JMIR Publications

ABSTRACT

Background:

Population-based administrative data can complement neonatal registry networks for benchmarking, but generating comparable cross-national estimates is challenging when databases are independently governed and differ in coding, linkage structures, transfer capture, outcome definitions, and hospital outcome attribution.

Objective:

We evaluated whether a common, non-pooled harmonization workflow using independently governed administrative databases could support comparison of preterm care systems in Ontario, Canada, and Japan, including resource use, care organization, outcomes, and risk-adjusted between-hospital variation.

Methods:

We conducted a population-based retrospective cohort study of liveborn infants delivered at 24–36 weeks’ gestation in Ontario, Canada, and Japan from April 2011 to March 2021. Data were analyzed separately within each jurisdiction under a common protocol aligning cohort definitions, coding-based variables, transfer handling, hospital attribution, and statistical modeling. We compared characteristics, resource use, and outcomes among infants born at 24–36 weeks and estimated risk-adjusted between-hospital variation among infants born at 24–31 weeks using multilevel logistic regression, summarized by median odds ratios. The primary outcome was in-hospital death or major surgical/interventional outcome for necrotizing enterocolitis, intraventricular hemorrhage, or patent ductus arteriosus.

Results:

The 24–36-week comparison population included 97,876 infants from 95 Ontario hospitals and 265,426 infants from 599 Japanese hospitals. At 24–31 weeks, Ontario had lower use of prolonged mechanical ventilation and greater concentration of high-risk care in tertiary centers. The primary outcome was higher in Ontario than in Japan (10.4% vs 9.1%), reflecting higher mortality (7.2% vs 4.1%) but lower surgical/interventional outcome rates (3.6% vs 5.4%). Median odds ratios showed greater hospital variation in Japan for the primary outcome (Ontario vs Japan: 1.24 [95% CI 1.04–1.55] vs 1.77 [1.65–1.92]) and surgical/interventional outcomes (1.38 [1.04–2.02] vs 2.26 [2.04–2.53]); mortality variation was less pronounced (1.27 [1.03–1.60] vs 1.50 [1.36–1.66]).

Conclusions:

A governance-preserving harmonization workflow supported non-pooled cross-national benchmarking of preterm care using independently governed administrative databases. It identified distinct benchmarking signals—survival among the most premature infants in Ontario and referral/procedural-care variation in Japan—and offers a feasible template for routine-data benchmarking in other transfer-sensitive care settings with institution-level outcome variation when patient-level data cannot be shared