Statistical model predicting a short duration to birth in women with preterm labor at 22–35 weeks' gestation: the importance of large vaginal Gram-positive rods

Abstract

To identify risk factors for a short interval to birth in women with preterm labor, and to construct a statistical model to predict birth within seven days from the diagnosis of preterm labor at 22-35 weeks of gestation. Vaginal flora was obtained from 126 singleton pregnant women hospitalized for preterm labor at 22-35 weeks' gestation. The amount of vaginal large Gram-positive rods (GPR) was counted in a bright field under x400 magnification and classified semiquantitively as loss of GPRs, decreased GPRs (<10), and normal flora (10 or more). The effects of vaginal GPRs, cervical dilatation, and previous history of preterm birth on the subsequent occurrence of birth were analyzed using proportional hazards model, and the effects on birth within seven days from the diagnosis of preterm labor were analyzed using multivariate logistic regression. Fifty-four women (42.9%) delivered preterm. Both loss of GPRs and decreased GPRs were independent risk factors for a short interval from threatened preterm labor to birth, after adjusting the effect of cervical dilatation and past history of preterm birth (hazard ratio 3.4 [95% CI 2.0-5.5] and 2.0 [95% CI 1.1-3.6], respectively). Cervical dilatation of <4.0 cm and 2.0-3.9 cm, and past history of preterm birth were also independent risk factors for a short interval to birth. Loss of GPRs and decreased GPRs, and cervical dilatation of <4.0 cm and 2.0-3.9 cm were independently associated with birth within seven days from the diagnosis of preterm labor (OR 26 [95% CI 5.3-130], 11 [1.9-69], 76 [8.0-720], and 6.4 [1.5-27], respectively). Loss of GPRs and decreased GPRs may be independently important for developing birth in women with preterm labor.