OBJECTIVETo identify risk factors for difficult delivery among nulliparas in the second stage of labor with continuous epidural analgesia, and to develop a multivariable model that is predictive of difficult delivery.METHODSThe database is derived from a multicenter randomized trial of delayed pushing for nulliparous women under continuous infusion epidural. Members of this cohort (n= 1862) were randomly divided into two groups: a “Model Development” and a “Model Validation” group. We used univariate and multivariable techniques to assess associations between anthropometric, sociodemographic, and obstetric variables and difficult delivery.RESULTSWith the referent defined as the category of lesser risk, the developed model showed that the risk of difficult delivery was increased for women with height less than 160 cm (odds ratio [OR] 2.1, 90% confidence interval [CI] 1.2, 3.4), prepregnancy weight greater than 65 kg (OR 1.6, 90% CI 1.0, 2.6), age greater than or equal to 35 years (OR 3.0, 90% CI 1.1, 8.1), and gestational age greater than or equal to 41 weeks (OR 1.8, 90% CI 1.1, 2.8). Induction of epidural analgesia late in labor (greater than or equal to 6 cm) was associated with a higher risk of difficult delivery than induction between 3 and 5 cm (OR 1.9, 90% CI 1.3, 2.8). An interval of greater than or equal to 360 minutes between epidural induction and full dilatation increased the risk of difficult delivery (OR 3.8, 90% CI 1.5, 9.5). Fetal station above +2 at full dilatation and a posterior fetal position were both strongly associated with difficult delivery (OR 2.7, 90% CI 1.4, 5.0, and OR 11.2, 90% CI 4.9, 25.6, respectively). For the multivariable predictive model, when the sensitivity was 57%, the specificity was 75%, and the positive predictive value was 35%.CONCLUSIONOur observations concerning maternal characteristics and obstetric variables are consistent with previous observations with the exception of time of induction of the epidural. The predictive model may be useful in defining high-risk populations for subsequent intervention studies designed to assess approaches to reduce difficult delivery.