[Electronic fetal monitoring and its relationship to neonatal and infant mortality in a national database: A sensitivity analysis].

OBJECTIVE

Sensitivity analysis of the association between electronic fetal monitoring and neonatal and infant mortality previously reported from a United States database.

MATERIALS AND

METHODS

Retrospective cohort study of 11,916,806 live births linked to neonatal and infant deaths during the years 1997-2002 from the United States Centers for Disease Control's National Center of Health Statistics (NCHS) linked birth and infant death data. Restrictions were performed to exclude deliveries occurring outside of a hospital, precipitous labors, breech deliveries, eleven risk factors of pregnancy, multiple gestations, deliveries before 24 and after 44 weeks, implausible birthweights, repeat cesarean sections, and congenital anomalies. An additional analysis explored the effect of further restrictions to term births, birth weight≥2500 g, absence of maternal fever (>38°C), and absence of labor induction or augmentation. For each year, adjusted relative risks (RR) were estimated with log binomial regression. A six-year pooled association was estimated by the generic inverse variance method using a random effects model.

RESULTS

For the six-year period, there was a significant reduction in risk in the group with electronic fetal monitoring for early neonatal mortality (RR=0.54, 95 % CI: 0.52-0.57), late neonatal mortality (RR=0.84, 95 % CI: 0.78-0.90), post-neonatal mortality (RR=0.86, 95 % CI: 0.83-0.90), and infant mortality from all causes (RR=0.75, 95 % CI: 0.73-0.77), from perinatal causes (RR=0.60, 95 % CI: 0.57-0.63), and from hypoxia (RR=0.67, 95 % CI: 0.54-0.84). In the additional analysis, which only examined the outcome of infant mortality from all causes, there was also a significant reduction in risk (RR=0.91, 95 % CI: 0.85-0.97).

CONCLUSION

Using the NCHS linked birth and infant death data over a 6-year period, electronic fetal monitoring was associated with decreased neonatal and infant mortality as has been previously reported. These results were robust to two levels of restriction applied on potential confounding variables.