HPTN 024 was a randomized, double-blinded, placebo-controlled Phase III trial that enrolled both HIV-infected and HIV-uninfected women. The primary objectives of this trial were to determine the efficacy of a low-cost antibiotic intervention for prevention of chorioamnionitis-associated MTCT of HIV and of preterm birth [8
]. This multicenter trial was conducted at clinical sites in 4 Sub-Saharan African cities: Blantyre and Lilongwe, Malawi; Dar es Salaam, Tanzania; and Lusaka, Zambia. Enrollment started in July 2001 and ended in February 2003. The design and results of the HPTN 024 trial have been described in detail elsewhere [8
]. Briefly, pregnant women (between 20 and 24 weeks of pregnancy) were randomized to receive either antibiotics (250 mg of metronidazole 3 times a day and 250 mg of erythromycin orally 3 times a day for 7 days) or placebo. The antibiotics and dosages were chosen because they were generally effective against the vaginal and intrauterine organisms and reduction in preterm birth may be achieved using a prolonged course of antibiotics (1 week or more) in the second trimester. Data regarding socio-demographic characteristics as well as obstetric and medical history were obtained through participant interviews. The external genitalia, vagina, and cervix were examined, and a bimanual examination was performed. In addition to physical examination, laboratory tests were performed (including HIV testing, quantification of CD4 cell counts and plasma viral loads, and screening for syphilis, trichomoniasis, candidiasis, bacterial vaginosis, gonorrhea and Chlamydia
]. In addition, at the 26–30 week visit, participants were given antibiotics (250 mg of metronidazole and 500 mg of ampicillin) or placebo to be taken at the onset of labor and then every 4 hours, continuing after delivery 3 times a day until the 1-week course was completed. Compliance with treatment was assessed using pill counts at the 26–30 week visit as described previously [8
]. Women were asked to bring the peripartum antibiotics/placebo blister packs to the next study visit at 4–6 weeks post partum. Adherence was assessed by the number of pills missing from the blister pack at the follow-up visit. If the blister packs were not returned, reported use was noted. Adherence to prenatal study drugs was assessed among 1562 (99%) of the 1573 women randomly assigned and delivered before March 5, 2003. Of these, 98% in each study arm took all 21 tablets of metronidazole and erythromycin prenatally. Adherence was also high at the time of labor: 1392 (88%) of the 1573 women took at least one dose before delivery. Physical examinations during pregnancy, at delivery, discharge, and within 3 months after delivery were performed to determine the mothers’ medical conditions and obstetric complications. The HPTN 024 trial revealed that, despite reducing vaginal infections, the antibiotic regimen used did not reduce preterm birth, increase the time to delivery, or increase birth weight [9
]. HIV-infected women and their infants were offered MTCT prophylaxis with nevirapine [10
]. Mothers’ medical conditions and obstetric complications (including postpartum hemorrhage) were managed according to the existing obstetrics management guidelines in the individual countries (including the use of active management of the third stage of labor and use of oxytocic agents). Data presented at an interim Data Safety Monitoring Board (DSMB) showed that the antibiotic regimen did not show a meaningful reduction of MTCT of HIV. Thus, further randomization and distribution of study drugs was halted on March 5, 2003 at all study sites.
Because of the early termination of the trial by the DSMB, this analysis includes only those HIV-infected and uninfected pregnant women who were eligible to receive both the prenatal and peripartum antibiotic/placebo regimens. These were the women who were randomized on or before October 15, 2002. This date was determined by subtracting 140 days from March 5, 2003, assuming a woman entered the study at a gestational age of 20 weeks and the normal course of pregnancy is 40 weeks.
We examined the effect of antibiotics versus placebo on a number of maternal medical conditions and physical examination findings including diarrhea (3 or more loose and watery stools per day over the last 2 weeks); cough (coughing all day for more than 2 weeks); fever (hot to touch for more than 2 weeks); recurrent itchy vaginal discharge; herpes zoster; oral thrush; hospitalization; weight gain or loss during pregnancy; mastitis; puerperal sepsis; and death. Information on diarrhea, cough, fever, itchy vaginal discharge, and hospitalization was collected at enrollment, the first and second prenatal visits, the intrapartum visit, and follow-up visits post delivery. Data regarding fever during labor or post partum; prenatal or postpartum hemorrhage; and hypertension with or without proteinuria were collected at the intrapartum visit. Information on puerperal sepsis and admission to the postnatal ward was collected at the time of maternal discharge from the hospital. Weight was measured at enrollment, the first and second prenatal visits, and at discharge. Mastitis was defined as inflammation of the breast occurring within 3 months post delivery. Data related to syphilis risk factors, prevalence, and birth outcomes in the HPTN 024 cohort have been published [11
This was a planned secondary analysis. The sample size in the original study design (i.e. 1560 women in each arm) would provide adequate power to detect a medium or large effect in the target medical conditions between the treatment arms. With the sample size used in this analysis (i.e. 779 HIV-infected women in each arm), the power is less adequate, especially when the medical condition is rare or the expected effect size is small. To assess differences in baseline characteristics (including demographics, laboratory results, obstetric and medical history, and medical conditions/physical examination findings) between the antibiotic and the placebo arms, Pearson χ2 or Fisher exact test P values were reported for categorical variables, and 2-sample t test or Wilcoxon rank sum test P values were calculated for continuous variables. For medical conditions for which data were collected on multiple occasions, generalized estimating equations (GEE) models were used to compute the odds of having a condition in the antibiotics arm versus the placebo arm, while adjusting for baseline conditions. A logistic regression model was utilized to analyze binary outcomes that were collected only once during follow-up. A Cox proportional hazards model was used to estimate treatment effect on death. Weight gain was defined as the average per-week weight change from enrollment to the second prenatal visit or the first one thereafter if a woman missed her second visit. A general linear model was used to ascertain the effect of treatment on weight gain adjusting for gestational age at enrollment (measured by fundal height in centimeters). Analyses on the effect of the intervention were performed separately for HIV-infected and uninfected cohorts. P <0.05 was regarded as statistically significant. All statistical analyses were performed using SAS version 9.1.3 on SunOS 5.9 platform (SAS, Cary NC, USA).
Written informed consent was obtained from all participants prior to HIV testing and inclusion in the study. The study protocol was approved by U.S. institutional review boards (Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD; School of Medicine, University of Alabama at Birmingham; School of Medicine, University of North Carolina at Chapel Hill; Harvard University School of Public Health, Boston, MA) and by institutional ethical review boards/committees (University of Malawi College of Medicine, Lilongwe, Malawi; Muhimbili University College of Health Sciences, Dar es Salaam, Tanzania; and the University of Zambia School of Medicine, Lusaka, Zambia).