KNOWLEDGE AND ATTITUDES TOWARDS AUTISM SPECTRUM DISORDERS AMONGST GENERAL PRACTITIONERS IN KARACHI, PAKISTAN By Khalid Ibrahim Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Epidemiology - Master of Science A THESIS 2019 ABSTRACT By Khalid Ibrahim KNOWLEDGE AND ATTITUDES TOWARDS AUTISM SPECTRUM DISORDERS AMONGST GENERAL PRACTITIONERS IN KARACHI, PAKISTAN Introduction: Autism Spectrum Disorder (ASD) is a complex lifelong neurodevelopmental and behavioral disorder that manifests in infancy or early childhood and is characterized by impairments in social interaction and communication, and by repetitive, stereotyped behavior. While the diagnosed prevalence of ASD is on the rise, reliable epidemiological data, including risk factors for ASD is lacking in many developing countries, including Pakistan. In this context, the main objective of this study is to evaluate whether there is a lag in information transfer between what is known about ASD in developed countries compared to that which is known amongst general practitioners (GP) in Karachi, Pakistan. More specifically, this research tests the following two hypotheses: (1) Physicians younger than 30 years of age are more likely to have knowledge of ASD, and (2) Among general practitioners who self-reported having an understanding of ASD, when questioned about old and new theories, those that graduated in the distant past (before 1990), will continue to hold antiquated misconceptions regarding ASD diagnosis and treatment. Methods: We utilized the Mediguide Comprehensive Medical Directory to establish a sampling frame for the city of Karachi (n = 5,849 practicing general practitioners), from which a final analytic sample of 348 physicians was derived. We used the statistical software, SAS® Version 9.4, for all analyses. Results: Of 348 physicians, 148 (44.6%) had heard of the term “autism.” Physicians younger than 30 years of age were more likely to self-report knowledge of autism compared to physicians who were older than 50 years of age (OR = 2.30; 95% Cl: 1.17, 4.55). During their medical school training, physicians who had ward rotations in pediatrics and internal medicine were more likely to self-report knowledge of autism (OR = 2.36; 95% Cl: 1.38, 4.03 and OR = 2.37 95% Cl: 1.31, 4.28), respectively. Physicians who graduated after 1990 were 4.28 times more likely to correctly respond to five or more of the eight questions regarding diagnosis and treatment of autism (OR = 4.28; 95% Cl: 1.00, 21.37). Discussion: This study adds to what is currently known about the knowledge disparity with respect to ASD that exists between developed countries and Karachi, Pakistan. As the global prevalence of ASD becomes a major public health concern, narrowing this knowledge gap takes on an increasingly important significance. It is imperative that general practitioners in Karachi gain a contemporary understanding of ASD so that timely interventions can be implemented. Copyright by KHALID IBRAHIM 2019 TABLE OF CONTENTS LIST OF TABLES ...................................................................................................................... vii LIST OF FIGURES ..................................................................................................................... ix KEY TO ABBREVIATIONS ......................................................................................................... x Introduction ........................................................................................................................... 1 Specific Aims .................................................................................................................................... 2 Background ...................................................................................................... 3 Historical Perspective .......................................................................................................... 4 Diagnosing Autism Spectrum Disorders ............................................................................... 6 Autism Screening .................................................................................................................................. 7 Standardized diagnostic instruments (ADI-R and ADOS) ...................................................................... 9 Treatment ......................................................................................................................... 10 1.2.1 1.2.2 Pakistan ......................................................................................................... 13 Profile of Pakistan and Karachi .......................................................................................... 13 Health Sector in Pakistan ................................................................................................... 18 Politics and federal government responsibility ................................................................................... 18 Pakistan, an epidemiologic transition in the works ............................................................................ 19 Chapter 1. 1.1 1.2 1.3 2.1 2.2 Chapter 2. 2.2.1 2.2.2 Chapter 3. 3.1 3.1.1 3.1.2 3.1.3 3.1.4 3.1.5 Research Design and Methods ....................................................................... 21 Study Location and Population .......................................................................................... 21 Description of Study Population ......................................................................................................... 21 Establishing Partnerships with Local Organizations ............................................................................ 22 Procedure for Establishing a Sampling Frame ..................................................................................... 23 Study Sample ...................................................................................................................................... 24 Sample Size ......................................................................................................................................... 25 IRB and Database development ......................................................................................... 26 Institutional Review Board .................................................................................................................. 26 Database development ....................................................................................................................... 26 Questionnaire Delivery ...................................................................................................... 27 Questionnaire Design and Flow ......................................................................................... 28 Demographics ..................................................................................................................................... 29 3.4.1 Exposure to Autism Spectrum Disorders ............................................................................................ 30 3.4.2 Knowledge .......................................................................................................................................... 30 3.4.3 3.4.4 Attitudes and Practices ....................................................................................................................... 31 3.4.5 Open Ended Questions ....................................................................................................................... 32 3.2.1 3.2.2 3.2 3.3 3.4 Results ........................................................................................................... 33 Demographic Analysis ....................................................................................................... 34 4.1 4.2 Clinical Care ....................................................................................................................... 36 4.3 Medical Education and Interests ........................................................................................ 37 4.4 Self-reported knowledge of the term “autism” and Logistic Regression ............................. 39 Chapter 4. v Chapter 5. 5.1 5.2 5.3 5.4 5.5 5.6 5.7 Discussion ...................................................................................................... 58 Factors most likely to predict knowledge of the term “autism” .......................................... 60 Continuing Medical Education and Other Sources of Medical Knowledge .......................... 61 Antiquated knowledge of ASD related topics ..................................................................... 63 Clinical Translation ............................................................................................................ 64 Pharmaceutical Companies ................................................................................................ 65 Bimodal distribution of wealth and health ......................................................................... 66 Findings from fieldwork ..................................................................................................... 67 Geographical Constraints .................................................................................................................... 68 Logistical barriers ................................................................................................................................ 68 Cultural congruence of instrument ..................................................................................................... 68 Incentive for physicians ....................................................................................................................... 69 Attitudes and Practices ...................................................................................................... 69 Human Subjects Review .................................................................................................... 70 Limitations ........................................................................................................................ 70 5.7.1 5.7.2 5.7.3 5.7.4 5.8 5.9 5.10 4.5 4.6 4.7 Knowledge ........................................................................................................................ 45 Source of Knowledge about Autism ................................................................................... 45 Attitudes and Clinical Practices .......................................................................................... 54 4.7.1 Open ended questions ........................................................................................................................ 56 APPENDIX ............................................................................................................................. 72 BIBLIOGRAPHY ..................................................................................................................... 80 vi Table 1 Demographic Variables .................................................................................................... 35 LIST OF TABLES Table 2 Mean age of GPs in the study sample .............................................................................. 36 Table 3 Variables describing the study subjects' clinical practice ................................................ 37 Table 4 Variables describing GPs' medical education .................................................................. 38 Table 5 Distribution of subjects who self-reported knowledge of the term "autism." ................ 39 Table 6 Logistic regression response profile ................................................................................ 40 Table 7 Logistic regression model fit statistics ............................................................................. 41 Table 8 Testing global null hypothesis .......................................................................................... 41 Table 9 Analysis of effects for logistic regression ......................................................................... 42 Table 10 Analysis of maximum likelihood estimates .................................................................... 43 Table 11 Odds Ratio estimates for variables in logistic regression model ................................... 44 Table 12 Association of predicted probabilities and observed responses ................................... 45 Table 13 Sources of knowledge about the term "autism" ........................................................... 46 Table 14 Media sources of knowledge about “autism” ............................................................... 46 Table 15 Frequency of subjects who attended symposia about ASD ........................................... 47 Table 16 Self-reported working knowledge of ASD ...................................................................... 48 Table 17 General knowledge questions about ASD ...................................................................... 48 Table 18 GPs’ responses to statements about ASD ...................................................................... 49 Table 19 Bivariate analysis of knowledge statements and whether the subject graduated before or after 1990 ................................................................................................................................. 50 Table 20 Bivariate analysis comparing low knowledge scores and high knowledge scores with demographic variables ................................................................................................................. 53 vii Table 21 Statements evaluating GPs’ attitudes towards ASD ...................................................... 55 Table 22 Open-ended response categories .................................................................................. 56 Table 23 Selected quoted comments from GPs ........................................................................... 57 viii Figure 1 Map of Pakistan (Central Intelligence Agency 2007) ...................................................... 15 LIST OF FIGURES Figure 2 Map of Karachi (Wikimedia Commons contributors 8 May 2016 ) ................................ 17 Figure 3 Questionnaire flow for GPs in this study ........................................................................ 29 Figure 4 Flow of subjects through interview ................................................................................ 34 Figure 5 Distribution of physicians' field of personal interest ...................................................... 39 Figure 6 A forest plot of odds ratio estimates from the logistic regression model ...................... 44 Figure 7 Frequency of correct responses to knowledge statements about ASD .......................... 52 ix AIC ASD COS GPs MSUIRB PPV SC SCQ SRS-2 KEY TO ABBREVIATIONS Akaike Information Criterion Autism Spectrum Disorder Childhood Onset Schizophrenia General Practitioners Michigan State University Institutional Review Board Positive Predictive Value Schwarz Criterion Social Communication Questionnaire Social Responsiveness Scale™, 2nd Edition x Introduction Autism Spectrum Disorder (ASD) is a complex lifelong neurodevelopmental and behavioral disorder that manifests in infancy or early childhood. It is characterized by impairments in social interaction and communication, and by repetitive, stereotyped behavior (Tchaconas and Adesman 2013). The diagnosed prevalence of ASD is on the rise in developed countries and has become a major public health concern (Matson and Kozlowski 2011, Baxter, Brugha et al. 2015). In these countries, resources have been invested to accurately assess prevalence and identify potentially related risk factors (Zwicker and Emery 2014). At the time of this study, reliable epidemiologic data on ASD were lacking in Pakistan. In most developing countries, the prevalence of factors associated with ASD are unknown (Daley 2002). To best understand how healthcare for ASD is delivered in Pakistan, we need to get a better understanding of Pakistani physician’s knowledge and views about ASD (Imran and Azeem 2014). As the first step in conducting population based epidemiological studies in Pakistan we proposed the Pakistan Autism Project. This project set out to administer a questionnaire to assess knowledge, attitudes, and practices of the General Practitioners (GPs) in Karachi, the largest city in Pakistan. The primary objective of this study is to evaluate whether there is a lag in information transfer between what is known about ASD in developed countries compared to that which is known by general practitioners in Karachi. Secondly, to assess the effectiveness of continuing medical education amongst practicing physicians to keep up-to-date with ASD research. The following two hypotheses are tested to achieve this overarching goal: (1) Physicians who 1 graduated from medical school more than 20 years ago lack a contemporary understanding of ASD, and (2) Some physicians who self-report knowledge of ASD might still be unqualified to treat their patients presenting with ASD and to properly refer them to appropriate care. Specific Aims 1. To assess the knowledge, attitude and practices of the general practitioners as it pertains to ASD in Karachi 2. To identify factors associated with knowledge, attitude and practices of the general practitioners in Karachi regarding ASD 2 Chapter 1. Background This introduction to the topic of ASD is distributed amongst two chapters: Chapter 1 will start with a brief history of ASD, standardized diagnostic criteria, and treatment options for people with ASD. Chapter 2 focuses on the global burden of disease for ASD with a focus on the developing world. Later in Chapter 1, I will also present the most current research findings in regards to ASD risk factors, surveillance, and the latest treatment options. Although, comparatively little has been written regarding the impact of ASD in developing countries Chapter 2 will compare and contrast the efforts of researchers and policymakers in the developing world and in industrialized nations. This historical perspective on ASD provides context for questions presented in the questionnaire we administered to study subjects. To best assess the knowledge of physicians surveyed, I asked them questions to confirm or dispel old and new theories regarding ASD. In Chapter 3, I will present the study design and explain how the sampling frame was established to derive our study sample from this population. I also present the methods of the developing the questionnaire design and flow. Then, Chapter 4, I present the statistical methods for analysis and the results that follow. Chapter 5 provides the discussion of our findings in the context of existing literature and the implications of our research for the clinical knowledgebase. 