THS- THESKB Joell LIBRARY Michigan State University This is to certify that the thesis entitled v/V' Chorioamniqnitiscand Cerebral Palsy . ‘. sL ”up“! .. w l ' ' ~ ’ presented by Sung Lee has been accepted towards fulfillment of the requirements for Master's . E idemiolo degree 1n p gy \_— 9-7 Major professor Date July 34 2001 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution PLACE IN RETURN BOX to remove this checkout from your record. ‘TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 6/01 cJClRC/DateDuopes-pts CHORIOAMNIONITIS AND CEREBRAL PALSY Sung G Lee A THESIS Submitted to Michigan State University In partial fulfillment of the requirements For the degree of MASTER OF SCIENCE Department of Epidemiology 200 l ABSTRACT CHORIOAMNIONITIS AND CEREBRAL PALSY By Sung G Lee Preterm infants (less than 37 gestational weeks) constitute more than one third of all new cases of cerebral palsy (CP). Seven studies were reviewed with a meta-analysis to evaluate the association between chorioamnionitis and CP. Unlike a previous meta-analysis by Wu, bias from misclassification was minimized and homogeneity was achieved by dividing articles into two groups based on the method of exposure ascertainment: clinical chorioamnionitis and clinicians’ impression of chorioamnionitis. Clinical chorioamnionitis was positively associated with CP (OR=1.79, 95% CI 0.96-3.36). Clinicians’ impression of chorioamnionitis showed a strong association with CP (OR=3.83, 95% CI 2.27-6.45). The association between chorioamnionitis and CP in the New Jersey Neonatal Brain Hemorrhage (NBH) data was tested in a prospective study design. Comparison was made between two cohorts of infants divided according to gestational age above and below 31 weeks. Among four definitions of clinical chorioamnionitis used in the recent literatures, the one most strongly associated with CP was discharge diagnosis of chorioamnionitis in the cohort of infants above 31 weeks (OR=6.61, 95% CI 1.61-27 .21), while abruptio placenta played a more important role in infants below 31 weeks afier controlling for infection, race, route of delivery, SGA, and preterm labor (OR=4.24, 95%CI 1.34-13.47). ACKNOWLEDGMENTS I thank my advisors, Dr. Paneth, Dr. Collins, and Dr. Holzman. I owe a huge debt to my colleague and my mentor, Dr. Paneth, who keeps encouraging me to think like an epidemiologist in my specialty of Ob-Gyn. Later in my training Dr. Collins inspired me that I could continue my work as a researcher and a clinician. Dr. Holzman introduced me to the critical thinking. I pay my sincere thanks to all teachers of my last six years. Without them I would not be able to come out of my small box and see the new vista that is so beautifirl and challenging. To Ms. Lora McAdams, Linda Fortin, and Geri Ziolkowski, I am thanlcful for their help in my 6 years as a student. I thank Dr. Hong Qiu, Ms. Madeleine Lenski and Mr. James Jetton for their help with NBH data. To Ms. Tammy Salman, Bonnie Hissong, Michelle Garn, and Joy DeBoer I owe much because without them I could not have been able to attend classes and see patients at the same time. To my wife, Sook Young, I owe more than I can express for her encouragement, kind advices, and patience. My parents instilled the dream when I was a kid that I could do whatever I want to do in my life. My dad trusted me in this more than anybody. iii TABLE OF CONTENTS ABSTRACT--- -- - - - ............. - - - - -- - -- - ------ II ACKNOWLEDGMENTS ...... - - - - ..... III GLOSSARY ...... - _ - - - - -- - VIII PART ONE -- 4 A SYSTEMATIC LITERATURE REVIEW: META-ANALYSIS - - 4 METHODS .................................................................................................................... 4 Study Identification .................................................................................................. 4 Study Selection ........................................................................................................ 5 Data Extraction ....................................................................................................... 8 STATISTICAL ANALYSIS: ............................................................................................... 8 DESCRIPTION OF STUDIES .............................................................................................. 8 RESULTS .................................................................................................................... 13 1) Association between chorioamnionitis and CP in preterm infants in 7 selected studies ................................................................................................................... 13 2) Association between clinical chorioamnionitis and CP in preterm infants in 4 selected studies ...................................................................................................... 14 PART TWO -- - - - - - - -- - l7 NBH DATA. - - - - . - ------ - --------------- --17 MATERIALS AND METHODS ........................................................................................ 17 iv Study Population ................................................................................................... 1 7 F allow-up Examination ......................................................................................... 18 Study Design and Data Analysis ............................................................................ 19 RESULTS .................................................................................................................... 20 1. Univariate analysis of risk factors for cerebral palsy: ........................................ 20 2. Univariate analysis of risk factors for chorioamnionitis: .................................... 24 3. Univariate analysis in Multiple Pregnancy ........................................................ 2 7 4. The association between DCP and 4 dtflerent definitions of chorioamninitis ...... 29 5. Multivariate analysis: Determinants of chorioamnionitis ................................... 31 6. Multivariate Analysis: Predictors of Disabling CP ............................................ 34 META-ANALYSIS wrm NBH DATA ADDED ............................................................ 37 1) Studies dealing with clinical chorioamnionitis plus NBH data-below 31 wee/cs group ..................................................................................................................... 37 2) Studies dealing with clinical chorioamnionitis plus NBH data-above 31 weeks group ..................................................................................................................... 38 3) Studies dealing with other chorioamnionitis and NBH data based on discharge diagnosis of chorioamnionitis-below 31 weeks group ............................................ 39 4) Studies dealing with other chorioamnionitis and NBH data with discharge diagnosis of chorioamnionitis-above 31 weeks group ............................................. 40 DISCUSSION... - - - - - -------- - - - - - 42 FUTURE RESEARCH DIRECTION: - 48 BIBLIOGRAPHY - - - -- - ----------------- ---49 LIST OF TABLES Table 1: Publications excluded from the meta-analysis ........................................... 6 Table 2: Studies selected for the meta-analysis ...................................................... 7 Table 3: Characteristics of seven studies comparing chorioamnionitis and CP ........ 10 Table 4: Diagnostic criteria of clinical chorioamnionitis in each study .................... 12 Table 5: Univariate analysis of association of variables with DCP .......................... 22 Table 6: Univariate analysis of association of variables with chorioamnionitis ....... 25 Table 7: Univariate analysis of risk factors with DCP in multiple pregnancy .......... 27 Table 8: The association between CP and 4 definitions of chorioamnionitis ............ 30 Table 9: Definition of infection .............................................................................. 30 Table 10: Spearman correlation coefficients of chorioamnionitis ............................. 32 Table 11: Model #1: Chorioamnionitis and its predictors (below 31 weeks) ............ 33 Table 12: Model #2: Chorioamnionitis and its predictors (above 31 weeks) ............. 33 Table 13: Model #3: Disabling CP and its predictors (below 31 weeks) ................... 35 Table 14: Model #4: Disabling CP and its predictors (above 31 weeks) ................... 36 LIST OF FIGURES Random Effects Model (REM) and graph in 7 selected studies .............................. 13 Fixed Effects Model (FEM) and graph in 4 selected studies ................................... 15 Fixed Effects Model (FEM) and graph in 3 selected studies ................................... 16 Study population ................................................................................................... 19 Graphs of 4 definitions of infection in NBH data (below 31 weeks) ....................... 30 Graphs of 4 definitions of infection in NBH data (above 31 weeks) ....................... 31 Fixed Effects Model (FEM) and graph in 5 selected studies (below 31 weeks) ...... 38 Fixed Effects Model (FEM) and graph in 5 selected studies (above 31 weeks) ....... 39 Fixed Effects Model (FEM) and graph in 4 selected studies (below 31 weeks) ...... 40 Fixed Effects Model (FEM) and graph in 4 selected studies (above 31 weeks) ....... 