A comparison of dual vs. single diagnostic code approaches for neonatal opioid withdrawal syndrome surveillance in neonates ≥ 35 weeks gestational age.

Abstract

Background: Using dual ICD-10-CM codes for exposure/withdrawal (P04.49 or P96.1, respectively) may be more sensitive for identifying Neonatal Opioid Withdrawal Syndrome (NOWS) than P96.1 alone. There are currently no studies that report on national prevalence using this dual code approach.

Objectives: We examined U.S. national NOWS prevalence trends comparing the dual vs. single code approach for surveillance.

Methods: 2016-2022 cycles of cross-sectional data from the U.S. Healthcare Cost and Utilization Project, Kids' Inpatient Database were used to estimate weighted NOWS prevalence in newborns ≥ 35 weeks excluding iatrogenic cases. Generalized linear models with predictive margins were used to estimate prevalence differences for select sociodemographic factors stratified by race/ethnicity. Select measures of morbidity and mortality were evaluated as secondary outcome measures for NOWS.

Results: The survey weighted population was 1,084,633 pediatric discharges and crude NOWS prevalence decreased across 2016-2022 cycles. Using dual codes, crude NOWS prevalence doubled overall at 19.05/1000 (95% CI: 18.33, 19.8) (vs. single: 9.48; 95% CI: 9.05, 9.93) and across most factors and was highest in White infants (dual: 25.15; 95% CI: 24.09, 26.26, single: 13.98; 95% CI: 13.27, 14.71). Crude prevalence for Black infants almost tripled (dual: 19.67; 95% CI: 18.46, 20.95, single: 6.19; 95% CI: 5.65, 6.78). Medicaid payer-type (dual: 35.93; 95% CI: 34.6, 37.3, single: 18.12; 95% CI: 17.29, 18.98), lowest ecologic income-quartile (dual: 29.43; 95% CI: 28.12, 30.79, single: 13.72; 95% CI: 12.92, 14.58), and rural hospital-location (dual: 33.22; 95% CI: 30.75, 35.88, single: 13.88; 95% CI: 12.31, 15.66) were factors with the highest crude prevalence. Trends for relative measures for within category comparison of crude prevalence were largely similar but showed greater magnitude of change for survey year and income quartile. In fully adjusted models, effect modification was evident by year, payer type, and income level, for all race/ethnicity comparisons with dual codes, and short-term morbidity risk appeared higher with dual codes (15.2% vs. single 6.2%) whereas length of stay was greater with single code (7.0 vs. 11.7 days).

Conclusions: The dual code approach provides an alternative strategy for NOWS surveillance.