3 1.1 Historical Perspective Although, we have known of autism as a developmental disorder for more than half a century, in a large part, ASD research remains in the descriptive stages of epidemiological studies. The term “autism” was first used by Eugen Bleuler a Swiss psychiatrist in 1911 to describe aloneness and ambivalence in adult schizophrenia. Almost a century later, this misconception in diagnostic clarity between ASD and Schizophrenia remains difficult to disabuse (Woodbury-Smith, Boyd et al. 2010). In 1943, Dr. Leo Kanner was first to publish a detailed description of autism as it is known today. Dr. Kanner’s article “Autistic Disturbances of Affective Contact” was based on a case study of eleven children compiled over five years (Kanner 1943). Kanner’s study subjects were children who had withdrawn human contact since the age of one. He noted detailed descriptions of many aspects of these children’s lives. He observed their interpersonal interactions, development since early childhood, and even eccentric habits specific to each child. Kanner observed five distinguishing criteria for autistic children; a lack of contact with other people, strong desire for “preservation of sameness,” a fascination with specific objects typically used for repetitive activities, language development not used for interpersonal communication, and “the retention of intelligent and pensive physiognomy and good cognitive potential” (Eisenberg and Kanner 1956). Seven decades later, there is scientific consensus “that autism occurs equitably across demographic groups” and ASD does not have racial, ethnic or social boundaries (Dyches 2004). In contrast, in 1943, when Kanner described his study sample and presented his patients with ASD symptomology, he noted that every child in his observation “[came] of highly intelligent 4 families” (Kanner 1943). Furthermore, he described his patients coming from households of intelligent families who were “strongly preoccupied with scientific, literary, or artistic work” (Kanner 1943). These statements lead to much controversy decades into the future (Dyches 2004). Just a year after Leo Kanner’s original publication, an Austrian pediatrician, Hans Asperger published an article titled “Autistic Psychopathy in Childhood”, describing “autistic psychopathy” in four children (Asperger 1944). Both Kanner and Asperger made similar observations in their patients but Asperger’s patients all developed speech before school age, while only one of Kanner’s patients developed speech (Frith 1991). Dr. Asperger referred to his patients with autistic psychopathy as “little professors” and they were described as “talking like grown-ups” even in early childhood (Asperger 1944, Frith 1991). Both Kanner and Asperger noted that patients with these behavioral traits had previously been misdiagnosed as either schizophrenic or feeble-minded (Kanner 1943, Asperger 1944). As Kanner explained, “several children of our group were introduced to us as idiots or imbeciles. One still resides in a state school for the feebleminded, and two had been previously considered as schizophrenic” (Kanner 1943). Before DSM-III, many childhood disorders such as autism, schizophrenia, and several other severe mental disorders were categorized together in clinical practices (American Psychiatric Association. 1980). The relationship between autistic disorder and childhood onset schizophrenia (COS) has evolved during the last decade (Reaven, Hepburn et al. 2008). The establishment of a diagnostic criteria was important in distinguishing between COS and ASD. Kolvin emphasizes that in the middle of the 20th century the lack of uniformity in diagnostic 5 criteria resulted in several childhood onset psychoses to be muddled together (Kolvin 1971). In exploratory studies, it was observed that psychotic disorders are dependent on age when symptoms emerge. Infantile autism was among other psychoses that were observed within the first two years after birth (Kolvin 1971). Kanner in his groundbreaking paper pointed out that late-onset psychoses, observed after three years of age more closely resembled adult schizophrenia (Kanner 1943, Kolvin 1971) Age of onset and the frequency of occurrence of different symptoms have been two of the most common defining characteristics that set apart schizophrenia and autism in children (Kolvin 1971, Ghaziuddin 2005). Children with childhood onset schizophrenia present with hallucination and delusions, (Russell 1994) while autistic children typically have deficits in social interaction, imaginative play, social/ emotional reciprocity, and communication (American Psychiatric Association 1994). As the scientific community started to agree upon this set diagnostic criteria, the original association between COS and ASD was dispelled. Nearly 50 years ago, autism and COS started being distinctly defined (Kolvin 1971, Petty, Ornitz et al. 1984, Green, Padron-Gayol et al. 1992). 1.2 Diagnosing Autism Spectrum Disorders In 1980, infantile autism was first included in the Diagnostic and Statistical Manual of Mental Health Disorders, Third Edition (DSM-III), within the category of Pervasive Developmental Disorders. It wasn’t until the revision of the DSM-III that Autistic Disorder was reclassified and was no longer just a contemporary classification. The clinical criteria for identifying ASD are defined in the Diagnostic Statistical Manual of Mental Health Disorders, Fourth Edition (DSM IV) and also the more recent Diagnostic Statistical Manual of Mental 6 Disorders, Fifth Edition (DSM-V) (American Psychiatric Association 1994, American Psychiatric Association 2000, American Psychiatric Association. and American Psychiatric Association. DSM- 5 Task Force. 2013). According to these manuals, ASD encompasses three developmental disorders; Autistic disorder (AD), Asperger’s Syndrome (AS) and Pervasive Developmental Disorders, Not Otherwise Specified (PDD-NOS). Although this study was conducted before the publication of DSM-V, the latest version of the diagnostic manual revises the criteria for autism, no longer includes Asperger’s Syndrome, but includes a new diagnosis for social communication disorder (SCD) (American Psychiatric Association. and American Psychiatric Association. DSM-5 Task Force. 2013). ASD are spectrum disorders and lie on a continuum extending from AS or mild social deficit to severe handicap. There is also a high level of variability with respect to the presence and intensity of symptoms (Johnson, Myers et al. 2007). It is suggested that due to such variability in symptoms there may be additional subtypes (Johnson, Myers et al. 2007). Each child is unique and their respective symptomology poses challenges in accurate diagnosis for ASD. There are no predictive pathognomonic features but early deficits in socialization are fairly reliable “red flags” (Johnson, Myers et al. 2007). 1.2.1 Autism Screening Over the last few decades, there have been considerable advances in research to understand early behavioral markers that can accurately provide an ASD diagnosis (Zwaigenbaum, Bauman et al. 2015). Translation of these discoveries into clinical practice is dependent upon parents, primary care physicians and teachers being able to use simple screening tools to identify children who may be on the autism spectrum. Furthermore, the 7 broad implementation of these tools results in optimal detection of early signs in children that may lead to proper diagnosis (Johnson and Myers 2007). Children who receive an early diagnosis and early intervention show substantial improvement in development (Robins and Dumont-Mathieu 2006). If these children are able to develop language and symbolic play by the age of five, they tend to have a higher probability of being enrolled in a regular classroom (Robins and Dumont-Mathieu 2006). ASD meet many of the World Health Organization’s criteria for screening: the condition poses an important health care problem, there is treatment available for the ASD, there is an acceptable test available for the condition, and most importantly when treatment is initiated early on, the patient outcomes are more favorable (World Health Organization 2010). A good screening test should be easy to administer, inexpensive, accurate, and reliable. A screening test should also be sensitive. Sensitivity and specificity for these screening instruments are the primary considerations as they get incorporated into practice. As a brief description, sensitivity of a screening instrument refers to how accurately an instrument can identify children on the autism spectrum when they are known to have ASD. Specificity, on the other hand, refers to how accurately a subject can be classified as normal when they truly don’t have ASD. A more sensitive screening test results in fewer false negatives at the cost of more false positives. Taking into consideration that there is a cost associated with a false positive, a screening instrument’s main purpose is to identify red flags for a clinician and therefore should have a higher sensitivity. From a clinician’s perspective, another important consideration is positive predictive value (PPV); a screening test’s PPV determines the probability that a subject has a 8 disease when a screening tests abnormally. The PPV is dependent on three things, the sensitivity, specificity, and prevalence of the disease. A disease with a very low prevalence will make the PPV small. Screening instruments alert a child’s caretaker to seek professional care and a confirmed diagnosis from a specialist. There are two general categories of ASD screening tools: level 1 screens apply to children regardless of risk status, while level 2 screening tools are designed for children who are at an increased risk for ASD (Johnson and Myers 2007). These screenings tools range from short checklists like the Checklist for Autism in Toddlers (M-CHAT) that can be administered by parents as both a level 1 screen or a level 2 screen (Robins, Fein et al. 2001) to the Social Communication Questionnaire (SCQ) which is more appropriate for the primary care setting (Allen, Silove et al. 2007). The Social Responsiveness Scale, 2nd Edition (SRS- 2) is also commonly used to detect subtle symptoms associated with ASD and to identify social impairment such as social awareness, social cognition, and social communication (Constantino and Gruber 2012). 1.2.2 Standardized diagnostic instruments (ADI-R and ADOS) At this point, there are no laboratory tests for diagnosing autism. The Diagnostic Statistical Manual of Mental Disorders (DSM) offers a standard criterion for diagnosing mental disorders. The DSM is published by the American Psychiatric Association and is updated based on updated scientific discoveries. The DSM-V provides mental health professionals with a diagnostic criteria to diagnose autism spectrum disorders. Most epidemiologic studies on ASD use the DSM criteria as the primary means to diagnose and identify autism although there is some controversy over its structure and exact definitions of mental disorders (Conner 1999). 9 Three characteristics symptoms that are commonly used to identify and define ASD include impairments in social interaction, verbal and nonverbal communication, and repetitive behaviors or interests. In the last two decades, there has been dedicated attention to developing standardized instruments to accurately identify ASD cases at an early age. Two such instruments have become the most common in ASD research and clinical practice: the Autism Diagnostic Observation Schedule ADOS (Lord, Risi et al. 1999) and the Autism Diagnostic Interview – Revised ADI-R (Lord, Rutter et al. 1994). Along with clinical judgment, these instruments have received much acclaim and have become the gold standard for diagnosis and case ascertainment for autism research (de Bildt, Sytema et al. 2004). ADOS and ADI-R have also demonstrated accuracy in distinguishing between autism and missed developmental milestones in a child’s development (Reaven, Hepburn et al. 2008). The ADOS is a semi-structured, play based observational interview of patients who are suspected of having ASD. ADOS has four modules that are organized in tasks and materials appropriate for the patient’s age and expressive language level. Observations during these modules are used to score the patient on their social and communication domains. ADI-R is also a semi-structured caregiver interview and it has algorithms scoring the patient in three domains; social reciprocity, communication, and repetitive behaviors. Both these instruments require substantial training and skill in administration and scoring. 1.3 Treatment Although there is no cure for ASD, there are many treatment options for patients who have been diagnosed with ASD. Therapy and treatment for ASD are very intensive and 10 comprehensively takes on the unique symptoms for each patient on the autism spectrum. Some treatment programs take place at a patient’s home while others are conducted with a trained therapist or a specialist (Autism Speaks Inc. 2010). Successful treatment for ASD is dependent on early diagnosis and a treatment plan that is specific for the patient. No single treatment works well for everyone. Several factors are taken into consideration when developing a treatment plan such as; the patient’s strengths and weaknesses, the severity of symptoms, subtype of autism, age of patient, access to health care services, and associated physical or psychiatric conditions (Ghaziuddin 2005). The list below presents the characteristics of a generic treatment plan for a patient with ASD. Essential characteristics of effective interventions for children with autism spectrum disorder 1) Start intervention programs as early as possible. 2) Intensive treatment, 5 days a week for a minimum of 5 hours per day. 3) Use of repeated planned teaching opportunities that are structured over brief periods of time. 4) Sufficient individualized adult attention on a daily basis. 5) Inclusion of a family component, including parent training. 6) Mechanisms for ongoing assessment with corresponding adjustments in programming (Brentani, Paula et al. 2013). With the help of a good treatment plan, a physician’s treatment goal should be to: allow a patient to have functional spontaneous communication; have social instruction or interpersonal communication in various settings; be able to maintain new skills and generalize 11 them to a real-life context; and to employ a behavior management system that results in non- punitive techniques to improve behavioral challenges (Brentani, Paula et al. 2013). Deficits in social skills are pervasive and one of the distinctive characteristics of ASD (American Psychiatric Association 1994). Social skills training is an integral component of treating patients with ASD and the majority of such training requires parental and family involvement (Brentani, Paula et al. 2013). Often social skills training programs are based in schools and focus on interpersonal communication. Normally developing children are integrated as peer-facilitators and help autistic children learn how to greet others, share toys, and engage in conversation (Ghaziuddin 2005). Behavioral programs first use methods like the applied behavior analysis (ABA) which develop a behavioral treatment plan and then launch into a series of intensive targeted behavioral reinforcement (Cooper, Heron et al. 2007). As compared to the general population, people on the autism spectrum have higher levels of social loneliness and anxiety (Gillott, Furniss et al. 2001, Bellini 2004). Cognitive- behavioral therapy (CBT) has been effective at treating anxiety and daily living skills in clinical trials (Wood, Drahota et al. 2009, Reaven, Blakeley-Smith et al. 2012). In one clinical trial, 78% of 7-11 year olds were positive responders to CBT (Wood, Drahota et al.). 