41 vii Abruptio: CP: DCP: FEM: GA: IUGR: IVH: LMP: LBW: PEL/VE: PVL: PROM: Previa: SGA: VLBW: GLOSSARY abruptio placenta: premature separation of placenta cerebral palsy: a nonprogressive impairment of motor firnction disabling CP: inability to walk ten steps unaided by 2 years of age fixed effects model: a model for a meta-analysis with homogeneity gestational age: age of fetus in weeks at birth intrauterine growth restriction: infants weighing less than tenth percentile of weight for that GA intraventricular hemorrhage last menstrual period: the first day of last regular period low birth weight: infants born with birth weight less than 2,500gm New Jersey Neonatal Hemorrhage study parenchymal lesion/ventricular enlargement periventricular leucomalacia premature rupture of membrane placenta previa: placenta located near or over internal cervical os of pregnant uterus random effects model: a model for a meta-analysis with heterogeneity small for gestational age: infants weighing less than tenth percentile of weight for that GA very low birth weight: infants born with birth weight less than 1,500gm white matter damage: parenchymal bleeding of germinal matrix in fetal brain viii INTRODUCTION The prevalence of cerebral palsy (CP) has not changed over past two decades and remains about 2.0 per 1000 live births [l]. The incidences of preterm births and of very low birth weight (VLBW) (infants weighing less than 1500g) are 8.7-10.5% and 1.17- 124 % respectively [2] [3]. There have been no significant changes in the birth rate of VLBW infants or in the incidence of CP among these infants (7.5%-8.S% among surviving VLBW infants) [4]. Of the approximately 4 million live births annually, about 50,000 VLBW infants are admitted to neonatal intensive care units. Since the introduction of steroids and surfactants in the 19803, there has been a continual decline in the mortality rate, so that approximately 85% of infants between 700 to 1500g birth weights now survive the neonatal period [5]. Now more than one third of CP (3,100 of 8,000 annual CP) comes from VLBW infants. VLBW infants are 27 times more likely to have CP than full-tenn infants [6]. Without improvement of CP incidence in this birthweight group, it will be difficult to realize significant progress in lowering the overall incidence of CP in this country. The annual cost of CP is $2.4 billion, which represents one third of the cost of the 18 most common birth defects [7]. Indirect expenditures include the life long care of CP which burdens the family, the health care industry, and society. One factor that seems to play an important role in CP among VLBW infants is chorioamnionitis, whether this be via an association with clinical chorioamnionitis or histological chorioamnionitis, or possibly via a combined effect of clinical and histological infection. The association of chorioamnionitis and CP is significant in some, but not in all studies. Some studies reported minimal or no association between chorioamnionitis and CP [8, 9]. Others showed a strong association [10, 11]. This inconsistency may come from considerable methodological difi‘erences brought about by various researchers’ differing definitions of chorioamnionitis. Because clinical chorioamnionitis is diagnosed based on signs and symptoms of infection recorded in medical records, some authors select some of them or others use all before they include them as an exposure. Whereas some studies [8, 12] used only fever for the diagnosis of chorioamnionitis, others [13-15] added clinical signs such as tachycardia, uterine tenderness, and tachycardia for the criteria of chorioamnionitis. In numerous studies that have researched a possible association between chorioamnionitis and CP in preterm infants, infants were categorized by their birth weights alone 1500g or less [8, 14, 16-18]. This method of sampling by weight alone could oversample older gestational ages (GA) infants with intrauterine growth restriction (IU GR). Thus categorization by birthweight alone might produce cohorts that are not optimal for causation studies [19]. Studies have suggested that infants less than 30 to 32 weeks gestational age (GA) have higher incidences of chorioamnionitis, premature rupture of membrane (PROM), and intraventricular hemorrhage (IVI-I), than those of more than 32 weeks GA. Chorioamnionitis is more common when PROM occurs before 30 to 32 weeks than later in pregnancy [20]. PROM occurred in 27-3 6% of infants born before 32 weeks and in 3-5% of those above 32 weeks [21, 22]. Not all authors agree with this large difi‘erence in the incidence of PROM by GA, but an inverse relation between GA and PROM has been recognized in several studies[22-24]. The rate of NH diminishes sharply as gestational age increases, with an incidence of less than 1% at gestational age of 32 weeks and older [25]. Few studies have compared the strength of association between infants born before 31 weeks and older GA groups. No studies have tested the effect of this association using different criteria for diagnosis of chorioamnionitis to the same cohort to compare the strength of association by diagnostic criteria. By conducting studies of premature infants by gestational week (below 31 and above 31 weeks) the confounding effect of GA and IUGR will be reduced for analysis. Difi‘erences of association in the two GA cohorts could help us to elucidate possible causal factors of disabling CP (DCP). The objective of the current study is to test the null hypothesis that the incidence of DCP is unrelated to chorioamnionitis after adjustment of relevant confounders in both infants above and below 31 weeks GA. This paper consists of two parts: part one describes a meta-analysis that is conducted to summarize the association of chorioamnionitis and CP in recent clinical and epidemiological researches. In part two, using the data from New Jersey Neonatal Brain Hemorrhage (NBH) study of low birth weight infants less than 2000g, the association of DCP and clinical chorioamnionitis is tested in two cohorts, stratified by gestational age less than 31 weeks (below 31 weeks) and equal to and above 31 weeks (above 31 weeks). Clinical chorioamnionitis using four different definitions commonly used in recent studies is analyzed to compare the strength of association by diagnostic criteria. Finally, the meta-analysis is re-run with the inclusion of the NBH data. PART ONE A Systematic Literature Review: Meta-Analysis A meta-analysis was conducted to acquire an understanding of the association between cerebral palsy and chorioamnionitis. Over the past two decades, several epidemiological studies have examined the association between cerebral palsy and intraamniotic infection. Varying definitions, difl‘erent modeling, various cohorts and location differences were noticeable in the several studies reviewed. Some recent case-control and cohort studies have reported positive associations between cerebral palsy and chorioamnionitis [10, ll, 13, 15, 26]. No association was shown in other studies [18, 27]. Thus it is still unclear whether there is an elevated risk for CP among women who are diagnosed with chorioamnionitis. It is also difficult to distinguish whether the risks of CP are associated with the exposure to clinically obvious chorioamnionitis, to histologically determined chorioamnionitis, or to combined (clinical and histologic) chorioamnionitis. It appears that differences in study designs and consequent difi‘erences in biases may influence the results of the studies and account for discrepancies in the literatures. To acquire a better understanding of the nature of the reported association between cerebral palsy and chorioamnionitis, a meta-analysis was conducted including all studies published fi'om 1966 to December 2000. Meghggg Study Identification Studies included in this review were identified from a larger subset of all identified studies of cerebral palsy-related etiology, epidemiology and prenatal risk factors for chorioamnionitis. This larger pool of studies was identified by cross-referencing the following Medical Subject Headings (MeSH) terms using MEDLINE from 1966 to 2000—cerebral palsy, chorioamnionitis with the following MeSH terms-etiology, epidemiology. In addition, every citation for cerebral palsy fi'om January 1990 to 2000 was examined and the abstracts of potentially relevant citations were reviewed on line. Study Selection The following selection criteria were used to identify studies for inclusion in this analysis. 1. Stuay design: Included studies were required to be either case control studies, nested case control studies, or cohort studies; the relative risk or odds ratio with 95% confidence interval was provided or could be calculated from the data presented in the paper. 