(200) Characterizing Infants in The Sentinel Distributed Database Mother-Infant Linkage Table

Publication Number: 1245

Background: Due to gaps in insurance coverage from birth until enrollment and the need to track neonatal/infant outcomes after prenatal drug exposure, the US FDA’s Sentinel Distributed Database contains a Mother-Infant Linkage (MIL) table.

Objectives: Among infants in the MIL table, we characterized demographic and enrollment characteristics, medication use, immunizations, health conditions, and major congenital malformations (MCMs).

Methods: We described the attributes [0, 28], [29, 365], and [0, 365] days after birth from January 1, 2000, to March 31, 2021. Among infants with paired mothers, we described these characteristics by searching both mother’s and infant’s record, and infant’s record alone. We quantified variation with standardized mean differences (SMDs).

Results: We identified 6,131,472 infants in the MIL table; (46.8% of infants belonged to a linked mother-infant pair). About 49% were female, ~90% were of unknown race and/or ethnicity, and ~60% had >1 year of continuous enrollment. The mean (SD) days from birth to enrollment was 3.3 (35.0) and 84.6 (121.5) days among infants in a mother-infant pair and those not in a pair, respectively.
    
When searching the infant’s record alone in the first 28 days, infants were similar regardless of whether they were part of a linked pair. The most common health conditions were low birth weight and gastroesophageal reflux disease (2.5-3.9%; 1.5-3.4%). However, adding the mother’s record to the search among paired infants significantly increased evidence of antibiotic and opioid administrations, most medication dispensings (especially antibiotics and glucocorticoids), and 6 of the 22 common health conditions (SMDs ~0.15, 0.1-0.5, and 0.1-0.3) in the neonatal period.
    
After the first month, infants paired to mothers had more medication dispensings (SMD ~0.2 for antiasthmatics, antibiotics, and glucocorticoids), immunizations (SMD ≥0.5 for six of the 11), and common health conditions (SMD 0.1-0.2 for allergies, asthma, influenza, otitis, and sinusitis) compared to infants without paired mothers.
Prevalence of MCMs recorded in the first year were similar regardless of whether the infant’s mother was matched (all SMD < 0.05).
    

Conclusions: We described nearly 2.9 million infants with- and 3.3 million without matched mothers, who were relatively similar in the first 28 days (suggesting little influence from selection bias) but diverged in key health indicators after the neonatal period. Adding the mother’s record to searches where available significantly increased capture of some elements but had little impact on others. This finding indicates that while leveraging linked mother-infant records can be beneficial, future investigators should use clinical judgment when doing so to avoid potential misclassification.