Maternal vaccination and preterm birth: using data mining as a screening tool

Abstract

Objective: The main purpose of this study was to identify possible associations between medicines used in pregnancy and preterm deliveries using data mining as a screening tool.

Settings: Prospective cohort study.

Methods: We used data mining to identify possible correlates between preterm delivery and medicines used by 92,235 pregnant Danish women who took part in the Danish National Birth Cohort (DNBC). We then evaluated the association between one of the identified exposures (vaccination) and the risk for preterm birth by using logistic regression. The women were classified into groups according to their exposure to vaccination. The regression analyses were adjusted for the following covariates: parity, infant's gender, maternal Body-Mass Index (BMI), age, smoking, drinking, job, number of inhabitants in the place of residence, infections, diabetes, high blood pressure and preeclampsia.

Main outcome measure: Preterm birth, a delivery occurring before the 259th day of gestation (i.e., less than 37 full weeks).

Results: Data mining had indicated that maternal vaccination (among other factors) might be related to preterm birth. The following regression analysis showed that, the women who reported being vaccinated shortly before or during gestation had a slightly higher risk of giving preterm birth (O.R. = 1.14; 95% CI 1.04-1.25) as compared to the non-vaccinated group.

Conclusion: Whether the association between maternal vaccination and the risk for preterm birth found here is causal or not deserves further studies. Data mining, especially with additional refinements, may be a valuable and very efficient tool to screen large databases for relevant information which can be used in clinical and public health research.