2. Outcome: cerebral palsy 3. Exposures: Clinical chorioamnionitis, histological or microbiological chorioamnionitis, or the attending (discharge) diagnosis of chorioamnionitis 4. Studies including singletons of gestational week less than or equal to 35 weeks, or birth weight less than or equal to 1,500g 5. Excluded Studies: unspecified gestational weeks, gestational week greater than 3 5, and if multiple births were included in the study. Through the literature search in MEDLINE (1966-2000), I identified a total of 128 English language studies with titles that discussed etiology or risk factors of cerebral palsy or periventricular leucomalacia (PVL) (believed to be a precursor of cerebral palsy in premature infants); 1350 studies had ‘chorioamnionitis’ in the titles, or abstracts, or the medical subject headings. Of 15 initially identified publications that discussed the association between chorioamnionitis and cerebral palsy or PVL, one study (Itakura)[28] was excluded because its primary hypothesis was to estimate the timing of brain damage involved in the onset of PVL in the perinatal period and analyzed neonatal electroencephalograms. One article (Eschenbach) [29] was an editorial and therefore excluded from this review. One study (Grether) [30] was excluded because it was in infants of normal birth weight. Two studies (Wu and Vigneswaran) [31, 32] were excluded due to publication type- one was a meta-analysis, the other a review paper. The outcome variable in Verma U [27] and Redline RW [33] papers were IVH and PVL and thus were excluded from the study. One paper (Perlman) [34] had PVL as the main outcome variable. We decided to focus only on CP as the outcome variable for the meta-analysis, and did not include Perlman’s study. For O’Shea et al [15, 17] and Yoon et a1 [26, 35] published papers that were updates of previously reported studies, only the paper with the most inclusive study population was selected, i.e., O’Shea 1998 and Yoon BH 2000 [15, 26]. Table-1 lists these 10 publications as reviewed but excluded from the meta-analysis. The remaining 5 independent papers [8, 10, 15, 18, 26] reported results pertaining to chorioamnionitis as the exposure of interest and cerebral palsy as the main outcome variable, and were included in the meta-analysis. Table 1: Publications reviewed but excluded from the meta-analysis Publi- Study cation Reasons for exclusion Year Itakura-A et al 1996 Primary hypothesis was not CP and chorioamnionitis Eschenbach-DA et a1 1997 Editorial type of publication Grether-1K et al 1997 Study in normal birth weight infants Wu YW et a1 2000 A meta-analysis Vigneswaran R et al 2000 A literature review paper Yoon BH et a1 (1) 1997 Gestational age up to 35 weeks Verma U et a1 1997 Outcome in periventricular leucomalacia (PVL) Redline RW et a1 1998 Outcome in PVL, the same cohort of Wilson-Costello O’Shea et al 1997 Cases from 1978-89 Perlman JM et al 1996 Outcome variable was PVL Two papers, Cooke’s and Allan’s, were cited in articles that were selected for this study (Table 2), although they had not been picked up through Medline key words search, so they were included in this review. Thus, in total, seven studies (Table 2) constitute the meta-analysis throughout this paper. Table 2. Studies selected for the meta-analysis Study Publication GA week/ # of # of Study year Year Study Design Birth Weight cases controls Murphy DJ 1984-90 1995 case-control <32 wks 59 234 Grether JK 1983-85 1996 case—control <1500 g 42 75 Wilson- Costello 1983-91 1998 case-control < 1500 g 50 50 O’Shea 1986-93 1998 case-control 500-1500 g 62 124 Cooke RW 1980-86 1990 case-control <1500 g 81 81 Yoon BH 1993-95 2000 cohort 26-35 wks 14 109 Allan WC 1989-92 1997 cohort 600-1250 g 36 339 Data Extraction Information was assembled in a standard form, including sample size in each case or control group and the exposure/outcome of interest. The following data were abstracted onto the standardized form: publication year, years of study, gestational age, method of obtaining exposure information, definition of chorioamnionitis, definition of cerebral palsy, minimum age at diagnosis of cerebral palsy, numbers permitting me to calculate OR or RR; source of cases and controls, and extent of controlling for potential confounding. Sgtigtiggl Anglygg’ : Meta-Analyst software [3 6] was used to combine all studies with both the DerSimonnian and Laird random effects method (D&L REM) and the fixed efi‘ects method (M-H FEM). The pooled risk estimates for these 7 studies are summarized by weighted averages in which the weight is the inverse of the variance of the estimate. A heterogeneity test Q-statistic, as well as random-effects and fixed-efi‘ects summaries, was computed. Heterogeneity and homogeneity describe the degree of between-study variability among studies. In order to test heterogeneity the Q-statistic and P value were calculated. IfP <0.05, then there is significant heterogeneity. In such case we use random efi‘ects models for a meta-analysis, which considers both between-study and within-study variability. The fixed effects model considers only within-study variability, and can be used when there is significant homogeneity (P>0.05). riinf i The general characteristics of the seven studies are summarized in Table 3. The 7 studies included 1356 subjects in total: 344 CP patients, and 1012 non-CP patients. Years of research were in the 19803 or early 19905. The study designs and settings were slightly difi‘erent. Murphy [10] and Grether [8] sought information from local registries and performed case-control studies in Oxford, UK and in the California Bay Area counties, respectively. O’Shea’s [15] and Allan’s [3 7] studies were in multi- site tertiary obstetric referral centers. The Wilson-Costello’s [18] and Yoon’s [26] studies were at University Hospital for Women in Cleveland Ohio, and at university hospitals in Seoul Korea. Cooke’s study [9] was a nested case-control study conducted in Mercy Regional Neonatal Intensive Care Unit at University of Liverpool, UK. mum-negates“. ammo—025nm: Amwioifiuotoso SET—0 0ch ace—B0 tono0 6.950080 6.380000 3550.030 3.9.8.080 3.9.8.030 summon. anew 50833 82 88 83 32 was 32 32 .8 so». uaoawamc :2: a me» R :08 a we» 5 a: on m N _ a: 8 m mu m 000° om... a Asa-Vs. 23.30: 3.0 3:8 3.38 .2er a $.00 2:20: $.30 3.039. 3.9 a s meaoasa a 3.32 23.30: 3.0 .6 3.20.: 33.20 a 3.5: 3?.st §§= $.3va 3.52 s acacia“ an a: S 42 on me a.” 28:50 0.. a on E S S on N.. a .88 .6 0o 3. 0058.88 an as as m; 42. a» a 355% a 00223 E 33 as an Bop 5. N: «4: was on: - ”8 Essence a .3080 .880 :02 .380 SE»... See... aeoaoe SE80 mmfluso hmfle ammoee .958 .580 .2 .880— .vS—wcm Jam—800 65.92% com—05$ new “802% .003. e828 .38m .333: 532 . 8-32 3-82 3-32 8.2:: 8-32 3-82 8&2 35a 0o a.» 03 .52 mm :8» 9.80 85b 5389:8159 235 Sea: 553.. Q0 93 mwiogotoao meta—coo «0:03» :05... no 88.58820 .m 050,—. 10 The main criteria of eligibility used in these 7 studies were roughly the same. Subjects were eligible for inclusion if they were singleton infants under 35 gestational weeks or with birth weight equal to or less than 1500 g singleton infants. All seven studies excluded infants of multiple births because they pursue different physiological maturation fi'om singleton infants. Gestational weeks or birth weight were not significantly different in CP and non-CP groups. Diagnostic criteria for chorioamnionitis in each study are summarized in Table-4. The diagnostic criteria for clinical chorioamnionitis used in 4 studies (Grether, Wilson- Costello, Cooke, Yoon) [8, 9, 18, 26] were, in principle, the same. They were based on 6 clinical symptoms or signs, either alone or in combination. Signs included were matemal fever during labor, uterine tenderness, maternal leukocytosis, malodorous amniotic fluid, and maternal or fetal tachycardia. However, other three authors combined clinical and histological signs of infection. Murphy [10] included histologic and microbial chorioamnionitis for exposure. Allan [3 7] used attendings’ diagnosis of chorioamnionitis for exposure. O’Shea included 23 cases of attendings’ diagnosis of chorioamnionitis that included histologic findings, antibiotic use, and clinical signs of chorioamnionitis. Therefore, Murphy, O’Shea, and Allan [10, 15, 37] used similarly defined chorioamnionitis as their exposures, because discharge or attendings’ diagnosis of infection included clinical, histologic and microbial findings. CP was the main outcome of interest among 7 studies. Grether [8] got CP information from postneonatal medical and service agency records and fi'om the birth defect registry, and excluded nondisabling CP. The definition of CP used in the registry in Murphy’s study [10] was ‘perrnanent impairment of voluntary movement or posture’. ll Table 4 Diagnostic Criteria for Clinical Chorioamnionitis in each study Clinical Murphy Grether Wilson- O’Shea Cooke Yoon Allan symptom/sign Costello Maternal fever >=2 Fevcr>38 >37.8 + PROM + PROM 37.8 + Attending’ Uterine clinical during 2 signs 2 sign, or + l 2 of diagnosis tenderness signs labor or or sign or signs of chorio— Leukocytosis> or foul >383 + l attendings’ more arnnionitis l 1000 microbiolo- smelling sign diagnosis Foul smelling gical or discharge of chorio- discharge hiStOIOSic arnniotis Maternal evidence Tachycardia >100 Fetal tachycardia >160 Cooke [9] evaluated the degree of disability and included only CP with noticeable disability. Children with a diagnosis of ‘clumsiness’ were not included. Allan [3 7] included the CP subtypes, diplegia, hemiplegia, and tetraplegia. Wilson-Costello [18] studied 72 infants with classically defined CP (50) as well as those with nondisabling CP (22). 50 infants with CP were included in this study. O’Shea [15] diagnosed CP with a multidisciplinary examination at 1 year adjusted age. Yoon [26] did not mention degree of disability for inclusion. Ages of diagnosis of CP in these studies were all under 5 years. 