Variation in the postpartum hemorrhage rate in a clinical trial of oral misoprostol.

OBJECTIVE: The main objective of this study was to identify factors associated with variation in the rate of acute postpartum hemorrhage (PPH), defined as blood loss >or= 500 mL within 2 hours of delivery, observed in a randomized clinical trial of misoprostol for the prevention of PPH, conducted in rural India. Although the women in the misoprostol group had a significantly lower probability of having a PPH, we also noted a reduction in the rate of PPH in the placebo group over the course of the study. We hypothesized that this was due to the changing skills of the auxiliary nurse midwives (ANMs) over the course of the study. METHODS: We conducted a post-hoc analysis examining variation in PPH rates over the duration of the trial among the women randomized to the placebo arm (n = 808). Descriptive, correlation analysis and generalized estimating equations (GEE) were used to predict PPH rates. With no direct measure of ANM skills, we used proxy measures, including: (1) the ANM's point of entry into the study (original ANMs at the initiation of the trial were less skilled than replacement ANMs); (2) the study duration, representing exposure of the ANM to ongoing training and monitoring; and (3) duration of the second stage of labor as a measure of improved delivery practices. RESULTS: As the study duration increased, the duration of the second stage of labor decreased (-0.12, p = 0.001) and as the duration of the second stage of labor decreased, the rate of PPH decreased (0.0282; 95% CI 0.0201-0.0363). For each 10-minute increase in the duration of second stage labor increased PPH odds by 7.1% and each 30-day duration of the trial decreased PPH odds by 3.4%. Additionally, a patient delivered by an original ANM was 3.14 times more likely to have a PPH compared to a patient delivered by a replacement ANM. CONCLUSIONS: Declining PPH rates were associated with improved skills and delivery practices that decreased duration of the second stage of labor. These improvements appeared to be consistent with the introduction of the more skilled replacement ANMs as well as ongoing training and monitoring for all ANMs over the duration of the trial.

Duke Scholars

Author Hrishikesh Chakraborty Biostatistics & Bioinformatics