assessed through exploring CD4 count as a potential reflection on HIV transmiss ion. Immunology (CD4 count) plays a role in understanding the adherence levels of the mother. For instance, if a pregnant woman/new mother shows decrease in CD4 count over time, it could be indicative of poor adherence. Non - HIV status disclosure among sexu al partners can be a potential barrier to PMTCT adherence, hence the need to explore how marital status vary among those who do not disclose and also understand the reasons for non - disclosure among pregnant and post - partum mothers. To get a different pers pective on HIV status disclosure, health care workers (n=84) from these study areas were interviewed. The use of mixed methods approach allowed for multiple perspectives and validation of findings (Varga et, 2006). Lack of HIV status disclosure, 81 do not sig nificantly increase the risk of HIV transmission, however, declining trends in CD4 counts are suggestive of poor adherence levels to PMTCT for both partners. 3. 4 . THEORETICAL AND CONCEPTUAL FRAMEWORK 3. 4 .1. Health Belief Model In this paper, the health be lief model is undertaken as a theoretical framework to draw upon [lack of] HIV status disclosure of pregnant and mothers living with HIV and enrolled on PMTCT. The Health Belief Model (HBM) is a psychological model that attempts to explain and predict indi vidual health behaviors by focusing on the attitudes and beliefs (Prochaska et al . , 2002). The HBM was first developed in the 1950s by social psychologists Hochbaum, Rosenstock and Kegels working in the U.S. Public Health Services (Prochaska et al . , 2002). The model was developed in response to the failure of a free tuberculosis (TB) health screening program (Prochaska et al . , 2002) and from that time, HBM has been used to investigate a variety of long - and short - term health behaviors, including sexual risk behaviors and the transmission of HIV/AIDS (Cox, 2009; Lindsay, 2009; Odu and Akante, 2009; Tenkarang et al . , 2009; Henderson, 2015). The HBM is based on the understanding that a person will take a health - related action (for instance, adhere to PMTCT ther apy) if that individual feels that a negative health condition (disclose HIV status to partner in order to prevent infection to both unborn baby and partner) can be avoided, has a positive expectation that by taking a recommended action, she will avoid a negative health condition (that is, disclosing HIV status will lead to the partner being supportive ), and believes that she can successfully take a recommended health action (disclose status so that she attends couple testing, takes the pills as recommend ed by doctor and attend clinic appointments). The HBM has four main constructs representing the perceived threat and net benefits being perceived susceptibility, perceived severity, perceived benefits, and perceived barriers (Prochaska et al . , 2002). These concepts were proposed so as to account for people's readiness to act and a recent added concept, of self - efficacy, usually termed as one's confidence in the ability to successfully perform an action. It was added by Rosenstock and colleagues in 82 1988 to h elp the HBM better fit the challenges of changing habitual unhealthy behaviors, such as being sedentary, smoking, or overeating (Rosenstock et al . , 1988). These concepts/constructs can then be linked back to knowledge that is accessible to those who design preventive strategies (Henderson, 2015). Perceived susceptibility stipulate that an individual has his/her own perception of the likelihood of experiencing a condition that could negatively affect one's health and it varies widely in their perception of s usceptibility to a disease. Those at low end of the spectrum are usual in denial of the possibility of contracting or even passing an adverse condition to others (HIV knowledge). Individuals in a moderate category admit to a statistical possibility of dise ase susceptibility. Those individuals at the high extreme of susceptibility feel there is real danger that they will experience a given disease (Prochaska et al . , 2002). The second construct being the perceived seriousness analyzes the beliefs a person hol ds concerning the effects a given disease would have on one's state of affairs. For example, financial burdens, social stigma, the prevention of di sease or toward dealing with an illness (in this case disclosure of HIV status) is the next step typically expected of individuals after they have accepted the susceptibility of a disease and recognized its severity and it is usually influenced by the beli efs regarding the action (Prochaska et al . , 2002). 3. 5 . STUDY DESIGN This study utilize d a n observational , cross - sectional survey ( winter , 2016) with retrospective responses. The survey included closed - ended and open - ended questions implemented among p reg nant women and mothers (n=67) living with HIV and enrolled on ART for PMTCT in urban, peri - urban and rural areas of Greater Gaborone. Both pregnant women and mothers were recruited because PMTCT continues after delivery thus calling for continuum in health care. The 67 women were taken as the main participants in this study, however, to understand the adherence levels among HIV positive women, all the 419 HIV 83 positive women attending 21 study HIV clinics and on ART in Greater Gaborone interviewed were inclu ded for comparative purposes only . Within these clinics, 84 health care workers were also surveyed for their perspectives in this study . The sampling methodology and mechanisms by which to optimize reliability and validity of the sample is described below. 3. 6 . DATA AND METHODS 3. 6 .1. Study Area This study took place in the Greater Gaborone which is made of the Gaborone city (as its own district) and three surrounding districts, South East, Kgatleng and Kweneng East Gaborone alone has a total populatio n of 231,592, while the larger metropolitan area has about 421,907 people (Botswana Census , 2011). During followed by urban villages and lower in rural area hence the selection of the Greater Gaborone. Gaborone has seen growing trends of HIV prevalence rates because of its strategic location, its proximity to the railway line and Johannesburg, South Africa, its already established administrative offices, its accessibility to most of the major tribes, its non - association with any particular tribe, and most importantly, its closeness to a major water source of Gaborone Dam. Gaborone and South East were in one district until recently (after the recent popula tion census of 2011) when Gaborone became an independent district on its one (Statistics Botswana, 2015). About 13% of males age 18 months and above were HIV positive, 19.8% were female, which gave the total district HIV prevalence to be 17% (BAIS IV, 2013 ). About 19% of males age 18 months and above were HIV positive, 14.6% were female, which gave the total district HIV prevalence to be 16.6% (BAIS IV, 2013). There were 33,332 people enrolled on ART in Gaborone district as of February 2017. The patients we re distributed across the 22 health facilities in the area (MoHW, 2017). South East district has a fewer ARV patients and clinics as compared to Gaborone with 4,659 people on ARV and six health facilities (MoHW, 2017). 84 Although the Kweneng East district is close to Gaborone, generally it experienced a modest growth of about three percent per annum between 2001 and 2011 (Statistics Botswana, 2015). A negative growth is usually due to outward migration, which is normally due to lack of employment opportunitie s in an area (Statistics Botswana, 2015). About 20% of males age 18 months and above were HIV positive, 22.1% were female, which gave the total district HIV prevalence of 21.5% (BAIS IV, 2013). There were 26,587 people enrolled on ART in Kweneng East distr ict as of February 2017. The patients were distributed across the 30 health facilities in the area (MoHW, 2017). About 16% of males age 18 months and above were HIV positive in Kgatleng District, 23.8% were female, which gave the total district HIV prevale nce to be 19.9% (BAIS IV, 2013). There were 11,028 people enrolled on ART in Kgatleng district as of February 2017. The patients were distributed across the 26 health facilities in the area (MoHW, 2017). Total fertility rates for the women across the distr ict was 2.88 and the mean age at child bearing was 20 years in 2016 (Statistics Botswana - Selected Statistical Indicators 1966 - 2016). 3. 6 .2. Primary Data Collection The data for this study were generated primarily from a survey administered to adult pregna nt women and mothers who are liv ing with HIV and enrolled on ART for PMTCT at public clinics ( n =21) (face to face interview) (n=67), as well as a semi - structured interview of key informants who are health personnel working (n=84) at the study areas. The st udy was a retrospective cross sectional survey which took place in May - August 2016. The questionnaires were administered in Setswana (national) and English (official) languages to the participants. For inclusion criteria, only 18 and above women who we re H IV positive and enrolled on ART for PMTCT therapy (WHO Option B+) at a public clinic took part in the study. The women had to also be pregnant or have a child who is at least 6 months and be on PMTCT program. Exclusion criteria, was that women younger than 18 and not on ART for PMTCT therapy be excluded from the study. The women were visited daily at their respective clinics until the maximum number of participants were reached. The questionnaires were administered by the principal researcher and four resea rch assistants. 85 The principal researcher took a training with the MSU IRB while the research assistants who are all obtained from the MSU IRB. All the assistants are fluent in both the languages. The self - administered questionnaire to pregnant women and mothers on PMTCT ( s ee Appendix B ) have 44 variables (with six sections) which range d from demographics to the water related concerns in the medication . The questions on demographics and maternal characteristics (first three sections) that included annual income, partner HIV status, nu mber of children and education we re close ended que stions. The remaining portions dealt with knowledge of HIV, treatment seeking behaviors, and water related concerns. These were more open - ended, allowing mothers (new and expectant) to provide detailed information. The sections on access a treatment seeking behaviors included PMTCT adherence ]. When women came to the clinics, they were required to bring their medical report card and this is where their CD4 count and viral load information was retrieved. Only the latest records were extracted and used in the analysis of the data. The information acquired from the participants was coded and preprocessed and analyzed using descriptives in SAS 9.4 st atistical software. To get a different perspective on non HIV status disclosure and PMTCT adherence , key informants were chosen and a semi structured interview (n=31 variables) used. Their responses were coded and evaluated using thematic analysis method t o generate informative themes that augment the survey data compiled from the patients. Their questions were all open - ended except for the demographics (first section) that included their qualifications, nationality and length at the health facility. The se ven remaining portions dealt with accommodation (in terms of PMTCT access), adherence , education to the patients and community on PMTCT as well as water related concerns. All these questions were open - ended allowing 86 health providers to provide detailed Inf ormation ( see Appendix C ) (doctor 1, nurse 2 and 3 [for the two nurses] and Pharmacist 4) before generation of themes for thematic analysis. 3. 6 . 3 . Sampling Methodology The (n=21) clinics were purposively selected fr om the three geographies (urban, peri - urban and rural areas) in the four districts (Gaborone, Kgatleng, Kweneng East and South East) of Botswana. These districts were selected because of their high HIV prevalence rates, high number of ART/PMTCT facilities and the (n=5) , peri - urban (n=5) and rural clinics (n=11) were selected. From all the clinics, a minimum of two pregnant and mothers who were living with HIV and on ART for PMTC T were purposively selected. The target was to get a total of 42 women (21 clinics * 2 women) from all the 21 clinics and this number was increased to 67 to account for under and over sampling. Every second pregnant and or mother to enter the clinic door o r the waiting area, was selected. For the key informants, a total of four from each clinic (doctor, two nurses and a pharmacist [ARV dispenser] were selected. As mentioned earlier, the comparative sample of 419 women, sampling methodology is explained in d etail from the previous work (Paper 1). No sampling was done on the key informants as the four mentioned are the ones that have daily and direct contact with mothers attending PMTCT clinics. 3.6.3.1. Reliability and Validity Two questionnaires were pre - tes ted with two pregnant women enrolled on PMTCT in the urban clinic of Old Naledi clinic the day before the official start of all the survey data collection. This was done to check how long it took the women to answer questions and how they understood the co ntents as well as the language they preferred. These two questionnaires were not included in the analysis of the data. All the pregnant females and mothers were visited at their respective clinics at the same time for the duration of the study (from 7 : 30 am until 2 :00 pm). In addition to the principal researcher, four experienced research 87 assistants assiste d with the data collection. These assistants were locals who were trained on IRB both from MSU and MOHW requirements and spoke both Setswana and English language s . Descriptive frequencies were run on all variables to inspect if they had enough values to perform statistical tests. 3.6.3.2. Ethical Review Permission to conduct the study was sought from IRB MSU and Human Research and Development Committee (H RDC) from the Ministry of Health and Wellness in Botswana to ensure that all rights, dignity and safety of study participants were safe guarded. After satisfying the requirements of the MoHW IRB, to get into the different clinics, separate approvals were s ought from the District Health Management Teams [DHMTs] in Botswana. MOHW provides primary health care services through DHMTs, which are responsible for running a network of health facilities. Participants were given information about the purpose and objec tives of the study. The interviewers also read the consent form to all participants. This form provided information about the freedom to participate or withdraw at any given point during the study without any consequences. A written consent was obtained fr om all pregnant and mothers living with HIV and on ART for PMTCT who were willing to participate in the study. Confidentiality and privacy actual addr ess es . Anonymity of respondents were assured by the use of codes (number ID) and aggregate analysis (relative addresses). 3. 6 .4. Descriptive Methods To answer the posed research questions, 17 variables from the questionnaire were used. These were categoriz ed into four groups being demographics, behavioral, physiological and geographical characteristics. These categories are important because PMTCT adherence is a complex measure that can and behavioral) and physical environment (place of residence) characteristics. From the 17, one behavioral variable (Know status) was a dependent variable. It was derived from question 14 ( s ee A ppendix B ) which asked the women if they 88 disclosed their HIV status to their sexual partner. It was a dichotomous variable, where (Yes) indicated disclosure and (No) a lack of disclosure of HIV status. As shown earlier by the literature p artner disclosure is a central concept in PMTCT pr ograms (Njunga and Blystad, 2 010) and the success of PMTCT amily (Igwegbe and Ugboaja, 2010). 3. 6 .5. Thematic Analysis In addition, to the women living with HIV and enrolled on PMTCT data, health care workers were also interviewed to argument the findings from the mothers. Thematic analysis is a method for identifying, analyzing, organizing, describing, and reporting themes found within a data set (Braun and Clarke, 2006). To conduct a thematic analysis , six steps recommended by Braun and Clarke (2006) were followed. The first step was to familiarize oneself with the data. This stage involved entering the data from the paper questionnaires into SPSS software (IBM SPSS [2017] Statistics for Windows, Versi on 25.0. Armonk, NY: IBM Corpas) a database management, and running frequencies in SAS 9.4 Software (SAS Institute Inc., 2002 - 2012, Cary, NC, USA). I nitial codes were generated by the clinics that women attended ( rural or urban areas ) . The third step was to search for themes and tying back to the theory (Health Belief Model) . T heme s were then review ed , where initially , four themes emerged based on the questions and the reoccurring terms that the health workers provided but after careful review, only two ma jor themes with four sub themes were used. This reduction was done because even though there were four themes in the beginning, they were interrelated which made them seem repetitive. The fifth step was on understanding the lack of HIV status disclosure to partners and therapy adherence thus the themes revolved around these concepts. The final step produce d a summary that tied the information to the theory. Two major themes emerged which were related to the adherence and lack of HIV status disclosure to mal e sexual partners . Direct quotes were used to emphasize meanings. By conducting all these steps, there was confidence in results generalization because a logical and falsifiable order was followed. 89 3. 7 . RESULTS 3. 7 .1. Non - Adherence Levels In general, moth ers were more likely to be non - adherent compared to non - mothers in the study area. The overall non - adherence was 32.2% for all the women with indications that pregnant women had the highest adherence levels (85.7%) followed by non - mothers (75 .0 %) and the l east for mothers at 60.4% (Table 2). There were 67 mothers living with HIV in Greater Gaborone. By age, there were more wome n aged 30 - 39 (41.0% mothers and 36.1 % non - mothers. Of these high non - adherence was for mothers. Majority of the women were single (5 9.7% and 51.9%) respectively and 37.5% of these single mothers were non - adherent compared to 25.5% of non - mothers. About 9 .0 % mothers had no education and half of them were non - adherent compared to 22.2 % of non - mothers. Only two mothers were university edu cated and w ere all adhering compared to 9 .9 % of college educated non - mothers of whom about a third was non - adherent. Majority of the non - mothers were unemployed (40.6%) followed by self - employed (16 .7%) while the reverse was true for mothers. Non - adherence was almost double for self - employed mothers (40 .0 %) compared to non - mothers (22 .0 %). Majority of the women were Christians (80.6% mothers, 87.7 % non - mothers). Christian mothers as well as those with other beliefs have both the highest non - adherence levels (37 .0 %). The one mother who practiced African beliefs was adherent (Table 1 7 ). Non - adherence levels by whether one was a mother or pregnant showed that of the 14 pregnant women , 14% (n=2) were non - adherent compared to mothers (n=53) who had children alrea dy of whom, 39. 6% were non adherent (Table 18 ). 90 Table 1 7 . Demographic Characteristics of Mothers (n=67) vs. Non - Mothers (n=3 52 ) Living with HIV in Greater Gaborone, Botswana 2016. Demographic Characteristics Mothers Non - Adherence Not - Mothers Non - Adher ence No. (%) (%) No. (%) (%) Age (Years) 18 - 29 10 14.9 50.0 71 20.2 21.1 30 - 39 28 41.8 32.1 12 7 3 6.1 29 .1 40 - 49 29 43 .3 31.1 1 07 30.4 24.3 50+ 0 0 0.0 47 13.3 21.3 Marital Status Married 5 7.5 20.0 54 15.3 25.9 Single 40 59.7 37.5 1 8 3 51.9 24.0 Cohabit 21 31.3 28.6 100 28.4 27 6.0 Widowed 1 1.5 100.0 14 3.9 28.6 Divorced 0 0.0 0.0 1 <1 0.0 Education No. Formal 6 8.9 50.0 18 5.1 22.2 Primary 24 35.8 29.2 75 21.3 28.0 Secondary 32 47.8 40.6 2 12 60.2 2 3 .6 Vocational 3 4.5 0.0 1 2 3 .1 25.0 University 2 2.9 0.0 35 9 .9 28.6 Occupation Unemployed 17 25.4 29.4 1 43 4 0.6 2 3.8 Self - Employed 20 29.9 40.0 59 16.8 22.0 Farmer 1 1.5 100.0 13 3.7 53.9 Private 21 31.3 38.1 81 23.1 25.9 Government 8 11.9 12.5 5 6 1 5.9 23.2 Religion Christianity 54 80.6 37.0 3 09 87.8 24.3 African Beliefs 1 1.5 0.0 11 3.1 27.3 Other 12 17.9 37.0 3 2 9. 1 31 . 3 Total 67 100.0 34.3 3 52 100.0 26.3 Table 18 . Maternal Characteristics by Adherence vs. Non - Adhere nce for a Sample of Women (n=419 ) in Greater Gaborone, 2016. Maternal Characteristics Women Non - Adherence Adherence No. (%) No. (%) No. (%) Mother - Pregnant 14 3.2 2 14.3 12 85.7 Mother - Post - Partum 53 12.6 21 39.6 32 60.4 All Other women 352 84.01 88 25.0 264 75.0 Total 4 19 100.0 111 32.2 308 67.8 Adherence = Mothers vs. All Other Women (Odd Ratio = 1.47, 95% CI 0.84 - 2.55). 