12 Results 1) Association between chorioamnionitis and CP in preterm infants in 7 selected studies Results from Random Effects Model (REM): Pooled OR=2.65 (95% CI: 1.59-4.43). The Q-statistic showed that the pooled analysis was heterogeneous (Q=11.54, p<0.025). : N Stud fiamaunw 9 3' 811m Total Pte - .9 - II. ...-3......- fleta-flnalyaie = Random Effects Hodel (88L) Exper1nent Control Odde 952 Year Obs Tet Ohe Tot Ratio Lou 1995 18 59 8 234 .77 2.16 1996 23 42 42 75 8.95 8. 1998 11 5 58 2.87 8.78 1998 12 62 11 124 2.47 1.82 1989 17 81 6 81 3.32 1.24 888 2 14 7 1 2.43 8.4 1997 18 36 26 339 4.63 2.8! 1356 85 344 186 1812 2.65 1.59 z - 3.7379 0 I 11.54 Tau“2 - 8.2228 Overall Heterogeneity: Z . ...-... - 14.4363 10 .291? a 12 842 Results based on REM presented in graph: CPRLL .110 Random Effects Model (0&L) Odds Ratio ‘ " . C 2 5 STUDY 1 nuuw 2 Grether 3 Hilson 4 O’Shee 5 Cook. 6 YoonC 7 Allan Overall YEN! 1 996 l 996 l 998 1 998 1 989 2000 1 987 An-r 0 01 0.02 0.05 01 0.2 0.5 I r I ALL In: 10 20 50 III) A All 293 117 100 186 162 123 z = 3.7‘ 211 = 0.00019 Eggle + Expand - Shrink Mgr quons 6r ' ds 1 Ole! Symbol Dual /UOF rams: Legends H Help 0mm The test of heterogeneity (Q=11.54 p <0.025) indicated that the 7 studies were significantly heterogeneous. Some of the heterogeneity seemed likely to come from using different criteria of exposure. Murphy, O’Shea, and Allan had difi‘erent definitions of infection, which were more sensitive than the remaining 4 studies. They were based on combination of clinical and histologic signs/symptoms. Therefore, I re-ran the meta- analysis excluding the Murphy, O’Shea, and Allan studies. The meta-analysis results are as follows: 2) Association between clinical chorioamnionitis and CP in preterm infants in 4 selected studies Results from Random Effects Model among 4 studies: Pooled OR = 1.79 (95% CI: 0.96- 3.36). The Q-statistic showed that the pooled analysis is homogeneous (Q=4.30, p>0.20). Therefore, the result from the Fixed Effects Model (FEM) in 4 studies is presented below: Pooled OR = 1.71 (95% CI: 1.04 — 2.80). A weak positive association was noted from the pooled results. - sum: 2.. was 4% Q:“' ‘s‘a x :ss‘w. s». Heta-flnalyeie 3 Fixed Effects Model (II-ll) Experiment Control Odde 952 CI , 8 Study Year Obe Tot Obe Tot Ratio Low High xflt : 1 Grether 1996 23 42 42 75 3.95 3.44 2.33 43.7543 2 Wilson 1998 11 S. 6 58 2.8? 8.78 6.12 21.4838 5 3 Cooke 1989 1? 81 6 81 3.32 1.24 3.92 25.831? 5 4 Yoonc aue 2 14 7 189 2.43 8.45 13.5 8.9385 Total Pte I 5“ 53 18? 61 315 1.71 1.84 2.88 z - 2 1248 2? I ...34 L. i... l4 Results based on FEM among 4 studies in graph: l‘f’fill .rm leud f- fit-«13 Hurlwl (11 H) ...-.10 -.. Ii 0.01 0.02 0.150.! 0.2 0.8 1 2 8 10 20 so too aux" imam. + Expand 1 tether 117 - Shrink 2 Hilton 3 Cooke S Grids 4 Yoenc S Plotstpbol 0 Dual/(IF L Legends Overall 3) Association between ‘combined’ chorioamnionitis and CP in preterm infants (Murphy, O’Shea, and Allan’s studies) The test of homogeneity showed that the other 3 studies were also homogeneous (Q=l .81, p=0.20). The results fi'om the fixed efl‘ects model showed OR=3.83 (95%CI 2.27-6.45) and a significant association between CP and infection. \W‘fi .. . (flax: KKK} \fi <~‘-- x.': \N.\":'\:'.\‘\ \\\M Hete-flnelyaie = Fixed Effects Hodel (H—fl) Experiment Control Odds 952 CI 8 Study Year Ohe Tot Ohe Tot Ratio Low High xflt . 1 Firefly 1995 18 59 8 234 5.97 2.16 15.36 27.645? 2 0’8 0 1998 12 62 11 124 2.4? 1.” 5.96 34.0003 3 allen 199? 18 36 26 339 4.63 2..18.64 38.3468 Total Pte - 054 32 157 45 697 3.03 2.2? 6.45 ‘ 0 - 5.0523 2? - (0.00001 15 In graph: I'l'illl ,MH at o -. . 0.01 0.02 0.“ 0.1 0.2 0.5 1 2 8 10 20 m 1“) m m “8‘ 1 1111111 1 1111111 1 1111111 1 1111111 Expand 1 mm 1000 : - Shrink 2 0.818. 18” : 3 811m 1887 8 ' smar- . M Overall 2- 0.0; 5.52.9200! _.__. L mow In summary, the meta-analysis with all 7 studies showed that the overall pooled result was statistically heterogeneous (p<0.025). The heterogeneity came from large between- study variances. Therefore, the analysis was divided into two halves. The first was restricted to the 4 studies that used only clinical signs and symptoms to determine clinical chorioamnionitis. A weak association was found based on the fixed efl‘ects model. The odds ratio was 1.71 and the 95% confidence interval was 1.04 — 2.80. The other half included the studies of Murphy, Allan and O’Shea that were homogeneous. In this group there was a strong positive association between ‘combined’chorioamnionitis and CP with an 0R=3.83 (95%01 2.27-6.45). 16 PART TWO NBH DATA Materials and Methods Study Population The Central New Jersey Neonatal Brain Hemorrhage Study (NBH study), a multi-center, long-term follow-up study of small infants, enrolled a geographically representative sample of 1,105 live-born infants weighing<=2,000g at birth, a cohort comprising about 85% of all births of that weight born from September 1984 to June 1987 in the central New Jersey counties of Ocean, Monmouth, and Middlesex. Their weight ranged fi'om 501 g to 2,000g and gestational ages fiom 22 to 41 weeks. Of these 167 died in the hospital and 37 died later, leaving 901 infants available for follow-up examination at age two. Twenty-six percent of the cohort was African-American and 25% were multiple births. In this study, prenatal and postnatal records were abstracted and the mother was interviewed shortly afier birth. Information about prenatal exposures including smoking, alcohol and medications, illness, behaviors, and reproductive history and practice was obtained and recorded. Data were abstracted in detail from mothers’ prenatal records and hospital charts [38, 39]. Gestational age in the NBH study was calculated from the last menstrual period (LMP) as reported at a postpartum interview or as abstracted from the prenatal record if there was no interview. When there was dating information from an ultrasound before 18 weeks, the ultrasound dates were used only if they were more than 14 days discrepant fi'om the menstrual dates. Postnatal GA assessments based on physical or neurological criteria were not used [40]. 17 Follow-up Examination Age at the time of assessment was two years following the expected date of confinement (EDC) based on the LMP of mothers. The examination took place at one of two testing sites in central New Jersey or at the home of the participants. Of the 901 survivors at age two, 10 children were adopted out, 67 moved out of state, and 103 children could not be located. The remaining 721 (80%) children were examined, and clinical information was obtained by mail or phone interview of parents on 56 children known to have moved out of state, yielding a total follow-up rate of 86% (n=777). Examination was focused on the detection of major developmental handicaps. Motor status was assessed by a specially trained nurse or nurse practitioner who examined the child’s motor tone, extrapyramidal movements, and tendon reflexes in all limbs. If she could not classify the child as clearly normal in motor function, the child was referred to one of four consultant child neurologists. For the purpose of this investigation, very low birth weight infants fiom twin and other multiple gestations were excluded because morbidities are unique to multiple pregnancies. Similarly, infants with congenital malformations, when information was available, were excluded for this study. The outcomes of interest were CP status assessed at 2 years of age. Criteria for disabling CP (DCP) included the following, in addition to specific neurologic findings [41]: 1. Inability to walk ten steps unaided by 2 years of age; 2. Bayley motor score greater than 1 SD lower than performance score; 3. Receipt of physical therapy for motor disability; and 4. Use of braces or physical assistance devices [38]. 18 Infants with nondisabling CP were included with non-CP infants. Therefore, Non- DCP infants in this study represented the total number of infants with nondisabling CP and without CP. Study Design and Data Analysis This is the prospective cohort study for 565 infants divided into two cohorts for the purpose of comparison. 237 singletons whose gestational ages were less than 31 weeks (below 31 weeks) and 328 babies with gestational weeks 31 and greater (above 31 weeks) were studied separately. These two subsets were analyzed for association between CP and chorioamnionitis (Figure 1). 