Michael Samawi, Hani Samawi, Gulzar H Shah, Majd Al-Saleh
Background/Objectives: We examined discharge-level associations between an administrative NQI03 numerator-code-positive bloodstream infection construct and hospital outcomes in very low birth weight (VLBW) discharges and whether the NQI03 short-stay criterion altered mortality estimates. Methods: Using the 2022 HCUP Kids' Inpatient Database, we constructed an independent 500-1499 g cohort from ICD-10-CM birth-weight codes. Mortality, procedure-defined mechanical ventilation, and prolonged hospitalization (≥108 days, cohort P90) were modeled with adjustment for birth weight, completed gestational age, patient/discharge characteristics, and hospital characteristics. Missing baseline covariates were addressed with 50 multiple imputations; survey-design, transfer/inborn, short-stay, and model-specification sensitivities were examined. Results: Among 39,630 discharges from 1750 hospitals, 1655 (4.18%) were NQI03 numerator-code-positive. Multiple-imputation aORs were 1.20 (95% CI 1.005-1.432) for mortality, 2.50 (2.14-2.93) for mechanical ventilation, and 2.02 (1.75-2.33) for prolonged hospitalization; corresponding adjusted risk ratios were 1.17, 1.29, and 1.60. Secondary complete-case record-level standardized probabilities were 7.5% versus 6.5%, 58.6% versus 42.6%, and 15.4% versus 9.5%, respectively; uncertainty for these estimates is reported using hospital-cluster bootstrap confidence intervals. Of 3041 deaths, 1528 (50.25%) occurred during stays <3 days; imposing LOS ≥ 3 days increased the mortality aOR to 1.96. Conclusions: NQI03 numerator-code-positive bloodstream infection showed a modest, imprecise association with mortality and more pronounced associations with mechanical ventilation and prolonged hospitalization. Mortality estimates were sensitive to short-stay eligibility. Because infection timing is unavailable, findings are non-causal discharge-level associations.