91 3. 7 .2. HIV Status Disclosure All of the women enrolled on PMTCT were diagnosed between 2002 - 2016. There were more (n=42) ompared to those who were not sure (n=21) and those who did not know (n=4) (Table 19 were for both those who did not know (Table 19 ). Of the 67 mothers, 45 women have told their partne rs their HIV status while 1 3 have not and 9 were still finding ways on how to. With regards to duration on PMTCT, 47 have been enrolled for less than 6 months and 20 for more than 6 months. The maximum number of children among the mothers was 6 (Table 20 ). Table 19 1 Status, Greater Gaborone, Botswana, 2016. Year Yes No Not Sure N=42 N=4 N=21 2016 S 1 8M 1 ,C 1 - C 1 2015 S 1 - S 1 ,C 1 201 4 S 1 ,M 1 ,C 2 - S 3 2013 C 1 - S 2 2012 S 3 ,C 4 - C 1 2011 S 3 ,C 1 S 1 C 1 C 2 2010 S 3 ,M 1 ,C 1 - C 2 2009 S 1 - S 1 ,C 1 2008 S 1 ,M 1 ,C 3 - C 1 2007 S 1 ,M 1 ,C 1 C 1 - 2006 S 1 ,M 2 ,C 2 - - 2005 S 1 ,M 1 ,C 1 - S 1 2004 - - C 2 2003 C 1 S 1 S 1 2002 S 1 ,C 1 - S 1 Total S 16 ,M 8 ,C 18 S 2 ,C 2 S 10 ,C 11 1 S=Single; M=Married; C=Cohabitate 92 Table 20 1 for Mothers Enrolled in PMTCT by Year of HIV Diagnosis and Number of Children. 1 No= No partner does not know, I am still finding out how to tell him (n=1 3 ) Not Sure= Not sure, I do not know how to tell him (n=9) Yes= Yes, partner knows of my HIV status (n=45) To further understand the relationships of mothers and their partners, age and marital status was assessed. Young mothers (18 - 29) who were sing le did not have older partners and the only one with older partner aged 45 - 50 was cohabiting. Mothers of middle ages (30 - 39) were more likely to be in unmarried relationships with men of 35 - 55 years (15 single and n=10 cohabiting, only 2 married). For the older mothers, their partners were of the same age (45 - 50) with majority (n=5) of married status (Table 21 ). HIV Diagnosis Year Enrolled in PMTCT < 6 Months (n=47) Enrolled in PMTCT >= 6 Months (n =20) Disclose HIV Status Disclose HIV Status No Not Sure Yes No Not Sure Yes 2016 - - 1 - 1 2 2015 1 1 1 - - - 2014 1 2 2 - - 2 2013 1 - - - 1 1 2012 1 - 2 - - 4 2011 2 1 2 1 - - 2010 1 - 6 - - - 2009 1 1 1 - - - 2008 - 1 4 - - 1 2007 - - 3 - - 1 2006 - - 3 - - 2 2005 1 - 3 - - - 2004 2 - - - - - 2003 - 1 2 - - - 2002 1 - 2 - - - Total 12 7 32 1 2 13 93 Table 21 To investigate the role of living with family members who are HIV positive and how that affected HIV status disclosure of the mothers, the followi ng analysis was conducted. About 67 .0 % of the mothers of lived in a household with a member who is HIV positive, have disclosed their status compared to the 33 .0 % who did not live with HIV positive family members (Table 22 ). Table 22 . HIV S tatus Di sclosure of Mothers and Family Member L iving with HIV. 3.7.3. CD4 Count and Year of HIV diagnosis Mothers started off with high CD4 count during early years ( 750 copies/m l in 2002) and gradually decreased with time. M others with HIV infected partners and those who did not know (Figure 10) had the declining trends of CD4 count compared to those with HIV negative partners (Figure 10 ). For mothers who Age Groups (Years) Marital Status Single No. Married No. Cohabitate No. Mother 18 - 29 Partner 25 - 34 6 - 11 Partner 35 - 44 2 1 4 Partner 45 - 50 - - 1 Total 8 1 16 Mother 30 - 39 Partner 25 - 34 1 - 1 Partner 35 - 44 15 2 10 Partner 45 - 50 1 - 2 Total 17 2 13 Mother 40 - 49 Partner 25 - 34 - - - Partner 35 - 44 - - 1 Partner 45 - 50 3 5 1 Total 3 5 2 Family with HIV Disclosur e Yes No Total Yes 30 (66.7%) 15 (33.3%) 45 No 17 (77.3%) 5 (22.7) 22 94 have disclos ed to their partners, they had a high baseline CD4 count compared to those who have not disclosed (Figure 11 ). Figure 9 : CD4 Count by Year of Diagnosis, Sample of Mothers (n=67) Living with HIV in Greater Gaborone, Botswana, 2016. 95 Figure 10 : CD4 Cou Status, Greater Gaborone, Botswana, 2016. Figure 11 : CD4 Count by Year of HIV Diagnosis for a Sample of Mothers (n=67) and HIV Disclosure Status to Partner, Greater Ga borone, 2016. 96 3. 7 .4. Thematic Analysis: Qualitative analysis To answer this research question, semi structured interview conducted with the health workers were used. Four questions surrounding PMTCT guidelines were used and these were on breastfeeding, PMT CT enrolment criteria, social support and compliance. Two major themes emerged which are (i) conflict in autonomous decision making and (ii) adherence and compliance. 3.7.4.1. Theme 1: Conflict in Autonomous Decision - Making This theme was named based on th e two questions on breastfeeding and social support. To avoid transmission of virus to the baby, the World Health Organization [WHO] recommends that women breastfeed and PMTCT program offers free formula milk to all mothers living with HIV. The WHO contin ues to caution that in cases where formula feeding is not a viable option, women living with HIV are recommended to exclusively breastfeed (rather than mixed feeding), providing that they are on ART. This is because, while formula feeding offers the safest option for postnatal HIV prevention, in resource poor settings it is not always easy for families to afford or access things such as clean water which are needed for it use (UNAIDS, 2013). Throughout the years, the government of Botswana has recommended t hat women living with HIV should exclusively formula feed their infants and provided infant formula free of charge until the infant is 1 year of age to support this recommendation ( MoH W 2013; Ndubuka et al . , 2013). However, earlier in 2011, MoH W recommende d exclusive formula feeding for the first six months of life only for women for whom formula feeding is acceptable, feasible, affordable, sustainable and safe (AFASS) (WHO, 2006; 2007; Ndubuka et al . , 2013). On the other hand, the society in Botswana, rega rds a the one who breastfeed. This task of decision making can be daunting to the new mother who is already battling with the fact that she is HIV positive. A female nurse in a rural clinic said , ause some women have high viral load, but babies on the other hand need mothers antibodies, but we [Health providers] make sure to give 97 them advice on both advantages and disadvantages of infant feeding (Female Nurse, Rural Clinic ). Another nurse in a ru ral clinic in Kweneng East district emphasised on the mothers defying the cultural We advise mothers to ignore the village talk that looks down on women who bottle feed (Fem Nurse, rural ). A male pharmacist at the s concerns when he said , period, so that they understand why the woman cannot breastfeed (Male Pharmacist, R ural Clinic ). Formula mi lk was provided for free to mothers who choose to, until the baby reaches the first age. According to the Government of Botswana (2011), If a mother chooses exclusive formula feeding she will be provided with free infant formula until the baby is 12 months of age. This aligned with what the doctor in a Gaborone clinic said when asked if mothers living with HIV had a choice of infant feeding. He stated , provide free formula milk for those who choose to bottle feed (Male Physician, Urban Clinic ). It is clear that different clinics and health workers, do have preferences on the type of infant feeding. A doctor in a peri - urban area emphasized that they wanted the mothers to use formula milk but those that choose to breastfeed due to societal pressure, we cautioned to avoid mixed feeding. He said , Those who choose to breastfeed because they want to please their in - laws, particulalry the married ones, we emphasize that they should breastfeed for atleast 6 months exclusively (Male Physician , Peri - urban Cl inic ). Social support is also important to mothers who are on PMTCT program for the successful execution. receive counselling concerning their HIV status, a 98 This implies that mothers living with HIV should disclose their status to at least their sexual partners, families and health workers particularly when they visit non - PMTCT clinics such as obstetric ians. A female nurse in an urban clinic emphasized that most of the mothers living with HIV that they see on daily basis were open to them about their status but the opposite was true when they visited other services (non HIV) health facilities . the mothers to communicate openly with their gynecologists at other clinics about [... non - disclosure] can affect their unborn children. If a gynecologist (F emale N urse in, Gaborone ) . As an extension of social support, w hen health workers were asked if the mothers /pregnant females were required to bring treatment partners trusted individuals that act as adherence partners particularly their d clinics. Ten health providers from different clinics responded that women were not required to bring their partners to the clinic appointments. Among those who said no, five were nurses [urban, rural], three pharmacists [urban, peri - urban] and two doctor s [in Gaborone]. Majority of the health workers indicated that women were required to bring their sexual partners and if not possible, a close member of the family. A female doctor in a rural clinic added that , We tell the women that, if their partners re fuse to come to the clinic with them usually on treatment counselling days, we require them to tag along their mothers, aunts, as they are the (confinement period ). It is important to do so, because their caretakers usually force them to breastfeed telling them that bottle feeding is torturing the baby by not providing the required 99 e most of (Female Physician, Rural Clinic) . A female nurse in rural clinic of indicated that even though they asked the mothers to bring their partners to the clinic, they did not deny them the services. She said , yes we ask the women to bring someone with them to the clinic, but it is not enforced really. It is within their rights to be given help regardless of whether they have brought their adherence (F emale Nurse, Rural Clinic). 3.7.4.2. Theme 2: Adherence and Compliance Two questions were posed to the health care workers and they included the criteria that they used to define eligibility for PMTCT and the compliance measures that they used to assess ad herence [or lack of] to the program. All the health workers were in agreement that all mothers living with HIV were enrolled in the PMTCT program but it was clear that there were slight differences among clinics. A male nurse in a rural clinic of Siga said that , that we do not get women who enroll in late stages. Particularly here in rural areas, women th month of pregnancy. If ( Male Nurse, Rural Clinic). I t is clear that since the adoption of option B+, most clinics were phasing out the CD4 count criteria. In the past, we used the CD 4 count criteria like in the ART program, but we are gradually phasing it out. Right now (2016), all the women who come to our health facilities, as long as they are pregnant and HIV positive, we enrol them in the PMTCT program (Female, Pharmacist , Peri - u rban Clinic ). 100 These sentiments were echoed by 53 other health care providers who stated that they did not use the CD4 count criteria as of the time of the interview (May - August 2016). On compliance with the program, the reoccurring terms among the health A doctor in Gaborone expressed concern that mothers living with HIV particularly in the low income area s presented to the clinic in the late stages of pregnancy. Most of these women s ometimes do not understand the importance of early enrolment, and this gives us very limited choices if an unknown HIV positive [..status] mother shows to the delivery room, unregistered , never attended antenatal care classes, HIV testing , .You can imagi ne the challenge she presents to us as health care providers. So we emphasize the benefits of having a baby born free of HIV. When they hear that their babies can have a different life from theirs, you can see big smailes from their faces (Male Physician in , Gaborone). Among the health care providers who emphasized that non - adherent women were offered counselling sessions, one emphasised the importance of patient - provider relationship. We encourage them [..women] to treat us as they would a friend, tha t way it becomes easy for them to confide in us and tell us what they do not understand when it comes to the PMTCT program. And we really understands that the program is intensive and sometimes, culturally foreign to them, but for their sakes and babies, i t is important to follow each and every guideline (Male Nurse, Rural Clinic ). Health workers added that the use of Setswana language among those women with little formal education was important to make sure the message reaches to everyone. He said , Fo r the women that we feel do not have the education[..formal training] , we always counsel and give all the instructions in Setswana. We realized that sometimes women 101 understan d, so Setswana comes to rescue in cases like those (Male N urse, Rural Clinic ). A phamacist in (rural) added that counselling should not be for particular days only as they currently have in their health facility, it should be done on daily basis. ngoi ng counselling is very important because these women battle HIV on daily basis. - patient counselling, it can also be patient - patient, family counselling too (Male P harmacist, R ural Clinic ). Adherence partners were shown to be the intergral part of the PMTCT program as they unloaded the burden of the health workers . A female nurse in Gaborone clinic indicated that , defaulters I mean the women that , no matter how much we counsel them to follow the program strictly, they will still miss something. It could be mixed feeding, missing clinic appointments for herself and the baby.Those are the ones that we tell them to bring someone for social suppo rt because it is very clear that they cant cope on their own. And sometimes we fear that pressurisng them to come with someone deter them from coming at all (Female N urse, in Gaborone). 3.8. DISCUSSION About 67.2% (n=45) of HIV - positive mothers enrolled i n the study reported having disclosed HIV positive status to their partner. This finding is comparable to most studies from Sub - Saharan Africa, with disclosure ranges from 16.7% and 86% (Nyandat and Van Rensburg, 2017). Bachanas et al. (2013) reported 80.0 % of the 3538 HIV - positive patients in Kenya, Tanzania and Botswana had disclosed their status to their partners. Similarly, 70.0% of 20 HIV - positive pregnant women in Kenya had disclosed their status (Walcott et al., 2013). In their study, Nyandat and Va n Rensburg found that 83.3% of the women in their Kenyan study have disclosed their HIV status to their male partners (Nyandat and Van Rensburg, 2017). 102 In this study findings showed that by age, there were more middle aged women (30 - 39) ( 32.1 % mothers and 29.1 % non - mothers. High non - adherence levels were observed for mothers. Majority of the women were single (59.7% and 51.9%) respectively and 37.5% of these single mothers were non - adherent compared to 2 4 % of non - mothers. About 9% mothers had no education and half of them were non - adherent compared to 22 % of non - mothers. Only two mothers were university educated and we re all adhering compared to 9.9 % of college educated non - mothers of whom about a third was non - adherent. Majority of the non - mothers were une mployed (40.3%) followed by self - employed (16 .7%) while the reverse was true for mothers. The single status of the women as well as lack of unemployment showed to be the risk factors of non - adherence. This finding was corroborated by findings from a Kenyan study (Kinuthia et al., 2018) on PMTCT uptake and non - disclosure that shorter duration of relationship and employment status were associated with non - di sclosure of HIV - positive status. Being single can compromise of many things not limited to lack of soci al and financial support. The same can be said about lack of unemployment. It means the woman has to be dependent on a family member or partner when it comes to transport fees for clinic checkups and food for nutrition. This was consistent with other studi es (Antelman et al., 2001; Kiula et al., 2013) that women who were unmarried were more likely not to disclose . All of the women enrolled on PMTCT were diagnosed with HIV in the recent years of 2002 - 2016. There were more (n=42) mothers who knew their partn know. It is usually easier for married couples to keep no secrets between themselves hence the reason why a lot of married men knew their partner HIV status. Even though majority of the mothers (n=45) have disclosed their HIV status to their partners, there were still the 22 who have not. This is a real concern because not only the women health is involved PMTCT as a comprehensive program, it involves couple testing. In a case where the mother has not told 103 the partner her HIV status, it means, she attends the clinic appointments by herself. Other Afr ican studies (Kenya - Kinuthia et al., 2018; Kenya Farquhar et al., 2004) have found that HIV - infected women who had not disclosed their results were less likely to use PMTCT services . Study conducted in Botswana, reported that there was a general concern o f low male partner testing (21 .0 % in 2017) (MoHW, 2017). Visser and colleagues (2008), emphasized that disclosure of HIV status involves a series of steps of decision making. These include (1) coming to terms with the HIV diagnosis, decision on the appropr iateness of disclosing to a particular person, weighing the anticipated impacts and then choosing the appropriate situation for disclosure (Visser et al . , 2008; Kimberly and Serovish, 1996). Other studies corroborated this study finds why most married moth ers disclosed. Simoni et al . , 1995; Kalichman, 2003 have shown that women who disclose their HIV status is in part to reduce stress and increase access to social support (personal reasons) and to maintain honesty in relationships as well as protecting othe rs from infection (interpersonal level) (Visser et al . , 2008). partner was HIV negative, their CD4 count steadily increased compared to those with HIV infect ed partners (Figure 10 ). partner. Low CD4 count can be indicative of poor medication adherence. Other studies have shown that not disclosing HIV results and lack of partne r support is associated with poor uptake of antiretroviral medication (Farquhar et al.,2001; Kiarie et al., 2003). mothers. in 2011, MoH W recommended e xclusive formula feeding for the first six months of life only for women for whom formula feeding is acceptable, feasible, affordable, sustainable and safe (AFASS) (WHO, 2006; 2007; Ndubuka et al . , 2013). On the other hand, the society in Botswana, regards 104 breast feeding when they should not in order to be accepted in the society as strong mothers. Meanwhile a study conducted in Johannesburg, S outh Africa on HIV pregnant women, showed the opposite that some women will go to uncomfortable measures such as HIV status disclosure to protect their infants and avoid breastfeeding (Varga et al . , 2006). A closer look at the relationships of mothers and their partners, revealed that young mothers (18 - 29) who were single did not have older partners and the only one with older partner aged 45 - 50 was cohabiting. Mothers of middle ages (30 - 39) were more likely to be in unmarried relationships with men of 35 - 5 5 years. For the older mothers, their partners were of the same age (45 - 50) with majority (n=5) of married status (Table 21 ). Looking back to Paper 1, men who were aged 30 - 39, were the ones with high risk of non - adherence and comorbidities. Also these moth ers earlier (Table 1 7 ) showed high non - adherence levels. With these combinations, the risk of further re - transmission is very high making it hard to reach viral us had the most declining trends of CD4 count over time (Figure 10 ) and the same trend was observed with the partner knowi ). Health workers also reported that women were selective of whom they told /disclosed their HIV stat us. Women tend to be open with ART health workers but the opposite was true when they visited non - HIV facilit ies. They did not tell their pr acti ti concern was that most of these women sometime s did not fully understand the importance of early enrolment, as they can only present to the ART facility at the time of delivery but as soon as they are made aware that enrolling in PMTCT early provides an opportuinity with their children being born HIV free, women tend to take better action. Other study confirmed these findings where they reported that PMTCT problems included late presentation for health care delivery service during pregnancy, inadequate ante - natal visits, poor opportunistic infection sc reening, non - disclosure of HIV - positive status to partners, and 105 inadequate family planning counseling (du Plessis et al . , 2014; Bucagu et al . ,2013; Mnyani et al . , 2014). Low HIV testing rates during pregnancy, suboptimal healthcare coverage, and poor patie nt retention were also some of the issues shared by other authors (Prendergast et al . , 2015; Adetokunboh and Oluwasanu, 2016). 