200 Died 1 105 128 lost of follow-up or missing data @nfants at aQE C 565 Singletons > I < 3‘1 m >- 31 wire #DCP=31(13%) #DCP=11 (3%) #Chorioamnionitis=13 (6%) #Chorioamnionitis=20 (6%) Figure 1: Study Population 212 Multiple Births Four criteria of clinical chorioamnionitis were used to test its association with disabling CP: 1) The diagnosis of amnionitis fi'om discharge medical charts; 19 2) O’Shea’s definition of clinical chorioamnionitis, i.e. PROM plus two of signs of infection (fever, uterine tenderness, foul fluid, antepartum use of antibiotics), and no other source of fever 3) Grether’s criteria, i.e. fever>38 C during labor or foul smelling discharge 4) PROM plus maternal fever (Cooke’s criteria) The association of risk factors with the risk of cerebral palsy and discharge diagnosis of amnionitis was described using the odds ratio (ORs) and 95%confidence limits. Univariate comparisons between two subsets difi‘ering in gestational ages were tested using the chi-square test and the Fisher exact test. If one cell in any two by two table was zero, then one was added to all cells to calculate the OR. Spearrnan correlation analysis was completed to find the importance of variables of chorioamnionitis. The univariate analysis of association of selected variables with DCP in the multiple births was also conducted and presented in the results section. Multiple logistic regression analysis was used to model the association between discharge diagnosis of chorioamnionitis and predictors after adjusting for the risk factors of DCP: race, abruptio, SGA, route of delivery, and preterm labor. To assess the confounding effect among the risk factors of DCP, multiple logistic regression analysis was used to model the association between discharge diagnosis of chorioamnionitis and DCP. The adjusted Odds Ratio was reported for each subset. Mn 1. Univariate analysis of risk factors for cerebral palsy: 20 Univariate analysis of demographic factors, obstetric factors, and brain damages is shown in Table 5. The demographic characteristics such as sex, race, and level of education were not statistically different between DCP and non-DCP in infants above 31 weeks. The same results were noted for sex and years of education in infants below 31 weeks. CP was significantly more fi'equent in White infants than in Black infants (OR 4.02, 95% CI l.18-13.17) in infants below 31 weeks. In infants below 31 weeks, disabling CP was associated with abruptio placenta (OR=4. 17, p=0.001). In infants above 31 weeks, there were associations between DCP and variables related to pregnancy history such as chorioamnionitis diagnosed from discharge medical charts (OR=6.61, p=0.02), but not in infants below 31 weeks (p=0.80). No association was found with status of parity, placenta previa, delivery route, tocolysis, and preterm labor. The associations between CP and preeclampsia were somewhat protective in both subsets (OR=0.17, p=0.06) but were not statically significant (P>0.05). Brain damages diagnosed by ultrasound, both IVH and PEL/VE were strongly associated with CP in all subsets: PEUVE (OR=14.58, 95% CI 598-3555) and IVH (OR=5. 12, 95% CI 2.23-11.72) in infants below 31 weeks, PEL/VE (OR=69.73, 95% CI 13.86- 355.02) and IVH (OR=7.11, 95% CI 1.79-29.64) in infants above 31 weeks. 21 Table 5 Univariate analysis of association of several variables with DCP in singletons below 31 weeks above 31 weeks Variable I DCP No DCP I DCP No DCP I N 31 206 11 317 Sex Male 14 98 6 158 Female 17 108 5 159 OR=1.10 (0.52 - 2.35) OR=0.83 (0.25 — 2.77) | Race I White 28 144 9 241 Black 3 62 2 76 I OR=4.02 (1.18 — 13.71) OR=1.42 (0.30 — 6.71) I Educatlon <=12 yr schooling 21 98 3 143 I >12 .yr 9 80 5 117 I schooling OR=1.90 (0.83 — 4.39) OR=0.49 (0.11 - 2.10) | Parity I <= 22 145 6 206 j | >2 8 33 2 57 F OR=0.63 (0.26 — 1.53) OR=0.83 (0.16 - 4.22) Chorioamnionitis Yes 2 11 3 17 No 29 195 8 300 . OR=1.22 (0.26 — 5.80) OR=6.61 (1.61 — 27.21) [NH | | Yes 19 | 52 4 37 I No 10 | 140 4 263 OR=5.12 (2.23 — 11.72) OR=7.11 (1.70 - 29.64) PEL/VE Yes 17 17 5 7 No 12 175 3 293 OR=14.58 (5.98 — 35.55) OR=69.73 (13.86 — 355.02) 22 below 31 weeks Cont’d above 31 weeks Variable DCP Non-DCP DCP Non-DCP Abruptio Yes 9 I 19 2 31 No l 20 ] 176 6 273 I 011184.17 (1.66 - 10.44) OR=2.94 (0.57 — 15.18) Previa Yes 1 12 1 13 No 28 183 7 291 OR=1.84 (0.23 — 14.67) OR=3.20 (0.37 — 27.94) Route Vaginal 17 106 4 148 C-section 14 95 6 161 0R=0.92 (0.51 — 2.33) OR= 0.73 (0.20 — 2.62) SGA Yes 18 6 137 No 30 188 5 180 OR=0.35 (0.04 - 2.70) OR=1.S8 (0.47 — 5.27) Tocolysis ' Yes 9 69 O 65 No 22 137 11 252 OR=0.81 (0.36 — 1.86) OR=0.32 (0.02-2.43) Hypertension Yes 2 29 0 103 No 28 170 10 199 OR=0.42 (0.15 — 1.76) OR=0.17 (0.01-1.33) Preterm Labor Yes 15 101 5 132 No 8 65 3 119 OR=1.21 (0.48 — 3.01) 23 OR=1.50 (0.35 - 6.42) 2. Univariate analysis of risk factors for chorioamnionitis: The rate of infection was not significantly difi'erent between whites and blacks in preterm infants (p>0. 10). The rate of chorioamnionitis in infants with abruptio placenta was not significantly different fi'om those without. SGA and preeclampsia (listed as hypertension in the table) was associated with a decreased risk of chorioamnionitis in infants above 31 weeks (OR=0.13, 95% CI 0.13-0.58). 24 Table 6 Univariate analysis of several variables with discharge diagnosis of chorioamnionitis in the NBH data 1below 31 weeks I above 31 weeks present absent 20 308 [DCP I Yes 2 29 3 8 No 11 195 17 300 0 =1.22 (0.26 — 5.80) OR=6.62 (1.61 - 27.21) Sex 3 109 12 152 10 115 8 156 OR=0.32 (0.08 — 1.18) OR=1.54 (0.61 — 3.87) 8 '1'] 8 E % G I White I7 I 165 16 234 Black 6 I 59 4 74 OR=0.42 (0.13 - 1.29) OR=1.27 (0.41 — 3.90) Education :I <=12 ,y’ 8 111 10 136 schoohng >12 yr 5 84 9 113 schoolrng l0R=121 (0.38 — 3.83) 0R=0.92 (0.36 — 2.35) Parity J <= 11 156 14 198 >2 2 39 5 54 OR=1.38 (0.29 — 6.46) OR=0.76 (0.26 — 2.21) IVH F Yes I 6 65 4 37 No 7 143 14 253 OR=1.89 (0.61 - 5.83) OR=1.95 (0.61 - 6.25) PEUVE Yes 3 31 1 11 No 10 177 17 279 L OR=1.71 (0.45 — 6.58) OR=1.49 (0.18 — 12.24) 25 rbelow 31 weeks above 31 weeks Chorioamnionitis“ present absent present absent Abruptio Yes 0 28 I 2 31 No 13 183 18 261 OR=0.49 (002-382) 0R=0.94 (0.21 — 4.22) Previa I Yes 1 | 12 0 14 I No 12 [ 199 20 278 OR=1.38 (0.17 — 11.53) OR=. | Route I Vaginal 7 116 9 143 C-section 6 103 11 156 | IOR=1.04 (0.34 — 3.18) OR=0.89 (0.36 — 2.21) I SGA ] [ Yes 1 | 18 2 141 [ No 12 | 206 18 167 | 0R=0.95 (0.12 — 7.76) 0R=o.13 (0.03 — 0.58) [Tocolysis Yes 8 70 6 59 No | 5 154 14 249 J OR=3.52 (1.11 — 11.14) OR=1.81 (0.67 — 4.90) Hypertension | Yes 1 | 30 1 102 No 12 I 186 19 190 OR=0.52 (0.06 — 4.12) OR=0.10 (0.01 -— 0.74) Labor 1 Yes 6 110 10 127 | No | 3 70 114 | OR=1.27 (0.31 - 5.25) * Chorioamnionitis = discharge diagnosis of chorioamnionitis by clinicians 26 8 OR=1.12 (0.43 — 2.94) I 3. Univariate analysis in Multiple Pregnancy Among a total of 280 infants of multiple births, 212 were examined. They represented all multiple births less than 2000g with no stratification in gestational ages (Table 7). The overall incidence of CP in multiple births was higher than in singletons (15% vs. 9.8%). Chorioamnionitis showed a significant association with CP (OR=11.24, 95% CI 1.5-84.9) even though numbers exposed (n=4) were small. Two indicators of brain damage, IVH and PEL\VE, showed strong associations with DCP (OR 23.2, 42.6 respectively). Abruptio placenta was strongly associated with CP as in singletons (OR=8. 1, 95%CI l.25-52.22). One third of multiple births showed intrauterine grth restriction compared with 0.8% in singleton infants below 31 weeks. Preterm labor was more significantly associated with CP than was no preterm labor (OR=8. 1, 95% CI 1.04- 62.96). Table 7 Univariate analysis of risk factors with DCP in multiple pregnancy Variable Sex Male Female White 14 150 Black S 43 OR=0.80 0.27-2.3 Education <=12 yr 7 47 >12 yr 3 47 OR=2.33 0.59-9.5 27 Cont’ Variable <=2 delivers 9 74 >2 1 20 OR=2.43 0.29-20.3 Chorioamnionitis Yes 2 2 No 17 OR=11.24 l 14 31 3 154 0R=23.l8 6.28-85.51 10 6 7 179 OR= 42.62 12.05-1 2 3 15 182 OR= 8.09(1.25-52.22) V 3 35 C-section 16 158 OR=0.85 0.23-3. SGA Yes 5 66 No 14 127 OR= 0.69 0.24-1.99 5 131 14 62 OR=O.75 0.26-2. 19 28 Variable No DCP Yes 1 139 No 18 54 OR=0.14 0.01- Preterrn Labor Yes 16 No l 4. The association between DCP and 4 different definitions of chorioamnionitis Chorioamnionitis illustrated by 3 of 4 common definitions showed no association with DCP. The exception was the discharge diagnosis of amnionitis, which showed a positive association between CP, but only in the subset of infants above 31 weeks (OR=6.61, 95% CI 161-27.21). The results based on O’Shea and Grether definitions were similar in terms of the strength and direction of association. Discharge diagnosis of amnionitis and PROM plus fever provided the same trend of the association. In infants below 31 weeks, a negative association was noted by using the O’Shea or Grether definition (OR=0.54 and 0.57, respectively), while a positive association was found in infants above 31 weeks (OR=1.04 and OR=1.17 respectively). But they were not significant statistically. 29 Table 8: The association between CP and using 4 different definitions of chorioamnionitis below 31 weeks above 31 weeks CP (n=31) No CP (206) CP (n=11) No CP (n=317) Amnionitis 2 11 3 l7 OR=1.22 (0.26 - 5.80) OR=6.62 (1.61 — 27.21) O’Shea 3 34 2 1 56 OR=0.54 (0.16 — 1.88) OR=1.04 (0.22 - 4.92) Grether 6 61 3 77 OR=0.57 (0.22 - 1.46) OR=1.17 (0.30 - 4.52) PROM+Fever 3 15 1 18 OR=1.36 (0.37 - 5.01) OR=1.66 (0.20 — 13.70) Table 9. Definition of Infection Amnionitis O’Shea Grether Prom plus Fever Attendings’ PROM and two: Maternal fever Maternal Fever>100 diagnosis of Fever, tenderness, greater than chorioamnionitis Foul-fluid 100.4 at discharge Include Antepartum use of Foul smelling Ruptured membrane histologic, antibiotics discharge microbial, and clinical infection [42] No UTI No UTI No UTI In graph (below 31 weeks) 1 hnienitie 2 981100 3 ðer 4 Meyer ”Ti 8.81 8.82 8.85 8.1 8.2 “deletie 957. CI 5 1 2 5 lllll 18 28 88 l I 8181 1. £1888 ......go -.--.- 1 1. 3O In graph (above 31 weeks) Odds Ratio 957. CI , 8.81 8.82 8.85 8.1 82 8.5 1 2 5 18 28 50 1. 