12 Chapter 2. Pakistan This study was conducted in a country that is geographically, economically, and culturally very different from industrialized western nations. To best understand the study population and the study sample, this chapter provides a brief overview of Pakistan and goes into detail about Karachi, the study site. 2.1 Profile of Pakistan and Karachi Pakistan proclaimed independence from Britain in 1947 and is located in South Asia. It borders Afghanistan and Iran on the East; China on the North; and India on the West (Figure 1). Pakistan’s boundary lines occupy the easternmost part of the Tigris Euphrates river system and has been home to Indus Valley civilizations since antiquity, 2800 BC –1800 BC. Throughout time the Indus civilizations were invaded by many diverse groups of people such as; the Huns (Jayapalan 2001), Persians (Art. 2000), Arabs (Wink 2002), Turks, Afghans (Bartel 1999), Mongols and the Europeans. As a product of this history, today’s Pakistani people are largely multicultural and multilingual. Pakistan is the sixth most populous country in the world (Directorate of Intelligence 2010) and has the second largest Muslim population in the world. Pakistan is divided into five provinces: Pakhtoon Khuwan, formerly known as the North Western Frontier Province (NWFP); Gilgit Baltistan, home to the Himalayan mountains; Punjab, the agricultural province; Sindh, the study site is the southernmost metropolis of this province; and Balochistan, located in the southwest of Pakistan and is the largest and the most scarcely populated province of Pakistan. Geographically, Pakistan is separated into three regions; the mountainous terrain of the North, 13 the arid plateaus of the west, and the well irrigated fertile land of Punjab and Sindh. Islamabad (the capital), Lahore and Karachi are the most populous cities of Pakistan. In Pakistan, over sixty languages are spoken but there are three official languages: Urdu, English, and Sindhi (Kachru, Kachru et al. 2008). Although, Urdu is the most commonly spoken language of Pakistan, English is the lingua franca of most government ministries and higher socioeconomic classes. Pakistani physicians are expected to be proficient in both Urdu and English but their medical education is conducted strictly in English. Since independence in 1947, Pakistan has sustained two major wars with neighboring India. This continued military tension results in heavy military expenditures for Pakistan (Directorate of Intelligence 2010). Three major factions have gained enough power to have influence on national policy; the military, civil services, and politicians. Power has shifted back and forth between military governments and democratically elected civilian governments. “As a result, many government policies have been made by and for the civil servants and politicians, who are either rich landlords or belong to the small coterie of rural or urban elites” (Ghaffar, Kazi et al. 2000). Due to the Pakistan’s expenditures on the military, healthcare expenditures remain low (0.8% of GDP) even compared to neighboring poorer countries such as Bangladesh (1.2% of GDP), Nepal (1.2% of GDP), and Sri Lanka (1.5% of GDP) (Ghaffar, Kazi et al. 2000). 14 Figure 1 Map of Pakistan (Central Intelligence Agency 2007) Karachi is the southernmost metropolis of Pakistan and is located in the Sindh Province (Figure 2). It is amongst the top ten largest cities in the world by population. As a model of diversity, Karachi accommodates residents with a wide spectrum of socioeconomic status and various ethnic backgrounds. Rapid rural-to-urban migration has resulted in slums in Karachi, whose residents typically face much heavier burdens of disease (D'Souza and Bryant 1999). Karachi boasts the highest per capita income in South Asia. This bimodal distribution of wealth has resulted in an enormous economic disparity. Karachi’s population size, economic, and 15 cultural diversity was the main justification for selecting it as an ideal study location for this observational study. 16 Figure 2 Map of Karachi (Wikimedia Commons contributors 8 May 2016 ) DISTRICT SOUTH: 1. Lyari Town, 2. Saddar Town, DISTRICT EAST: 3. Jamshed Town, 4. Gulshan Town, DISTRICT CENTRAL: 5. Liaquatabad Town, 6. North Nazimabad Town, 7. Gulberg Town, 8. New Karachi Town, DISTRICT WEST: 9. Kemari Town, 10. SITE Town, 11. Baldia Town, 12. Orangi Town, DISTRICT MALIR: 13. Malir Town, 14. Bin Qasim Town, 15. Gadap Town. DISTRICT KORANGI: 16. Korangi Town, 17. Landhi Town, 18. Shah Faisal Town, CANTONMENTS: A. Karachi Cantonment, B. Clifton Cantonment, C. Korangi Creek Cantonment, D. Faisal Cantonment, E. Malir Cantonment, F. Manora Cantonment. 17 2.2 Health Sector in Pakistan 2.2.1 Politics and federal government responsibility Before 1947, in pre-partition India the British Empire had established a strong health care system for government employees while on the other hand the health care system largely overlooked the civilian population (Mushtaq 2009). The British government built a few hospitals that still stand today; in 1860, Lahore Medical College was established and later named King Edward Hospital and Medical School (Mushtaq, 2009). Kind Edward Hospital’s establishment launched a network of hospitals throughout the Indo-Pak sub-continent. The federal government is responsible for establishing health care policies but the provincial governments are largely accountable for implementation of these policies (Ghaffar, Kazi et al. 2000). Due to scarce resources, the federal government provides a very basic health care framework, but it has proven difficult to coordinate these initiatives nationally. These difficulties have hampered large scale improvements in health care for the general public (Ghaffar, Kazi et al. 2000). Health care provisions are divided into two parts; the public and private sector. The private sector is comprised of private hospitals and clinics spread throughout the rural and urban landscape. These services are typically provided based on a fee for service system. The private sectors serves nearly 70% of the Pakistani population (World Bank 1993). Neither the private nor the public sectors work under a regulatory framework and there is a lack of information regarding descriptive information about the Pakistani health care system. Consequently, there is little information relating to the financial resources involved in 18 maintaining the private sector or the public sector of the health care system (Ghaffar, Kazi et al. 2000). This makes it much more difficult to gauge where attention is needed to improve the system. “Healthcare problems are directly linked to the prevailing social, economic, and political systems that determine the allocation of resources in and outside the health sector” (Khowaja 2009). The public sector is comprised of 10,000 facilities ranging from basic health care to tertiary referral centers (Ministry of Health 1995). Health care facilities in rural areas have three to four times the patient load of basic health care units in urban areas (Ministry of Health 1995). As of 1995, there were 22 tertiary care facilities in Pakistan but they are typically associated with academic institutions located in major cities (Ghaffar, Kazi et al. 2000). 2.2.2 Pakistan, an epidemiologic transition in the works The theory of epidemiologic transition deals with large-scale changes in the demographic or sociological makeup of a region and their effect on disease occurrence and death. These complex patterns of disease and death correlate with how the demographic, social and economic makeup of a nation changes as it transgresses from a pre-industrial nation to an industrialized one. Nations that are in early stages of development tend to have high levels of fertility. During this transitions, “a long term shift occurs in mortality and disease patterns whereby pandemics of infection are gradually displaced by degenerative and man- made diseases as the chief form of morbidity and primary cause of death” (Omran 1971). Later in this transition, as pandemics recede, death rates for young children and women of reproductive age are reduced compared to other age groups (Omran 1971). 19 Pakistan has a young age structure with high fertility and low life expectancy; high maternal and child mortality in addition to high incidence of infectious and communicable disease (Afzal and Yusuf 2013). Understanding the stage of epidemiologic transition and the demographic makeup of a nation is important to epidemiologists and healthcare professionals because in developing countries healthcare professionals treat more patients with infectious disease while in industrialized countries physicians generally tend to patients with chronic disease. This balance is further complicated when working in countries that are in the midst of this epidemiologic transition (Afzal and Yusuf 2013). In addition to being amidst a demographic transition, Pakistan has a bimodal distribution of wealth. This has resulted in two very different health care systems working in parallel: a private health care system for the higher socioeconomic classes; and a public and clinic-based healthcare system for the lower socioeconomic classes. In low and middle-income countries, quality in both public and private provider groups have proven to perform poorly but comparatively the private sector has better drug availability and performs better in some aspects of clinical care (Berendes, Heywood et al. 2011). The epidemiologic transition of a country is directly related to a physician’s medical awareness. In a developing country, physicians may be less aware of neurodevelopmental disorders such as Autism because the healthcare community is lagging behind in efforts to curb diseases that are immediately fatal and are focusing on infectious disease. Additionally, the large majority of patients are getting healthcare with a fee for service and may not be able to afford healthcare for children for social and communication disorders such as ASD. 20 Chapter 3. Research Design and Methods 3.1 Study Location and Population 3.1.1 Description of Study Population The study sample for this project compromised of 348 general practitioners practicing in Karachi, Pakistan. Practicing physicians in Pakistan are required to be licensed and registered with the Pakistan Medical and Dental Council (PMDC). The year this study was conducted 116,009 were registered as practicing physicians in Pakistan; out of this number, 49,009 were registered in Sindh province and Karachi is the biggest city in Sindh (Khan 2004). To establish a sampling frame, it was important to find a trusted database of registered physicians in Pakistan. The PMDC database is the most comprehensive and up to date record of clinicians in Pakistan but it wasn’t available for our research purposes. Aga Khan University (AKU) has a small database of physicians who had attended Continuing Medical Education (CME) seminars at AKU. This registry was incomplete and would have led us to a misrepresentation of general practitioners (GPs) in Karachi; physicians who attend CMEs may be different from physicians who are not in AKU’s database. Another source of data on practicing clinicians could be found within pharmaceutical companies’ records. These pharmaceutical databases are used for marketing pharmaceutical products to all physicians and are regularly updated with new and existing records. We acquired the latest version of this database (Medi Publications: House of Medical Publications 2006) for establishing our sample frame and our sample population. This database of physicians listed 5,849 practicing physicians in Karachi. 21 3.1.2 Establishing Partnerships with Local Organizations In the early stages of study planning and development, we established a working relationship with Aga Khan University in Karachi Pakistan. Aga Khan University is a prominent clinical research institution in Pakistan and has consistently received the highest rankings by the Higher Education Commission of Pakistan (Higher Education Commission 2012). Within the medical college at AKU we collaborated with the department of psychiatry and they were eager to provide support. Dr. Ehsan Ullah Syed, a faculty member in the department of psychiatry, agreed to be a local supervisor. He helped with establishing a research network and provided assistance with hiring an on-site research team. Establishing a relationship with AKU helped us with developing credibility within the medical community and a rapport with hospitals, pharmaceutical organizations, and study subjects. We posted ads on the university noticeboards to hire a team of medical students who would help with conducting the study. These medical students were from Aga Khan University Medical College and Dow Medical College. They were instrumental in interviewing physicians for this study and for finding clinics around the city. The study team had local know-how about cultural practices and were knowledgeable about local transit, so we could get around the metropolitan city. Dr. Hossein Rahbar was the off-site principal investigator on this project. I was the project coordinator and co-investigator on this project. My responsibilities included helping design the study, designing the questionnaire, getting research funding, managing study staff, training, conducting physician interviews, study logistics, data cleaning, data entry, and analysis. 22 I was also required to make presentations to AKU physicians about preliminary findings at the end of my time there. 3.1.3 Procedure for Establishing a Sampling Frame We acquired official maps from the Karachi municipal headquarters. These maps did not provide enough detail to identify inner-city streets where many of the neighborhood clinics and smaller hospitals were located. In fact, some parts of the map were uncharted. Neighborhood residents often assigned names to smaller streets. Enrolling a representative sample was going to require going deeper into neighborhoods where we could find smaller clinics. Most physicians’ clinics didn’t use street address to help patients locate their clinics. Instead, physicians used common landmarks to identify their location or were already well known around a neighborhood that they didn’t need to list a specific address. To find these clinics, we needed help from Karachi’s residents and people who were very familiar with commonly used shops, relatively prominent street corners, and other cultural landmarks. Our research team was based in Aga Khan University, which is located centrally in Karachi. To establish a representative sample of general practitioners in Karachi we needed to cluster sample recruitment from all of Karachi’s twenty towns. Due to having limited resources such as; restricted transportation, cost of rental cars, personnel, and logistical constraints we drew a thirty-minute driving radius around Aga Khan University (AKU) and started recruitment within this radius. Two other considerations were taken into account when establishing this travel radius: the safety of the interviewers and enough geographic range for us to enroll enough subjects without oversampling from the same towns. Mid-study, we expanded the subject recruitment radius to encompass all twenty towns of Karachi. By this time, the study team felt 23 more comfortable getting themselves around the city and had become more familiar with how to locate inner-city clinics using local landmarks. 3.1.4 Study Sample By utilizing a local research network, a study team that was familiar with the geographical layout of Karachi, and a database of practicing physicians, we were able to establish a sampling frame. The source database had names, clinic addresses, and phone numbers of physicians categorized by neighborhoods and towns. From our established study frame, subjects were drawn from each town of Karachi, as long as they had a listed address or general clinic location. In addition to clinics, general practitioners were also recruited from both government and private hospitals. This systematic convenience sample ensured that we encompassed the diverse demographic and geographic regions of the city of Karachi and there was representation from rural, urban, large hospital, and smaller clinic settings. To prioritize study team organization and to streamline subject enrollment, each morning, we had a team meeting and identified which town we were going to focus on that day. After this, using our database of physicians working in Karachi, we identified which neighborhood within the respective town we were going to recruit subjects from. Every other physician was identified as a potential subject, but we went back to the top of the list if there weren’t enough subjects in a neighborhood. During this study team meeting, each team member identified physicians’ names with complete addresses from the database. A team member was sent to a study interview only if they were able to clearly understand a location from the database. Many of the addresses in this database had ambiguous addresses because most inner-city streets did not have names. The database used city landmarks to identify clinic locations. After a list of 24 potential subjects was identified, the study team went out in the field to conduct physician interviews. The physicians did not have any preconceived notions that they were going to be interviewed on a respective day nor were any public announcements made about this study being conducted. In addition to the team meeting, twice a week the study team met with me to review interview data, turn in paper forms for data entry into an electronic medical record system, and to work through any difficulties obtaining physician interviews. 3.1.5 Sample Size To calculate the sample size for this project we needed to estimate how many GPs in Karachi would be aware of ASD. This sample, within a 5% margin of error and 95% confidence needed to be able to estimate the proportion GPs in Karachi who are knowledgeable of Autism. Taking for granted the established margin of error and the confidence interval, we also need the population standard deviation to estimate a sample size. We did not know the actual percentage of physicians were aware of ASD in Karachi nor were there any previous studies to help with this estimate. For this reason, we chose the largest estimate of standard deviation with a population proportion of 50%. This calculation resulted in a sample size estimation of n = 384 GPs. The actual number of interviewed GPs was n = 348. In our post-hoc power analysis, the actual sample size was sufficient in detecting significant observed odds ratios with a power of at least 80%. 