A clinical and microbiologic analysis of risk factors for puerperal endometritis.

Predictors of postpartum endometritis were identified in 607 asymptomatic, laboring women. One hundred (16.5%) developed postpartum endometritis. Multivariate analysis using stepwise logistic regression identified cesarean delivery (relative risk 12.8; P less than .0001) as the dominant overall predictor. In patients with cesarean delivery (N = 124), prophylactic antibiotics (relative risk 0.54; P less than .0002) and high-virulence bacteria or Mycoplasma hominis (relative risk 1.4; P less than .01) predicted the incidence of endometritis, and in patients with vaginal delivery (N = 483), "bacterial vaginosis organisms" (relative risk 14.2; P less than .001) and aerobic gram-negative rods (relative risk 4.2; P less than .01) predicted endometritis. Despite significant associations found on univariate analysis, clinical variables such as duration of labor, rupture of membranes, and internal monitoring were not predictive of endometritis in the multivariate analysis. Our findings show that cesarean delivery and certain organisms, such as bacterial vaginosis or high-virulence organisms, predict endometritis, and that clinical variables may be facilitators rather than predictors of endometritis.