Purpose: Following the Agency for Healthcare Research and Quality (AHRQ) review of the risks and benefits of providing cesarean delivery upon maternal request (CDMR), we sought to use decision analysis to prioritize future research by identifying which factors have the greatest influence on pregnancy outcomes.

Methods: A decision-analytic model was designed from the perspective of a nulliparous women contemplating one or more births, comparing two delivery strategies: 1) CDMR and 2) labored delivery, including non-elective cesarean delivery (NECD) or vaginal delivery (VD). We grouped the factors identified by AHRQ as key components in the decision to offer CDMR into modeled complications as follows: maternal mortality (probabilities: CDMR=0.000024, NECD=0.0001, VD=0.000024), neonatal mortality (probabilities: CDMR=0.0008, NECD=0.0008, VD=0.0002), neonatal injury (respiratory distress syndrome, transient tachypnea of the newborn, encephalopathy, or brachial plexus injury) (probabilities: CDMR=0.038, NECD=0.018, VD=0.006), maternal short-term outcomes (hemorrhagic, infectious, or anesthetic complications) (probabilities: CDMR=0.163, NECD=0.244, VD=0.071), maternal long-term outcomes (urinary or fecal incontinence) (probabilities: CDMR=0.072, NECD=0.178, VD=0.183), and subsequent obstetric outcomes (placenta previa or uterine rupture) (probabilities: CDMR=0.020, NECD=0.028, VD=0.011). Limited information regarding probabilities and disutilities were extracted from the literature, and thus sensitivity analyses addressed values from 0.0 to 1.0 for each one. To ascertain factors that were influential in the decision, one-way sensitivity analyses were undertaken to identify thresholds at which the preferred strategy changed. To examine potential interactions, two-way sensitivity analyses were conducted for factors identified as having a threshold in one-way analyses

Results: Given the lack of clinical evidence for disutilities, no determination of the preferred method could be made (CDMR versus labored delivery). Thresholds at which the decision changed were observed for the probabilities of: NECD (threshold=0.04), short term maternal outcomes after CDMR (threshold=0.2), and long term maternal outcomes after VD (threshold=0.2). Sensitivity analyses of estimated disutilties indicate that three have strong influence on the decision regarding CDMR: neonatal injury (threshold=0.5), short-term outcomes after CDMR (threshold=0.02), and long-term outcomes after vaginal delivery (threshold=0.2). The probability of requiring a NECD interacted strongly with all factors in this model.

Conclusions: Future research should focus on the determining the probability of non-elective cesarean delivery during labor, short-term maternal outcomes due to CDMR, and long-term maternal outcomes after vaginal delivery, as well as the disutilities for these health states.