25 3.2 IRB and Database development 3.2.1 Institutional Review Board The study questionnaire and study protocol were granted an expedited review by the Michigan State University Institutional Review Board (MSUIRB). Expedited status is granted to study proposals that pose no more than minimal risk to human subjects. Upon approval of the study proposal by the MSUIRB, I submitted the proposal to the Aga Khan University Institutional Review Board (AKUIRB). The AKUIRB meets to review research proposals on a monthly basis and this project needed to be approved before I arrived on site. The AKUIRB approved this study upon their first review. 3.2.2 Database development After the study was approved by both the MSUIRB and the AKUIRB, I developed a web- based electronic data capture system to be able to enter study data from Pakistan. The Michigan State University Biomedical Research and Informatics Core (BRIC) provided server space and access to a proprietary electronic data capture system called RIX. Within this system, I developed an electronic version of the study questionnaire. Data entry into a web-based data capture system provided several benefits. I was traveling back to the United States after collecting all the data and I did not want to rely on the transit of paper forms within the airline luggage system. These data were entered into RIX and sent to the BRIC servers from Pakistan. At this stage, the paper forms served as backups or our electronic data. Secondly, electronic data capture allowed me to improve data integrity and data quality. The RIX system allowed minimum and maximum values for string fields. Such 26 features prevented typographical errors during the data entry phase. Lastly, upon the completion of the data entry phase, the RIX system allowed me to directly extract these data to SAS for further cleaning and analysis. 3.3 Questionnaire Delivery Survey design is considered the cornerstone of epidemiologic research and the “goal is to turn the process of everyday conversation into one of rigorous measurement, and the questionnaire is our protocol for that conversation” (CDC 2003). Pakistan is a British Commonwealth country and medical training is conducted in English. Physicians in Pakistan are expected to speak English. In developing the survey instrument for this study, it was important that questionnaire language was culturally congruent and used appropriate language with the study population in mind. There are differences between the English spoken in the US and the English spoken in Pakistan. The questionnaire was pilot-tested with two physicians who received their medical training in Pakistan but were working in the United States. Our pilot testers were able to advise on diction and on improving the delivery of the questionnaire. Secondly, it was also important that the surveying procedure was reproducible and ensured maximum validity. To make certain that we were recording valid responses from each physician, we implemented the questionnaire in-person and recorded the responses on paper. The questions were delivered identically to each subject. The questionnaire and the interviewer prompts were defined explicitly on a paper form. The interviewers had designated prompts for clarification of questions. Having a specified questionnaire facilitated the interview process. Since we were conducting a knowledge, attitudes, and practices survey, an in-person questionnaire also ensured that the subject didn’t skip ahead in the interview. Some 27 downstream questions were only administered based on how physicians answered knowledge related questions. Lastly, administering the questionnaire in one language was likely to improve inter-rater reliability. 3.4 Questionnaire Design and Flow The study questionnaire had four main sections; demographic information about the general practitioner, knowledge about ASD, attitudes towards autism, and the physician’s practices or diagnostic methodology. All study subjects were asked the demographic information and whether they had heard of the term “autism”. Out of the full sample, only GPs who had heard of the term were asked to proceed with questions regarding sources of exposure to the term “autism.” We also asked this subsample of GPs some general knowledge questions about ASD. Next, the GPs were asked whether their understanding of ASD allowed them to make an appropriate diagnosis or referral, in other words, whether they had a working knowledge of ASD. Subjects who responded affirmatively completed the whole questionnaire including questions to access their knowledge and gauge their attitudes and practices about ASD. The Results section goes into further detail about this questionnaire flow but Figure 3 presents the general questionnaire flow for subjects in this study. 28 Figure 3 Questionnaire flow for GPs in this study Demograhic questions and self- reported knowledge of the term "autism" General knowledge questions and self reported working knowledge of ASD Knowledge, attitudes, and practices questions 3.4.1 Demographics The demographics section of the questionnaire was designed to better understand the GP’s medical background and to paint a clearer picture about their current medical practice. Specifically, the demographic section had questions about: the general practitioner’s choice of medical school, ward rotations during medical school, areas of interest in the field of medicine, daily clinical workload, whether they belonged to a medical panel and how the majority of their patients funded their medical care. Questions selected for this section were concise and to the point; the physician’s time commitment was a major consideration. For this section, the study team and I utilized various methods to accurately gauge a general practitioner’s daily workload. We asked how many children they saw weekly, the average time they spend with each patient, hours they work daily 29 and finally the total number of patients they see on a daily basis. These questions were distributed throughout this section of the questionnaire and were used to quantify physician workload. 3.4.2 Exposure to Autism Spectrum Disorders This section gauged the GPs exposure to ASD and ASD related topics. If GPs had been exposed to ASD, we further probed about their source of knowledge about ASD. We asked all physicians whether they had heard the term “autism”. To investigate the best mode of communication to disseminate medical information we asked where the physician had heard about "autism. Next, we asked the physician whether they knew the prevalence rates of autism globally and in the United States to understand what their perception of the ASD’s burden of disease. We asked how many children they suspect to be autistic out of every hundred children they treat. We also inquired about any governments, pharmaceutical or academic institutions that may have contacted them regarding ASD. Lastly, we asked whether they had a “working knowledge” about autism. Only physicians who thought they knew enough about autism to diagnose on the spectrum proceeded with the questionnaire. 3.4.3 Knowledge This section assesses the subject’s knowledge about ASD. Only subjects who had indicated that they have been exposed to ASD during their medical career were directed to this section. The interviewer guided the GPs through various statements regarding ASD and the general practitioners were asked to respond whether they agree or disagree. The GPs 30 responded on a five-point Likert scale from 1(strongly agree) to 5(strongly disagree). These statements were a combination of theories about ASD that are agreed upon in the scientific community and outdated or recently discarded theories regarding the causal pathway of autism. The interviewers were trained to ask each question in the same tone and at the same speed. We hope that these subtle practices kept questionnaire delivery as objective as possible. Statements that accurately represent ASD and ones that have been scientifically disabused were delivered in a similar manner. Likert scales were the most ideal format for these questions because they are an easily understandable method for people being interviewed and they also facilitate statistical analysis for population studies. Secondly, as compared to dichotomous response options, Likert scales allowed physicians to respond on a spectrum without being forced to take a strong stance. We hope that this method helped us better gauge the degree of agreement each subject had regarding questionnaire prompts. Lastly, using Likert scale response options dramatically reduced how much time we spend with each physician and the coded responses saved us from having to conduct natural language processing rules on the final data. 3.4.4 Attitudes and Practices This section mimics the structure of the knowledge section of the questionnaire. Questions in this section were directed towards the subject’s personal views towards ASD based on their exposure and knowledge. We made statements regarding ASD and asked whether the GP agreed or disagreed on a Likert scale. 31 3.4.5 Open Ended Questions The questionnaire ended with two open-ended questions. The interviewees were informed that the questionnaire has ended and asked whether they would like to say anything in regard to ASD in Pakistan. The study staff that was administering the questionnaire then transcribed what the subject said. In the early design stages of this study, we knew that little had been written about ASD in Karachi in medical journals nor in the lay media. An open-ended section was needed to capture data we might have missed in the structured questions. This question was also needed to be able to extract more information from the respondents regarding their feelings and understanding of ASD. The questionnaire was designed to allow comparative analysis amongst study subjects in regard to their knowledge, attitudes, and practices. This left a need for any free form responses the study subject wanted to give us for future studies. 32 Chapter 4. Results This chapter covers the analysis of the study. The dataset was cleaned by Dr. Hossein Rahbar, the Principal investigator (PI) of the study and my advisor. I will start by asking demographic questions from the dataset to better understand the univariate distributions of GPs in our study. Next, I will explain patterns of these distributions using bi-variate correlations. After that, I will explore bivariate patterns that help evaluate the hypotheses. Then, I will build a logistic regression model to better understand risk factors for the outcome (knowledge of autism spectrum disorders). Lastly, to provide a more comprehensive picture of how ASD are diagnosed and treated in this population, I will do a subgroup analysis of general practitioners who said they are knowledgeable about ASD. This subgroup analysis will help us best evaluate the knowledge, attitudes, and practices of subjects who self-report to be knowledgeable about ASD. Figure 4 presents the general flow of this questionnaire. The subsequent sections in this study will present data collected for this study sample. 33 Figure 4 Flow of subjects through interview Demograhic Data n = 348 Subjects who answered main outcome question about self reported knowledge of ASD n = 332 Self-reported knowledge of the term "autism" n = 148 Working knowledge of ASD n = 49 4.1 Demographic Analysis As detailed in the Chapter 3: Research Design and Methods, the study team recruited and interviewed GPs from Karachi. At the beginning of the study period, the GPs were recruited from a 30-minute driving radius around AKU and then this radius was expanded to encompass the all towns of Karachi. This provided a comprehensive and generalizable sample of general practitioners who were providing clinical care in the city of Karachi. These general practitioners had varying workloads and a diverse set of patients. Of the 332 subjects who answered the main outcome question about self reported knowledge about ASD, 224 (67%) of the subjects were men and 108 (33%) of subjects were women (Table 1). Nearly half of the subjects were older than 40 years (49%). At the time of the interview, the average age of the subjects was 41 years old, with the youngest practitioner being 25 years old and the oldest physician was 77 years old. Years since GPs graduated medical school has a bimodal distribution; the majority of the subjects (60%) either graduated within five years or more than 20 years ago. More 34 specifically, 83 subjects (25%) graduated within the last five years and 117 subjects (35%) graduated more than 20 years ago. The remaining subjects had graduated between five years and 20 years ago (40%). There was a similar distribution of years that the physicians had been practicing. More than half of the subject (53.6%), where born in Karachi and stayed local to establish their clinical care. The rest of the interviewed subjects were born elsewhere but moved to Karachi to practice medicine. Table 1 Demographic Variables Frequency Percent 108 224 78 83 158 83 56 33 42 117 113 49 39 30 99 178 154 32.5 67.5 24.4 26.0 49.5 25.1 16.9 10.0 12.7 35.4 34.2 14.9 11.8 9.09 30.0 53.6 46.4 Variable Sex Age Group Frequency Missing = 13 Years since graduating medical school at the time of interview. Frequency Missing = 1 Years practicing medicine at the time of interview. Frequency Missing = 2 Birthplace Response Female Male < 30 30 - 40 ≥ 40 ≤ 5 years 5 - 10 years 10 - 15 years 15 - 20 years > 20 years ≤ 5 years 5 - 10 years 10 - 15 years 15 - 20 years > 20 years Karachi Other 35 Table 2 Mean age of GPs in the study sample Age of General Practitioners at time of interview n 319 Mean 41.0783699 12.3849473 Std Dev Minimum Maximum 25 77 4.2 Clinical Care We asked our GPs whether the majority of their subjects were coming from rural settings, urban settings, or if they had a mix of patients from urban and rural settings. Karachi is an urban metropolis but borders larger agricultural populations. Often, patients come to Karachi from these rural settings to get clinical care. The vast majority of the subjects’ (80.3%) primary patient load was from urban settings but 19.7 % of GPs had patients from either a mix of rural and urban settings or from rural settings alone (Table 3). Next, we explored how patients were getting clinical care paid for. Pakistan does not have a socialized healthcare system and the majority of the Pakistan’s clinical care is paid for out of pocket. When we asked GPs to identify the distribution of source of payment to the clinics and hospitals by patients, 79.9% of GPs reported that more than half of their patients were paying for clinical care themselves: 49 (12.2%) reporting 50%-75% and 208 (64.6%) reporting 75%-100% (Table 3). 36 Table 3 Variables describing the study subjects' clinical practice Variable n = 325, Frequency Missing = 7 Distribution of a physicians' overall patient load that is personally funded. Response Rural Patients from rural vs urban settings. Rural/Urban Urban 0% - 25% 25% - 50% 50% - 75% 75% - 100% 0% - 25% 25% - 50% 50% - 75% 75% - 100% 0% - 25% Distribution of a physicians’ overall 25% - 50% patient load insured by the military. 50% - 75% 75% - 100% n = 322, Not specified = 10 Distribution of a physicians' overall patient load insured by the private insurance. n = 217, Not specified = 115 n = 209, Not specified = 123 Frequency Percent 11.1 8.6 80.3 7.8 12.4 15.2 64.6 71.9 13.8 13.8 0.5 97.1 1.0 1.9 0 36 28 261 25 40 49 208 156 30 30 1 203 2 4 0 4.3 Medical Education and Interests A large majority of our subjects had graduated from government schools (93.6%), with the biggest proportion coming from Dow Medical College (31.6%). Only, 21 subjects (6.4%) had graduated from private medical schools (Table 4). To better gauge a physician’s personal interest in medicine we asked what area of clinical care they were most interested in. They were allowed to specify more than one field of personal interest amongst the following response options: surgery, OB/GYN, psychiatry, pediatrics, internal medicine, geriatrics, and emergency medicine. These questions gauge a physician’s interest in medical topics outside of their educational curriculum (Table 4). More than a third (32%, n = 105) of the subjects, said they were interested in pediatrics. On the other end of the spectrum, only three subjects (<1%) were interested in geriatrics and 13 subjects (4%) were interested in psychiatry (Table 4). 