Recent work on the epidemiology of preterm birth

Abstract

Preterm birth proportions and outcomes are strongly influenced by the criteria for inclusion of births and deaths in national data systems and Registers. The European Commission Health Monitoring Program includes development of indicators to monitor and evaluate perinatal health, and a recent journal supplement provides results from this PERISTAT project. The analysis by subgroups of gestational age, birthweight, and plurality, shown in the supplement (1), demonstrates the impact of these registration differences. The second development is the recognition that the analysis of preterm birth and fetal and infant deaths would best be thought of as occurring within a cohort of conceptions which reach 20 weeks gestation (2). All those still in utero are at risk of being born in any given week. This approach has already provided an explanation for the paradox of intersecting perinatal mortality curves between babies of smokers and non-smokers, and between singleton and twin infants (2, 3). One research group has carried out a series of systematic reviews of the accuracy of methods for predicting preterm birth, either in threatened preterm labor or among asymptomatic women. Their review of published risk-scoring systems found that none could be recommended (4). A systematic review of controlled studies of perinatal outcomes for singletons and twins after assisted conception concluded that singletons had a significantly worse outcome than non-assisted singletons: the relative risk (RR) was 3.27 [95% confidence interval (CI) 2.03–5.28] for birth <32 weeks and 2.04 (95% CI 1.80–2.32) for any preterm birth. There were no differences in preterm birth between assisted and unassisted twin births (5). Evidence that intimate partner violence (IPV) is a major risk factor for preterm birth has been strengthened. IPV reported to police during pregnancy was followed by a large increase in births before 32 weeks [adjusted odds ratio (aOR) 3.71, 95% CI 1.80, 7.63) and 1.61 (95% CI 1.14, 2.28) for any preterm birth (6). A large population-based study found an increased risk of antepartum hemorrhage [aOR 3.79 (95% CI 1.38, 10.4)] and even larger odds of perinatal death [8.06 (95% CI 1.42, 45.6)] (7). The large European case–control study (EUROPOP) found no excess risk of preterm birth among women employed in pregnancy, but an increased risk for women working more than 42 hr a week, standing more than 6 hr a day, and among those with low job satisfaction (8). The possibility of longer term effects has been demonstrated most clearly by the association of a prior cesarean birth with antepartum hemorrhage and placenta praevia, both risks for preterm birth (9). An analysis of linked records from first and second births in Scotland using the fetuses-at-risk approach showed other adverse effects of prior cesarean, including an increased risk of late preterm birth (33–36 weeks) and fetal death at and after 34 weeks (10). EUROPOP took advantage of different patterns of abortion use in three groups of countries to measure a significant association of prior abortion with preterm birth in all three groups. The risk increased with the number of prior abortions and was larger for very preterm birth. The clinical pathways to preterm birth were spontaneous preterm labor, preterm PROM, and antepartum hemorrhage (11) Records of women with cervical dysplasia were linked to subsequent births in New Zealand to compare preterm birth in those treated and untreated. An increase in preterm PROM was seen with the loop electrosurgical excision procedure and laser conization, though not with laser ablation, and the risk increased with the height of tissue removed (12). The findings were not assessed against population-based data. These three examples demonstrate the need for a broader approach to thinking about epidemiology, etiology, and pathways to preterm birth.