8m" ”Ts 1 1 1 L1111 1 1 1 1 1 1 1 1 14111 1 1 1 1 1 111 1 1.11.1011: : 2 981100 e 3 Grether : 4 Mover : S. Multivariate analysis: Determinants of chorioamnionitis The Spearman correlation analysis was completed to ascertain the importance of variables that were found to be associated with the chorioarrmionitis diagnosis based on the discharge chart. Factors significant statistically (p<0.05) in both groups were leukocytosis, maternal fever, maternal tachycardia, and fetal tachycardia. Chorioamnionitis was correlated with PROM only in infants above 31 weeks (Table 10). When logistic regression models included 4 signs, PROM, and demographic factors (sex, race and education) to examine the relative importance of each of these for the definition of chorioamnionitis, the models indicated that demographic factors did not significantly contribute to the models (data not shown). Therefore, models were re-constructed alter excluding those three demographic factors. The results (Tables 11 and 12) showed that leukocytosis was important for chorioamnionitis in infants below. and above 31 weeks groups (OR=3.56; 95%CI: 105-12.08) (OR=3.75; 95%CI: 131-1071). In addition, fetal tachycardia was a significant factor for chorioamnionitis in below 31 weeks group (OR=4.34; 95%CI: 1.23-1496) while maternal fever and PROM played significant roles 31 for chorioamnionitis in above 31 weeks group (ORMd fem=3.23 95%CI: 1. 15-9. 12; ORpROM=3.71 95%CI: 132-10.37). Table 10. Univariate Analysis: Spearman correlation coefficients of chorioamnionitis and apparent signs of infection Chorioamnionitis (below 3 lweeks) Chorioamnionitis (above 31weeks) Spearman r p-value Spearman r p-value Leukocytosis < > 15000 0.17 0.007 0.26 0.001 Maternal Fever . . 7 . . >37.8°C 015 001 019 0001 Maternal Tachycardia 0.20 0.002 0. 16 0.005 > 100 Fetal Tachycardia 0.21 0.001 0.19 0.001 > 160 PROM 0.01 0.86 0.19 <0.001 32 Table 11. Model #1 Chorioamnionitis and its predictors (below 31 weeks) Regression Standard Variable Coefficient Error OR 95% CI Intercept -5 .00 0.85 - - Leukocytosis > 15000 Yes 1.27 0.62 3.56 105-12.08 No - - Maternal Fever > 37.8 °C Yes 0.70 0.66 2.02 0.56-7.29 No - - Maternal Tachycardia > 100 Yes 1.56 0.82 4.77 096-2374 No - - Fetal Tachycardia > 160 Yes 1.47 0.63 4.34 126-1496 No - - PROM Yes -0.41 0.77 0.66 0.15-2.99 No - - Table 12. Model #2 Chorioamnionitis and its predictors (above 31 weeks) Regression Standard Variable Coefficient Error OR 95% CI Intercept -4.69 0.58 - - Leukocytosis > 15000 Yes 1.32 0.54 3.75 1.31-10.71 No - - Maternal Fever > 37.8 °C Yes 1.17 0.53 3.23 1.15-9. 12 No - - Maternal Tachycardia > 100 Yes 0.55 0.57 1.74 056-534 No - - Fetal Tachycardia > 160 Yes 1.04 0.59 2.82 0.90-8.90 No - - PROM Yes 1.31 0.53 3.70 1.32-10.37 No - - 33 6. Multivariate Analysis: Predictors of Disabling CP Multiple logistic regression analysis was performed to examine the association of CP and chorioamnionitis stratified by gestational week 31 less or 31 and above. DCP was the dependent variable. Chorioamnionitis recorded at discharge on the medical chart was the exposure variable of main interest. The covariates were selected based on the knowledge of biological plausibility or the results from the univariate analysis. The covariates included in the models were race, abruptio placenta, route of delivery, SGA, and preterm labor. Adjusted for the covariates mentioned above, in below 31 weeks group, DCP was not significantly associated with chorioamnionitis (OR=3.53, 95%CI: 0.63-19.68). Abruptio placenta was significantly associated with DCP in this group (OR=4.24 95%CI: 1.34- 13.47). In above 31 weeks group, a strong association was demonstrated between chorioamnionitis and DCP after controlling for race, abruptio, route of delivery, SGA, and preterm labor (OR=17.04; 95%CI: 275-10550) (Table 13 and 14). 34 Table 13. Model #3 Predictors of DCP (below 31 weeks) Regression Standard Variable Coefficient Error OR 95% CI Intercept -3.29 0.75 - - Chorioamnionitis Yes 1.26 0.88 3.52 0.63-19.68 No - - Race White 1.13 0.66 3.08 0.85-11.19 Black - - Abruptio Yes 1.44 0.59 4.24 1.34-13.47 No - - Route C-section 0.32 0.51 1.38 0.51-3.78 Vaginal - - SGA Yes -0.57 1.12 0.56 0.06-5.07 No - - Preterm Labor Yes -0.05 0.51 0.95 0.35-2.58 No 35 Table 14. Model #4 Predictors of DCP (above 3 lweeks) Regression Standard Variable Coefficient Error OR 95% CI Intercept -4.70 1. 19 - - Chorioamnionitis Yes 2.84 0.93 17.04 2.75-105.50 No - - Race White -0.24 0.88 0.78 0.14-4.42 Black - - Abruptio Yes 1.51 0.97 4.51 068-3010 No - - Route C-section 0.33 0.80 1.39 0.29-6.65 Vaginal - - SGA Yes 0.92 0.86 2.51 0.47-13.52 No - - Preterm Labor Yes 0.09 0.82 1.10 0.22-5.47 No 36 META-ANALY I ' NBH Finally, an attempt was made to compare the result of the meta-analysis after NBH data was incorporated. 1) Studies dealing with clinical chorioamnionitis plus NBH data-below 31 weeks SrWP Results fi'om Random Effects Model: Pooled OR = 1.70 (95% CI: 1.04-2.7 8). The Q statistic showed that the pooled analysis was homogeneous (Q=4.31, P>0.25). Results from the Fixed Effects Model are shown below: Pooled OR = 1.69 (95% CI: 1.07 — 2.69). Chi-square test showed the pooled analysis was homogeneous (X2 =4.34, p>0.25). A weak positive association was again found fiom the pooled analysis. : . .__ :3 -:.55: $5555: \Rfi§%.:u.....§ '5 \{xx , .. . . ' ' ‘wi‘ w. .91: 5' \fi-f‘i‘ 5.55:55 55:51:55; .35- \...." 3"“:‘15- 5"? fleta-flnalyeie : Fixed Effects Hodel (H—H) Experiment Control Odds 952 CI H Study Year Oh. Tot Obs Tot Ratio Low High th , ‘_-- .... ...-I. ...... III... .....I ...-I II... ...-I ...-... 2 1 Grether 1996 23 42 42 75 . 8.95 8.44 2.83 38.8699 1 2 Hileon 1998 11 58 6 58 2.87 8.78 6.12 18.6928 3 3 Cooke 1989 17 81 6 81 3.32 1.24 8.92 22.4758 4 YoonC 2888 2 14 7 189 2.43 8.45 13.85 7.7784 f 5 PROHPeuer-31 2881 3 31 15 237 1.59 8.43 5.82 12.9911 1 Total Pte - 778 56 218 76 552 1.69 1.87 2.69 r z - 2.2382 2? - 8.885 a ............ In 37 Fixed Effect: Model (H-H) I’m! 'etlo " 0 01 0.02 0.05 01 112 0.6 2 - 5 13 20 50 100 S‘I'LDY YERR =97- . . . . . . . . + Expand 1 Brother 1896 117 -- Shrink 2 Hilton 1998 100 3 Cooke 1&3 162 : Plot Quone G Sn 3 4 YoonC 2000 123 c 5 91015111111001 0 Owl/LU 5 Wever-Zh 2001 268 t L Legends menu 77 z - 2.24 2F - 0.025 -—0-—- 2) Studies dealing with clinical chorioamnionitis plus NBH data-above 31 weeks group Results from Random Efi‘ects Model: Pooled OR = 1.72 (95% CI: 1.03-2.88). The Q statistic showed that the pooled analysis was homogeneous (Q=4.30, P>0.25). Results fiom the Fixed Effects Model was presented below: Pooled OR = 1.71 (95% CI: 1.06 — 2.76). Chi-square test showed the pooled analysis was homogeneous (X2 =4.35, p>0.25). A weak positive association was found from the pooled analysis. llete-Hnalyeie = Fixed Effects Nodal (H) Experinent Control Odd: 952 01 0110 Int N Study Year 0110 'l'ot Ratio Low High 1 Grether 1996 23 42 42 '5 8.95 8.44 2.83 41.4854 2 Uilaon 1998 11 58 6 58 2.87 8.78 6.12 28.3386 3 Cooke 1989 17 81 6 81 3.32 1.24 8.92 24.4451 4 YoonC 2000 2 14 7 189 2.43 8.45 13.85 8.4512 5 Meyer031 2H1 1 11 18 328 1 .72 8.21 14.28 5 .3677 Total Pte - 841 54 198 79 643 1 .71 1 .86 2 .76 z - 2.1813 21' - 0.029 38 In graph CPRI l. .119 F 1xe11 Effect: Nude! (Fl-H) Ratio 925/. CI 0.01 0.02 0.05 0.1 0.2 0.5 2 5 10 20 50 100 STLDY YEAR :33 . . . . 1 . 1... 1 . .... Sc: 4 Expand 1 Grether 1996 117 - Shrink 2 Hllson 1888 1(1) 3 Cooke 199 162 . Plot 11‘ ans 0 Bri 4 YoonC 2000 123 c S Plotwaol 13 Bust All" 5 Wever+31 2001 333 c L Legends Overall 841 z = 2.10 2G I 0.029 —e--— 3) Studies dealing with combined chorioamnionitis and NBH data based on discharge diagnosis in hospital charts-below 31 weeks group Results from Random Efi‘ects Model: Pooled OR = 3.56 (95% CI: 2.12-5.99). The Q statistic showed that the pooled analysis was homogeneous (Q=3.33, P>0.25). Results from the Fixed Effects Model are shown below: Pooled OR = 3.44 (95% CI: 2.11 — 5.58). A positive association between CP and chorioamnionitis was found. Hate-analysis : Fixed Effects Hodel (ll-H) Experimnt Control Odds 952 CI I Study Year Obs 'l'ot Obs Tot Bat 10 low H igh th 1 lin- hy 1995 18 59 8 234 5. 77 2.16 15.36 24. 9136 2 O'Sfiea 1998 12 62 11 2. 47 1.I 5.96 38. 6474 3 Hllen1997 18 36 26 339 4. 63 2. 18.64 34.5564 4 Cherie-31 2I1 2 31 11 237 8.38 6.71 9.8826 Total Pts - 1122 34 188 56 934 3.44 2.11 5.58 z - 4.9846 21’ I (8.“ 39 In graph: I 1’”! | NH - o . . 0.01 0.02 0.05 0.1 0.2 0.8 1 2 8 to 20 80 IN W M “I... 4 W 1 term 1'1- arm 2 0’31». 1” 3 anon 187 m 4 Cheri 31 2001 :31“. e- or! 0 MW L Legends Overall 4) Studies dealing with combined chorioamnionitis and NBH data based on discharge diagnosis in hospital charts -above 31 weeks group Results from Random Effects Model: Pooled OR = 4.23 (95% CI: 4.23-6.87). The Q statistic showed that the pooled analysis was homogeneous (Q=2.3 1, P>0.25). Results fi'om the Fixed Effects Model presented below: Pooled OR = 4.02 (95% CI: 2.45 — 6.58). In contrast to the other group, a strong positive association between CP and chorioamnionitis was found. "eta-analysis : Fixed Effects flodel (H—fl) Experiment Control Odds 952 CI 8 Study Year Obs Tet Obs Tot Ratio Low High th 1 by 1995 18 59 8 234 5.77 2.16 15.36 24.4863 1 2 0'8 a 1998 12 62 11 124 2.47 1.1% 5.96 38.8833 § 3 Illan 1997 18 36 26 339 4.63 2.8! 