37 Next, we asked GPs what ward rotations they had during their medical training. This question starts to investigate the subject’s academic curriculum during medical school. During medical school, only 43 subjects (13.0%) had a ward rotation in psychiatry, while almost half (44.28%) of the subjects had a pediatric ward rotation. The most popular ward rotation was surgery (62.05%) while the least number of subjects (1.8%) were in a geriatric care ward Table 4 Variables describing GPs' medical education rotation (Table 4). Variable Medical College Attended Government vs Private Medical school. Frequency Missing = 6 Field of personal interest in medicine Ward Rotation in Medical School Frequency Percent 105 85 42 25 75 305 21 105 66 45 28 26 13 3 206 232 147 117 74 43 6 31.63 25.6 12.65 7.53 22.59 93.56 6.44 31.63 19.88 13.55 8.43 7.83 3.92 0.90 62.05 69.88 44.28 35.24 22.29 12.95 1.81 Response Dow Medical College Sindh Medical College Liaquat Medical College Chandka Medical College Other Government Private Pediatrics Internal Medicine OB/GYN Surgery Emergency Medicine Psychiatry Geriatrics Surgical Internal Medicine Pediatrics OB/GYN Emergency Medicine Psychiatry Gerontology 38 y c n e u q e r F 120 100 80 60 40 20 0 Figure 1 Distribution of physicians' field of personal interest 105 66 45 28 26 Pediatrics Internal Medicine OB/GYN Surgery Emergency Medicine 13 3 Psychiatry Geriatrics 4.4 Self-reported knowledge of the term “autism” and Logistic Regression The first part of the questionnaire presented demographic data for the study sample and presented an overview of the study sample. This overview included the subjects educational background. The knowledge section of this questionnaire delved further into the GPs’ self-reported knowledge of the term “autism.” Of the sample of 332 subjects, 148 subjects had heard of the term “autism,” while 184 subjects (55.4%) had never heard of the term and 16 (4.8%) didn’t respond to this question (Table 5). Table 5 Distribution of subjects who self-reported knowledge of the term "autism." Self-reported knowledge of the term “autism” Self-reported knowledge of the term "Autism" Missing = 16 Response Yes No Frequency 148 184 Percent 44.6 55.4 39 After getting a good understanding of the demographic makeup of the study subjects and calculating what percent of GPs who self-reported having knowledge of the term autism, I wanted to fit demographic variables to a logistic regression model. The dependent variable was self-reported knowledge of the term “autism” and I started with independent variables that included everything from the demographic section of the questionnaire. In a logistic regression model, records are included in the regression analysis if all variables in the final model have responses. Taking this into consideration, only 312 observations were used in this model. Table 6 Logistic regression response profile Ordered Value Self-reported Knowledge of ASD Response Profile 1 2 Total Yes No Total Frequency 139 173 312 The model’s convergence was satisfied demonstrating that the model output could be used to predict an outcome. In this case, we could use our independent variables to predict if a physician is likely or unlikely to have heard of the term “autism.” Below, the model fit statistics were used to develop the best model. After using forward, backward, and stepwise selection processes removed all non-significant variables were removed from the original model. At this point, certain variables were included back into the model because some variables, although not statistically significant, made clinical sense to be in the model. Variables included in the final model are described in more detail later in this section. As variables were being added back in, I used both the model fit statistics to determine whether a good model was being developed. The model fit statistics include Akaike Information Criterion (AIC) and Schwarz Criterion (SC). AIC measures goodness of fit in a model but prevents 40 overfitting by penalizing the inclusion of too many independent variables in a model. The Schwarz Criterion is also used for model selection but penalizes the number of parameters in a model more strongly compared to AIC. In summary, the model fit statistics allow us to compare statistical models to each other and prevents underfitting or overfitting (Table 7). Table 7 Logistic regression model fit statistics Model Fit Statistics Intercept Only 430.811 434.554 428.811 Intercept and 7 Covariates 406.393 432.594 392.393 Criterion AIC SC -2 Log L Testing the global null hypothesis demonstrates that the independent variables in my model are able to provide a prediction for the outcome in the final model. The global null hypothesis of no effects of independent variables was rejected (Table 8). Table 8 Testing global null hypothesis Testing Global Null Hypothesis: BETA=0 Chi-Square 36.4187 34.7213 31.3396 DF 6 6 6 Pr > ChiSq <.0001 <.0001 <.0001 Test Score Wald Likelihood Ratio The resulting logistic regression model has the following variables: GP’s age group; and whether they had ward rotations in psychiatry, pediatrics, or internal medicine. Using forward, backwards and stepwise selection, only two variables were statistically selected to be in the model: a ward rotation in pediatrics and a ward rotation in internal medicine. The GPs’ age group, and whether they had ward rotation in psychiatry were later added back into the model because both of those variables are likely to inform their knowledge of ASD. The resulting, final 41 logistic regression model maintained its predictive ability and these latter variables were kept in the model (Table 9). Effect Age Group Pediatrics Ward Rotation Psychiatry Ward Rotation Internal Medicine Ward Rotation DF 3 1 1 1 Table 9 Analysis of effects for logistic regression Type 3 Analysis of Effects Wald Chi-Square 7.8911 9.8076 0.6414 8.1418 Pr > ChiSq 0.0483 0.0017 0.4232 0.0043 The analysis of maximum likelihood estimates breaks down age group, the one categorical variable we have in our sample and examines statistically significant effect on the model (Table 10). The reference category was the age group of GPs who were older than 50 years. The self-reported knowledge of autism was higher amongst GPs ages 25-29 years relative to the reference group. The other two age groups, 30-39 years and 40-49 years did not have a statistically significant difference relative to the reference category. The same is true for GPs who had a ward rotation in pediatrics and internal medicine relative to those who did not (Table 10). As indicated in the Table 10, a positive coefficient indicates that the respective response presents an increased likelihood of self-reported knowledge about ASD. 42 Table 10 Analysis of maximum likelihood estimates Analysis of Maximum Likelihood Estimates Estimate Standard 25-29 years+ 30-39 years- 40-49 years- Yes+ Yes+ Yes+ D F 1 1 1 1 1 1 1 -0.2451 0.5719 -0.0407 -0.2693 0.4283 0.1661 0.4308 Error 0.2213 0.2086 0.2043 0.2176 0.1368 0.2074 0.1510 Wald Chi-Square 1.2262 7.5171 0.0398 1.5322 9.8076 0.6414 8.1418 Pr > ChiSq 0.2681 0.0061 0.8419 0.2158 0.0017 0.4232 0.0043 - Indicates that this response is the same or reduces likelihood for self-reported knowledge of ASD + Indicates that this response increases likelihood for self-reported knowledge of ASD Parameter Intercept Age Group Age Group Age Group Pediatrics Ward Rotation Psychiatry Ward Rotation Internal Medicine Ward Rotation Odds ratio estimates from a logistic regression model demonstrate how all the variables are contributing to the model. In this case, there is a significant difference in the model outcome when we compare physicians who are 25- 30 years old as compared to physicians who are older than 50. In other words, the youngest age group is predictive of self-reported knowledge of “autism” while GPs in the 30-39 and 40-49 categories did not differ significantly in their self-reported knowledge of the term “autism” from those who were older than 50 (Table 11). Similarly, a medical school ward rotation in pediatrics and internal medicine is predictive of a physician’s exposure to the term “autism.” In Figure 6, it is easier to see a graphical representation of how these odds ratio estimates are presented within this model. Notice that, point estimates for physicians who had ward rotations in pediatrics and internal medicine are greater than 1 and their 95% confidence intervals do not encompass 1. 43 Table 11 Odds Ratio estimates for variables in logistic regression model Effect Odds Ratio Estimates Point Estimate Age Group 25-29 vs >50 years Age Group 30-39 vs >50 years Age Group 40-49 vs >50 years Pediatrics Ward Rotation Yes vs No Psychiatry Ward Rotation Yes vs No Internal Medicine Ward Rotation Yes vs No 2.302 1.247 0.993 2.355 1.394 2.367 95% Wald Confidence Limits 1.166 0.638 0.494 1.378 0.618 1.310 4.545 2.439 1.993 4.026 3.143 4.278 Figure 6 A forest plot of odds ratio estimates from the logistic regression model Logistic Regression Odds Ratio Estimates Age Group 0-29 vs >50 years Age Group 30-39 vs >50 years Age Group 40-49 vs >50 years Pediatrics Ward Rotation Psychiatry Ward Rotation Internal Medicine Ward Rotation 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Odds Ratios In general, the association of predicted probabilities and observed responses shows how good this logistic regression model is at correctly predicting an outcome. In this case, we can see that the model has a much higher percent concordant than percent discordant. In other words, the model is able to predict an outcome accurately much more often than it predicts an outcome inaccurately. Similarly, the c-statistic, also known as the concordance statistic is 44 helpful in demonstrating the strength of a model. The c-statistic picks a randomly selected subject that has experienced an outcome (disease or condition) and gives the probability that their risk is higher than a subject who did not experience the same outcome. The c-statistic has to be greater than 0.5; if the c-statistic is close to 0.5, it is demonstrating that the resulting binary outcome variable, is akin to tossing a coin. For this logistic regression model, the c- statistic is 0.681; a high c-statistic indicates a better model (Table 12). Table 12 Association of predicted probabilities and observed responses Percent Concordant Percent Discordant Percent Tied Pairs Association of Predicted Probabilities and Observed Responses 64.2 28.0 7.8 24047 Somers' D Gamma Tau-a c 0.361 0.392 0.179 0.681 4.5 Knowledge As described in Figure 4, only subjects who self-reported having knowledge about the term “autism” were asked about their source of knowledge about ASD. Then, they were asked subsequent questions to evaluate a deeper knowledge of ASD than just the term. This section presents their responses to questions regarding their source of exposure to the term “autism” and their knowledge about ASD. 4.6 Source of Knowledge about Autism The most common source of self-reported knowledge of the term was from the general media. More than half of the study subjects (51.75%, n = 74) heard of “autism” from the media (Table 13). In Pakistan, the term “media” encompasses, television, print media, radio, and the internet. Television (41.9%), print media (39.2%), and the internet (35.1%) were the more 45 common sources for exposing our subjects to the topic of ASD, while radio (6.8%) was less commonly used for disseminating information on this topic (Table 14). After media, medical journals were the second most common (48.9%, n = 69) source of knowledge about the term “autism” (Table 13). Next, 67 (46.5%) subjects from our study sample had heard about the term “autism” in medical school (Table 13). I was also interested in understanding whether non-profit organizations were conducting ASD related seminars in Karachi. This questionnaire revealed that if there were seminars conducted by external organizations, the subject material was not being presented to our study sample. Only, 4.9% or 7 subjects reported hearing about “autism” from an external organization (Table 13). Table 13 Sources of knowledge about the term "autism" Source of knowledge about the term “autism” Media Frequency Missing = 5 Medical Journal Frequency Missing = 7 Medical School Frequency Missing = 4 External Organization Frequency Missing = 5 Response Yes No Yes No Yes No Yes No Frequency 74 69 69 72 67 77 7 136 Percent 51.8 48.3 48.9 51.1 46.5 53.5 4.9 95.1 Table 14 Media sources of knowledge about “autism” Media sources of knowledge about the term “autism” Television Print Media Radio Internet Frequency 31 29 5 26 Percent 41.9 39.2 6.8 35.1 When asked whether general practitioners had attended a symposium on autism, 19 subjects (13.5%) indicted that they had and 122 subjects (86.5%) said that they had not (Table 46 15). The subjects who responded affirmatively were asked about the sponsor of these symposia and 9 subjects indicated that they had attended a symposium on autism at a university. Six indicating that the university was Aga Khan University. Out of the four subjects who attended a symposium with a pharmaceutical company, one of them indicate that the symposium was sponsored by Novartis. Two subjects mentioned the National Institute of Child Health as being a government sponsor. One subject mentioned Ida Rieu Welfare Association as being a Non- Governmental Organization sponsor. Table 15 Frequency of subjects who attended symposia about ASD Frequency of subjects who attended symposia on ASD Frequency 19 122 3 4 9 2 Response Yes No Government Organization Pharm Company University NGO Autism Symposium Frequency Missing = 5 Percent 13.5 86.5 15.8 21.1 47.4 10.5 After establishing, how many subjects had heard of the term “autism,” (n = 148) we asked whether they had a working knowledge of autism. From this sub-sample, 49 subjects (29.3%) reported having working knowledge of autism (Table 16) . The next set of general knowledge questions gauged a deeper understanding of autism (Table 17). First, we asked our study subjects whether they could provide the prevalence of autism in the United States; five subjects (3.5%) responded to this question with an average response of 2,504/100,000. Next, we asked GPs the same question in regard to the global prevalence; eight subjects (5.6%) responded and had a mean estimate of worldwide prevalence at 2,524/ 100,000. Lastly, in this series of questions, we directed the topic to the subject’s own practice and asked how many 47 patients in their practice seem to be on the autism spectrum. In response to this question, 91 subjects responded; within their respective practices on average 1.4/100 subjects showed signs of ASD which translates into 1,400/100,000. Table 16 Self-reported working knowledge of ASD Working Knowledge of Autism Working Knowledge Response Yes No Frequency 49 83 Percent 29.3 49.7 Table 17 General knowledge questions about ASD General Knowledge Questions Do you have an estimate of the prevalence of autism in the US? Mean response to estimated prevalence in United States What about the prevalence of autism in children Worldwide? Mean estimated prevalence in the World Out of every 100 patients you see, how many of them seem to be on the autism spectrum Response Yes No 2,504 / 100,000 Yes No 2,524 / 100,000 1.4 (prevalence of 1,400/100,000) Frequency 5 138 5 8 134 8 91 Percent 3.5 96.5 5.63 94.37 To further evaluate the subjects’ knowledge, in this questionnaire we presented the subjects with several statements and asked them to respond on a Likert scale: Strongly Agree, Agree, Undecided, Disagree, and Strongly Disagree. Table 18 presents GPs’ responses to these statements and outlines the percentage of subjects who answered correctly and incorrectly. In other words, if a statement was correct and the subject both agreed or strongly agreed with the statement, they are presented with a correct response in this table. If the statement was 48 incorrect and they disagreed or strongly disagreed, they are presented with the correct response in this table. If an individual was undecided thy were not included in these analyses. On average, 58.7% of the time subjects answered these responses correctly. Table 18 GPs’ responses to statements about ASD GPs’ responses to statements about ASD Correct Response to this question Autistic children show detachment from their parents. Autism is more prevalent in higher socioeconomic classes Autism is more prevalent in higher educational classes. Autistic children are not affectionate Autism is a precursor for schizophrenia It is hard to distinguish between autism and schizophrenia Autism in children is mostly attributed to neglect in early childhood parenting Children tend to psychologically mature and grow out of autism Total Average Response Frequency Percent Agree Disagree Disagree Agree Disagree Disagree Disagree Disagree Responded Correctly Responded Incorrect Responded Correctly Responded Incorrect Responded Correctly Responded Incorrect Responded Correctly Responded Incorrect Responded Correctly Responded Incorrect Responded Correctly Responded Incorrect Responded Correctly Responded Incorrect Responded Correctly Responded Incorrect Average Correct Responses Average Incorrect Responses 30 16 22 18 24 15 17 27 21 15 31 11 24 22 28 15 65.2 34.8 55.0 45.0 61.5 38.5 38.6 61.4 58.3 41.7 73.8 26.2 52.2 47.8 65.1 34.9 58.7 41.8 49 Table 19 Bivariate analysis of knowledge statements and whether the subject graduated before or after 1990 Bivariate analysis of knowledge statements and whether the subject graduated Knowledge Statement Graduated after 1990 before or after 1990 Graduated before 1990 OR (95% CI) P Value Autism is more prevalent in higher socioeconomic classes (n = 40) Correct response Incorrect response 17 (77.27) 5(27.78) 5 (27.73) 13 (72.22) 8.84 (2.11,37.11) 0.002 Autism is more prevalent in higher educational classes (n = 39) Correct response Incorrect response 17 (80.95) 4 (19.05) 7 (38.89) 11 (61.11) 6.68 (1.58, 28.29) 0.0076 Autistic children are not affectionate (n = 44) Correct response Incorrect response 10 (58.82) 18 (66.67) 7 (41.18) 9 (33.33) 0.71 (0.20, 2.50) Autism is a precursor for schizophrenia (n = 36) Correct response Incorrect response 14 (66.67) 8 (53.33) 7 (33.33) 7 (46.67) 1.75 (0.45, 6.82) It is hard to distinguish between autism and schizophrenia (n = 42) Correct response Incorrect response 19 (61.29) 5 (45.45) 12 (38.71) 6 (54.55) 1.9 (0.38, 9.70) * NS NS NS Autism in children is mostly attributed to neglect in early childhood parenting (n = 46) Correct response Incorrect response 15 (62.50) 12 (54.55) 9 (37.50) 10 (45.45) 1.39 (0.42, 4.51) NS Children tend to psychologically mature and grow out of autism (n = 43) 6.89 (1.68, 28.10) 0.0053 Correct response Incorrect response 20 (71.43) 4 (26.67) 8 (28.57) 11 (73.33) * Exact Confidence Limits 50 In a bivariate analysis, we evaluated if there is a statistically significant correlations between whether a GP responded correctly to a statement and whether they had graduated before or after 1990 (Table 19). This bivariate analysis revealed three statistically significant correlations. Physicians who correctly disagree with the statement that autism is more prevalent in higher educational classes were more likely to have graduated after 1990 (Table 19). In reverse, physicians who graduated before 1990 were more likely to incorrectly think autism is common in higher socioeconomic classes. Similarly, GPs who incorrectly thought that ASD is more prevalent in highly educated families were also less likely to have graduated after 1990 (Table 19). Lastly, physicians had graduated medical school before 1990 also thought that children tend to grow out of autism (Table 19). Figure 7 displays the data from Table 18 as a histogram. From this chart, you can see that the majority of the subjects responded correctly to 4 of the 8 statements. For additional bivariate analysis, GPs’ response rates were organized into two categories: a high score and a low score. A high score was defined as five or greater accurate responses and a low score was defined as four or fewer accurate response. Then, in Table 20 these high and low scores to knowledge questions were compared to whether a GP graduated before or after 1990. These analyses revealed one statistically significant correlation. Physicians who were younger were more likely to score higher in their responses to these statements (OR (95% CI) 4.28 (1.00,21.37), p-Value 0.02). Even though the lower 95th percentile of this calculations reaches 1, this variable demonstrates a strong correlation (Table 20). 