(287) Assessing neonatal outcome misclassification using mother-infant dyad claims

Publication Number: 945

Background: Newborns’ medical expenses may be charged to their mothers’ insurance since they might not have coverage immediately after birth. To optimize sensitivity, researchers may identify newborn outcomes from mothers’ and infants’ claims, potentially leading to outcome misclassification.

Objectives: To assess the dynamics of infant insurance enrollment and explore neonatal outcome misclassification via contrasts of mother and infant claims.

Methods: We enrolled 2 mother-infant linked cohorts in MarketScan 2012-2018: 1) mother with ≥28 days continuous enrollment (CE) since delivery and no CE requirement for infant and 2) both mother and infant with ≥28 days CE.
Outcomes included ICD-9/10-CM codes for neonatal seizure or convulsions (individually or combined [NS/C]); and small for gestational age (SGA), including neonate-specific codes (NSGA), and maternal (delivery)-specific codes (MSGA) (e.g., ICD-10 O36.5910). We assessed the contribution of maternal claims to overall case capture and potential misclassification. For the latter, we assumed that cases in cohort 2 identified from maternal claims were mother and not infant outcomes.

Results: We identified 1,574,268 mother-infant linked pairs. Cohort 2 retained 770,659 (49%) dyads due to enrollment requirements.
In cohort 1, we identified 5701 NS/C (incidence 0.4%), 3,017 neonatal seizures (0.2%), 4,241 convulsions (0.3%), 69,632 SGA (4.4%), 39,056 NSGA (2.5%), and 45,767 MSGA (2.9%). Infant-only claims identified 70% NS/C, 95% neonatal seizures, 61% convulsions, 31% SGA, 90% NSGA, and < 1% MSGA. Mother-only claims identified 28% NS/C, 2% neonatal seizures, 37% convulsions, 47% SGA, 6% NSGA, and 99% MSGA. Only a fraction of cases were identified on both mother and infant claims (up to 4%).
In cohort 2, we identified 2840 NS/C (incidence 0.4%), 1,495 neonatal seizures (0.2%), 2,135 convulsions (0.3%), 34,143 SGA (4.4%), 20,172 NSGA (2.6%), and 21,628 MSGA (2.8%). Infant claims identified 72% NS/C, 96% neonatal seizures, 64% convulsions, 34% SGA, 93% NSGA, and < 1% MSGA. Mother claims identified 26% NS/C, 1% neonatal seizures, 34% convulsions, 44% SGA, 5% NSGA, and 99% MSGA. Cases identified in both remained small for each outcome (up to 4%).

Conclusions: Different infant outcomes may require different approaches to optimize sensitivity while not compromising specificity. For example, restriction to neonate seizure-specific codes may underestimate incidence by half even if considering mother and infant claims, but the use of maternal claims to capture convulsions may reduce specificity. In contrast, both MSGA and NSGA contribute to the comprehensive capture of SGA, but due to low risk for misclassification both maternal and infant claims could be considered.