3.8.1. Limitations Adherence measure used in this study was not specifically tailored to PMTCT, a general ART measure was used w hich could have maybe provided specific results. W omen were selected based on attendance at the ART study facilities, and thus this sample does not include the population of potentially most at - risk women who do not attend these clinic s . Another limitation is that the study relied on self - report for HIV disclosure status, introducing the risk of social desirability bias. Despite all these, the study showed that there is non - adherence among mothers enrolled on PMTCT and there was evidence of lack of disclosu re among both female and male partners which is a drawback to the therapy because there is a risk of lack of viral suppression. 3. 9 . CONCLUSIONS AND RECOMMENDATIONS This study explored the levels of adherence and HIV status disclosure among pregnant and ne w mothers living with HIV and enrolled on PMTCT in Greater Gaborone. Study has shown that there are high levels of non - adherence among mothers in Greater Gaborone as well lack of status disclosure. Post - natal women had the highest levels of non - adherence c ompared to pregnant women. Out of the 67 women, 22 have not disclosed their HIV status to their partners. Majority of the mothers were not married and were involved with males who were aged around 30 - 45. This is the age cohort where previously we saw that were non - adherent and were sick from other diseases. As potential HIV transmitters, mothers aged 30 - 39 were at the risk of being receivers. The impact of HIV - related stigma and discrimination was also reported as a barrier of disclosure to family members and medical professionals at non - HIV facilities as mothers preferred to keep their status confidential. Impacts of stigma have several implications in PMTCT 106 program when it comes to decisions about HIV testing, accepting the HIV+ test results, adherence to medications and recommended child feeding options (Moshi, 2016). There is a need to develop interventions that respond to the social and cultural contexts of HIV positive mothers to improve disclosure decision making in PMTCT programs. Although the counse lling sessions in the PMTCT program are focused on the pregnant woman, the raising of a child in the African context is family oriented which has significant implications for the PMTCT program (Madiba and Letsoalo, 2013) . It is important that the PMTCT pro gram is geared towards community based interventions that promote awareness and inclusion of male partners, family members and the community leaders about PMTCT adherence . Furthermore, PMTCT interventions in the community should aim at improving the social context of disclosure to reduce stigma and promote HIV disclosure to significant others (Madiba and Letsoalo, 2013). F uture directions will be to continue focusing on studying these middle aged women and their partners, as well as addressing issues of sta tus disclosure, breastfeeding, and unemployment. There is a need to learn about and heat issues that are rampant in Botswana. There is a need for ministri es such as Health, Environment and Labour to work together to empower the women and minimize the non - adherence associated risk factors in Botswana. 107 CHAPTER 4: Impacts of High Ambient Temperatures and Water Rationing on Antiretroviral (AR T ) and Preventi on of Mother - to - Child Transmission (PMTCT) Therapies Adherence in Greater Gaborone, Botswana 4. 0. INTRODUCTION In 2006 the and Hygiene (WASH) were implemented in to HIV programs . The key objective of PEPFAR is to reduce HIV - related morbidity and mortality in affected communities ( United States Agency International Development [USAID], (2011 ). It is important to recognize the significance of safe water, sanitation, and hygiene (WASH) promotion in protecting and caring for people living with HIV and integrate WASH improvement into HIV and AIDS policies and programs. PEPFAR has developed a preventive care package that identifies three key hygiene improvement practices : safe drinking water, washing hands with soap, and safe handling and disposal of feces, and suggests integrating these into all HIV and AIDS programs (USAID, 2011). Many life - threatening opportunistic infections are caused by exposure to unsafe drinking w ater, inadequate sanitation, and poor hygiene. Diarrhea, a very common symptom affects 90% of people living with HIV, especially in HIV - positive children (United Nations International Children's Emergency Fund [UNICEF], 2008). Diarrheal illness can inter fere with the absorption of antiretroviral (ARV) drugs and can contribute to developing HIV strains that are resistant to treatment (USAID, 2011). Good access to safe water and sanitation is important for people living with HIV and for the provision of ho me based care to AIDS patients. Water is necessary for bathing patients and washing soiled linen. Close latrines are necessary for weak patients who cannot walk long distances. Water also plays a significant role in keeping the house environment and the to ilet clean in order to reduce the risk of opportunistic infections such as diarrhea. Water and sanitation provision really plays a significant impact on the dignity of both patients and their caregivers (Haftu et al . , 2017). Safe drinking water is needed for absorption of ARVs and to make food easier to swallow for weak patients. Although there are many water needs for HIV patients, access 108 to WASH may prove difficult for households caring for HIV patients due to declining physical health, stigma and socioe conomic status (USAID, 2007; Haftu et al. , 2017). 4.1. PURPOSE OF THE STUDY To investigate the role of environmental factors (high temperatures and water scarcity and rationing) on ART and PMTCT non - adherence, for adults living with HIV in Greater Gaboron e, Botswana. The study sample was comprised of 658 adults living with HIV enrolled on ART (n=419 women, n=239 men) and PMTCT (n=67) and 84 health workers at 21 government clinics, in urban, peri - urban and rural areas of Greater Gaborone. The study sample i s explained in detail how it was derived elsewhere (Chapter s 2 and 3 ). Even though ART adherence is important in the assistance of elimination of new infections, physical (built environment and weather variables) that are understudied in the ART adherence literature, can be significant barriers to accessibility and utilization of antiretroviral drugs. 4. 2. BACKGROUND Southern Africa has experienced steadily increased trends in mean, maximum and minimum temperatures with the most significant warming occurrin g during the recent decades (Zhou et al . , 2010; Collins, 2011). In comparison to maximum temperatures, the minimum one have rather increased rapidly over inland Southern Africa (New et al . , 2006). Projected temperatures show that mean land surface warming is likely to surpass the global mean land surface temperature increase in all seasons (Sillmann and Roeckner, 2008; Watterson, 2009). It is further projected that towards the end of the 21 st century, warming of between 3.4 and 4.2c above the 1981 - 2000 aver age under the A2 scenario will exceed natural climate variability (Moise and Hudson, 2008). The semi - arid southwestern parts of the region which cover northwestern South Africa, Botswana and Namibia will experience high warming rates (WGIAR5 Annex 1; Shong we et al . , 2009). Figure 1 2 shows the steadily increasing temperature trends from 1991 - 2016 in the whole of Africa. Precipitation is also showing to be fluctuating throughout the period (Figure 1 2 ). 109 Figure 1 2 : Average M o nthly Temperature and Rainfall T r ends in Africa from 1991 - 2016. All over Southern Africa, there is an observed reduction in late austral summer precipitation extending from Namibia through Angola and towards the Congo (Hoerling et al . , 2006; New et al . , 2006). This dryness is associated with an upward trend in tropical Indian ocean sea surface temperatures. Modest declining trends in rainfall are found in Botswana, Zimbabwe and western South Africa. Changes have also been observed in the onset, duration, dry spell frequencies and rainfall intensity (Tadross et al . , 2005, . 2009; Thomaes et al . , 2007; Kniveton et al . , 2009). The increasing frequency of dry spells is also commonly accompanied by an increasing trend in daily rainfall intensity. This has consequences for run - off characteristics (New et al . , 2006). There are also precipitation projections across the Southern African region. Rainfall projections are more uncertain than temperature projections (Rowell, 2012) because they demonstrate higher spatial and seasonal dependence when couple d to temperature projections (Orlowsky and Seneviratine, 2012). The Coupled Model Intercomparrison Project Phase 3 (CMIP3) projections indicate a drying signal in the annual mean in the dry southwestern extending northeast from the desert areas in Namibia and Botswana (Moise and Hudson, 2008; James and Washington, 2013). Drier winters are projected over a large area in 0 5 10 15 20 25 30 0 10 20 30 40 50 60 70 80 Jan Feb Mar Apr May June July Aug Sep Oct Nov Dec Celcius mm 1991 - 2016 Average Monthly Rainfall (mm) Average Monthly Temp (C) 110 the region due to the poleward displacement of mid - latitude storm tracks (Shongwe et al . , 2009; Seth et al . , 2011). There are also projected declines in rainfall during austral spring months implying a delay in the onset of seasonal rains over a large part of the summer rainfall region of Southern Africa (Seth et al . , 2011). Besides volumes, rainfall patterns are also expected to change in int ensity and frequency, resulting in more extreme events and longer periods between rainfalls. River runoff and water availability are projected to decrease by 10 30 .0 % in the dry tropics ( IPCC , AR5 ). 4. 2 .1. Water, Sanitation and Hygiene (WASH) Barriers Comm unity based cross - sectional quantitative study among 413 participants in Arba Minch town, southern Ethiopia in 2016 assessed water, sanitation and hygiene (WASH) practices and associated factors among people living with HIV (Haftu et al . , 2017). A structur ed questionnaire and observational checklist was used to collect data on households socioeconomic, environmental conditions and behavioral characteristics. Logistic regression in SPSS was used to measure association between WASH practices and other associa ted factors. Socioeconomic status was derived through running a principal component analysis (PCA) on household asset data. Findings showed that majority (n=403, 97.6%) indicated that water source was tap water and 165 (40 .0 %) had the drinking water source within their compound (Haftu et al . , 2017). By water status, 76.5% (n=316) households had improved water status (piped water into residence, public tap and protected well/spring) compared to 23.5% (n=97) with unimproved water status. Factors associated wi th WASH practices showed that diarrhea was 10 times likely to affect those with unimproved water status. All other socioeconomic and demographics such as education, sex and age did not have significant association (Haftu et al . , 2017). The authors emphasiz ed that special attention should be focused towards people living with HIV as opportunistic illness such as diarrhea can prove to be dire when compared to the general population (Haftu et al . , 2017). Another Ethiopian study conducted by Yallew and colleagu es (2012) in Gondar city among 296 people living with HIV assessed WASH practices and other factors. Logistic regression and thematic analysis were 111 used for data analysis. About 60 .0 % (n=170) of the HIV patients had a toilet facility like traditional pit a nd flush latrine and majority 96 .0 %(n=163) had latrines that did not have hand washing facilities package (Yallew, Terefe, Herchline et al . , 2012). Only 59 .0 % (n=73) washed their hands with soap after using a latrine. When interviewed, one HIV positive fem ale widow reported that her neighbors did not allow her to use their tap as they believed she transmitted diseases so she was forced to travel long distances for water package ( Yallew et al. , 2012). Another female reported that her landlord prohibited her from using the compound latrine so she had to use a public one a far distance away. In conclusion, almost 50% of the people living with HIV had poor water and sanitation conditions. The authors recommended that WASH should be part of the ART package (Yalle w et al . , 2012). A qualitative study was conducted in the north - central of Zimbabwe in 2016 among 8 focus groups and three key informants in 9 study villages (Mbereko et al. , 2016). The goal was to assess the relationship between HIV and AIDS and water sca rcity (Mbereko et al. , 2016 ). A household (n=20) directly affected by HIV/AIDS was defined as household that 2 months prior to the study has had HIV patient or HIV related death. The indirect affected households (n=20) were the ones where one of their memb ers were involved in assisting a directly affected household. Semi structured interviews were held with households while thematic approach was used with the focus groups. Key informants indicated that at the clinic, 400 liters (20 buckets) of water was nee ded for caring for HIV bedridden patient while HIV/AIDS affected households used between 2 - 12 buckets. All the 40 households reported that having an AIDS patient required high quantities of water for washing linen and clothes, cook, bathe and utensils. Dir ectly affected households indicated that average there was a shortage of 103 liters per day while for indirectly affected households was 65 liters. On access to water issues, households in the upland villages, boreholes were privately owned thus often lock ed and water rationing was frequent. On cultural factors and stigma, communities did not want to mix with HIV patients particularly who were co - infected with TB as they 112 deemed it contagious. In conclusion, the authors recommended that Zimbabwe and African policy makers should treat HIV/AIDS and water as one issue in rural interventions (Mbereko et al., 2016). A study in the fishing villages (n=22) of Ngamiland, Botswana assed the effects of HIV/AIDS related stressors at the household level (n=248). Data was collected and analyzed through observational and thematic analysis of interviews (Ngwenya and Mosepele, 2007). Findings showed that the Botswana National Settlement Policy defined a village as a place with minimum of 500 people (Ngwenya and Mosepele, 2007 ). In Ngamiland, about 50% of the population lived in settlements with less than 500 people therefore could not receive health services directly except through primary centers such as Maun and Gumare. About 53% of the 248 households, reported that they hav e continuously cared for an HIV patient in the last 5 years (Ngwenya and Mosepele, 2007). Of the households with HIV patients, 29 .0 % said the illness had a serious impact on fishing activities. In conclusion, the authors recommended that social protection measures should protect households by reducing the vulnerability to HIV infection. The interventions should prevent the poor from becoming more poor by giving them the capacity to grow (Ngwenya and Mosepele, 2007). 4.2.1.1. Physical Infrastructure Barrier s In a qualitative study conducted at Victoria Hospital in Alice, a town in the Eastern Cape, South Africa, 23 isiXhosa - speaking participants (including both men and women) between the ages of 18 and 60 years were interviewed on adherence to antiretroviral drugs (Kheswa, 2017). The study was guided by the social - ecological framework of Bronfenbrenner (1979), which emphasizes that the health - seeking behavior of people living with HIV/AIDS (PLWHA) is influenced by their social, institutional and physical envi ronments. The author found that food insecurity, financial constraints, poor service from health workers, unfair dismissal, fear of the consequences of disclosure, and rejection by church members were the main factors influencing non adherence to medicatio n (Kheswa, 2017). The study suggested the need for supportive community initiatives, and collaborative approach at both local and national level, and 113 particularly in impoverished communities (Kheswa, 2017). Educating communities and making treatment and co unselling facilities more accessible to PLWHA was also recommended (Kheswa, 2017). In another South African study, using systematic sampling, 735 HIV - positive patients were selected prior to commencing on antiretroviral therapy [ART] from outpatient depart ments from three hospitals and followed - up at six months and interviewed with a questionnaire (Peltzer et al, 2010). Important factors of ART adherence included urban area of residence and adequate physical environment including transport and access to hea lth services (Peltzer et al . , 2010). Living in an urban area was likely to be associated with lower transport costs and fewer disruptions in access to medications (Peltzer et al . , 2010). These findings were supported by other studies that found to be a fac ilitators of adherence [Berg et al . , 2004 New York; Kip et al . , 2009 Botswana; Sanjobo et al . , 2008 Zambia). After adjusting for health - related variables, for both factors, better adherence was associated with low depression and poorer adherence was associ ated with poor environmental factors (Peltzer et al . , 2010). Tuller and colleagues (2010) conducted semi - structured interviews with 41 patients living with HIV in Mbarara, Uganda (Tuller et al . , 2010). The focus of the paper was on transport costs as a bar rier to adherence. All participants reported that they used public transit or personal cars. From the findings, one female with 4 children reported that she sometimes missed to refill her ARV prescription because she did not have transport money to the cli nic. Another patient (male with 5 children) indicated that since he got HIV, he was weak so could not work in the fields thus did not make any money for transport to the ARV clinic. Other participants said that they borrowed transport money from family, fr iends and social clubs while the rest having to sometimes sell their assets in order to afford transport costs. In conclusion, the authors showed that patients had a high level of commitment to ART adherence although the high cost of transport for monthly clinic visit was a serious concern. They recommended that there should be creative interventions to address transportation costs barriers (Tuller et al . , 2010). 114 A retrospective cohort study with 537 adults on ART was conducted in two rural districts of Rwa nda at 21 health centers (Munyaneza et al . village of residence and their ART clinic. The cost surface distance which is the distance between location of village and ART clinic weighed by the earth surface was measured using ArcGIS software. Multivariate logistic regression model was used to assess the relationship between the outcome and distance using 2 - 3 km as a reference. Results showed that median cost surface distance was 5km. cost s urface distance to the clinic was significantly associated with missing an appointment (p=0.022) with patients living 0 - 1km from the clinic having the highest proportion of absence (55.4%). Even though the authors did not assess stigma, they attributed it to those who lived closer by missing because of lack HIV status disclosure. Authors concluded that further studies were needed to better understand why patients living close by missed appointment as these may improve early adherence (Munyaneza et al . , 2018 ). 