51 Figure 7 Frequency of correct responses to knowledge statements about ASD Number of Knowledge Questions Answered Correctly by Each Subject y c n e u q e r F 14 12 10 8 6 4 2 0 8 2 5 3 4 1 1 0 12 10 4 5 6 6 Number of Question Correct 2 7 1 8 52 Table 20 Bivariate analysis comparing low knowledge scores and high knowledge scores with demographic variables Bivariate analysis comparing low knowledge scores and high knowledge scores vs demographic variables Variable Male Female After 1990 Before 1990 Rural Urban Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No High Knowledge score (5-8 correct) Low Knowledge score (0-4 correct) OR (95% CL) P Value 11 (31.43) 8 (57.14) Sex 24 (68.57) 6 (42.86) 0.34 (0.10,1.23) Medical School graduation Year * 15 (51.72) 4 (20.00) 14 (48.28) 16 (80.00) Demographic makeup of majority of patients 4.28 (1.00,21.37) 2 (28.57) 16 (40.00) 11 (35.48) 8 (44.44) 9 (52.94) 10 (31.25) 4 (40.00) 15 (38.46) 13 (40.63) 6 (35.29) 14 (38.89) 5 (38.46) 6 (54.55) 13 (34.21) 5 (71.43) 24 (60.00) Surgery Ward Rotation 20 (64.52) 10 (55.56) OB/GYN Ward Rotation 8 (47.06) 22 (68.75) 0.60 (0.10, 3.47) 0.68 (0.21, 2.25) 2.47 (0.74, 8.31) Psychiatry Ward Rotation * 6 (60.00) 24 (61.54) 1.07 (0.19, 5.40) Pediatrics Ward Rotation 19 (59.38) 11 (64.71) 1.25 (0.37, 4.25) Internal Medicine Ward Rotation 22 (61.11) 8 (61.54) 1.02 (0.28, 3.75) Emergency Medicine Ward Rotation 5 (45.45) 25 (65.79) 2.31 (0.59, 9.02) Self-reported exposure to autism from medical school 1.27 (0.40, 4.02) 14 (58.33) 16 (64.00) 10 (41.67) 9 (18.37) Self-reported exposure to autism from medical journal 0.36 (0.11, 1.19) 8 (28.57) 11 (52.38) Self-reported exposure to autism from media (TV, Radio, Internet) 10 (41.67) 9 (37.50) 14 (58.33) 15 (62.50) 20 (71.43) 10 (47.62) 1.19 (0.37, 3.79) 0.07 0.02 0.29 0.2 0.08 0.28 0.23 0.26 0.13 0.21 0.06 0.22 * Exact Confidence Limits 53 4.7 Attitudes and Clinical Practices This last section of the questionnaire examines physician attitudes and practices towards their patients who are presenting with symptoms of ASD. Most physicians agreed that children on the autism spectrum should receive special education (80%, n = 48) and though that autism holds a social stigma in their respective communities (61.7%, 37) (Table 21). GPs in this study informed us that most parents in Pakistan did not think that their children were at risk of having ASD (70%, n = 42). When we asked whether GPs thought that diagnosing a child with autism would lead to discrimination against the child, we got mixed responses. From within this sub-sample, 29 subjects (48.3%) thought their ASD patients were likely to be discriminated against while 28 subjects (46.6%) thought their ASD patients would not face discrimination (Table 21). 54 Table 21 Statements evaluating GPs’ attitudes towards ASD Statements Evaluating GPs’ Attitudes Towards Autism Spectrum Disorders In your opinion should autistic children receive special education? Do you think autism holds a social stigma in this community? Do you believe that diagnosing a child with autism will lead to discrimination against this child? In general, there is a negative opinion towards children diagnosed with autism. Parents in Karachi Pakistan tend to think their children are at risk of autism. Response Strongly Agree Agree Undecided Disagree Strongly Disagree Strongly Agree Agree Undecided Disagree Strongly Disagree Strongly Agree Agree Undecided Disagree Strongly Disagree Strongly Agree Agree Undecided Disagree Strongly Disagree Strongly Agree Agree Undecided Disagree Strongly Disagree Percent Frequency 23 25 2 6 4 13 24 8 12 3 6 23 3 20 8 11 33 4 10 2 0 9 9 27 15 38.3 41.7 3.3 10.0 6.7 21.7 40.0 13.3 20.0 5.0 10.0 38.3 5.0 33.3 13.3 18.3 55.0 6.7 16.7 3.33 0 15.0 15.0 45.0 25.0 55 4.7.1 Open ended questions At the end of the interview and after a series of questions regarding GP’s attitudes and practices towards ASD, I included two open ended questions. The first question asked if the GPs had any general opinions about the prevalence of autism in Pakistan (Table 22). The last question asked GPs if they had any additional comments as we closed out the interview (Table 23). In response to these questions, 68 subjects responded to the first and 16 left a comment for the last question. I read through all the responses and established that there are four categories that encompassed physicians’ thoughts on ASD: increasing general community awareness campaigns in Pakistan (38, 56%), dealing with the social stigma of ASD (5, 7%), improving medical infrastructure to treat ASD (23, 34%), reducing vaccine administration because they may cause ASD (2, 3%), and reducing consanguineous marriages in Pakistan (4, 6%) Table 22. Table 22 Open-ended response categories Open ended response categories Open ended response category Awareness about ASD Reducing the social stigma of ASD Improving medical infrastructure Not giving children vaccines Reducing consanguineous marriages Frequency of coded responses 38 (56%) 5 (7%) 23 (34%) 2 (3%) 4 (6%) 56 Table 23 Selected quoted comments from GPs Quoted comment text Autism is largely unreported. In psychiatric practice since (sic)10 years he has not seen a single autistic child. However, a few days ago parents brought a child to him for information. The doctor though the child was autistic based on behavioral symptoms note -considers this disease to be of affluent nations (Europe, USA) -believes that a disease such as autism does not exist in Pakistan because small nucleated families allow for plenty of interaction 1. Has a grandson who is autistic and is receiving successful speech and 2. Perhaps antibiotics and premature birth of grandson may have played a behavioral therapy in U.K role in his becoming autistic 3. Other two siblings are normal (firborn (sic) was autistic I am a physician, but I am barely getting by because of my wife's medical bills. I work 15 hrs/ day and so do many other GPs in Pakistan. We don't have time to go to CMEs [continuing medical education seminars] Subject 2046 1009 2008 785 57 Chapter 5. Discussion ASD affect millions of people worldwide. Even though we do not understand the etiology of ASD, early intervention has proven to show improvement in social, communication, and behavioral outcomes. The global healthcare community has expressed increasing concern regarding the burden of disease due to ASD worldwide (Wing and Potter 2002, Barbaresi, Katusic et al. 2006). Before this study was conducted, epidemiologists in the United States and other developed countries around the world were establishing the prevalence of ASD to better understand the burden of disease from ASD (Fombonne 2002, Newschaffer, Falb et al. 2005). To better understand the burden of disease of ASD in Pakistan, first, we need to establish how knowledgeable Pakistani general practitioners are about ASD. Then, we must explore their attitudes and clinical practices around the diagnosis of ASD. Only after understanding if Pakistani GPs are able to diagnose or refer an autistic child properly could future research establish the burden of disease of ASD in Pakistan. In regard to mental health, in South Asian countries GPs are the first line of clinical care in a public health setting (Chisholm, James et al. 2000). This is especially true in Pakistan because bigger hospitals are centralized, and the majority of clinical care is provided by neighborhood health practitioners (Uplekar, Pathania et al. 2001). For a large majority of clinical care needs, a patient can get outpatient treatment at a GP’s office and return home to recover. For more intensive clinical care, patients are typically referred to a specialist or a hospital. This path to receiving healthcare is typical even when public services are available (Bennett, McPake et al. 1997). In an ideal scenario, a child on the autism spectrum is diagnosed early in their childhood and then referred to respective treatment programs to receive a 58 customized regimen of care (Al-Qabandi, Gorter et al. 2011). A combination of early diagnosis and treatment can yield an ideal outcome and improved quality of life for a patient with an ASD (Bennett, McPake et al. 1997). On the other hand, a child may end up spending the rest of their life not being able to function in society if they do not receive a diagnosis in time or are misdiagnosed. This is especially true when an early diagnosis results in access to services which may include early intervention (Al-Qabandi, Gorter et al. 2011). In this study sample, the GPs’ self-reported knowledge of the term “autism” was much lower than expected; a majority (n = 184, 55%) of the GPs had never heard of the term, and of the GPs that self-reported knowledge of “autism” (n = 148, 42%), only 29% (n = 49) reported that they had a working knowledge of ASD. A working knowledge of ASD indicated that the GP would be able to diagnose or refer an autistic patient to a specialist. These findings are especially alarming because at the time of these interviews, many countries around the world had widespread public health campaigns built around ASD, and ASD were a major conversation in the global health care community (Fombonne 2002). Physicians in the developed world were establishing and refining diagnostic criteria to properly treat patients with ASD (Newschaffer, Falb et al. 2005). To better gauge Pakistani physicians’ knowledge, attitudes, and practices in regard to ASD, we needed to evaluate how they are learning about ASD, how continuing medical education seminars are conducted, how pharmaceutical companies impact their medical education and lastly, what phases of clinical translation Pakistan is in when it comes to ASD research. 59 5.1 Factors most likely to predict knowledge of the term “autism” For a primary care physician, even knowing the term “autism” can lead to a greater understanding of ASD or lead a GP to refer a patient to a specialist. As an initial approach to understanding physicians, a simple question about a GPs’ knowledge of the term “autism” was incredibly informative about the knowledge transfer of medical information in this healthcare community. As addressed later in this chapter, a GPs’ medical school education was highly predictive of their current knowledge about ASD. As I was modeling the outcome variable of knowing or not knowing the term “autism,” a medical school ward rotation in internal medicine and pediatrics ward rotations were both predictive independent variables. Similarly, the youngest category of physician age groups, 25–29-year-old physicians, were more likely to know of the term. These correlations demonstrate that in this population if a physician was taught about ASD in their medical school curriculum, they were most likely to take that knowledge with them to their respective clinical practice. Both pediatrics and internal medicine ward rotations are more likely to facilitate the topic of mental health, especially when compared to ward rotations in surgery or obstetrics and gynecology. There are two things to consider in regard to a GP’s age being a predictive factor in their general knowledge of ASD: younger physicians might have better recall of a lesser-known topic in regard to medicine, or older physicians may not have been taught about ASD in their curriculum. Surgery was the second most common ward rotation (206, 62.1%) after internal medicine (232, 69.9%). Our study GPs’ medical school interests showcase their preference for surgery and primary care for adults. Their medical school curriculum lacked attention to specialized 60 care for elders, only 1.8% of physicians had a gerontology ward rotation (6, 1.8%) and psychological health with only 13.0% of physicians having a psychiatry ward rotation (43, 13.0%). Psychiatry is a crucial component of a patient’s medical care. Having learned about psychiatry along with primary care in a medical school setting results in holistic medical care for a patient’s needs. In a pre-industrialized nation, the popularity of certain ward rotations informs us where there is a need for health care practitioners in the country. Even though this study sample is comprised of GPs, their interests were relatively homogeneous and skewed toward surgery, pediatrics, and internal medicine. Shifting medical education to encompassed marginalized populations namely patients with specialized psychiatric needs would result in much better public health outcomes. 5.2 Continuing Medical Education and Other Sources of Medical Knowledge When evaluating the medical knowledge for GPs in our study population, there are two primary variables to consider: a GPs’ willingness to keep updated with the latest medical research and modes of access to medical knowledge. From the sub-sample of GPs who had already heard of “autism,” the biggest proportion—79 subjects (25% of the sub-sample)—were exposed to the topic from the media, and only 30 (10%) had read about ASD in a medical journal. If GPs are mainly being exposed to new studies via the media, but then have limited resources to gain a deeper understanding of a new medical topic, it is less likely that they will be able to keep up with the latest research. Healthcare information is communicated in two 61 ways: general information is relayed to lay people via the media, and in parallel, it is disseminated to healthcare professionals via medical journals and continuing medical education seminars (CMEs). In this case, even if a physician had heard about the term “autism,” there was a lack of resources for them to explore the topic further. Medical journals have been moving in the direction of an open-access platform compared to the traditional subscription-based model. However, at the time of this study, the majority of clinical research published in open-access journals had lower impact factors (Björk and Solomon 2012). In industrialized nations, physicians have access to medical journals mainly because the subscription rates are established to accommodate comparatively affluent hospitals and university subscriptions. For example, this study was published in the Journal of Autism and Developmental Disorders; the annual subscription rate for that journal is $199. For a GP in a private practice in Pakistan, subscription rates for multiple medical journals inhibit access to up-to-date medical knowledge. CMEs are another important source of information for GPs to gain knowledge about advances in clinical care and help them learn about new, more beneficial methods of patient care. The objective of CMEs is to discontinue the use of outdated methods of patient care that may cause harm to patients (Bloom 2005). In more developed healthcare systems, GPs are exposed to updated medical education topics through many sources, but CMEs tend to be more comprehensive reviews of specialized medical topics. For example, in the US, these seminars are part of a physician's medical practice and a method for them to stay up-to-date with the latest research. In Pakistan, such seminars are available; 19 (13% of sub-sample) of the physicians knowledgeable of the term “autism” had attended such a seminar, but these CMEs 62 are not mandatory, nor regulated (Vakani, Jafri et al. 2011). Only 9 of these subjects had attended a seminar that was sponsored by a university rather than an external organization; this tells us that there is a chance that CME topics have a bias toward industry motivations. These findings demonstrate that the healthcare system in Pakistan cannot rely on GPs’ medical school training alone; CMEs are imperative to provide the most current medical care for their patients. In this study, lack of access to CMEs and medical journals resulted in GPs practicing treatment solely based on their medical school curriculum, which places an emphasis on only four years of clinical training. Furthermore, medical school was decades ago for some GPs; recent graduates had more exposure to the term “autism” and were subsequently more likely to accurately answer deeper knowledge questions about ASD. In summary, this study found that physicians who graduated before 1990 scored lower on questions with a deep understanding of ASD than recently-graduated GPs. I will explore this finding further in a subsequent section. 