Feasibility of assessing breastfeeding status in electronic health records

Abstract

Background: Although breastfeeding status is routinely assessed at postpartum and infant health care visits, data on breastfeeding are difficult to obtain in retrospective studies. Claims databases lack this information and queries to participants long after the postpartum period are subject to recall bias. Electronic health records (EHRs) could possibly fill this knowledge gap. Objectives: To assess the feasibility of determining postpartum breastfeeding status at the time of intrauterine device (IUD) placement in preparation for an IUD safety study. Methods: Postpartum women with an IUD insertion were included in the study from 4 health care systems with EHRs: 3 west coast Kaiser Permanente (KP) sites—Northern California (KPNC), Southern California (KPSC), and Washington (KPWA)—and Regenstrief Institute (RI) in Indiana. A random sample of 125 women from each site who underwent IUD placement within 52 weeks postpartum was included—up to 25 women in each of 5 time intervals based on time since delivery. Breastfeeding status was classified as breastfeeding, not breastfeeding, or undetermined at the time of IUD insertion, and the percentage of women in each postpartum time interval was calculated. Evidence of breastfeeding was derived from manual review of mother and infant records (well‐child care visits) and clinical notes extracted from EHRs between delivery and 52 weeks postpartum. Breastfeeding classification was validated by clinical experts. Results: Across all sites, 92% of those sampled with insertion in the first 3 days postpartum were classified as breastfeeding; 90% in the interval >3 days to <4 weeks postpartum; 69%, ≥4 to <6 weeks; 71%, ≥6 to ≤14 weeks; and 41%, >14 to ≤52 weeks. For all postpartum intervals combined, the percentage of women categorized as breastfeeding at each site was 72%, 86%, 80%, and 45% at KPNC, KPSC, KPWA, and RI, respectively. Overall, 94% of the women sampled could be classified as either breastfeeding or not breastfeeding at the time of IUD insertion, with a range of 79%‐100% across the sites. Conclusions: As expected, a high percentage of women were determined to be breastfeeding in the first 4 weeks postpartum. There appeared to be geographic differences in the proportion breastfeeding, which is consistent with breastfeeding patterns reported by the CDC. Breastfeeding status could be ascertained for most women at these sites with EHRs. This study provides evidence that a retrospective study evaluating breastfeeding at the time of IUD insertion can be conducted successfully across sites using EHRs.