4.2.1.2. Gender Differences in HIV W omen are more vulnerable to HIV than men because of biological and cultural factors. B iological factors increase a woman's risk of becoming infected. For instance , women can have reproductive tract infections which make them more vulnerable to infection (UNAIDS, 1998). W omen are more susceptible to HIV infection than men male to female transmission of HIV is between two and four times more efficient than female to male (UNFPA, 2002). Socio - cultural factors can increa se the risk of HIV infection among women in other societies that promote male dominance. In most Southern African cultures , young women tend to have older, more experienced partners who are more likely to have sexually transmitted diseases from previous se xual activity. Violence against women, particularly coerced sex increases women's vulnerability to HIV/AIDS (Turmen, 2003) . Migration is another social factor that often result in disruption to social and family relationships and increases the likelihood of HIV infection. It is common in Southern Africa for men to leave their wives and families behind to work in mining towns and military camps. This has the potential to increase the 115 possibility that they will get involved in multiple and concurrent relati onships which put both themselves and their families at risk when they return home. 4.2.1.3. Gender and Water The role of both men and women in the water sector is important. However, there are gender differences and inequalities in the use and management of water resources, which inherently affect women in their efforts to have access to, and manage water resources (Botswana Integrated Water Resources & Water Efficiency Plan, 2013). In households, women are responsible for managing domestic water supplies and these may conflict with productive water needs such as agriculture hence creating tensions for women (Johnson et al . , 1997). With regard to access and control over how water resources should be used and allocated, the responsibility often rests with me n hence ownership of resources by women is negligible (Johnson et al . , 1997). Therefore, women rarely have equal access to water for productive use and are the first to be affected in times of water shortages. Water is interlinked with health impacts and w omen play a central role in family health and hygiene, especially those of their children. Therefore, if there is increased participation of women in water control, there is likelihood increase in the better handling of environmental health issues associat ed with water (Botswana Integrated Water Resources & Water Efficiency Plan, 2013). Asaba and colleagues conducted a cross sectional study in rural Makondo Parish in Uganda investigated the gender differences in water collection. They used a mixed method a pproach covering 602 households across the 15 villages in the parish (Asaba et al. 2013). In - depth interviews were conducted with key actors and focus groups with 5 women and men respectively and participation observation. Quantitative analysis was done us ing chi - square test while qualitative were analyzed using content analysis (Asaba et al., 2013). Results showed that majority of water fetching was women responsibility (54.7%) followed by female children (19.4%). Although women had the fetching responsibi lity, women in household with children tended to be less involved with water collection as they sent the children instead. Stage of relationship 116 also affected water collection as women who just got married or in the early days of cohabiting were less invol ved with water collection. Distance travelled to a water source was cited as a barrier. About 20.0% of the household respondents said their homes were located about 2km or more from water sources and 41.5% said they collected water from unimproved sources such as open wells and ponds. Waiting a long period of time queuing at a water source was also reported as a constraint as 26.7% of the respondents indicated (Asaba et al., 2013). During the long dry season (June - August), queuing was reported to be even lo nger with children spending on average between 2 - 6 hours. Authors explored water collection in the rural area of Uganda and found that time and distance were crucial in understanding what women and children face during water collection (Asaba et al., 2013) . With regards to gender, water and HIV, w omen and children bear a disproportionate amount of the time and resource burden of both water/ sanitation management and caring for people living with HIV in Sub Saharan Africa (West et al . , 2013) . The physical, e motional, and opportunity costs to women are substantial, and while the social arrangements that place women as primary resource managers and organized aro und gender that structures both labor and power (Connell, 1987) . Women spend long hours daily fetching water and travel long distances, sometimes multiple times per day, to water sources (Buor, 2004; Rathgeber, 1996; Upadhyay, 2005) . The physical strain as sociated with such task may have negativ e ollecting water can be particularly burdensome for elderly women or young girls when water supply points are far away, pump handles are heavy, and if they are living with HIV (Wegelin - Schuringa and Kamminga, 2003; West et al . , 2013). Physical strain is further compounded by heat and long queues at water sources (Upadhyay, 2005). A Botswana study by Ngwenya and Kgathi (2006) in five rural villages of Ngamiland assessed acces s to portable water in HIV/AIDS households (n=129). Face to face interviews were used as well as participant observations and secondary data. Qualitative data was reduced to themes. The findings showed that 117 73 .0 % of the households obtained water from commu nal standpipes (Ngwenya and Kgathi, 2006). The average time spent on water collection was 21 minutes in a day and 93 .0 % travelled less than 500 meters. In one village of Shorobe, participants (70 .0 % compared to 52 .0 % in other four villages) attributed that access to water was unreliable as humans and animals competed for the same source. They further indicated that water can be unavailable for up to a week. Caregivers of HIV patients had perceptions that provision of ARVs to HIV patients would reduce water demand as ARVs reduces the number of HIV related bed ridden patients (Ngwenya and Kgathi, 2006). About 80 .0 % of caregivers said they used more water than non - sanitize they did not bathe the patients but wipe the essential body parts and wounds (Ngwenya and Kgathi, 2006). When asked how regularly they washed their hands aft 51% reported that they did so regularly. In conclusion the authors stressed that water should not be strictly treated as an economic good (Ngwenya and Kgathi, 2006). 4. 2 .2. Botswana Botswana is a semi - ari d country that experiences low rainfalls resulting in scarce water resources. T he country continues to experience drought years (progressively i ntensified by climate change) and increased pressure for fresh - water supply due to rapidly increasing urbanizati on and climate change ( Department of Water Affairs - Ministry of Minerals, Energy & Water Resources, 2013). characterized by wide spatial variability, extreme scarcity, and a high dependency on trans - boundary waters (Governme nt of Botswana, 2012). Most of the water is located in the northwest which is remote from the population center in the eastern side of the country. The dependency ratio, reflecting that part of the total renewable water resources originating outside the co untry, is the highest in southern Africa The Limpopo River basin, occupying about 14 .0 % of the country in the east; (2) The Orange - Senqu River 118 basin, (1 2 .0 %) in the south; (3) The Zambezi River basin,(2%) in the north; (4) The Okavango Delta, which is an endorheic basin [stores water and allows no outflow to other external bodies such as rivers] (9 .0 %) in the northwest; (5) The South Interior, which also is an endorheic basin the remaining area (about 63 .0 %) including the Kalahari Desert and the Makgadikgadi Pans (Government of Botswana, 2012). Botswana surface water resources are restricted to ephemeral and perennial rivers and water stored in reservoir s. They are also limited and unevenly distributed over the country. Most sources are in northern Botswana, while most people live in south eastern Botswana in and around Gaborone. The average annual run - off is 1.2mm, ranging from 0 in western and central B otswana to over 50 mm per annum in the north averaging annual run - off of 696Mm 3 . Most of the run - off cannot be captured due to the lack of suitable dam sites, high variability of run - off in time and high evaporation. With regard to reservoirs, there about 100 reservoirs in the country most of them small and primarily used for agriculture. The Ministry of Agriculture has listed 92 dams of small to medium size, mostly for livestock use ( Department of Water Affairs - Ministry of Minerals, Energy & Water Resour ces, 2013) . 3 with an average annual recharge at 1 600 Mm 3 /annum (Department of Surveys and Mapping, 2001 and SMEC et. al., 1991). Groundwater collects in aquifers, and is abstract ed through well fields, mostly for use in villages, mines, power plants, and irrigation; individual boreholes serve the livestock sector as well as small villages. Only a small part of the groundwater resources in Botswana can be economically abstracted du e to high abstraction costs, low yields, poor water quality, and remoteness of aquifers in relation to consumer centers (SMEC et. al., 1991, Masedi et. al., 1999). Recharge is virtually zero in western Botswana, rising to 40 mm in the north. Groundwater de pth is related to rainfall level (and recharge) at around 20 m in northern Botswana to more than 100 m deep in south western Botswana ( Department of Water Affairs - Ministry of Minerals, Energy & Water Resources, 2013) . 119 Water Utilities Corporation (WUC) is responsible for water production in Botswana. There are sixteen management centers throughout the country through which WUC provides portable water. The management centers treat and provide water sourced from both surface and groundwater ( Botswana Environ ment Statistics Water & Climate Digest , 2017 ) . The estimated combined sustainable yield of national population census estimate of 2,024,904. This is less th an the current water demand of around 200 Mm 3 /a or 262L/person/day ( Department of Water Affairs - Ministry of Minerals, Energy & Water Resources, 2013) . In terms of water consumption by different sectors, agriculture is the largest water user (63Mm 3 ), foll owed by domestic use/households (57Mm 3 ), mining (27Mm 3 ) and government (12Mm 3 ). Figur e 13 shows the historical increase in mean monthly temperatures in Botswana (inland southern Africa), where the highest recorded was in t he years 2005 and 2016 (Figure 13 ) . The year 2016 shows the same high maximum monthly temperature in Gaborone, Botswana (Figure 14 ). Figure 13 : Average Monthly Temperature T rends in Botswana from 1991 - 2016. 21.40 21.60 21.80 22.00 22.20 22.40 22.60 22.80 23.00 23.20 1985 1990 1995 2000 2005 2010 2015 2020 Celsius Year 120 Figure 14 : Maximum Monthly Temperature T rends in Gaborone, Botswana from 19 85/86 - 2015/16. 4.2.2.1. Water Rationing Water rationing is limiting water use due to concerns regarding water supply. Reasons for water rationing can be due to drought and maintenance of infrastructure involved in the water distribution and it can be shor t or long term. In the beginning of 2015/16, due to climate conditions, the risk of drought remained high all over the country. Despite the good rains towards the end of the year, the exposure to drought remained high as Botswana is generally a water stres sed country and rains are erratic (Water Utilities Corporations Botswana ([WUC], 2016]). The rains came late, to dams which had hit rock bottom levels Shashe , Dikgatlhong, Ntimbale, Lotsane, Thune, Nnywane and Bokaa dams only six reached levels over 90 .0 %. Insignificant inflows were experienced by Gaborone (supplies the Greater Gaborone area) and Thune (Bobonong areas in the Central District) dams which went u p to 19 .0 % and 39 .0 % respectively. Gaborone and Bokaa (supplies Kgatleng district) Dams failed completely during the 2015/16 due to the worst hydrological drought in decades. The dams replenished in March 2016 resulting in the resuscitation of their operat ions, which brought significant reappearance to the Greater Gaborone area water supply 25.00 25.50 26.00 26.50 27.00 27.50 28.00 28.50 29.00 29.50 30.00 30.50 Celcius Year 121 situation. The slight recovery of the Gaborone Dam brought relief to the Greater Gaborone area which had been under water rationing for six months. Following these rains , water rationing was suspended and water supply restored to the area. However, some water restrictions were continued in some areas in the Greater Gaborone area and surrounding areas. These restrictions prohibited the use of potable water for many common activities which use a lot of water (WUC, 2016). According to the Ministry of Minerals, Energy and Water Resources (MMWER , surpasses supply by 32.9 million liters a day (MMWER, 2015). Even with rationin g, the demand surpasses supply by 18.2 million liters a day (MMWER, 2015). Due to this, the area continues to experience erratic low pressure to no water supply and should expect unannounced interruptions at any given time. Areas which were supplied by the Gaborone Dam were the most affected because of the layout of the water network which is not efficient in bringing water from the northern part of the city to the southern part for re - distribution (MMWER, 2015). 4.2.2.2. Power I nfrastructure and W ater E xtr action in Botswana Power plays a significant role in extraction of water from boreholes and dams countrywide. Botswana currently generates the bulk of its power from coal, and sits on large coal reserves of around 200 billion tons . Botswana Power Corporati on (BPC) is a public company for electrical power generation, transmission and distribution in Botswana. It is currently the only electricity supplier in the country . Only a portion (450 MW) of installed capacity is available to produce power, and addition al demand is met through electricity imports, primarily from South Africa which is also grappling with severe power leads to frequent and untimely power c uts. They can go from hours in the urban areas and days in rural areas. Unlike the Water Utilities Corporation, BPC hardly announces warnings ahead of time on scheduled power cuts. The overall current power access rate is 60 .0 % , which is rural: 54 .0 % and u rban: 65 .0 % . As of 122 2018, there were 211,000 people in Botswana without power and the target is to reach 100% a ccess by 2030 (USAID Power Africa, 2018). 4. 2 .3. Gaps in the Literature In almost all of the literature on climate and HIV medication in the re gion, the environmental barriers were cited as food insecurity (physical environment) and transportation system to access services (built environment). In addition, the studies were also looking at the relationship between HIV/AIDS and water keeping focus on sanitation and hygiene (Obi et al., 2006; UNICEF, 2006; Nkongo & Chonya, 2009). The assumed benefit of improved sanitation and potable water was mainly to reducing diarrhea morbidity (Obi et al., 2006). Diseases associated with fecal contamination of wa progression from being HIV positive to full - blown AIDS due to the rapid weakening of the immune system ( Fewtrell et al. , 2005; Yates et al . , 2015; Nkongo & Chonya, 2009). Most of these studies assumed the cause and effect rela tionship of the interaction of HIV and AIDS and water scarcity, yet the factors and impacts of the complex interaction of multiple stressors have multiple social realities including high ambient temperatures, low antiretroviral adherence, food insecurity, poverty, and other structural barriers (Mbereko et al. , 2016 ). Multiple social realities are the products of the interaction of multiple et al. , 2016 ). 4.3. GOAL OF THE S TUDY The aim of this study is to assess the impacts of environmental factors that include ambient temperatures and water rationing on ART and PMTCT adherence at individual and health care levels for adult men and women, and pregnant women and new mothers. This assessment will be carried out by investigat ing the adherence levels on both adult men and women who are living with HIV and on ART and PMTCT therapies in relation to water rationing and heat. Clinic attendance will also be assessed to understand if it is impacted by environmental factors. Finally, unpack the specific environmental barriers that affect mothers on PMTCT in particular. 123 4. 4 . DATA AND METHODS 4. 4 .1. Study Area Botswana is a landlocked country in southern Africa characterized by a semi - ar id climate. With an area covering 224,607 sq uare mi les , Botswana is sparsely populated with 2.2 million people. The country is mainly flat, dominated by the Kalahari Desert, which covers up to 70% of its land surface (Figure 15 ) . The Okavango Delta, one of the world's largest inland deltas, is in the northwest district (Figure 15 ) . The Makgadikgadi Pan, a large salt pan, lies in the north. The capital city of Gaborone is located in the south east where around 10% of the population reside (Figure 15 ) . It is hot and dry for much of the year but there is a rainy season which runs through the summer months. The summer season begins in November and ends in March. It usually brings very high temperatures. The winter season begins in May and ends in August. This is also the dry season when virtually no rainfall occurs. Winter days are invariably sunny and cool to warm; however, evening and night temperatures can drop below freezing point in some areas, especially in the southwest. Rainfall tends to be erratic, unpre dictable and highly regional. Often a heavy downpour may occur in one area while 10 kilometers away there is no rain at all. The mean annual rainfall varies from a maximum of over 650mm in the extreme northeast area of the Chobe District to a minimum of le ss than 250mm in the extreme southwest part of Kgalagadi District where the Kalahari Desert is located (Botswana Tourism Organization [BTO], 2019). Showers are often followed by strong sunshine which means that a good deal of the rainfall does not penetrat e the ground but is lost to evaporation and transpiration. Summer days are hot as temperatures rise to the 38°C (100.4 F) and now more often than not reaching a blistering 44°C (111.2 F). Winters are clear - skied and dry, warm during the daylight hours but, because there is no cloud cover, cold at night and in the early mornings (BTO, 2019). 124 Figure 15 : Map of Botswana Showing Wet (Okavango Delta) and Dry (National Parks and Kalahari Desert) Areas and Mining Places. Source: (Kelepile and Grady, 2019). 125 4.4.1.1. Greater Gaborone This study took place in the Greater Gaborone which is made of the Gaborone city (as its own district) and three surrounding districts, South East, Kgatleng and Kweneng East Gaborone alone has a total population of 231,592, whi le the larger metropolitan area has about 421,907 people (Botswana Census 2011). During followed by urban villages and lower in rural area hence the se lection of the Greater Gaborone. Gaborone has seen growing trends of HIV prevalence rates because of its strategic location, its proximity to the railway line and Johannesburg, South Africa, its already established administrative offices, its accessibility to most of the major tribes, its non - association with any particular tribe, and most importantly, its closeness to a major water source of Gaborone Dam. Gaborone and South East were in one district until recently (after the recent population census of 201 1) when Gaborone became an independent district on its one (Statistics Botswana, 2015). About 13% of males age 18 months and above were HIV positive, 19.8% were female, which gave the total district HIV prevalence to be 17% (BAIS IV, 2013). About 19% of ma les age 18 months and above were HIV positive, 14.6% were female, which gave the total district HIV prevalence to be 16.6% (BAIS IV, 2013). There were 33,332 people enrolled on ART in Gaborone district as of February 2017. The patients were distributed acr oss the 22 health facilities in the area (MoHW, 2017). South East district has a fewer ARV patients and clinics as compared to Gaborone with 4,659 people on ART and six health facilities (MoHW, 2017). Although the Kweneng East district is close to Gaborone , generally it experienced a modest growth of about three percent per annum between 2001 and 2011 (Statistics Botswana, 2015). A negative growth is usually due to outward migration, which is normally due to lack of employment opportunities in an area (Stat istics Botswana, 2015). About 20% of males age 18 months and above were HIV positive, 22.1% were female, which gave the total district HIV prevalence of 21.5% (BAIS IV, 2013). There were 26,587 people enrolled on ART in Kweneng East district as of February 2017. The patients were distributed across the 30 126 health facilities in the area (MoHW, 2017). About 16% of males age 18 months and above were HIV positive in Kgatleng district, 23.8% were female, which gave the total district HIV prevalence to be 19.9% (B AIS IV, 2013). There were 11,028 people enrolled on ART in Kgatleng district as of February 2017. The patients were distributed across the 26 health facilities in the area (MoHW, 2017). Access to sanitation facilities is measured at household level by asse ssing whether it has access to flush toilet either in - house, communal or shared with a neighbor or had access to a sanitary pit latrine (Statistics Botswana, 2015). Overall, the proportion of households in Gaborone with access to flush toilets was 50%, pit latrine 2.1% . In total, 99.1 per cent of the total population in Gaborone had access to toilet facilities (Statistics Botswana, 2015). With access to portable water in Gaborone, 58.4% of the population had piped water indoors, 31.9% piped outdoors, 7.7% c ommunal tap with a total access of 99.4% (Statistics Botswana, 2015). About 74% of the population in Gaborone used electricity for lighting and heating (Statistics Botswana, 2015). 