5.3 Antiquated knowledge of ASD related topics The knowledge, attitudes, and practices series of questions delved deeper into GPs’ understanding of ASD beyond their self-reported knowledge. This investigation evaluates whether a physician would be able to make an accurate diagnosis and then be able to refer a patient to an appropriate specialist. To better understand the intricacies of how knowledge transfer takes place within a public health framework, we need to first understand clinical translation. Then we can explore a comparative analysis of other healthcare populations where a successful translation has taken place for ASD. This comparison helps establishes a frame of reference to understand where Pakistan is in their clinical translation of ASD. 63 5.4 Clinical Translation Clinical translation is defined as the transfer of knowledge from the research bench to clinical practice (Waldman and Terzic 2010). Clinical practice often lags behind cutting edge clinical research. Even in industrialized nations, there is a heavy emphasis on reducing this lag between new discoveries and changes in clinical practice (Waldman and Terzic 2009). Clinical translation is delayed when there are barriers in communicating the most updated research with the clinical care community at large. These barriers are further amplified when the lay media is not making the general population aware of a new discovery. As addressed in the background section, the US healthcare system went through its own clinical translation for ASD research in the 1950s and 1960s (Schopler 1978, DeMyer, Hingtgen et al. 1981). It took many years for the lay media to disabuse antiquated theories about ASD. Some would argue that the US is still in this transitional phase when it comes to best explaining ASD to the general public (Rai, Lewis et al. 2012). Learning from the transition of knowledge about ASD in the US, we applied some of the same antiquated ideologies to the questions we asked Pakistani GPs. Physicians in this study sample showed a poor understanding of the epidemiology of ASD even when they had self-reported having a working knowledge of ASD. Follow-up questions for physicians who reported having a working knowledge of ASD were comprised of a series of statements. Some of these statements were correct and some were incorrect. These knowledge questions revealed that GPs had outdated knowledge about the intricacies of diagnosing and referring patients on the autism spectrum. There were 8 statements to test deeper knowledge about ASD; on average, subjects answered correctly 56% the time. From 64 within this sub-sample of GPs who reported having a working knowledge of ASD, almost half, 47% (15, n = 36), thought that autism was a precursor for schizophrenia, and 48% (22, n=46) of the subjects thought that autism is due to neglect in early childhood. Both of these statements about autism demonstrate that a noteworthy proportion of physicians are working based on outdated knowledge about ASD. A significant proportion of subjects, 40% (20), thought that autistic children tend to mature or grow out of autism; such a medical opinion could be obstructive to early diagnosis of ASD and prevent a patient from getting treatment. To better understand the characteristics of GPs who were scoring lower on these knowledge questions, I developed a score to quantify how well physicians were doing and to evaluate other characteristics of physicians who were scoring low. Even though it was marginally significant in this case, physicians who graduated after 1990 were 4.2 times as likely to correctly answer 5 question out of 8 (95% CI [1.00, 21.37], p-value 0.02). Lack of confidence about such a disorder leads to lack of diagnoses, misdiagnoses and a reduced number of appropriate referrals. If at the point of first contact with a healthcare professional, a child with ASD is turned away or misdiagnosed, it is less likely that they will get proper treatment throughout their lifetime. Even though the etiology for ASD is not well understood, there are many treatment options for children if they are diagnosed properly. This study demonstrates that a substantial size of the represented GPs are not adequately treating their patients who present with ASD. 5.5 Pharmaceutical Companies As detailed in the methods chapter, in the planning stages of this study I needed an established database of GPs working in Pakistan. The most comprehensive database is 65 maintained by the Government of Pakistan and keeps records of the physicians who are registered. This database was not available to our study staff, but the second-best option was a database maintained by pharmaceutical companies in Karachi. From an early stage, it was obvious that pharmaceutical representatives were the primary source of education for a lot of physicians working in Karachi; while conducting our interviews, the waiting rooms were often crowded with pharmaceutical representatives. These pharmaceutical representatives are in GPs’ offices to sell medications and provide updated pharmaceutical research. Often GPs’ only contact with recent research is through these pharmaceutical representatives and if a pharmaceutical company is not investing in making drugs for a specific disease, the GPs are less likely to find out about it. In this study sample, only 4 subjects had attended a pharmaceutical sponsored-seminar about ASD. 5.6 Bimodal distribution of wealth and health Pakistan is a pre-industrialized nation with a bimodal distribution of wealth. As we set out to conduct an observational study looking at how the general population of Pakistan is being treated, we learned that we were investigating two different healthcare systems: one serving the small, higher socioeconomic classes, and another healthcare system serving the much larger lower socioeconomic classes. Access to healthcare resources and care are very different when comparing the poorest quintile to the riches (Alam, Nishtar et al. 2010). Taking prenatal care as an example, women in the “richest quintile had a 35%, 38% and 20% higher probability of getting prenatal care, delivery by skilled provider and emergency obstetric care, respectively” (Alam, Nishtar et al. 2010). 66 When establishing a representative sample for this study, it was important to have representation from GPs serving the lower socioeconomic classes and GPs working at larger hospitals with a well-defined referral system. Our study sample was representative of large government hospitals, private hospitals, and private clinics. Physicians working in private hospitals have a well-established referral system, as well as more specialists who can treat referred patients. Not only is medical care for patients different amongst these two income brackets, but the opportunities for CMEs are very different. In other words, physicians who have access to bigger hospitals also have access to medical journals, lectures, specialists who are experts in a field, and a well-established referral system. 5.7 Findings from fieldwork This study was the first project to investigate the Knowledge, Attitudes, and Practices (KAP) of ASD amongst health care workers in Pakistan. KAP studies are often the first in line of an observational research approach to better understand a problem dealing with the delivery of clinical care. The KAP study’s theoretical framework contributes to the knowledgebase to better understand a sub-population, helps generates questions for future research and also establishes research methods for conducting observational studies in a new setting. As expected, this study generates more questions than it answers. Conducting health survey research in Karachi posed several barriers and a successful execution to these barriers required that our team find creative timely solutions. I have listed these barriers and my solutions to conducting this study in all towns of the biggest metropolis in Pakistan. 67 5.7.1 Geographical Constraints Locating physicians’ offices in rural or inner-city urban city settings without the help of accurate maps or addresses was difficult. Often the doctors’ offices in our database only had cultural and geographical landmarks to identify the location and these offices were usually found on unmarked roads. Such constraints often keep researchers from doing work within inner-city neighborhoods. I overcame this barrier by hiring study staff who were knowledgeable about different towns within Karachi. I also worked closely with a driver who was able to get access to city streets by talking fluently in several local languages. 5.7.2 Logistical barriers When I arrived on-site, in Pakistan, I advertised this research opportunity at medical school campuses with a project description and a call for medical students to apply for research staff positions. Logistically, it was difficult to manage the availability of study staff to match our schedule for conducting subject interviews. After the study staff was hired, I conducted weekly staff meetings to coordinate staff and to tend to difficulties in data collection. These meetings were imperative in successfully being able to enroll subjects for this study. 5.7.3 Cultural congruence of instrument Before I launched this study, I was aware that our questionnaire needed to have language that was culturally congruent with the study population. As described in the methods section, the instrument was pilot tested with Pakistani physicians. After launching this study, I learned that a successful study was also dependent on some less obvious cultural practices. For example, many female GPs only preferred speaking with and treating female patients. Two of 68 our study staff members were women and I made sure they were conducting the interviews with female physicians. 5.7.4 Incentive for physicians GPs in our study sample were giving us precious time out of their day so we could administer these questionnaires. In exchange for their time, we needed to give them a token of appreciation and some incentive for them to work with us. The IRB alerted us that this incentive could not be coercive, in other words, the incentive could not be so expensive that a physician would be forced to work with us. In the end, we commission a leather bag designer to make custom leather bookmarks that said “Pakistan Autism Project” in gold foil. These bookmarks were wrapped and presented to our study subjects at the end of the interview. We also hoped that after we left the physician, they would use these bookmarks as a reminder to become more aware of ASD. 5.8 Attitudes and Practices We have spent all of this time evaluating physicians’ knowledge and delving into their understanding of topics related to ASD, but a huge consideration in the execution of clinical care is a physician’s attitude or general perceptions of a specific disorder. In this case, even though many of our physicians believed that there needs to be special education for patients with ASD, there were split responses as to whether a child would be discriminated against if diagnosed an ASD. This could lead to physicians intentionally not referring a patient to a specialist out of fear of what could happen to the child in their respective communities. Physicians also alerted us that parents weren’t concerned that their child was on the autism 69 spectrum; in other words, most of the patients they were treating came in presenting other primary care needs. Lastly, as described in the background section, Pakistan is a pre- industrialized nation. After finishing an interview with one physician, he stated that if he is spending most of his day treating cholera, he is less concerned about non life-threatening disorders. These attitudes and practices are important factors when evaluating a healthcare system different than in the United States. 5.9 Human Subjects Review IRB approvals were obtained from Michigan State University (MSU) and from the Ethics Board at Aga Khan University (AKU), and the University of Texas Health Science Center at Houston (UTHSC-H). 5.10 Limitations As described in the methods section, the sample was not randomized. Even though, I went through due diligence in acquiring a representative sample it is possible that the bias was introduced when identifying physicians for this study. Secondly, this study was based on an interview that relied on self-reporting knowledge about a topic. There could be recall bias in regard to some of the independent variables in our model. Furthermore, it is difficult to evaluate truthfulness of physicians who responded to follow up questions in regard to their self-reported knowledge of the term “autism.” Even though, study team members used written scripts to conduct this interview different members of the study team visited assigned parts of the city This may have caused inconsistency in delivery of the interview. Taking these issues into account, many of the 70 interviews were conducted by me and strict supervision was provided for each study member. We also had weekly study team meetings throughout the interviewing phase of the study. 71 APPENDIX 72 M/F YYYY QUESTIONNAIRE: 1. Demographic Information Q1.1) Year or birth: Q1.2) Sex: Q1.3) Birthplace: Q1.4) Place of Residence for Most of Childhood: Q1.5) On average, how many children (aged 2-12 y) are treated in your practice on a weekly basis? Q1.6) Generally, from what setting do your patients come? Rural / urban Q1.7) In the city of Karachi, in which town is this clinic / hospital: Q1.8) What is the name and location of the medical school you attended? Q1.9) What year did you graduate from medical school? YYYY Q1.10) What was your ward rotation during medical school? City, Country NUM 1. Surgical 2. OB/GYN 3. Psychiatry 4. Pediatrics 5. Internal medicine 6. Geriatrics 7. Emergency Medicine 8. Other ________ 1. Military a. 0% - 25% b. 25% - 50% c. 51% - 75% Q1.11) Did you receive any further training post medical school? Q1.12) How long have you been practicing? (In Years) Q1.13) What percentage of your patients have their medical care funded by the following sources 73 d. 76% - 100% 2. Private insurance a. 0% - 25% b. 25% - 50% c. 51% - 75% d. 76% - 100% 3. Personally funded a. 0% - 25% b. 25% - 50% c. 51% - 75% d. 76% - 100% 4. Government Sponsored 5. other: ____________ a. 0% - 25% b. 25% - 50% c. 51% - 75% d. 76% - 100% a. 0% - 25% b. 25% - 50% c. 51% - 75% d. 76% - 100% Q1.14) Are you on the medical care panel of a specific company/organization? IF YES Approximately, how many of your patients are from this company/organization? HRS Q1.15) On a typical day, how many hours do you practice? Q1.16) Although in general practice, do you have a current field of personal interest in medicine? 1)) Surgical 2) General Practice 3) OB/GYN Q1.17) How many patients (children and adults) do you see on a typical day? Q1.18) On average, how much time do you spend with each patient? 7) Geriatrics 8) Emergency Medicine 9) Other 4) Psychiatry 5) Pediatrics 6) Internal Medicine HRS NUM 74 2. Knowledge & Exposure to information about ASD 2(a) Exposure to information about ASD Q2.1)Have your heard of the term “Autism”? IF YES Q2a.2)At medical school, were you exposed to the topic of Autism or Autism Spectrum Disorders (ASD)? (Y/N)__ Q2a.3)Have you read about Autism in medical journals? (Y/N)__ Q2a.3)Have you been contacted by any national or international organizations regarding autism or ASD? (Y/N)__ If YES, which organization _________. Q2a.3)Have you been exposed to autism or ASD via national or international media? (Y/N)__ If YES, which mode of communication a. Television b. Print Media c. Radio d. Internet e. Other: Please specify ________ Q2a.4)Do you have an estimate of prevalence rates of autism in the US? (Y/N)__ If YES, what is that estimate? ___ / 100,000 Q2a.5)What about the prevalence of autism in children worldwide? (Y/N)__ If YES, what is that estimate? ___ / 100,000 Q2a.6)Out of every hundred children being treated by you, how many of them are suspected to have autism? Q2a.7)Have you been exposed to autism at a symposium or seminar? If YES, exactly who organized it: a. Government organized_________ b. Pharmaceutical company_________ c. University_________ d. NGO_________ e. Other, please specify:_________ Q2a.8) Did this source (s) give you a working knowledge about autism? If YES, Next section If NO, END 75 Q2a.1) Have you been exposed to a disorder that fits this description? If YES, Next question If NO, END Q2a.2)At medical school, were you exposed to the topic of such a disorder (ASD)? (Y/N)__ Q2a.3)Have you read about Autism in medical journals? (Y/N)__ Q2a.4)Have you been contacted by any national or international organizations regarding autism or ASD? (Y/N)__ If YES, which organization _________. Q2a.5)Have you been exposed to autism or ASD via national or international media? (Y/N)__ If YES, which mode of communication f. Television g. Print Media h. Radio i. j. Other: Please specify ________ Internet IF NO (PROBE: “Until the very recently there has been little research about autism. In fact, we still do not have a cure for autism. You are not alone if you have not heard of this disorder. “ DEFINE AUTISM: Autism is a disorder of brain function that appears early in life, generally after the age of two. Children with autism have problems with social interaction, communication, imagination and behavior. Autistic traits persist into adulthood, but vary in severity. Some adults with autism function well, earning college degrees and living independently. Others never develop the skills of daily living, and may be incorrectly diagnosed with a variety of psychiatric illnesses. The causes of autism are unknown. ) Q2a.6)Do you have an estimate of prevalence rates of autism in the US? (Y/N)__ If YES, what is that estimate? ___ / 100,000 Q2a.7)What about the prevalence of autism in children worldwide? (Y/N)__ If YES, what is that estimate? ___ / 100,000 Q2a.8)Out of every hundred children being treated by you, how many of them are suspected to have autism? Q2a.9)Have you been exposed to autism at a symposium or seminar? 