18.64 33.8528 2 4 Gboeie§31 2881 3 11 17 328 6.86 1.67 28.21 11.7176 _ total Pte - 1193 35 168 62 1885 4.88 2.45 6.58 s - 5.5239 21' - 0.25; QM; =3.33; p>0,25). Clinical chorioamnionitis showed a weak association with DCP (OR=1.71, 95%CI 1.05-2.79), whereas the combined chorioamnionitis demonstrated stronger association with CP (OR=3.83, 95%C12.27-6.45). Another meta-analysis in a recent article showed a significant association between chorioamnionitis and CP [31]. Wu et al showed that clinical chorioamnionitis had a positive association with CP (ORcunim mmfl .9 (95%CI: 1.4 - 2.5)) in preterm neonates. Among the 11 studies Wu used to analyze the association of clinical chorioamnionitis and CP, some studies did not have a clear definition of clinical chorioamnionitis [43]; Gray’s study actually combined clinical and histologic chorioamnionitis into a single category of clinical chorioamnionitis [44]. Misclassification could have existed when the author tried to address the effect of clinical 42 chorioamnionitis on CP, and yet she included articles defining chorioamnionitis based on both clinical signs and histologic evidence. The association between clinical chorioamnionitis and CP was not statistically significant (OR=1.6; 95%CI: 0.9 — 2.7) in Wu’s meta-analysis. When Wu tried to evaluate the association between the combined chorioamnionitis (clinical or histologic) and CP, a positive association was found (OR = 1.8; 95%CI: 1.5 —2.3)[3 1]. However, the p-value for the test of homogeneity was 0.07 (If p<0. 10 the studies are considered heterogeneous and variability in study results is more likely not due to chance alone). In the NBH data, 31 weeks was used for the categorization of cohorts by GA. Unlike studies done on the basis of birth weight of 1500g, each cohort is homogeneous for analysis because the maturity of organ system is a function of duration of pregnancy and not of weight. Another reason favoring our approach is that 90 percentile of weight for GA 30 weeks intersects 2000g. Therefore, the below 31 weeks group would capture more than 90% of infants less than 31 weeks [19]. Thus below 31 weeks group would sample fewer older neonates with IUGR. In the NBH data, as seen in meta-analysis, discharge diagnosis of chorioamnionitis showed a stronger association with CP than did three other definitions of chorioamnionitis. As mentioned before, discharge diagnosis of amnionitis in NBH data could represent the combined chorioamnionitis in the meta-analysis. Thus it showed a stronger association with DCP than did the other three definitions in both cohorts (discharge diagnosis of chorioamnionitis will be called as chorioamnionitis in the following discussion). 43 Among infants below 31 weeks, 13 (5%) of 237 infants had chorioamnionitis. Two of the 13 had DCP, and so the infection did not have a significant association with DCP (OR=1.22, 95% CI 0.26-5.80). After controlling for race, abruptio, route, SGA, and preterm labor, chorioamnionitis showed a slightly stronger association with CP (OR=3.52, 95%CI 0.63-19.68) in this group, but this was not significant. This finding showed that infection is not an important risk factor for DCP in this age group. Chorioamnionitis appeared to be a weaker risk factor in this group. An important risk factor in this cohort was abruptio placenta (OR=4. 17 95%CI l.66—10.44), and there was a significant association between abruptio placenta and CP. The association between CP and abruptio placenta persisted even after controlling for infection, race, route of delivery, SGA, and preterm labor in infants below 31 weeks (OR=4.24, 95% CI 1.34- 13.47). Basic science research and clinical studies indicate a strong association among chorioamnionitis, brain white matter damage (WMD) and CP. Gravett et a1 produced intraamniotic infection with Group B streptococcus in rhesus monkeys in1994. They showed a sequential and predictable increase of cytokines in amniotic fluid [45]. Introduction of bacteria into newborn kittens was followed by a cytokine cascade and brain white matter damage [46]. Yoon showed white matter damage in rabbits by creating intraamniotic infection in 1997 [47]. Clinical studies, as seen in this meta- analysis, indicated a significant association between chorioamnionitis and CP. Chorioamnionitis appears in some studies to be potentially one of the most important contributors to the development of DCP [15, 48]. The NBH study did not show the association, but this might be from the fact that infants with chorioamnionitis died before they reached two years of age. There were 113 infants below 31 weeks who died and were singletons. Seven infants among them had chorioamnionitis diagnosed. In order to measure the possible effect of chorioamnionitis on brain damage, PEL/VE was measured as an outcome variable. 21 of 113 dead infants had PELNE. Thus another analysis was conducted to measure the combined effect of chorioamnionitis among dead and live infants below 31 weeks. When the outcome variable was defined as PEL/VE in died infants or DCP in live infants, the summary OR showed the following result. The association was still not statistically significant (OR=1.89, 95% CI 0.62-5.73). Several studies have reported the importance of subclinical infection in the research of chorioamnionitis. Abruptio placenta has shown a significant association with placental chorioamnionitis in some studies [49]. Clinical decision about the presence of infection was based on clinical signs and symptoms alone as in some studies, particularly when histological findings of placenta were not available. However, studies of infection in rhesus monkeys showed that none of the monkeys with firll-blown intraamniotic infection 45 showed the signs of infection such as fever or leukocytosis [45]. Only 25% of patients who had histologic chorioamnionitis at delivery presented with clinical signs of infection [50]. Bacteria have been recorded at transabdominal amniocentesis from 16.1 percent (0- 47.8%) of women in preterm labor with intact membranes [51-53]. When polymerase chain reaction (PCR) was employed to detect bacteria in amniotic fluid in women with preterm labor with intact membranes, a total 55.5 percent of samples were positive, whereas 9.5% of amniotic fluid culture results were positive [54]. These indicate the possibility that many infants with chorioamnionitis have subclinical chorioamnionitis, and do not develop clinical signs of infection. This subclinical infection may be one of the reasons that chorioamnionitis among infants below 31 weeks showed no association with CP. Infants in this age group might have more cases of subclinical chorioamnionitis than older infants. It has been speculated that subclinical chorioamnionitis have contributed to the development of abruptio placenta. Therefore, abruptio placenta with subclinical chorioamnionitis showed a significant association with CP in infants below 31 weeks. The research above helps to explain the reason why the number of cases exposed to clinical chorioamnionitis ranged widely when difi‘erent definition of Grether, O’ Shea, or PROM plus fever, was used to ascertain the exposure of infection in the NBH data. It appears that there is a need to have a strict definition of chorioamnionitis, which would reflect the real efl‘ect of infection on the risk of CP. In infants above 31 weeks, 20 (6%) of 328 infants had chorioamnionitis. Eleven infants had DCP and three suffered from chorioamnionitis. Thus chorioamnionitis showed a 46 strong association with CP (OR=6.61, 95% CI 161-27.12), and older infants may have severe infection, compared with that of infants below 31 weeks. Unlike many studies [14, 37, 48], preeclampsia did not show a significant protective effect for DCP in either subset of infants in the NBH data. Most studies have reported a protective effect of preeclampsia when analysis was done in neonates of 500-1500g weights (OR. 0.3—0.6). When NBH data was analyzed by weights (1500g) rather than weeks (31 weeks), a significant protective efl‘ect in neonates of 500-1500 g infants was also noted (OR=0.23, 95% CI 0.07-0.77). The association between CP and preeclampsia may be confounded by IUGR because many are growth retarded and of higher gestational age. This confounding effect of IUGR and older gestational ages was not apparent in the NBH analysis when neonates were grouped by gestational age. Intrauterine growth retardation has been shown to be a risk factor for CP in term or near term infants [55]. This study, however, showed that SGA was not a significant risk factor for CP (OR=0.35, 95%CI 0.04-2.70) in infants below 31 weeks. Incidence of PEL/VE was higher in below 31 weeks group (14 %) than in above 31weeks group (3%). Finally, this NBH research showed two possible clinical implications. In infants below 31 weeks, the relative contribution of prematurity to the risk of CP outweighs the risk of chorioamnionitis. Some infection in this age group is speculated to be subclinical and mild. Thus, as long as there are no signs of abruptio placenta, all efl‘orts at prolonging gestation would seem warranted by using generous use of broad-spectrum antibiotics. However, in infants above 31 weeks, the contribution of chorioamnionitis to CP is more significant because infection here seems to be severe and symptomatic. 47 Prompt delivery would be a reasonable choice if there were any signs of infection, or when lung maturity is obtained. Future research direction: In order to address the relationship between chorioamnionitis and cerebral palsy more efi’ectively, we need to have a consensus of definition in exposure (chorioamnionitis) and outcome (cerebral palsy). A wide range of false positives and false negatives may exist if subjective symptoms such as uterine tenderness and foul discharge are used to define the exposure. Therefore, fiJture studies of this relationship should focus on objective and repeatable findings from histology reports, microbial findings of chorioamnionitis, or diagnosis based on objective signs of infection such as leukocytosis, fever, tachycardia and PROM. Blinded observers are recommended when evaluating the presence of clinical chorioamnionitis. The status of CP should be defined for the study. It is highly recommended that prospective studies be conducted with the involvement of multi- centers and multi-discipline experts. 