4.4.1.2. Study clinics (n=21) The (n=21) clinics were purposively selected from the three geographies (urban, peri - urban and rural villages ) in the four districts of Botswana. These districts were selected because of their high HIV tial place. The greater Gaborone area is located on the southern part of Botswana whereas mentioned earlier is dry and receives little rainfall so vegetation is minimal in this side of the country. From the urban area of Gabor one, five clinics were select ed. A mong these urban clinics, three clinics (Broadhurst, Bontleng and Old Naledi) are located in low - income neighborhoods. Old Naledi is mainly a squatter settlement in the outskirts of Gaborone characterized by small businesses and relatively high popula tion density. This neighborhood is also close to Gaborone dam which is the city is major source of water, so minor fishing is evident as well as less dry landscapes. Five peri - urban are located in the peri - urban villages close to Gaborone. Due to the grow ing population of Gaborone, most people commute and reside in these peri 127 villages. Due to high densities of people, clinics in these areas, are usually located around residential tarred roads close to the peri - urban clinics. The eleven rural clinics usually had limited tarred roads and people use dirt roads to reach to these facilities. They are usually not as busy and congested as urban and peri - urban clinics, so tend to be relat ive small in p hysical structure . The images were sourced from Google Earth using a radius of 2km. All the images were updated as of November 2018 (summer in Botswana) except for three in the Kgatleng district (Makakatlela, Mochudi1 and Morwa) which were fr om February 2019 (still the summer season). 4. 5. STUDY DESIGN This study utilize d a cross - sectional survey ( winter , 2016 ) with retrospective responses. It is a mixed method study with quantitative data from survey administered face - to face 6 58 women and men living with HIV attendin g 21 HIV clinics in Greater Gabo rone. Within these clinics, 84 health care workers were also surveyed using qualitative semi - structured questionnaire for their perspectives. The sampling methodology and mechanisms by which to op timize reliability and validity of the sample is described below. 4. 5 . 1 . Primary Data Collection The data for this study was collected during the winter in Botswana . A sample of adults living with HIV, who were enrolled in ART (n= 6 58 ) and PMTCT (n=67) th erapies at government clinics, in urban, peri - urban and rural clinics were interviewed for this study . The data for this study were generated primarily from a survey (quantitative) administered to adult men and women who are living with HIV and enrolled o n ARV and pregnant and mothers enrolled on PMTCT at public clinics (face to face interview), as well as a semi - structured interview (qualitative) of key informants who are health personnel working at the study area. The questionnaires were administered in in Setswana (national) and English (official) languages to the participants. All participants were asked to sign an informed consent before they can partake in the study 128 to ensure confidentiality. The use of Setswana language (local) was useful in this stu dy because patients understood it better than English. For inclusion criteria, only 18 and above participants who are HIV positive and enrolled on ARV therapy at a public clinic took part in the study. The questionnaires were administered by the principal researcher and four research assistants. The principal researcher took a degree were trained by the principal researcher with materials directly obtained from the MSU IRB. All the assistants are fluent in both the languages. The self - administered questionnaire to pregnant women and mothers on PMTCT have 44 variables (with six sections) which range from demographics to the water related concerns in the med ication. The questions on demographics and maternal characteristics (first three sections) that included annual income, partner HIV status, number of children and education are close ended questions. The remaining portions that dealt with knowledge of HIV, treatment seeking behaviors, and water related concerns. These were more open - ended, allowing mothers (new and expectant) to provide detailed information. en questions are open - ended and important because they allow for g aining in - depth information. Biological data such as CD4 count (the blood cells in the body that fights the infections which indicate the health of the immune system), viral load (amount of HIV virus in the body which is used to monitor the efficiency of required to bring their medical file each time they visit their ART facility. Only the latest readings were taken as the patients take new readings at least three tim es a year. From all the clinics, a total of four from each clinic (doctor, two nurses and a pharmacist [ARV dispenser] key informants were selected. No 129 sampling was done on the key informants as the four mentioned are the ones that have daily and direct co ntact with patients attending ARV clinics. 4. 5 . 2 . Sampling Methodology The districts were selected using purposive sampling technique. The research focus was on the adult men and women living with HIV who are also on ART. The selected study area has a hi gh HIV prevalence, high number of HIV positive patien ts as well as high number of ART clinics. The study covered both urban, peri - urban and rural villages . Gaborone (urban), South East (peri - urban), Kgatleng and Kweneng (both rural) are close to the resea HIV prevalence. In total, four districts were selected based on the finding that the prevalence in urban areas is higher (19.2 percent) than rural areas (17.4 percent). R ural areas had consistent prevalence rate over the years, 17.1 percent in 2008 and 17.4 percent in 2013 (BAIS IV, 2013). To calculate the number of clinics needed, first the national prevalence (22%) was used since the study did not consider district as ri gid boundaries. There are 84 ARV health facilities in the study area, 22 Gaborone, 6 in South East, 26 in Kgatleng and 30 in Kweneng East . Using the national prevalence (.22*84 clinics), 18 clinics was recommended. To account for over and under sampling ot her facilities, the numbe r was increased to 21 clinics. S election of adult men and women in the chosen districts (Equation 1.1) was calculated using the formula below: The formula was adopted from a national stigma associated with HIV/AIDS survey conducted in 2014 ( Botswana Stigma Index Survey Report, 2014). N= (t 2 pq)/ (d 2 ) [Equation 1.1] Where; n = minimum sample size required t = z value of confidence level desired (95% confidence level) p = estimated prevalence in target population (the national adult pr evalence rate of 22%). 130 d = precision. The higher the precision desired the larger the sample size will be. 0.05 (5%) is the most common d value. q = 1 - p. Assuming 95% confidence and a precision of 0.05 Formula: Equation 1.1 N = (1.96 2 )*0.22*(1 - 0.22))/ (0.05 2 ) = 263.6. Design effect = 263.6 x 2 = 527.2 = 527 was the minimum sample size required for adult men and women living with HIV and taking AR T and the interview continued until the last patient was interview at the close of the day giving the final sample to be 658. Selection of pregnant women and mothers in the chosen clinics involved purposive sampling, where from each clinic, two women were selected giving a total of 67. In some instances, where there was comparison of mothers and non - mothers, a full s ample of all the women (n= 419) was used. This sample is explained in detail in Chapter 2. Following the selection of clinics and participants, a systematic random sampling was employed to select the ART/PMTCT patients at their respective clinics. Within t he context of this study, at each clinic, every second ART/PMTC patient/mother to enter the door was selected until the minimum required sample each clinic was reached. Participants were visited daily at their respective ART /PMTCT facilities in the study area. T o complement the information that was gathered from patien ts on AR T key informants were chosen and a semi structured interview (30 - 60 minutes interviews) used. From all the clinics, a total of four from each clinic (doctor, two nurses and a pharmac ist [ART/PMTCT dispenser] health care workers were selected (n=84 for the 21 clinics). The responses collected from the key informants during the semi - structured interview were grouped into categories to find out emerging themes. Direct quotes were used wh ere necessary. 131 4.5.2.1. Reliability and Validity of Data A sample of 658 ART patients including 67 mothers on PMTCT is relatively adequate for statistical analysis of factors influencing ARV/PMTCT adherence among adult men and women living with HIV in t he urban, peri - urban and rural areas of Botswana. For reliability purposes, all the patients were visited at their respective clinics at the same time (from 7.30 am until 2 :00 pm). In addition to the principal investigator, four experienced assistants help ed with the data collection. The study was also voluntary and participants signed a consent form before taking part in the survey. The data was coded only by the principal researcher to remove the differences which could be incurred if al l the assistants c oded the data . To avoid or minimize multi - collinearity when using a multivariate model in this case measuring ART/PMTCT adherence, variables were tested one by one. Variables that turn out to be highly significant, further statistical analysis were perform ed. 4.5.2.2. Ethics Overview The survey took place in the urban, peri urban and rural villages of Botswana from May until August 2016. Given the sensitivity of this research, an approval from the IRB was secured from both Michigan State University, and the Ministry of Health in Botswana prior to administering the survey. After satisfying the requirements of the MoHW IRB requirements, to get into the different clinics, separate IRBs were sought from the District Health Management Teams in Botswana. 4. 5.3 . Non - Adherence Definition Adherence in this paper is a dichotomous composite variable defined by three questions . Antiretroviral therapy. When a patien t had CD4 count of 350 copies/ml and more [CD4_REC], taking the pills as instructed by the doctor (100% dosage) [ARVPrescribe] and having missed one a dosage sometimes [MissDose] (95% or at least missed 1 pill a month). If all of these variables were satis fied, one was considered to be adherent (code 0), but if one or more were not met then a patient was considered non 132 adherent (code 1). Modelling in this paper, is non adherent (code1 ) . Non - adherence in this study was therefore defined as either having a C D4 count less than 350 copies, not taking antiretroviral (ARV) drugs as prescribed by the doctor and missing ART dosage sometimes. In summary: non adherenc e means (1) CD4 >= 350 copies/ml AND (2) ARVPrescribe = No AND (3) MissDose = Yes. In this study, si x independent variables were chosen to be included in the models as a way to investigate low ARV adherence levels among adult men and women. Explanatory variables potentially associated with the dependent variable included: Sex (female vs. male). Gender wa s defined as a dichotomous variable with females being at risk of non - adherence hence the code of 1 for females versus 0 for males. If participants had missed an appointment due to heat [AppHeat], this variable was coded as 1, and 0 for not missing the app ointment. The same was done for variable on [AppWater} which asked participants if they have missed an appointment due to lack of water at home. As mentioned earlier, patients are required to attend the clinic every month and this is where they refill thei r prescription. This means that, missing an appointment can be detrimental because it means not collecting a dosage for a month. To affected their healt h [Ration Health], was asked, and if one felt no side effects, the response was coded as a 0. The final variable [ClinicWater], asked patients if they have been turned away from acquiring clinical services because the clinic did not have water. If they ans wered yes, it was coded a 1 and 0 for those who have not been turned away. The mode of transport [TransportMode] was a variable to assess the type of transit patients used to attend their various clinics. Public transport was coded as 1, walking as 2 and p rivate car as 3. The final variable was the type of clinic patients attended. If the clinic was in an urban area, it was coded as 1, peri - urban as 2 and 3 for rural clinic. 4. 5.4 . Descriptive Methods All the continuous variables were assessed to check fo r distribution and normality through histograms - 133 skewed with many outliers, so patients with high numbers (<7000 copies / m l ) were removed. Further th e - which the number of partners that one had over the past year, had a couple of extreme values such as 16 partners in 12 months, so two patients were removed, leaving the n=658 in the analysis. To prepare the variables for logistic re gression, frequencies and cross tabulations [descriptive statistics] were run for all the dichotomous variables. This was helpful because it was a measure to check if all the assumptions of logistic regression were being meant. 4. 5.5 . Analytical Methods L ogistic regression and multilevel modeling were used as analytical methods to assess physical environment barriers to ART adherence. The PROC MIXED technique in SAS 9.4 statistical software was used. PROC MIXED is used to estimate (a) two - level organizatio nal models, (b) two - level growth models, and (c) three - level organizational models. Both random intercept and random intercept and slope models we re estimated. When building the model, the first thing is to always start with an unconditional model (i.e., a model that has no predictors) as this model is used to calculate the infraclass correlation coefficient (ICC) which estimates how much variation in the outcome exists between level - 2 units and gradually adding independent variables while checking for imp rovement in model fit after each model is estimated (Bell et al . , 2013). The full maximum likelihood (ML) was used examining the improvement in model fit the degrees of freedom, Kenward - Roger DF methods was used as it is well - suited as is intended to be used in models with unbalanced designs and can handle complex covariance structures (Bell et al . , 2013). 4. 5.6 . Thematic Analysis (Qualitative) The health - ended questions that enables for rich analysis and the emergence of themes that help in answering the posed research questions. To conduct a thematic analysis, steps recommended by Braun and Clarke (2006) were foll owed. The interest was on understanding levels, and water scarcity and ambient temperatures as barriers of ARV/PMTCT adherence, 134 so the themes revolved around this concept. By con ducting all these steps, there wa s a confidence in results generalization beca use they followed a logical and falsifiable order. 4. 6 . RESULTS T he environmental characteristics in relation to non - adherence among surveyed women and men in Greater Gaborone are presented in Table 23 . Of the 21 women who missed an appointment due to heat , 12 were non adherent and for men (n=14), 11 were non - adherent. The relationship between gender and missing appointment was significant for non - adherence . Most women (57.1 % ) indicated that water rationing affected their health and the same was observed fo r the men (57.7%) even though the association with non - adherence was not significant (X 2 =0.11, p 0.72). Fourteen women compared to 13 men reported that they have been turned away by the clinic because there was no water available (Table 23 ). This relation ship was significant for men but not for women. Table 23 . Environmental Characteristics and ARV Adherence vs. Non - Adherence 1 in a Sample of Women (n=419) and Men (n=239) in Greater Gaborone, Botswana, 2016. Environmental Characteristics Women Non - Adhe rence Men Non - Adherence Women 1 Chi - Square Men 1 Chi - Square No. (%) (%) No. (%) (%) p - value p - value Missed Appointment due to Heat 8.4 (0.003) 6.3 (0.011) Yes 21 5.0 2.9 14 5.9 4.6 No 398 94.9 26.2 225 94.1 41.4 Missed Appointment due to Water S carcity 4.4 (0.035) 1.9 (0.171) Yes 8 1.9 1.2 7 2.9 2.1 No 411 98.1 27.9 232 97.1 43.9 Missed Appointment due to Water Rationing 0.11 (0.729) 0.02 (0.892) Yes 239 57.1 16.2 138 57.7 26.4 No 180 42.9 12.9 101 42.3 19.6 Turned away from Clinic due to Water Scarcity 3.06 (0.80) 5.28 (0.021) Yes 14 3.3 1.7 13 5.4 4.2 No 405 96.7 27.4 226 94.6 41.9 Total 419 100 29.1 239 100 46.1 135 A sample of 658 men and women who were surveyed were asked how the four environmental variables affected th eir adherence to ART. These were compared by their location (Table 2 4 ). Majority of the 5.7 % urban), 6 1.7 % peri and 54.7 % rural were affected by water rationing in their homes (Table 2 4 ) . There were n= 126 patients who lived in urban area s and of those 52 were non - adherent. Of the 1 60 patients who lived in peri - urban areas, n= 6 0 were non - adherent and 1 20 of rural patients (n= 3 72 ) were non - adherent. Association was significant for rural areas by missing appointment due to heat and being tur ned away by the clinic due to lack of water (Table 2 4 ) . Table 2 4 . Envir onmental Characteristics and ART Non - Adherence Stratified by Location in Greater Gaborone, Botswana, 2016. Environmental Factors Urban Peri - Urban Rural Adherence Non - Adherence Adh erence Non - Adherence Adherence Non - Adherence 1 Appointment Missed due to Heat Yes 9 (7.1) 3 (5.7) 3 (3.0) 5 (8.3) 3 (0.81) 15 (12.5) No 68 (91.9) 49 (94.2) 97 (97.0) 55 (91.7) 249 (99.8) 105 (87.5) 2 Appointment Miss due to Water Rationing Yes 1 (0.8) 1 (0.8) 2 (2.0) 6 (10.0) 2 (0.8) 3 (2.5) No 73 (98.6) 51 (98.1) 98 (98.0) 54 (90.0) 250 (99.2) 117 (97.5) 3 Water Ration affects Health Yes 39 (52.7) 29 (55.7) 57 (57.0) 37 (61.7) 150 (59.5) 65 (54.7) No 35 (47.3) 23 (44.2) 43 (43.0) 23 (38.3) 102 (40. 5) 55 (45.8) 4 Clinic Turned You away due to Lack of Water Yes 2(2.7) 0 (0.0) 3 (3.0) 6 (10.0) 5 (1.9) 11 (9.2) No 72 (97.3) 52 (100.0) 97 (97.0) 54 (90.0) 247 (98.0) 109 (90.8) Total 74 (17.4) 52 (22.4) 100 (23.5) 60 (25.9) 252 (59.2) 120 (51.7) 1 Urb an (Chi - square=0.252, p - value=0.616); Peri - urban (Chi - square=2.24, p - value=0.134); Rural (Chi - square=22.6, p - value=<0.0001). 2 Urban (Chi - square=0.064, p - value=0.800); Peri - urban (Chi - square=5.05, p - value=0.025); Rural (Chi - square=1.78, p - value=0.182. 3 Ur ban (Chi - square=0.116, p - value=0.734); Peri - urban (Chi - square=0.337, p - value=0.562); Rural (Chi - square=0.956, p - value=0.328). 4 Urban (Chi - square=1.428, p - value=0.232); Peri - urban (Chi - square=3.46, p - value=0.063); Rural (Chi - square=10.188, p - value=0.001). To understand water rationing of women and men at home, patients grouped by clinic they attended were asked how water rationing affected their health (Figure 16 ). Water rationing affected more patients (n=28) than not (n=5) attending clinic 10 compare d to all other clinics (Figure 16 ). They were followed by 136 clinic 4 with n=26 affected and n=10 not affected patients. For patients attending clinic 13, there was no difference in how water rationing affected their health (n=15 respectively). Figure 16 . The R on w h ether Water Rationing Affected t heir Health at Home. abilities to adhere to their ARV regime, so in this study, part icipants were asked the mode of transport they used to reach their respective clinics controlling for missing appointment due to heat. Patients who used public transport and miss ed an appointment due to heat compared to those who walked and did not miss ha d h igh significant odds ratio of being non - adherence levels [OR 8.48, 2.32 - 31.06] (Table 25 ). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Yes 14 22 22 26 11 17 9 20 25 28 21 14 15 18 19 9 21 15 16 18 17 No 21 10 9 10 25 14 24 12 8 5 9 16 15 11 9 21 9 12 17 12 12 0 5 10 15 20 25 30 Frequency Clinics Yes No 137 Table 25 . Unadjusted Estimated Odds of Non - Adherence if Missed an Appointment due to Heat by Travel Model, Greater Gaborone, Botswana, 2016. 138 Tables 26 and 27 show the predictor variables that were used in the model adherence by patients nested within clinics. Missing appointment due to heat, wat er, being turned away by clinic, controlling for male gender and distance all reduced adherence but did not vary by clinic. Only missing due to heat ( - 0.22 p 0.008) varied by clinic (Table 26 ). Previous table (Table 2 4 ) showed that rural areas had signif icant non - adherence levels due to missing due to heat and being sent away by clinic due to lack of water, further investigation of fixed effects by rural location were explored (Table 27). All the variables in Table 26 reduced adherence but variation by cl inic was evident through missing appointment due to heat ( - 0.41 p 0.004. Residing in the rural without the interaction increased adherence ( 0.10 p 0. 