29: The MFMU cesarean registry: propensity score analysis for bias reduction in comparing elective repeat cesarean delivery with trial of labor after a previous cesarean


      Although multiple studies have compared outcomes of elective repeat cesarean delivery (ERCD) with trial of labor (TOL) after cesarean, these studies have been criticized as women who elect one approach over another may be inherently different. Propensity score (PS) analysis may limit this bias and identify groups with similar baseline patient characteristics. The objective of this analysis was to determine if PS analysis can generate ERCD and TOL groups with similar baseline patient characteristics to determine the odds of maternal and infant outcomes by mode of delivery with minimal bias.

      Study Design

      We performed a secondary analysis from a 1999-2002 prospective study of women eligible for a TOL at 37 weeks gestation or more with one low transverse incision from a prior cesarean. Women were categorized according to whether they underwent TOL or ERCD. ERCD consisted of women without medical or obstetrical indications for a repeat cesarean, who had either a cesarean without labor at 39 weeks or more, or who had spontaneous labor or rupture of membranes at 37 weeks or more and chose to proceed with cesarean delivery. TOL patients were matched to ERCD patients using 43 baseline variables comprising the PS. If no match was found the patient was not included. From the matched pairs, maternal and infant outcomes associated with each delivery approach were assessed using conditional logistic regression.


      The two matched study groups (n=3981) were balanced in baseline characteristics according to the standardized difference criterion. The likelihood of TOL success was 68.1%. The frequencies of endometritis, operative injury, respiratory distress syndrome and newborn infection were lower, while those of hysterectomy and wound complication were higher in the ERCD group. The odds of either a maternal or an infant composite adverse outcome were reduced by 33% in those who underwent an ERCD (p<.001 for both).


      Although absolute risks are low, even with well-balanced matched groups, overall maternal and infant morbidity is lower in the ERCD group.
      Tabled 1
      Table thumbnail grt3