48 Bibliography 49 10. 11. 12. 13. 14. 15. BIBLIOGRAPHY Stanley, F. and E. Alberman, eds. The Frequency of Cerebral Palsy: A Review of Population Studies in Industrialized Nations Since 1950,. The Epidemiology of the Cerebral Palsies. 1984, The Lavenham Press Ltd: London. 46-56. Goldenberg, Survival of infants with low weight and early gestational age. Am J Obstet Gynecol, 1984. 149(508-11). Wilcox. A, J., Birthweight and perinatal mortality-the effect of gestational age. Am J Public Health, 1992. 82: p. 378. Lorenz, J.M., A quantitive review of mortality and developmental disability in extremely premature newborns. ARCH PEDIATR ADOLESC MED, 1998. 152(MAY): p. 425-435. Horbar. J, D., Variability in 28 day outcomes for very low birth weight infants: an analysis of 11 neonatal intensive care units. Pediatrics, 1988. 82: p. 554. Ellengerg, Birth weight and gestational age in children with cerebral palsy or seizure disorders. Am J Dis Child, 1979. 133: p. 1044-8. MMWR, Economic costs of birth defects and cerebral palsy. MMWR weekly report, 1995. 44: p. 694-699. Grether, J.K., et al. , Prenatal and perinatal factors and cerebral palsy in very low birth weight infants [see comments]. J Pediatr, 1996. 128(3): p. 407-14. Cooke, R.W., Cerebral palsy in very low birthweight infants. Arch Dis Child, 1990. 65(2): p. 201-6. Murphy, D.J., et al. , Case-control study of antenatal and intrapartum risk factors for cerebral palsy in very preterm singleton babies. Lancet, 1995. 346(8988): p. 1449-54. O'Shea, T.M. and O. Dammann, Antecedents of cerebral palsy in very low-birth weight infants. Clin Perinatol, 2000. 27(2): p. 285-302. Morales, W.J., S.R.d. Washington, and AJ. Lazar, lhe effect of chorioamnionitis on perinatal outcome in preterm gestation. J Perinatol, 1987. 7(2): p. 105-10. Alexander, J .M., et al. , Clinical chorioamnionitis and the prognosis for very low birth weight infants. Obstet Gynecol, 1998. 91(5 Pt 1): p. 725-9. Cooke, Cerebral Palsy in very low birthweight infants. Archives of Dis in Childhood, 1990. 65: p. 201-206. O'Shea, T.M., K.L. Klinepeter, and KO. Dillard, Prenatal events and the risk of 50 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. cerebral palsy in very low birth weight infants. Am J Epidemiol, 1998. 147(4): p. 362-9. Allan, WC. and M. Riviello, J r., Perinatal cerebrovascular disease in the neonate. Parenchymal ischemic lesions in term and preterm infants. Pediatr Clin North Am, 1992. 39(4): p. 621-50. O'shea, Pregnant events and the risk of cerebral palsy in very low birth weight infants. Am J of Epidemiology, 1997. 147(number 4): p. 362-369. Wilson Costello, D., et al. , Perinatal correlates of cerebral palsy and other neurologic impairment among very low birth weight children. Pediatrics, 1998. 102(2 Pt 1): p. 315-22. Reuss. M, L., Clark. J, G, Paneth. N,, Efiiciency of sampling: Birthweight and gestational age distributions in two cohorts, <31 weeks and 5 00-1 499 grams. Paediatric and Perinatal Epidemiology, 1995. 9: p. 341-350. Gabbe. S.G. Niebyl. J, R., Simpson. J, L,, Premature Rupture of Membrane. Textbook: Obstetrics-normal and problem pregnancy, 1996: p. 795. Verber, I.G., et al. , Prolonged rupture of the fetal membranes and neonatal outcome. J Perinat Med, 1989. 17(6): p. 469-76. Guzick, The association of chorioamnionitis and preterm delivery. Obstet Gynecol, 1985. 65: p. 11-15. Gunn. G, C., Premature rupture of the fetal membranes. Am J Obstet Gynecol, 1970.106zp.469. Druzin. M, L., Nonintervention in premature rupture of the amniotic membranes. Surg Gynecol Obstet, 1986. 165: p. 5. Kilbride, H.W., J. Yeast, and D.W. Thibeault, Defining limits of survival: lethal pulmonary lgpoplasia after midtrimester premature rupture of membranes. Am J Obstet Gynecol, 1996. 175(3 Pt 1): p. 675-81. Yoon, B.H., et al. , Fetal exposure to an intra-amniotic inflammation and the development of cerebral palsy at the age of three years. Am J Obstet Gynecol, 2000. 182(3): p. 675-81. Verma, U., et al. , Obstetric antecedents of intraventricular hemorrhage and periventricular leukomalacia in the low-birth-weight neonate. Am J Obstet Gynecol, 1997. 176(2): p. 275-81. Itakura, A., et al. , Timing of periventricular leukomalacia using neonatal electroencephalography. Int J Gynaecol Obstet, 1996. 55(2): p. 111-5. 51 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. Eschenbach, D.A., Amniotic fluid infection and cerebral palsy. Focus on the fetus. Jama, 1997. 278(3): p. 247-8. Grether, J .K. and KB. Nelson, Maternal infection and cerebral palsy in infants of normal birth weight [see comments]. Jama, 1997. 278(3): p. 207-11. Wu, Y.W. and J .M. Colford, Jr., Chorioamnionitis as a risk factor for cerebral palsy: A meta-analysis. Jama, 2000. 284(11): p. 1417-24. Vigneswaran, R., Infection and preterm birth: evidence of a common causal relationship with bronchopulmonary dysplasia and cerebral palsy. J Paediatr Child Health, 2000. 36(4): p. 293-6. Redline, R.W., et al. , Placental lesions associated with neurologic impairment and cerebral palsy in very low-birth-weight infants. Arch Pathol Lab Med, 1998. 122(12): p. 1091-8. Perlman, J .M., R. Risser, and RS. Broyles, Bilateral cystic periventricular leukomalacia in the premature infant: associated risk factors. Pediatrics, 1996. 97(6 Pt 1): p. 822-7. Yoon, B.H., et al. , Amniotic fluid inflammatory cytokines (interleukin-6, interleukin-Meta, and tumor necrosis factor-alpha), neonatal brain white matter lesions, and cerebral palsy. Am J Obstet Gynecol, 1997. 177(1): p. 19-26. Lau, J ., Meta-Analysis Software, . 1997: New England Medical Center, Boston. Allan, W.C., et al. , Antecedents of cerebral palsy in a multicenter trial of indomethacin for intraventricular hemorrhage. Arch Pediatr Adolesc Med, 1997. 151(6): p. 580-5. Paneth. N, J .J ., Pinto-Martin. J ,, Magnesium Sulfate in Labor and Risk of Neonatal Brain Lesiona and Cerebral Palsy in Low Birth Weight Infants. Pediatrics, 1997. 99(5). Pinto-Martin. J, C.A., Zhao. 11,, Short interpregnancy interval and the risk of disabling cerebral palsy in a low birth weight population. The Journal of Pediatircs, 1997. 132: p. 818-21. Holzman. C, P.N., Little. R Pinto-Martin. J,, Perinatal Brain Injury in Premature Infants Born to Mothers Using Alcohol in Pregnancy. Pediatrics, 1995. 95(1): p. 66-73. Pinto-Martin. J, C.A., Riolo. S, Holzman. C, Susser. M, Paneth. N,, Cranial Ultrasound prediction of disabling and nondisabling cerebral palsy at age two in a low birth weight population. Pediatrics, 1995. 95: p. 249-254. O'Shea, T.M., J. Kothadia, and DD. Roberts, Periantal events and the risk of 52 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. intraparenchymal echodensity in very-low birthweight neonates. Paediatric and Perinatal Epidemiology, 1998. 12: p. 411. Nelson, KB. and J .H. Ellenberg, Predictors of low and very low birth weight and the relation of these to cerebral palsy. Jama, 1985. 254(11): p. 1473-9. Gray, P.H., et al. , Survival and neonatal and neurodevelopmental outcome of 24- 29 week gestation infants according to primary cause of preterm delivery. Aust N Z J Obstet Gynaecol, 1997. 37(2): p. 161-8. Gravett, M.G., An experimental model for intraamniotic infection and preterm labor in rhesus monkeys. Am J Obstet Gynecol, 1994. 171: p. 1660-1667. Gilles, F.H., A Leviton, and C. S. Kerr, Endotoxin Ieucoencephalopathy in the telencephalon of the newborn kitten. J Neurol Sci, 1976. 27(2): p. 183-91. Yoon, B.H., et al. , Experimentally induced intrauterine infection causes fetal brain white matter lesions in rabbits. Am J Obstet Gynecol, 1997. 177(4): p. 797- 802. Spinillo, A., et al. , Obstetric risk factors for periventricular leukomalacia among preterm infants. Br J Obstet Gynaecol, 1998. 105(8): p. 865-71. Darby, M.J., S.N. Caritis, and S. Shen-Schwarz, Placental Abruption in the preterm gestation: An association with chorioarrrnionitis. Obstet Gynecol, 1989. 74: p. 88. Hawrylyshyn, P., et al. , Premature rupture of membranes: Ihe role of C-reactive protein in the prediction of chorioarrrnionitis. Am J Obstet Gynecol, 1983. 147: p. 240-246. Romero, R. and M. Mazor, Infection and Preterm labor. Clinical Obstet Gynecol, 1988. 31(3): p. 553-584. Wallace, R.L. and ON. Herrick, Amniocentesis in the evaluation of premature labor. Obstet Gynecol, 1981. 57: p. 483. Bobbit, J .R. and CC. Hayslip, Amniotic fluid infection as determined by transabdominal amniocentesis in patients with intact membranes in premature labor. Am J Obstet Gynecol, 1981. 140: p. 947. Markenson, G.R., Ihe use of the polymerase chain reaction to detect bacteria in amniotic fluid in pregnancies complicated by preterm labor. Am J Obstet Gynecol, 1997. 177(6): p. 1471-7. Blair, E. and F. Stanley, Intrauterine growth and spastic cerebral palsy. 1. Association with birth weight for gestational age. Am J Obstet Gynecol, 1990. 162(1): p. 229-37. 53 MICHIGAN STATE LIEBRARIES U WWWWWWWIMIHHIHIWW HI 31293 02177 6681 ‘