02) (Table 27 ). Table 26 . Multilevel Model Estimating the Effects of Environmental Variables on Adherence among Adults Living with HIV attending Clinics in Gr eater Gaborone, 2016. Covariance Parameters Estimate Intercept Clinic 0.0068 Intercept Patient ID(ClinicID) 0.0276 Residual 0.1819 Fixed Effects Estimate S.E. DF t - value p - value Intercept 0.7388 0.0467 157 15.80 <.0001 Missed Appointm ent due to Heat - 0.2224 0.0845 647 - 2.63 0.0087 Missed Appointment due to Lack of Water - 0.1927 0.1260 650 - 1.53 0.1268 Does Water Rationing Affect your Health 0.4330 0.0375 650 1.15 0.2494 Clinic Turned Away due to Lack of Water - 0.1636 0.0945 637 - 1.7 3 0.0839 Sex (Men=1) - 0.1637 0.0374 641 - 4.38 <0.0001 Distance - 0.0013 0.0011 647 - 1.11 0.2658 Clinics: N= 21, Patients 36 each clinic total N=658, AIC 887.9. 139 Table 27 . Multilevel Model Estimating the Effects of Environmental Variables on Adherence a mong Adults Living with HIV attending Rural Clinics in Greater Gaborone, 2016. Covariance Parameters Estimate Intercept (Clinic) 0.0051 Intercept (Patient nested within Clinics) 0.0322 Residual 0.1751 Fixed Effects Estimate S.E. DF t - va lue p - value Intercept 0.7167 0.04479 108 16.00 <.0001 Missed Appointment due to Heat*Rural - 0.4161 0.1165 650 - 3.57 0.0004 Missed Appointment due to Lack of Water*Rural - 0.2258 0.1217 651 - 1.86 0.0640 Clinic Turned Away due to Lack of Water*Rural - 0.21 17 0.1217 648 - 1.74 0.0824 Sex (Men=1) - 0.1641 0.0371 642 - 4.42 <0.0001 Distance - 0.0018 0.0011 585 - 1.64 0.1025 Rural 0.1092 0.0458 55.5 2.38 0.0207 Clinics: N= 21, Patients 36 each clinic total N=658, AIC 876.8. 4. 6 .1. Descriptive statistics Mothers (n=67) were asked the environmental factors that impacted their adherence. Majority (n=55) indicated that year 2016 was drier and very hot (n=47) than previous years. Three mothers reported that they have experienced water rationing ; sent home without ser vice at clinic due to lack of water (n=1), not preparing baby formula because of lack of water (n=1) and miss appointment due to lack of water (n=1). The mothers (n=5) also expressed that they have experienced loss of children due to HIV related illnesses, malnutrition, traditional causes and undisclosed causes (Table 28 ). Table 28 . Environmental Variables of Concern to Mothers (n=67), Children and Infants, Greater Gaborone, Botswana, 2016. Environmental Variables No. Very Hot Compared to Previous Years 47 Drier Compared to Previous Years 55 Sent Home from Clinic due to No Water 1 Missed Appointment due to No Water 1 No Prepared Baby Formula due to Water Scarcity 1 No. Children Die 5 Causes of Death: HIV - related illnesses (1 - year, 4 ye ars) Malnutrition (2 - years) Traditional Causes (12 - years) Other (5 - years) 140 4. 6 .2. Health Workers Analysis on Water issues, High Temperatures and Non - adherence: Qualitative Results 4.6.2.1. Clinic Operations and Water Issue s In order to complement the findings related to lack of water and non - adherence, the health practitioners were asked using a semi - structured interview their water rationing opinions. There were three questions that were compiled to come up with the releva nt theme of clinic operations and water issues. The first question was whether in their respective clinics, has water been a problem with regards to provision of clinical services. The second one was on whether they had water backup in case of no water sup ply and finally if they ever denied a patient service because of water scarcity. Responses were categorized by provider (doctor, pharmacist and nurse) as well as by the clinic location (urban, peri - urban and rural villages ). Direct quotes were used to emph asize the important points relayed by the health practitioners. Out of the 84 providers, only 17 from 7 different clinics indicated that water was not an issue in service provision. Of all the 7 clinics, 5 were located in the city of Gaborone and 2 in the rural districts of South East and Kgatleng. The providers reported that they did not encounter water issues because they had a water reservoir in their clinic which they relied on rainfall and purchasing of water. They also added that they never turned ba ck a patient without providing service because of lack of water. The rest of the health workers ( n =67) lamented that water ration ing was a hindrance in their ART facilities. Doctors across the board complained about the poor water supply that can be shutdo wn at any point without prior notification from Water Utilities Corporation. Pharmacists decried of unreliable water supply that could sometimes go for days with just slow drops from the taps. While the nurses indicated that they longed for a backup water plan such a Jojo (water reservoir locally called by its brand name) that are found in urban clinics and hospitals. When asked how water rationing interrupted their day to day activities at the clinic, a doctor in a rural clinic said, 141 We tend to see weak p atients when there is no water because facilities such as toilets are unusable, and this forces patients to use unsanitary toilets that are a breeding places for opportunistic infections (Doctor, Rural Clinic ). Other services such as blood screening wer e halted on water rationing days. do not screen for blood which is important for their CD4 count and viral load when there is no water, so it really throws us out of schedule (Nurse, Rural C linic ). Even though the health workers complained about slo w services due to lack of water, all of them denied ever turning back a patient without a service. 4.6.2.2. Clinic Attendance and the Impact of Heat This theme was made up of five questions which were surrounding on how high temperatures affected the serv ices received by patients. They included whether clinic attendance decreased due to heat and the type of temperature controls available in the respective clinics. From what the health workers said, high temperatures posed as a constraint in their health fa cilities. A pharmacist in Gaborone clinic reported that , O ur drugs expire really quickly because of being exposed to high temperatures (Pharmacist, in Gaborone). Meanwhile a doctor in peri - urban clinic of Mopane said on very hot days, blood samples e xpire even before we have the chance to send them to the technical labs (Doctor, Peri - Urban Clinic ). High temperatures also slowed down other departments such as the women section ed to provide a lot of fluid samples (Nurse, Rural C l inic ). Health workers were asked about clinic attendance on hot days as missing clinic appointments can be one of the factors influencing non - adherence. Only 19 of the health providers indicated that h eat did not have 142 any impact on attendance as the patients visited the clinic on scheduled appointments. However, 65 differed as they reported that heat had a significant impact on clinic attendance. Yes, they miss appointments because of the heat particu larly the elderly and women with small babies Mothers also tend to skip too hot because most of these women walk to this clinic ( Nurse, Rural Clinic ). In the same district, a nurse from rural clinic reported that babies seem to be irritated in the summer, as they cry a lot during the appointmen ts with their mothers. This most of the time delays the que ( Nurse, Rural Clinic ). Finally, on how they controlled the temperatures in their respective clinics, fans and air conditioning were common temperature controls. All the urban clinics had air co nditioning but was only limited to the consultation room (nurse/doctor office) not at the patient waiting area. At, a p eri - urban facility, there was a fan where the ARV drugs and blood samples were kept. For the rural areas, whenever there was air conditio ning or fan, it was reserved for drugs and blood sample storage areas. 4. 7 . DISCUSSION These findings confirm that even when patients have a high level of commitment to antiretroviral therapy adherence, environmental factors in particular, the high freque ncy of water rationing and elevated day temperatures - 90 - 90 of eliminating HIV/AIDS by 2030. From the findings, it is evident that environmental factors play a role in ART non - adherence i n Great er Gaborone. From Table 23 , 21 women missed a clinic appointment due to heat on that particular day and of those 12 were non - adherent. For the 14 me n, 11 were non - adherent (Table 23 ). In summary, missing appointment due to heat was a risk factor for both g enders, while missing due to water was associated with non - adherence in women and being turned away by clinic for lack of water was associated 143 with males. Table 24 showed that more patients lived in rural areas (n= 3 72 and 120 of them non - adherent ) than in peri - urban and urban locations. Other factors that were associated with non - adherence in rural areas was missing due to heat and being turned away by clinics due to lack of water. Majority of the patients reported that water rationing negatively affected their health and this was evident that it was a common factor across all locations even though peri - urban residents were mostly affected (6 1 .7%) (Table 2 4 ). This agrees with the fact that condition of the physical infrastructure that supports human settlem ents also influences health risks. This includes power supply, roads, provision of water for drinking and washing, waste management and sanitation (Smith et al . , 2014). Patients in the rural areas are usually the ones who carry the most burden of water rat ioning because sometimes frequency of water cuts is exacerbated by the unreliable power supply to the villages as shown previously that this is common in rural Botswana . The mode of transport that patients use to reach their respective clinics played a sig nificant role in adherence levels of ART. Those who used public transport and missed appointment due to heat were 8 times likely to be non - adheren t compared to those who walked and did not miss an appointment. These findings agree with a South African stud y where important factors of ART adherence included urban area of residence and adequate physical environment including transport and access to health services (Peltzer et al . , 2010). In their study, living in an urban area was likely to be associated with lower transport costs and fewer disruptions in access to medications (Peltzer et al . , 2010). In this study, the patients were coming from almost similar places meaning that their locations were more clustered than dispersed, so tend to have similar geogra phic characteristics. They also attended public clinics that are somewhat regulated the same by district health management teams. This makes it a little difficult to tease out the variations by facilities. However, the multilevel modelling showed that ther e were variations by missing clinic appointments due to heat when controlling for male gender (Tables 26 - 27 ). It showed that patients within clinics that did not miss an appointment due to heat, had a better adherence level. Mothers on the 144 other hand did n ot only risk their health due to lack of water, they were lives in danger if they do not have water to prepare sanitary formula milk. They can be forced to mix feed (breastfeed and bottle feeding) which is not advisable wh en a mother has unsuppressed viral load (Table 28 ). Other studies agree d with this finding of the study that poor adherence which could be due to cultural and other practices were associated with a risk of mixed infant feeding as a result of feeding prefer ences and poor maternal adherence to recommended infant feeding guidelines (Lawani et al . , 2014; Adetokunboh and Oluwasanu, 2016). Mothers from this study also reported that they felt that 2016 was the drier and hotter year compared to other years. This wa s corroborated by the earlier temperature trends of 2016 in Gaborone and across the country (Figures 13 - 14 ). The health workers are the caretakers of the patients thus their opinions mattered as much as the patients. They complained that water rationing w as a hindrance to their day - day running of the health facilities and this impacted the adherence levels of the patients because activities such as blood screening stops. Blood screening is important because it assists in monitoring CD4 and viral load level s. However, all of them contradicted the patients to say that they have never sent back patients without service even no water days. They also mentioned that when there was no water, patients tend to use dirty toilets which are conducive places for opport unistic infections such as diarrhea. These findings are corroborated by those from an Ethiopian study where they found that factors associated with water, sanitation and hygiene (WASH) practices showed that diarrhea was 10 times likely to affect those with unimproved water status (Haftu et al . , 2017). The authors emphasized that special attention should be focused towards people living with HIV as opportunistic illness such as diarrhea can prove to be dire when compared to the general population (Haftu et a l., 2017). With regards to the impact of heat on no n - adherence, the health practitione rs in Greater Gaborone emphasized that heat was a constraint because some patients missed their clinic appointments specifically the elderly and mothers. The clinics also incurred additional expenses and delays because the 145 ARV drugs expired quickly due to the exposure to the sun. As showed previously on the climate change and impacts on human health (IPCC 5 Smith et al . , 2014), the public health capacity and adaptation pla ys a role on indicating the extent to which the three categories of exposure translate to the actual health burden and moderated by factors such as accessibility of public health and the socioeconomic conditions of the people involved as well as the adapta tion measures. With direct exposure to heat, in clinics which did not have adequate resources such as air conditioning, clinic operations tend to be limited because necessities such as drugs expire quickly. Socioeconomic status also increases vulnerability to those with already compromised immune systems such as the elderly, and mothers with young babies who cannot stand long periods of being exposed to the intense heat. 4.7.1. Limitations The survey was done in health facilities, and may have been unable to capture the experiences of the adult women and men who never utilize the ART facility. The findings may as well not be representative of the general population of adult women and men enrolled on ART and PMTCT. The issue of recall bias cannot be ignored in this study. On the questions related to missing an appointment due to heat, it could prove to be that answers on 2016 being very hot could be attributed to the fact that patients were interviewed during the winter season. This could exaggerate their per spectives. Multilevel modelling to show variations by clinics indicated small variations because patients came from clustered locations that were somewhat similar. They also attended public clinics that had almost same resources. Despite all these, the stu dy showed that water rationing and elevated temperatures increased non - adherence in Greater Gaborone. 4. 8 . CONCLUSIONS AND RECOMMENDATIONS the rural dwellers receiving the short end of the stick. Elevated temperatures also increased the non - adherence because both men and women who reported missing the clinic appointments due to heat were 146 non - adherent. It has shown that water rationing affects more wome n compared to males and this can potentially reduce their adherence on ART/PMTCT. This study also highlighted that even though heat burdened with the respon sibility of caring for their babies which means sometimes bringing them to the clinics. The study also highlighted the importance of location where findings showed that patients living in rural villages, were more affected by water rationing as compared to their counterparts. This shows that, geography is important in understanding health because this can explain why rural patients could be defaulting on their clinic appointments. It is adding to the limited literature on environmental factors that affect A RT adherence particularly at low scale (clinics and districts within the country). Weather does not only impact human health through direct exposure, human systems such as physical structures (built environment) do play a significant role. The location (wh ether rural or urban) influences adherence to ART. Future directions call for continuing to study the rural populations as they are likely to they practice unsafe WASH practices which puts them at a higher risk of opportunistic illnesses. The intervention therefore will be for the relevant authorities to warn patients well in advance if there is going to be water cuts. Clinics can have showers at their facilities to provide for basic needs of rural populations. Heat was also a constraining factor at the va rious clinics because health workers indicated that sometimes drugs expired due to exposure to the heat. This decreases the supply of drugs and can in turn reduce the services of the clinics. On sunny days, patients particularly the elderly and mothers wer e likely to miss the clinic appointments because patients waiting areas are not equipped with fans or air conditioning. F or patients such as mothers and elders, services such as adherence counselling that did not require patients to be present at the clini c can be done at home. Health workers can dedicate at least one day a week to visit patients at home. As long as we keep investigating the constraints of ART adherence, we are definitely on the right path to eliminate new HIV infections and achieve the goa l of having a free HIV generation sooner than the target year of 2030. 147 CHAPTER 5: Discussions and C onclusions 5. 0. DISCUSSION This study sought out to assess three objectives which were to investigate (i) barriers to ART adherence and their potential impac t(s) on immunodeficiency (as a reflection of HIV transmission potential) for adult women and men (ii) as well as mothers on PMTCT and (iii) the role of environmental factors (high temperatures and water scarcity and rationing) on ART and PMTCT non - adherenc e in Greater Gaborone, Botswana. The first objective which was tested using three research questions found that there were main contributions that this study made. The first was learning about the HIV life course which was calculated as the time period to when the patient was first diagnosed with HIV and the length before being first initiated on ART and the age of the patient considered. Men as compared to women tend to survive longer without ART going for as long as 22 - 30 years. Results further showed tha t for both the women and men, majority were diagnosed with HIV during their reproductive years. The same men had low adherence rates and CD4 count. All these factors could be due to not used to taking medication for a long time and low CD4 count taking lo w to pick up because the body has been run down for a long time. Whichever the reason is, this study teased out the need to further investigate HIV disease progression by looking at historic narratives. The study further show ed that men have lower adhere nce to ART than women; and men 30+years are the primary transmitters. Non - adherence in men 30 - 40 years was mainly due to having low CD4 count, having other diseases such as Tuberculosis, high blood pressure and anemia. These men were also taking other dru gs in addition to ARVs. Several authors corroborated this study finding that men tend to initiate ART at older ages and with more advanced HIV disease than women (Braitstein et al., 2008; Muula et al., 2007; Stringer et al., 2006) and markers of advanced H IV disease at the time of ART initiation strongly predict early mortality on ART . 148 With increasing age, lower adherence among men 50+ years is largely explained by single marital status and the need to seek other companions for support later in life. By l ocation, p atients in rural areas had low adherence (51.7% non - adherence) rates when compared to their counterparts (22.4% urban; 25.9% peri - urban) . Peltzer and colleagues in their 2011 study found that generally, adherence was better in urban vs. rur al are as (Peltzer et al . , 2011). Among pregnant women and mothers, a majority cohabit with HIV infected men (n=45) and many (n=2 2 ) - adherent and aged 30 - 39 years. These age cohort was cohabiting with males of the same age which we previously saw that they were non - adherent and had other comorbidities such as TB and taking other drugs in addition to ARVs. Re - infection is very high in an environment like that as there is a high risk of in creased viral loads which in turn can lead to ART failure. In their study, Damian et al (2019) found that HIV status disclosure was higher among women who were married/cohabiting (Damian et al . , 2019). Post - partum, many women do not disclose their own HIV processing their own HIV status while trust is evolving in the relationship. Majority of these women were not married, so trust issues among partners can be limited. A st udy conducted in Tanzania corroborated this finding where in their study, found that prevalence of disclosure to a partner started at 22% within 2 months and increased to 40% after nearly 4 years (Antelman et al . , 2001). They further explained that women were less likely to disclose to their partners if they were in cohabiting relationships, had low wage employment, had previously disclosed to a female relative, or reported use of contraceptive method (Antelman et al . , 2001). Finally, high ambient tempera ture is an important barrier to monthly clinic visits for both men and women living in urban, peri - urban and rural areas; and water rationing is a significant risk factor for non - adherence in rural areas. Being turned away by the clinic because of no water availability, is another important barrier for women including, pregnant and post - partum mothers and men living with HIV in Greater Gaborone. 