Article type
Year
Abstract
Background: We carry out systematic reviews to minimize bias and increase validity. For this reason, the preferred design for included studies is generally randomized controlled trial (RCT). This elegant and powerful design has some weaknesses, however. For some topics, limiting systematic reviews to RCTs may impose serious threats to validity.
Objectives: This presentation identifies threats to validity of RCT-only systematic reviews that compare risks of cesarean and vaginal birth.
Methods: The Maternity Center Association carried out a systematic review to identify harms that differ between cesarean and vaginal birth to inform childbearing women, their caregivers and others in an environment of rapid change in belief and practice without benefit of a systematic weighing of best evidence. This best identified evidence approach included observational research and identified dozens of outcomes that appear to differ by mode of delivery. The present analysis assesses the capacity of RCTs to measure this set of outcomes.
Results: Most outcomes that differed in the observational review would not be measured at all or well in RCT-only systematic reviews, under-representing cesarean and over-representing vaginal birth harms. Measurement challenges include follow-up for outcomes that may occur many years into the future, valid data for many rare life-threatening outcomes (insufficient power, type II errors), difficulty distinguishing between effects of vaginal birth per se and vaginal birth management practices, and troubling pelvic floor outcome measurement standards. High rates of protocol violation of women randomized to vaginal birth groups in the present maternity care environment would compound power requirements. Even if very large trials with good follow-up were feasible, full results would not be available for 20-25 years, though 125 million+ babies are born globally every year.
Conclusions: RCT-only systematic reviews that compare risks of vaginal and cesarean birth will consistently under-represent harms of cesarean section and over-represent harms of vaginal birth. The conclusion of an RCT-only systematic review may be equipoise when a valid risk balance sheet would strongly favor vaginal birth. To shed best possible light on the full range of outcomes of interest in this comparison, a best available/feasible design approach not limited to RCTs is required.
Objectives: This presentation identifies threats to validity of RCT-only systematic reviews that compare risks of cesarean and vaginal birth.
Methods: The Maternity Center Association carried out a systematic review to identify harms that differ between cesarean and vaginal birth to inform childbearing women, their caregivers and others in an environment of rapid change in belief and practice without benefit of a systematic weighing of best evidence. This best identified evidence approach included observational research and identified dozens of outcomes that appear to differ by mode of delivery. The present analysis assesses the capacity of RCTs to measure this set of outcomes.
Results: Most outcomes that differed in the observational review would not be measured at all or well in RCT-only systematic reviews, under-representing cesarean and over-representing vaginal birth harms. Measurement challenges include follow-up for outcomes that may occur many years into the future, valid data for many rare life-threatening outcomes (insufficient power, type II errors), difficulty distinguishing between effects of vaginal birth per se and vaginal birth management practices, and troubling pelvic floor outcome measurement standards. High rates of protocol violation of women randomized to vaginal birth groups in the present maternity care environment would compound power requirements. Even if very large trials with good follow-up were feasible, full results would not be available for 20-25 years, though 125 million+ babies are born globally every year.
Conclusions: RCT-only systematic reviews that compare risks of vaginal and cesarean birth will consistently under-represent harms of cesarean section and over-represent harms of vaginal birth. The conclusion of an RCT-only systematic review may be equipoise when a valid risk balance sheet would strongly favor vaginal birth. To shed best possible light on the full range of outcomes of interest in this comparison, a best available/feasible design approach not limited to RCTs is required.
PDF