76 If YES, exactly who organized it: f. Government organized_________ g. Pharmaceutical company_________ h. University_________ i. NGO_________ j. Other, please specify:_________ Q2a.10) Did this source (s) give you a working knowledge about autism? If YES, Next section If NO, END 77 2 (b) Knowledge Answer the following questions from 1 (strongly agree) to 5 (strongly disagree) (1) --------------- (2) ---------------(3) --------------- (4) --------------- (5) Strongly Agree Agree Undecided Disagree Strongly Disagree Q2b.1) Autistic children show detachment from their parents Q2b.2) Autism is more prevalent in higher socioeconomic classes Q2b.3) Autism is more prevalent in higher educational classes Q2b.4) Autistic children are not affectionate Q2b.5) Autism a precursor for schizophrenia Q2b.6) It is hard to distinguish between autism and schizophrenia Q2b.7) I have a high level of accuracy in diagnosing autism spectrum disorders Q2b.8) Autism in children is mostly attributed to neglect in early childhood parenting. Q2b.9) Children tend to psychologically mature and grow out of autism 3. Attitudes Answer the following questions from 1 (strongly agree) to 7 (strongly disagree) (1) --------------- (2) ---------------(3) --------------- (4) --------------- (5) Strongly Agree Agree Undecided Disagree Strongly Disagree 1) In your opinion should autistic children receive special education? 2) Do you think autism holds a social stigma in this community? 3) Do you believe that diagnosing a child with autism will lead to discrimination against this child? 4) In general, there is a negative opinion towards children diagnosed with autism. 5) Parents in Karachi Pakistan tend to think their children are at risk of autism. 6) Autism is preventable. 4. Practices Q4.1) How do you diagnose Autism Spectrum Disorders? a. Use ADI-R or ADOS b. General evaluation c. Parents are given a questionnaire to fill out and note observation d. DSM IV criteria e. Other, specify _________ Q4.2) Please list some early indicators of autism in a 2 year old child Q4.3)In Pakistan, are there any early diagnosis questionnaires in place for parents of autistic children? (Y/N)___ Q4.4) In diagnosing children with autism are the following symptom necessary, not necessary but helpful, and not helpful in diagnosis (1)-----------------------------------(2)--------------------------------(3) Helpful but not necessary Not Necessary Necessary Impaired social interaction Impaired communication Hallucinations Restricted and repetitive behavior Lack of eye contact Confirmed Schizophrenia 78 Hearing voices Language Disturbances Hypersensitivities to certain environment Depression Anxiety Q4.5) What other diagnostic behaviors do you use in your assessments? Q4.6) What in your opinion can be done in decrease the prevalence of autism in Pakistan? 79 BIBLIOGRAPHY 80 BIBLIOGRAPHY Afzal, U. and A. Yusuf (2013). "The State of Health in Pakistan: An Overview." The Lahore Journal of Economics 18: 233–247. Al-Qabandi, M., J. W. Gorter and P. Rosenbaum (2011). "Early Autism Detection: Are We Ready for Routine Screening?" Pediatrics 128(1). Alam, A. Y., S. Nishtar, S. Amjad and K. M. Bile (2010). "Impact of wealth status on health outcomes in Pakistan." Eastern Mediterranean Health Journal 16: S152-158. Allen, C. W., N. Silove, K. Williams and P. Hutchins (2007). "Validity of the Social Communication Questionnaire in Assessing Risk of Autism in Preschool Children with Developmental Problems." Journal of Autism and Developmental Disorders 37(7): 1272-1278. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV). Washington, DC, American Psychiatric Publishing. American Psychiatric Association (2000). Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, Text Revision (DSM-IVTR). Washington, DC, American Psychiatric Publishing. American Psychiatric Association. (1980). Quick reference to the diagnostic criteria from DSM- III. Washington, D.C., The Association. American Psychiatric Association. and American Psychiatric Association. DSM-5 Task Force. (2013). Diagnostic and statistical manual of mental disorders DSM-5. Washington, D.C., American Psychiatric Association,: xliv, 947 p. Art., D. o. A. N. E. (2000). "The Achaemenid Persian Empire (550–330 B.C.)." In Heilbrunn Timeline of Art History., 2016, from http://www.metmuseum.org/toah/hd/acha/hd_acha.htm. Asperger, H. (1944). " Die „Autistischen Psychopathen” im Kindesalter." European Archives of Psychiatry and Clinical Neuroscience 117(1). Autism Speaks Inc. (2010). "How is Autism Treated?", from https://www.autismspeaks.org/sites/default/files/documents/100-day- kit/treatment_version_2_0.pdf. Barbaresi, W. J., S. K. Katusic and R. G. Voigt (2006). "Autism: A Review of the State of the Science for Pediatric Primary Health Care Clinicians." Arch Pediatr Adolesc Med. 160(11): 1167- 1175. 81 Bartel, N. (1999). "Ibn Battuta's Trip: Part Seven - Delhi, capital of Muslim India." The Travels of Ibn Battuta - A Virtual Tour with the 14th Century Traveler, from https://web.archive.org/web/20100612001214/http://www.sfusd.k12.ca.us/schwww/sch618/I bn_Battuta/Battuta's_Trip_Seven.html. Baxter, A. J., T. S. Brugha, H. E. Erskine, R. W. Scheurer, T. Vos and J. G. Scott (2015). "The epidemiology and global burden of autism spectrum disorders." Psychological Medicine 45(03): 601-613. Bellini, S. (2004). "Social skill deficits and anxiety in high-functioning adolescents with autism spectrum disorders." Focus on Autism and Other Developmental Disabilities 19(2): 78-86. Bennett, S., B. McPake and A. Mills (1997). Private health providers in developing countries : serving the public interest? London ; Atlantic Highlands, N.J., Zed Books. Berendes, S., P. Heywood, S. Oliver and P. Garner (2011). "Quality of Private and Public Ambulatory Health Care in Low and Middle Income Countries: Systematic Review of Comparative Studies." PLOS Medicine 8(4): e1000433. Björk, B.-C. and D. Solomon (2012). "Open access versus subscription journals: a comparison of scientific impact." BMC Medicine 10(1): 73. Bloom, B. S. (2005). "Effects of continuing medical education on improving physician clinical care and patient health: A review of systematic reviews." International Journal of Technology Assessment in Health Care 21(3): 380-385. Brentani, H., C. S. Paula, D. Bordini, D. Rolim, F. Sato, J. Portolese, M. C. Pacifico and J. T. McCracken (2013). "Autism spectrum disorders: an overview on diagnosis and treatment." Rev Bras Psiquiatr 35 Suppl 1: S62-72. CDC (2003). Questionnaire Design. Reproductive Health Epidemiology Series. CDC. Atlanta, GA, CDC. Module 4. Central Intelligence Agency (2007). Map of Pakiistan. M. o. Pakistan, Central Intellegence Agency. 704.6KB. Chisholm, D., S. James, K. Sekar, K. K. Kumar, R. S. Murthy, K. Saeed and M. Mubbashar (2000). "Integration of mental health care into primary care: Demonstration cost–outcome study in India and Pakistan." British Journal of Psychiatry 176(6): 581-588. Conner, M. G. (1999). "Criticism of America's Diagnostic Bible - The DSM." Retrieved November, 2006, from http://www.oregoncounseling.org/Diagnosis/CriticismOfDSM.htm. Constantino, J. N. and C. P. Gruber (2012). Social responsiveness scale : SRS-2. Torrane, California, Western Psychological Services. 82 Cooper, J. O., T. E. Heron and W. L. Heward (2007). "Applied behavior analysis." D'Souza, R. and J. Bryant (1999). "Determinants of Childhood Mortality in Slums of Karachi, Pakistan." Journal of Health and Populations in Developing Countries 2(1): 33-44. Daley, T. C. (2002). "The Need for Cross-Cultural Research on the Pervasive Developmental Disorders." Transcultural Psychiatry 39(4): 531-550. de Bildt, A., S. Sytema, C. Ketelaars, D. Kraijer, E. Mulder, F. Volkmar and R. Minderaa (2004). "Interrelationship between Autism Diagnostic Observation Schedule-Generic (ADOS-G),Autism Diagnostic Interview-Revised (ADI-R), and the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) Classification in Children and Adolescents with Mental Retardation." Journal of Autism and Developmental Disorders 34: 129-137. DeMyer, M. K., J. N. Hingtgen and R. K. Jackson (1981). "Infantile Autism Reviewed: A Decade of Research." Schizophrenia Bulletin 7(3): 388-451. Directorate of Intelligence. (2010, March 4th 2010). "CIA World Fact Book." Retrieved March 2010, from https://www.cia.gov/library/publications/the-world- factbook/rankorder/2119rank.html. Dyches, T. T. (2004). "Multicultural issues in autism." Journal of autism and developmental disorders 34(2): 211-222. Eisenberg, L. and L. Kanner (1956). "Early infantile autism, 1943 - 1955." American Journal of Orthopsychiatry. Fombonne, E. (2002). "Epidemiological trends in rates of autism." Molecular Psychiatry 7: S4. Frith, U. (1991). Autism and Asperger Syndrome, Cambridge University Press. Ghaffar, A., B. M. Kazi and M. Salman (2000). "Health care systems in transitions III. Pakistan, Part I. An overview of the health care system in Pakistan." Journal of Public Health Medicine 22(1): 38-42. Ghaziuddin, M. (2005). Mental Health Aspects of Autism and Asperger Syndrome. Philadelphia, Jessica Kingsley Publishers. Gillott, A., F. Furniss and A. Walter (2001). "Anxiety in high-functioning children with autism." Autism 5(3): 277-286. Green, W. H., M. Padron-Gayol, A. S. Hardesty and M. Bassiri (1992). "Schizophrenia with childhood onset: A phenomenological study of 38 cases." Journal of the American Academy of Child and Adolescent Psychiatry 31: 968 - 976. 83 Higher Education Commission. (2012, 03/14/2012). "Quality and Research based Ranking of Pakistani HEIs." Retrieved 10/11/2014, 2014, from http://www.hec.gov.pk/INSIDEHEC/DIVISIONS/QALI/OTHERS/RANKINGOFUNIVERSITIES/Pages/ CategoryWise.aspx#. Imran, N. and M. W. Azeem (2014). Autism Spectrum Disorders: Perspective from Pakistan. Comprehensive Guide to Autism. B. V. Patel, R. V. Preedy and R. C. Martin. New York, NY, Springer New York: 2483-2496. Jayapalan, N. (2001). Economic history of India : ancient to present day. New Delhi, Atlantic. Johnson, C. and S. Myers (2007). "American Academy of Pediatrics Council on Children With Disabilities. Identification and evaluation of children with autism spectrum disorders. ." Pediatrics 120(5): 1183–1215. Johnson, C. P., S. M. Myers and Council on Children With Disabilities (2007). Identification and Evaluation of Children With Autism Spectrum Disorders, American Academy of Pediatrics. Kachru, B. B., Y. Kachru and S. N. Sridhar (2008). Language in South Asia, Cambridge University Press. Kanner, L. (1943). "Autistic Disturbances of Affective Contact." Nervous Child 2: 217 - 250. Khan, A. J. (2004). "Scope of medical colleges in private sector." J Ayub Med Coll Abbottabad 16(1): 1-3. Khowaja, K. (2009). "Healthcare systems and care delivery in Pakistan." J Nurs Adm 39(6): 263- 265. Kolvin, I. (1971). "Studies in childhood psychoses I: Diagnostic criteria and classification." British Journal of Psychiatry 118: 381 - 384. Lord, C., S. Risi, L. K. Lambrecht, E. H. Cook, B. L. Leventhal, P. C. DiLavore, A. Pickles and M. Rutter (1999). "The Autism Diagnostic Observation Schedule – Generic: A standard measure of social and communication deficits associated with the spectrum of autism." Journal of Autism and Developmental Disorders 30: 205–223. Lord, C., M. Rutter and A. Le Couteur (1994). "Autism Diagnostic Interview – Revised: A revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders." Journal of Autism and Developmental Disorders 24: 659–685. Matson, J. L. and A. M. Kozlowski (2011). "The increasing prevalence of autism spectrum disorders." Research in Autism Spectrum Disorders 5(1): 418-425. Medi Publications: House of Medical Publications (2006). Mediguide: Comprehensive Medical Directory (2004-2005). S. M. S. Yunus. Karachi, Pakistan. 84 Ministry of Health (1995). Situation analysis of health sector in Pakistan. Islamabad, Government of Pakistan. Mushtaq, M. U. (2009). "Public Health in British India: A Brief Account of the History of Medical Services and Disease Prevention in Colonial India." Indian Journal of Community Medicine : Official Publication of Indian Association of Preventive & Social Medicine 34(1): 6-14. Newschaffer, C., M. Falb and J. Gurney (2005). "National Autism Prevalence Trends From United States Special Education Data." Pediatrics 115(3). Omran, A. R. (1971). "The Epidemiologic Transition: A Theory of the Epidemiology of Population Change." The Milbank Memorial Fund Quarterly 49(4): 509-538. Petty, L. K., E. M. Ornitz, J. D. Michelman and E. G. Zimmerman (1984). "Autistic children who become schizophrenic." Archives of General Psychiatry 41: 129-135. Rai, D., G. Lewis, M. Lundberg, R. Araya, A. Svensson, C. Dalman, P. Carpenter and C. Magnusson (2012). "Parental Socioeconomic Status and Risk of Offspring Autism Spectrum Disorders in a Swedish Population-Based Study." Journal of the American Academy of Child & Adolescent Psychiatry 51(5): 467-476.e466. Reaven, J., A. Blakeley-Smith, K. Culhane-Shelburne and S. Hepburn (2012). "Group cognitive behavior therapy for children with high-functioning autism spectrum disorders and anxiety: A randomized trial." Journal of Child Psychology and Psychiatry 53(4): 410-419. Reaven, J. A., S. L. Hepburn, R. G. Ross and (2008). "Use of the ADOS and ADI-R in Children with Psychosis: Importance of Clinical Judgment." Clinical Child Psychology and Psychiatry 13: 81-94. Robins, D., D. Fein, M. Barton and J. Green (2001). "The Modified Checklist for Autism in Toddlers: an initial study investigating the early detection of autism and pervasive developmental disorders. ." Autism Dev Disord. 31(2): 131-144. Robins, D. L. and T. M. Dumont-Mathieu (2006). "Early Screening for Autism Spectrum Disorders: Update on the Modified Checklist for Autism in Toddlers and Other Measures." Journal of Developmental & Behavioral Pediatrics 27(2): S111-S119. Russell, A. T. (1994). "The Clinical Presentation of Childhood-onset Schizophrenia." Schizphrenia Bulletin 20(4): 631-646. Schopler, E. (1978). "On confusion in the diagnosis of autism." Journal of autism and childhood schizophrenia 8(2): 137-138. Tchaconas, A. and A. Adesman (2013). "Autism spectrum disorders: a pediatric overview and update." Curr Opin Pediatr 25(1): 130-144. 85 Uplekar, M., V. Pathania and M. Raviglione (2001). "Private practitioners and public health: weak links in tuberculosis control." The Lancet 358(9285): 912-916. Vakani, F. S., W. Jafri, A. Amin and M. Sheerani (2011). "Continuing medical education and pharmaceutical industry." Journal of the College of Physicians and Surgeons Pakistan 21(6): 360-361. Waldman, S. A. and A. Terzic (2009). "Translational Medicine in the Era of Health Care Reform." Clinical and Translational Science 2(2): 96-97. Waldman, S. A. and A. Terzic (2010). "Clinical and Translational Science: From Bench-Bedside to Global Village." Clinical and Translational Science 3(5): 254-257. Wikimedia Commons contributors (8 May 2016 ). Map of Karachi. Wing, L. and D. Potter (2002). "The epidemiology of autistic spectrum disorders: is the prevalence rising?" Ment Retard Dev Disabil Res Rev 8: 151-161. Wink, A. (2002). Al-Hind: Early medieval India and the expansion of Islam, 7th-11th centuries Brill. Wood, J. J., A. Drahota, K. Sze, K. Har, A. Chiu and D. A. Langer (2009). "Cognitive behavioral therapy for anxiety in children with autism spectrum disorders: A randomized, controlled trial." Journal of Child Psychology and Psychiatry 50(3): 224-234. Woodbury-Smith, M. R., K. Boyd and P. Szatmari (2010). "Autism spectrum disorders, schizophrenia and diagnostic confusion." Journal of Psychiatry & Neuroscience : JPN 35(5): 360- 360. World Bank (1993). Health Sector Study. Islamic Republic of Pakistan: key concerns and solutions. Washington, DC, World Bank: 65-80. World Health Organization. (2010). "WHO Recommendations on the Diagnosis of HIV Infection in Infants and Children. ." ANNEX 4, Characteristics of a screening test., from https://www.ncbi.nlm.nih.gov/books/NBK138555/. Zwaigenbaum, L., M. L. Bauman, D. Fein, K. Pierce, T. Buie, P. A. Davis, C. Newschaffer, D. L. Robins, A. Wetherby, R. Choueiri, C. Kasari, W. L. Stone, N. Yirmiya, A. Estes, R. L. Hansen, J. C. McPartland, M. R. Natowicz, A. Carter, D. Granpeesheh, Z. Mailloux, S. Smith Roley and S. Wagner (2015). "Early Screening of Autism Spectrum Disorder: Recommendations for Practice and Research." Pediatrics 136(Supplement 1): S41. Zwicker, J. D. and J. C. Emery (2014). "Autism research funding allocation: can economics tell us if we have got it right?" Autism Res 7(6): 704-711. 86