149 There were clinics in the study area particularly in rural areas which did not have air conditioning and fans and on hot days reduced the clinic capacity because ARV drugs and blood samples for CD4 count monitoring expired. Mothers and elderly patients tend to miss the appointments on these hot days due to the heat discomfort anticipated to be experienced at the clin ics. Other factors included the distance and mode of transport used to reach the respective clinics. It was evident that those who used public transport (insinuating that they lived far from clinics) had lower adherence rates compared to those who walked ( lived close by). 5.1. CONCLUSION This study showed that non - adherence was high among men living with HIV in Greater Gaborone particularly the older men (30 - 50+) with the middle aged men as the highest risk because they had other comorbidities and low CD4 count. The study showed that there were people who survived long periods of time without ART and most were men and as the major transmitters of HIV infection, these men require additional attention in terms of ensuring ARV adherence, to reduce HIV infe ction. There were also mothers who were still not disclosing their HIV status to their partners and also not knowing their - adherence in Greater Gaborone was also exacerbated by the climate direct exposures such as heat and indirec t human systems such as water rationing both at home and at the clinic. Earlier it was mentioned that Botswana have met its 90 - 90 - 90 programmatic goals already (UNAIDS, 2019); however , significant number of new infections and AIDS - related deaths persists ( n=8,500 incidence rate, 4.36 new HIV infections ) . The strength of this study is that it utilized the large population estimates from secondary sources to understand the global scale. I t then went further to follow the people on the ground by zooming into t he 658 adults on ART and PMTCT in Greater Gaborone to further investigate the potential causes of increased new infections. Without ART adherence, an HIV free generation is nothing but a dream. 150 5.2. RECOMMENDATIONS With regards to these findings the 90 - 9 0 - 90 Treat All Program can achieve its maximum potential through the following recommendations: Initiate and implement more HIV and ART programs that are geared towards men in order to increase awareness and openness on health issues Regular teachings on C D4 count and viral load at counselling sessions within the ART facilities Promotion of PMTCT education in a general population, including sexual partners and not only limited to pregnant women at antenatal setting to increase the understanding of the purpo se and benefits of PMTCT program Mothers counselling is needed to address issues of HIV status disclosure and this should not only be the burden of health workers. Communities and social networks where these mothers spend more time such as churches and co mmunity meetings should be involved. As climate change is something that cannot be eliminated, mitigation and adaptation measures and sending warning water cut messages well in time for people to prepare. ART health facilities can also invest in providing showers facilities for patients particularly mothers. 5.3. FUTURE RESEARCH There is a need to study the people who survived long periods of time without ART to better understand HIV disease progression. The time is now because in the near future, this cohort of people will be gone a lower CD4 count to be enrolled on ART). 151 When individual, social structural and environmental barriers are minimized, it is a step in the right e nsure high ART adherence, achieve viral suppression and eliminate new HIV infections by 2030. There is also a need to further research on UNAIDS indicators such as the Incidence Mortality Ratio (IMR) which identifies a point at which HIV related health car e costs diminish, with an IMR < 1.0 (in a scenario of low mortality) indicating a decline in the HIV population, thereby, the ability to reduce the costs of ART services. It is common recently fo r donors to pull HIV funds from affected countries simply b ecause there are reduced number of HIV cases with time. More funds are needed to educate people on prevention and importance of ART so as to not see people die off but to eliminate HIV by not getting it at all. 152 APPENDICES 153 A PPENDIX A : Adults on ART Survey Questionnaire for the Patients My name is Matlhogonolo Bene. My home is Botswana. I am a PhD student in the Geography Department eatment Seeking Behaviors for Antiretroviral (ARV) and Prevention of Mother - To - Child Transmission (PMTCT) in treatments for men and women and mothers with HIV in the urban, peri - villages and rural areas of Botswana. The findings from this study will be used to help people in the future to receive ARV and PMTCT therapy. Your participation in this study will be voluntary. I would really appreciate it if you can tak e your time to answer the following questions. Thank you! Date: ID: Checkbox [ ] A. Demographics 1. Sex/Gender: Female [ ] Male [ ] 3. How long have you lived in this community? 4. Age: 18 - 24 [ ] 25 - 30[ ] 31 - 39 [ ] 40 - 49 [ ] 50 - 59 [ ] 60 and above [ ] 5. Marital Status Single: [ ] Married [ ] Living together/cohabiting [ ] Widowed [ ] 7. Occupation: Gov ernment employee [ ] Private sector [ ] Farmer [ ] Self - employed [ ] Unemployed [ ] 9. What is the highest education you have completed? Neve r attended school [ ] Primary school [ ] Secondary [ ] Technical/Vocational [ ] University [ ] B. Health Status 12. How often do you feel sick or unhealthy? 13. Have you been sick or unhealthy in the last two months? 14. How many times have you felt unhealthy this year? C. HIV/AIDS Knowledge 15. Do you use a condom each time with a sexual partner? Yes [ ] No [ ] 16. Where do you get condoms? 17. How many sexual partners have you had over the last 12 months? 18. Have you ever had sexual intercourse while drunk (intoxicated)? 154 19. Do you think a person can get HIV/AIDS through witchcraft? Yes [ ] No [ ] 20. Do you think a person can get HIV/AIDS through a mosquito bite? Y es [ ] No [ ] D. Treatment Seeking Behaviors 21. Where do you go for help or treatment whenever you are feeling sick? 23. Where do you get ARVs from? Government hospital [ ] Private Doctor [ ] Government clinic [ ] other 24. Are you taking any other drugs apart from ARVs? Yes [ ] No [ ] E. Access to ARV Therapy 25. Is this closest ARV facility to you? Yes [ ] No [ ] If not, why not go to other closer facility? 26. How far is the ARV facility from your house? 27. How do you get to the ARV facility? 28. How often are you required to go for the treatment? 29. How long do have to wait before you can be a ttended to at the ARV facility? 30. Which days and times do you generally go to the clinic? F. Family/Household/Community 31. Does anyone else in your family/household have HIV? 32. Is anyone else in your family/household on ARV treatment? 33. Should chil dren be taught about condoms, HIV/AIDS and ARVs? G. Adherence to the Medication 34. Do you take the ARV medication as prescribed by the doctor? Yes [ ] No [ ] 35. Do you miss doses sometimes? Yes [ ] No [ ] 36. Do you feel better after taking ARVs? Yes [ ] No [ ] H. Water Related Concerns 37. Yes: [ ] No: [ ] What was the temperature on that day? 38. 39. Has the clinic ever sent you back home without offering a service because they did not have water? Yes: [ ] No [ ] 40. How many days of water rationing have you experienced in a week at home? 41. Does water rationing affect your health? 42. Have you ever missed your pill dosage at home because there was no water? 43. How hot was it this year comp ared to other years? 44. Since January, how many times it rained this year? 155 45. 46. Did you feel it was drier this year than other years? I. Anything else you would like to tell me? Thank you for your t ime! CD4 Count ARV Initiation Date Height Weight 156 APPENDIX B : Questionnaire for Mothers My name is Matlhogonolo Bene. My home is Botswana. I am a PhD student in the Geography Department at Michigan Stat Behaviors for Antiretroviral (ARV) and Prevention of Mother - To - Child Transmission (PMTCT) in to ARV and PMTCT treatments for men and women and mothers with HIV in the urban, peri - villages and rural areas of Botswana. The findings from this study will be used to help people in the future to receive ARV [and PMTCT] therapy. Your participation in t his study will be voluntary. I would really appreciate it if you can take your time to answer the following questions. Thank you! Date: ID: Checkbox [ ] A. Demographics 2. How long have you live d in this community? Months [ ] Years [ ] 3. Age 18 - 24 [ ] 25 - 30 [ ] 31 - 39 [ ] 40 - 49 [ ] 4. Marital Status Single [ ] Married [ ] Living together/cohabiting [ ] Widowed [ ] 5. Religion Christianity [ ] Islam [ ] African Tradition 6. Occupation Government employee [ ] Private sector [ ] Farmer [ ] Self - employed [ ] Unemployed [ ] 7. Annual income: How much do you make in a year? If you are not emplo 9. What is the highest education you have completed? Never attended school [ ] Primary school [ ] Secondary [ ] Technical/Vocational [ ] University [ ] B. Maternal Characteristics (Mother) 10. Y ear of HIV diagnosis 11. Do you have any other illness? 12. Is your partner HIV positive? 13. How old is your partner? 14. Does he know your HIV status? 15. Who is the decision maker in your relation for using contraception? What are the reasons for usi ng contraception? 16. Do you know about female condoms? C. Maternal Characteristics (Child) 15. Do you have any children? If no, do you plan on having a family in the future? 157 16. Do you know about Prevention of Mother To - Child Transmission (PMTCT) therapy? If yes, As part of the antenatal care visit for last birth, were you offered HIV test between time for delivery and before/after baby was born? If yes, how many do you have? If yes, how old are the children? 17. Where were the child [ren] born? Government Hospital/Clinic [ ] Private Hospital/Clinic [ ] Home [ ] other? 18. Are any of your child [ren] HIV positive? D. Treatment Seeking Behaviors 19. Are you enrolled in PMTCT? 20. How long have you been enrolled? 21. Did you know about PMTCT bef ore child birth/pregnancy? 22. Are you taking ARV drugs in addition to the PMTCT therapy? 23. Where do you go for help if your child is sick/not feeling well? 24. Where do you go for help if you are sick/not feeling well? 25. How long does it take you to g et to this facility in Q 23and 24? 26. How far is any health facility from where you stay? 27. Did any of the children die? If yes, how old was the child when s/he died? E. HIV/AIDS Knowledge 28. W hen was the last time you had sexual intercourse while drunk (alcohol intoxicated)? 29. Do you think a healthy looking person can have HIV? 30. Tell me ways that you know of which HIV can be transmitted from mother to her unborn child. F. Water Related Co ncerns 31. Yes: [ ] No: [ ] What was the temperature on that day? 32. Yes: [ ] No [ ] Explain 33. 34. 35. Has the clinic ever sent you or your baby back home without offering a service because of no water? 36. Have you ever missed your pill dosage at home because there was no water? 37. How many days of water rationing have you experienced in a week at home? 38. Does water rationing affect your health? 39. Have you or the baby ever gotten an infection because of water scarcity? 40. 41. How hot was it this year compared to other years? 158 42. Since January, how many times it rained this year? 43. Have yo 44. Do you feel it was drier than other years? G. Do you have anything you would like to share with me? Thank you for time! 159 APPENDIX C : Survey Semi - Structured Interv iew for Key Informants (Doctors, Nurses and Pharmacists) My name is Matlhogonolo Bene. My home is Botswana. I am a PhD student in the Geography Department Seeking Behaviors for Antiretroviral (ARV) and Prevention of Mother - To - Child Transmission (PMTCT) in treatments for men and women and mothers with HIV in urban, peri - villages and rural areas of Botswana. The findings from this study will be used to help people in the future to receive ARV and PMTCT therapy. Your participation in this study will be voluntary. I would really appreciate it if you can take your time to answer the following questions. Thank you! Date: ID: A. Demographics Facility name: Provider Gender: Nationality: Highest education attainment: How long have you worked at this facility? B. Accommodation and Affordability 1. What are the operating h ours of this facility? 2. With regards to staffing resources in your facility, can you say that they are adequate? And if not, why? 3. Do patients pay for the services? 4. Are the needs of the patients met? How do you assess that? C. Distance and Access 5. Which places do your patients usually come from? 6. Are there any steps to helping those patients coming from far places? Explain. D. Socioeconomic Status of Patients 7. How will you describe those who are mostly enrolled in the therapy? E. Education on HIV & ARV 8. Do you counsel patients? If yes, explain: 9. Do you provide patients with condoms? If yes, explain: 160 10. Do you educate the patients about alcohol and HIV? 11. How do you educate both the patients and community about HIV and the ARV therapy? 12. Are there instru ctions given to the patients on how to take the treatment? If yes, in which language (s)? 13. Are the ARV patients required to bring others for consultation and why? If yes, explain: 14. How do you ensure adherence to ARV therapy among patients? 15. How do y ou deal with issues of loss to - follow up? F. Education and Adherence on PMTCT 16. Which criteria do you use to enroll pregnant women/mothers on PMTCT? 17. How do you deal with issues of breastfeeding and PMTCT? 18. Are the women on PMTCT required to bring thei r spouses during consultation? 19. What measures do you use for women who do not comply? G. Test Results given to the patient? 21. How long is th e patient given before they are told to come back? 22. H. Water Related Concerns 23. Has water been a problem this managing the clinic? Explain: 24. What do you do when there is a lack of water? 25. Have you ever turned a patient away because of lack of water? 26. Has the temperature affected the operations of the clinic? 27. Do you see fewer patients come in on a hot day? 28. 161 29. How do you control tempe rature in the clinic? 30. Where do you get water? 31. Do you think it was drier this year? I. Do you have any other questions for me? Thank you for your time! 162 REFERENCES 163 REFERENCES Allison, P. D. (2012). Logistic regressio n using SAS: Theory and application. SAS Institute. Lemeshow, S., & Hosmer Jr, D. W. (1982). A review of goodness of fit statistics for use in the development of logistic regression models. American Journal of Epidemiology , 115(1), 92 - 106. Antelman, G., S mith Fawzi, M. C., Kaaya, S., Mbwambo, J., Msamanga, G. I., Hunter, D. J., & Fawzi, W. W. (2001). Predictors of HIV - 1 serostatus disclosure: a prospective study among HIV - infected pregnant women in Dar es Salaam, Tanzania. AIDS (London, England), 15(14), 1 865 1874. Amico, K.R., Simoni, J., Wilson, I., Malow, R.M., Devieux, J., Deschamps, A. (2009). Monitoring Patient Adherence: Recommendations from and for Practice. Ankomah, A. (1992). Premarital sexual relationships in Ghana in the era of AIDS. Health Po licy and Planning, 7,135 143. Acosta - Michlik, L., U. Kelkar, and U. Sharma, 2008: A critical overview: local evidence on vulnerabilities and adaptations to global environmental change in developing countries . Global Environmental Change , 18(4), 539 - 542. A detokunboh, O. O., & Oluwasanu, M. (2016). Eliminating mother - to - child transmission of the human immunodeficiency virus in sub - Saharan Africa: The journey so far and what remains to be done. Journal of Infection and Public Health , 9(4), 396 - 407. Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179 - 2 11. Ajzen, I. (2001). Attitudes. Annual Review of Psychology, 52,27 - 58. ol, and the theory of planned behavior. Journal of Applied Social Psychology, 32(4), 665 - 683. Anderson, K., Mitchell, J. (2000). Differential access in the receipt of antiretroviral drugs for the treatment of AIDS and its implications for survival. Archiv es of Internal Medicine , 160 (20), 3114 - 3120 . Andersson, N., Cockroft, A. & Shea, B. (2008). Gender - based violence and HIV: Relevance for HIV prevention in hyper - endemic countries of Southern Africa. AIDS, 22 (suppl. 4), S73 - S86. Barry, F., & Grady, S.C . (2018) Africana womanism as an extension of feminism in political ecology (of health) research. Geoforum, doi.org/10.1016/j - geoforum.2018.09.024. Birbeck, G., Kvalsund, M., Byers, P., Bradbury, R., Mang'ombe, C., Organek, N., . . . Malama, C. (20 11). Neuropsychiatric and socioeconomic status impact antiretroviral adherence and mortality in rural zambia. American Journal of Tropical Medicine and Hygiene, 85(4),782 - 789. 164 Bezabhe, W., Chalmers, L., Bereznicki, L., Peterson, G., Bimirew, M., & Kassie, D. (2014). Barriers and facilitators of adherence to antiretroviral drug therapy and retention in care among adult HIV - positive patients: A qualitative study from ethiopia. Plos One, 9(5), e97353. Berg, K. M., Demas, P. A., Howard, A. A., Schoenbaum, E. E., Gourevitch, M. N., & Arnsten, J. H. (2004). Gender differences in factors associated with adherence to antiretroviral therapy. Journal of general internal medicine, 19(11), 1111 - 1117. Braun, V., & Clarke, V. (2006). Using thematic analysis in psychol ogy. Qualitative Research in Psychology , 3(2), 77 - 101. Braitstein, P., Boulle, A., Nash, D., Brinkhof, M. W., Dabis, F., Laurent, C., ... & Hosseinipour, M. (2008). Gender and the use of antiretroviral treatment in resource - constrained settings: findings from a multicenter collaboration. Journal of women's health , 17 (1), 47 - 55. Bachanas P, Medley A, Pals S, et al. Disclosure, knowledge of partner status, and condom use among HIV - positive patients attending clinical care in Tanzania, Kenya, and Namibia. AI DS Patient Care STDS. 2013;27(7):425 435. Bucagu, M., de Dieu Bizimana, J., Muganda, J., & Humblet, C. P. (2013). Socio - economic, clinical and biological risk factors for mother - to child transmission of HIV - 1 in Muhima health centre (Rwanda): a prospectiv e cohort study. Archives of Public Health , 71(1), 4. Bagnol, B., Chamo, E. (2004). Intergenerational relationships in Mozambique: what is driving young women and older men? Sex Health Exchange, 3:10 11. Barros, A. J., & Hirakata, V. N. (2003). Alternativ es for logistic regression in cross - sectional studies: an empirical comparison of models that directly estimate the prevalence ratio. BMC Medical Research Methodology, 3(1), 21. Bisson, G. P., Rowh, A., Weinstein, R., Gaolathe, T., Frank, I., & Gross, R. (2008). Antiretroviral failure despite high levels of adherence: discordant adherence response relationship in Botswana. Journal of Acquired Immune Deficiency Syndromes , 49 (1), 107. Boonstra, E., Lindbaek, M., Khulumani, P., Ngome, E., & Fugelli, P. (2002 ). Adherence to treatment guidelines in primary health care facilities in Botswana. Tropical Medicine & International Health , 7 (2), 178 - 186. Buseh, A., Glass, L., McElmurry, B. (2002). Cultural and gender issues related to HIV/AIDS prevention in rural Swa ziland: a focus group analysis. Health Care Women International, 23,173 184. Charurat, M., Oyegunle, M., Benjamin, R., Habib, A., Eze, E., Ele, P., . . . Blattner, W. (2010). Patient retention and adherence to antiretrovirals in a large antiretroviral the rapy program in nigeria: A longitudinal analysis for risk factors. Plos One, 5(5), e10584 . Centers for Disease Control and Prevention (2019). Stages of HIV infection. Retrieved from https://wwwn.cdc.gov/hivrisk/what_is/stages_hiv_infection.html on July 21, 2019. 165 Chesney, M.A. (2000). Factors affecting adherence to antiretroviral therapy Clinical Infectious Diseases , Cole, J. (2004). Fresh contact in Tamatave, Madagascar: sex, money, and intergenerational transformation. An Ethnologist , 31,573 588. titative review of 50 years of research Medical Care Dunkle, K.L., Jewkes, R., Brown, H., Gray, G., McIntyre, J., Harlow, S. (2004). Transactional sex among women in Soweto, South Africa: Prevalence, risk factors and associations with HIV i nfection. Social Science & Medicine, 59,1581 1592. du Plessis, E., Shaw, S. Y., Gichuhi, M., Gelmon, L., Estambale, B. B., Lester, R., ... & Avery, L. S. (2014). Prevention of mother - to - child transmission of HIV in Kenya: challenges to implementation. BMC Health Services Research , 14(1), S10. Ebuy, H., Yebyo, H., & Alemayehu, M. (2015). Level of adherence and predictors of adherence to the Option B+ PMTCT programme in Tigray, northern Ethiopia. International Journal of Infectious Diseases , 33, 123 - 129. E llece, S. E. (2016). Gender and HIV/AIDS in Botswana: Thirty years later. Pula: Botswana Journal of African Studies, 30(2),268 - 282. Fishbein , M., Ajzen , I. Predicting and Changing Behavior: The Reasoned Action Approach. Psychology Press (Taylor & Francis) ; New York, NY, USA: 2010. Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behavior: An introduction to theory and research. Reading MA: Addison - Wesley. Feucht, U. D., Meyer, A., & Kruger, M. (2014). Missing HIV prevention opportunitie s in South African children a 7 - year review. BMC Public Health , 14(1), 1265. Farahani, M., Price, N., El - Halabi, S., Mlaudzi, N., Keapoletswe, K., Lebelonyane, R., & Gabaake, K. (2016). Trends and determinants of survival for over 200 000 patients on ant iretroviral treatment in the Botswana National Program: 2002 2013. AIDS (London, England) , 30 (3), 477. Farquhar, C., Kiarie, J. N., Richardson, B. A., Kabura, M. N., John, F. N., Nduati, R. W., ... & John - Stewart, G. C. (2004). Antenatal couple counseling increases uptake of interventions to prevent HIV - 1 transmission. Journal of acquired immune deficiency syndromes (1999), 37(5), 1620. Farquhar, C., Mbori - Ngacha, D. A., Bosire, R. K., Nduati, R. W., Kreiss, J. K., & John, G. C. (2001). Partner notificati on by HIV - 1 seropositive pregnant women: association with infant feeding decisions. AIDS (London, England), 15(6), 815 - 817. Ferradini, L., Jeannin, A., Pinoges, L., Izopet, J., Odhiambo, D., Mankhambo, L., ... & Carrieri, M. P. (2006). Scaling up of highl y active antiretroviral therapy in a rural district of Malawi: an effectiveness assessment. The Lancet, 367(9519), 1335 - 1342. . 166 Fung, (2015) moderating effects of project management experience, project team size, project duration and project value size on t effectiveness in Malaysia. Gaolathe, T., Wirth, K. E., ScD, Holme, M. P., Makhema, J., , Moyo, S., Chakalisa, U., , . . . Botswana Combination Prevention Project study team. (2016) . Botswana's progress toward achieving the 2020 UNAIDS 90 - 90 - 90 antiretroviral therapy and virological suppression goals: A population - based survey. Lancet HIV, 3(5), e221 - e230 . Genn, L., Chapman, J., Okatch, H., Abell, N., Marukutira, T., Tshume, O., . . . Lowenthal, E.D. (2018). Pharmacy refill data are poor predictors of virologic treatment outcomes in adolescents with HIV in Botswana . AIDS and Behavior, doi.org/10.1007/s10461 - 018 - 2325 - 9. Gill, C.J. et al. (2005). No room for complacency about adherenc Africa. AIDS Gillespie, S., Greener, R. (2006). Is poverty or wealth driving HIV transmission? Working paper for UNAIDS Technical Consultation on Prevention of Sexual Transmission of HIV. Geneva: UNAIDS; 19 September 2006. Gisslén, M., Svedhem, V., Lindborg, L., Flamholc, L., Norrgren, H., Wendahl, S., & Sönnerborg, A. (2017). Sweden, the first country to achieve the Joint United Nations Programme on HIV/AIDS (UNAIDS)/World Health Organization (WH HIV Medicine , 18(4), 305 - 307. Glynn JR, Carael M, Auvert B, Kahindo M, Chege J, Musonda R, et al. (2001). Why do young women have a much higher prevalence of HIV than young men? A study in Kisumu, Kenya, and Ndol a, Zambia. AIDS, 15 (Suppl. 4),51 60. Gamell, A., Letang, E., Jullu, B., Mwaigomole, G., Nyamtema, A., Hatz, C., ... & Tanner, M. (2013). Uptake of guidelines on prevention of mother - to - child transmission of HIV in rural Tanzania: time for change. Swiss M edical Weekly: Official Journal of the Swiss Society of Infectious Diseases, the Swiss Society of Internal Medicine, the Swiss Society of Pneumology , 143. Government of Botswana, (2011 ) Kweneng District Council PMTCT retrieved from http://www.gov.bw/en/Mi nistries -- Authorities/Local - Authorities/Kweneng - District - Council/Tools - and - Services/Services/PMTCT/ February 2019 Government of Botswana (2012). Botswana national water policy. Ministry Minerals, Energy and Water Resources, Gaborone, Botswana. Grady, S.C ., Frake, A.N., Zhang, Q., Bene, M., Jordan, D.R., Vertalka, J., Fan, Y. et al. (2017). Neonatal mortality in East Africa and West Africa: a geographic analysis of district - level demographic and health survey data. Geospatial Health , 12(1). Hallman, K. (2 004). Socioeconomic disadvantage and unsafe sexual behaviors among young women and men in South Africa. Working paper no. 190. Washington DC: Policy Research Division, Population Council. 167 Haftu, D., Gebremichael, G., Ajema, D., Gedamu, G. & Agedew, E. (20 17) Water, sanitation and hygiene practice associated factors among HIV infected people in Arba Minch town, Southern Ethiopia. Journal of Water and Health 15(4),615 - 625. Haynes, (2005). Interventions for enhancing medication adherence. The Cochrane Databas e of Systematic Reviews , Issue 4. pub2. Hippocrates. (1978). Airs, waters and places. An essay on the influence of climate, water supply and situation on health. In: Lloyd GE, editor. Hippocratic Writings . London, England: Penguin; pp. 148 69. Hunter, M. African Studies , 61,99 120. Jewkes, R.K., Levin J. B. & Penn - Kekana, L. (2003). Gender inequalities, intimate partner violence and HIV preventive practices: Findings of a South Afri can cross - sectional study. Social Science and Medicine, 56(1), 125 - 134 . Jones, L. (2006). Sexual decision - making by urban youth in AIDS afflicted Swaziland. African Journal of AIDS Research, 5,145 157. Kambou, S., Shah, M., Nkhama, G. (1998). For a penci l: sex and adolescence in peri - urban Lusaka. In: Guijt I, Shah M, et al., editors. The myth of community: gender issues in participatory development. London: Immediate Technology Publications , 110 120. Karlyn, A. (2005). Intimacy revealed: sexual experime ntation and the construction of risk among young people in Mozambique. Culture, Health Sexuality, 7,279 292. urban South Africa. Culture, Health Sexua lity, 6,377 391. Kelly, R., Gray, R., Sewankambo, N., Serwadda, D., Wabwire - Mangen, F., Lutalo, T., Wawer, M. (2008). Age differences in sexual partners and risk of HIV - 1 infection in rural Uganda. Journal of Acquired Immunodeficiency Syndrome 2003 ; 32:4 41 51. S24 AIDS , 22 (suppl 4). Kharsany, A. B., & Karim, Q. A. (2016). HIV infection and AIDS in Sub - Saharan Africa: current status, challenges and opportunities. The Open AIDS Journal , 10, 34. Kinuthia, J., Singa, B., McGrath, C. J., Odeny, B., Langat, A., Katana, A., ... & John - Stewart, G. (2018). Prevalence and correlates of non - disclosure of maternal HIV status to male partners: a national survey in Kenya. BMC public health, 18(1), 671. Kiula, E. S., Damian, D. J., & Msuya, S. E. (2013). Predictors of HIV serostatus disclosure to partners among HIV - positive pregnant women in Morogoro, Tanzania. BMC public health, 13(1), 433. Kiarie, J. N., Kreiss, J. K., Richardson, B. A., & John - Stewart, G. C. (2003). Compliance with antiretroviral regimens to prev ent perinatal HIV - 1 transmission in Kenya. AIDS (London, England), 17(1), 65. 168 Kip, E., Ehlers, V. J., & van der Wal, Dirk M. (2009). Patients' adherence to anti - retroviral therapy in botswana. Journal of Nursing Scholarship: An Official Publication of Sig ma Theta Tau International Honor Society of Nursing, 41(2), 149 - 15. Krawczyk , C . S . , Funkhouser , E . , Kilby , J . M . , Vermund , S . H . (2006). Delayed access to HIV diagnosis and care: Special concerns for the Southern United States. AIDS Care , 18 (Suppl 1): S35 - S44 . Kristensen , S . , Sinkala , M . , Vermund , S. Transmission of HIV. In: Essex M, Mboup S, Kanki P, Marlink R, Tlou S, editors. AIDS in Africa . New York: Kluwer Academic Press; 2002. King, G., & Zeng, L. (2001). Logistic regression in rare events data. Pol itical analysis , 9(2), 137 - 163. Tabachnick, B. G., & Fidell, L. S. (2007). Multivariate analysis of variance and covariance. Using multivariate statistics, 3, 402 - 407. retrov iral therapy in Botswana. Journal of Nursing Scholarship , 41 (2), 149 - 157. Kirsten, I., Sewangi, J., Kunz, A., Dugange, F., Ziske, J., Jordan - Harder, B., ... & Theuring, S. (2011). Adherence to combination prophylaxis for prevention of mother - to - child - tran smission of HIV in Tanzania. PloS One , 6(6), e21020. Lilian, R. R., Kalk, E., Technau, K. G., & Sherman, G. G. (2013). Birth diagnosis of HIV infection in infants to reduce infant mortality and monitor for elimination of mother - to - child transmission. The Pediatric Infectious Disease Journal, 32(10), 1080 - 1085. Leclerc - Madlala S. (2002). Youth, HIV/AIDS and the importance of sexual culture and context. Social Dynamics, 28,20 41. Leclerc - Madlala, S. (2003). Transactional sex and the pursuit of modernity. S ocial Dynamics, 29,213 233. Longfield, K., Glick, A., Waithaka, M., Berman, J. (2004). Relationships between older men and younger women: implications for STIs/ HIV in Kenya. Studies in Family Planning, 35,125 134. Leclerc - Madlala, S. (2008). Age - disparat e and intergenerational sex in Southern Africa: The dynamics of hypervulnerability. AIDS (London, England), 22 Suppl 4, S17 - S25. sexual behavior in urban Ken ya. International Family Planning Perspectives, 31,6 14. Lule, E., & Haacker, M. (2011). The Fiscal Dimension of HIV/AIDS in Botswana, South Africa, Swaziland, and Uganda: Experiences from Botswana, South Africa, Swaziland, and Uganda . World Bank Publica tions. Machel, J. (2001). Unsafe sexual behaviour among schoolgirls in Mozambique: a matter of gender and class. Reproductive Health Matters, 9,82 90. McLean, P. (1995). Sexual behaviours and attitudes of high school students in the Kingdom of Swaziland. Journal of Adolescents Research, 10,400 420. 169 Mofenson, L. M. (2010). Prevention in neglected subpopulations: prevention of mother - to - child transmission of HIV infection. Clinical Infectious Diseases , 50(Supplement_3), S130 - S148. Mnyani, C. N., Simango, A., Murphy, J., Chersich, M., & McIntyre, J. A. (2014). Patient factors to target for elimination of mother - to - child transmission of HIV. Globalization and Health , 10(1), 36. Mugasha, C., Kigozi, J., Kiragga, A., Muganzi, A., Sewankambo, N., Coutinho, A., & Nakanjako, D. (2014). Intra - facility linkage of HIV - positive mothers and HIV - exposed babies into HIV chronic care: rural and urban experience in a resource limited setting. PLoS One , 9(12), e115171. Ministry of Transport and Communications (MTC), Gover nment of Botswana, 2016 UNAIDS (2012). Technical guide for countries to set targets for universal access to HIV prevention, treatment and care for injecting drug users (2012). Ministry Minerals, Energy and Water Resources (2015) greater Gaborone area wate r supply situation 24 December 2015. Gaborone, Botswana. Munyaneza, F., Ntaganira, J., Nyirazinyoye, L., Birru, E., Nisingizwe, M. P., Gupta, N., . . . Hedt - Gauthier, B. L. (2018). Community - based accompaniment and the impact of distance for HIV patients n ewly initiated on antiretroviral therapy: Early outcomes and clinic visit adherence in rural Rwanda. AIDS and Behavior , 22(1), 77 - 85 Marukutira, T., Stoové, M., Lockman, S., Mills, L. A., Gaolathe, T., Lebelonyane, R., . . . Hellard, M. (2018). A tale of t Journal of the International AIDS Society , 21(3), doi:10.1002/jia2.25090. Ministry of Health and Wellness, Botswana. (2011). HIV prevalence in adult females . R etrieved from http://www.gov.bw/en/Citizens/Sub - Audiences/Women/Women -- HIV/ on January 7, 2019 . Muula, A. S., Ngulube, T. J., Siziya, S., Makupe, C. M., Umar, E., Prozesky, H. W., ... & Mataya, R. H. (2007). Gender distribution of adult patients on highly active antire troviral therapy (HAART) in Southern Africa: a systematic review. BMC public health , 7(1), 63. - retroviral treatment: A survey correlating pharmacy refill records and pill counts wit h immunological and virological indices. International Journal of Nursing Studies , 48(11), 1323 - 1329. Nkomazana, O., Peersman, W., Willcox, M., Mash, R., & Phaladze, N. (2014). Human resources for health in Botswana: the results of in - country database and reports analysis. African Journal of Primary Health Care & Family Medicine , 6 (1), 1 - 8. Nkosana, J., Rosenthal, D. (2007). Saying no to intergenerational sex: the experience of schoolgirls in Botswana. Vulnerable Child Youth Studies, 2,1 11. Nkosana, J., Rosenthal, D. (2007). The dynamics of intergenerational sexual relationships: the experience of schoolgirls in Botswana. Sexual Health, 4,181 187. 170 Nnko, S., Pool, R. (2007). Sexual discourse in the context of AIDS: dominant themes in adolescent sexualit y among primary school pupils in Magu district, Tanzania. Health Transactions Reviews, 7 (Suppl. 3),85 90. National AIDS Coordinating Agency. (2015) Progress Report of the National Response to the 2011 Declaration of Commitments on HIV and AIDS. Gaborone, Botswana Ndubuka, J., Ndubuka, N., Li, Y., Marshall, C. M., & Ehiri, J. (2013). Knowledge, attitudes and practices regarding infant feeding among HIV - infected pregnant women in Gaborone, Botswana: a cross - sectional survey. BMJ Open , 3(11), e003749. Nyand at J, Van Rensburg G. (2017) Non - disclosure of HIV positive status to a partner and mother - to - child transmission of HIV: Evidence from a case control study conducted in a rural county in Kenya. S outhern Afr ican Journal of HIV Med icine. 18(1), 1 - 10 Orrell, C., Bangsberg, D. R, Badri, M., & Wood, R. (2003). Adherence is not a barrier to successful antiretroviral therapy in South Africa. Journal of AIDS, 17: 1369 - 1375. O'Brien, S. J., Nelson, G. W., Winkler, C. A., & Smith, M. W. (2000). Polygenic and multif actorial disease gene association in man: lessons from AIDS. Annual review of genetics , 34(1), 563 - 591. Ojikutu, B., Makadzange, A. T., & Gaolathe, T. (2008). Scaling up ART treatment capacity: lessons learned from South Africa, Zimbabwe, and Botswana. Cu rrent Infectious Disease Reports , 10 (1), 69 - 73. Peltzer, K., Friend - du Preez, N., Ramlagan, S., & Anderson, J. (2010). Antiretroviral treatment adherence among HIV patients in KwaZulu - natal, South Africa. BMC Public Health , 10(1), 111 Prochaska, J. O., R edding, C. A., Evers, K. E., Glanz, K., Lewis, F. M., & Rimer, B. K. (2002). Health behavior and health education: theory, research, and practice. The transtheoretical model and stages of change, 60 - 84. Poulin, M. (2007). Sex, money, and premarital partne rships in southern Malawi. Social Science Medicine, 65,2383 2393. Plummer, M., Ross, D., Wight, D., Changalucha, J., Mshana, G., Wamoyi, J., Todd, J., Anemona, A., Mosha, exual behavior data collected in rural northern Tanzania using five methods. Sexually Transmitted Infection, (supp 2), ii49 ii56. Rosenstock, I. M., Strecher, V. J., & Becker, M. H. (1988). Social learning theory and the health belief model . Health Educat ion Quarterly , 15(2), 175 - 183. Condom Use: A Panel Study. Journal of Applied Social Psychology , 26(9), 749 - 772. Samson, J., D. Berteaux, B.J. McG ill, and M.M. Humphries (2011): Geographic disparities and moral hazards in the predicted impacts of climate change on human populations. Global Ecology and Biogeography , 20(4), 532 - 544. Schwartländer, B., Stover, J., Hallett, T., Atun, R., Avila, C., Gouw s, E.& Alsallaq, R. (2011). Towards an improved investment approach for an effective response to HIV/AIDS. The Lancet , 377 (9782), 2031 - 2041. 171 Shelton, J.D., Cassell, M., Adetunji, J. (2005). Is poverty or wealth at the root of HIV? Lancet, 366,1057 1058. S Vulnerable victims and active social agents. Social Science and Medicine, 52,1815 1826. Smith, K.R., A. Woodward, D. Campbell - Lendrum, D.D. Chadee, Y. Honda, Q. Liu, J.M. Olwoch, B. Revich, and R. Sauerborn, 2014: Human health: impacts, adaptation, and co - benefits. In: Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and Sectoral Aspects. C ontribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Field, C.B., V.R. Barros, D.J. Dokken, K.J. Mach, M.D. Mastrandrea, T.E. Bilir, M. Chatterjee, K.L. Ebi, Y.O. Estrada, R.C. Genova, B. Girma, E .S. Kissel, A.N. Levy, S. MacCracken, P.R. Mastrandrea, and L.L. White (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, pp. 709 - 754. Spangler, S. A., Onono, M., Bukusi, E. A., Cohen, C. R., & Turan, J. M. (2014). HIV - po sitive status disclosure and use of essential PMTCT and maternal health services in rural Kenya. Journal of Acquired Immune Deficiency Syndromes (1999), 67(Suppl 4), S235. Sherr, Lorraine, and Natasha Croome. Involving fathers in prevention of mother to c hild transmission initiatives what the evidence suggests. Journal of the International AIDS Society 15 (2012): 17378. Stringer, J. S., Zulu, I., Levy, J., Stringer, E. M., Mwango, A., Chi, B. H., ... & Saag, M. S. (2006). Rapid scale - up of antiretroviral therapy at primary care sites in Zambia: feasibility and early outcomes. Jama, 296(7), 782 - 793. Tandon, A., Murray, C.J.L., Lauer, J.A., Evans, D.B. Measuring Overall health system performance for 191 countries. World Health Organization GPE Discussion Pa per Serices No. 30 (2016). Tapp, C., Milloy, M. J., Kerr, T., Zhang, R., Guillemi, S., Hogg, R. S., ... & Wood, E. (2011). Female gender predicts lower access and adherence to antiretroviral therapy in a setting of free healthcare. BMC Infectious Diseases , 11(1), 86. Thupayagale - Tshweneagae, G. (2010). Behaviours used by HIV - positive adolescents to prevent stigmatization in Botswana: HIV - positive adolescents. International Nursing Review , 57(2), 260 - 264. Tawfik, L. (2003). Soap, sweetness, and revenge: p atterns of sexual onset and partnerships amidst AIDS in rural southern Malawi. Baltimore, MD: John Hopkins University, Bloomberg School of Public Health. Tuller, D. M., Bangsberg, D. R., Senkungu, J., Ware, N. C., Emenyonu, N., & Weiser, S. D. (2010). Tra nsportation costs impede sustained adherence and access to HAART in a clinic population in southwestern uganda: A qualitative study. AIDS and Behavior , 14(4), 778 - 784. Türmen, T. (2003). Gender and HIV/aids. International Journal of Gynecology & Obstetrics , 82(3), 411 - 418. United Nations International Children's Emergency Fund [UNICEF] (2008). Why improved sanitation is important for children. www.sanitationyear2008.org. United States Agency International Development [USAID] (2011). Programming Water, Sani tation and Hygiene (WASH) Activities in U.S. Government Country Operational Plans (COPs). 172 United States Agency for International Developent. (2007) Integration of Water Sanitation and Hygiene into HIV/AIDS Home - Based Care Strategies. CRS, Malawi. UNAIDS 19 98 Gender and HIV/AIDS. Geneva, World Health Organization UNFPA (2002) Addressing Gender Perspectives in HIV Prevention. HIV Prevention Now: Programme Briefs No. 4, February 2002.New York, United Nations Population Fund (UNFPA). Walcott MM, Hatcher AM, Kw ena Z, Turan JM. (2013). Facilitating HIV status disclosure for pregnant women and partners in rural Kenya: A qualitative study. B MC Public Health Weinreb, A. (2002). Lateral and vertical intergenerational exchange in rural Malawi . Journal of Cross - Culture Gerontology, 17,101 138. Wojciki, J.M. (2005). Socioeconomic status as a risk factor for HIV infection in women in east, central and southern Africa: a systematic review . Journal of Biosocial Science, 37 ,1 36. Wight, D., Plummer, M.L., Mshana, G., Wanoy i, J., Shigongoc, Z.S., Ross, D.A. (2006). Contradictory sexual norms and expectations for young people in rural Northern Tanzania. Social Science and Medicine, 62,987 997. Wojciki, J.M. (2002). Commercial sex work or ukuphanda? Sex - formoney exchange in S oweto and Hammanskraal Area, South Africa. Culture. Medical Psychiatry , 26,339 370. Van Klaveren, M., Tijdens, K., Hughie - and employment in Botswana. Amsterdam: AIAS Working Paper , 09 - 81. Water Utilities corporation (2016) 2015/16 Annual Report. Sedibeng House, Gaborone, Botswana. World Health Organization (WHO)/CDC Prevention of Mother - to - Child Transmission of HIV Generic Training Package is a comprehensive approach to the training of heal thcare workers (2008). World Health Organization (2011). Joint strategic action framework to accelerate the scale - up of voluntary medical male circumcision for HIV prevention in eastern and southern Africa (2012 2016), Geneva. Wynn, A., Ramogola - Masire, D., Gaolebale, P., Moshashane, N., Agatha Offorjebe, O., Arena, K. & Morroni, C. (2016). Acceptability and feasibility of sexually transmitted infection testing and treatment among pregnant women in Gaborone, Botswana, 2015. BioMed Research International . Yates, T., Lantagne, D., Mintz, E., & Quick, R. (2015). The impact of water, sanitation, and hygiene interventions on the health and well - being of people living with HIV: A systematic review. J AIDS Journal of Acquired Immune Deficiency Syndromes, 68 Supp l 3, S318 - S330.