Background Variation in readmission rates may assess the quality of a provider through the quality of inpatient care or transitions from inpatient to outpatient providers. The aim of this project was to validate readmission rates as a measure of NICU quality.

Methods Using birth certificates linked to maternal and infant hospital discharge records, a cohort was constructed of 23–34 week gestation infants who survived to hospital discharge at a California hospital discharging over 50 such infants per year between 1995–2009 (N = 296,509 at 141 hospitals). Unadjusted variations in hospital readmission rates within 7, 14, 30, 90, and 365 days after discharge were compared to rates adjusted for hospital casemix, including patient gestational age, insurance status, race/ethnicity, and maternal education, and BPD, IVH, NEC, and ROP as measures of chronic health conditions.

Results Unadjusted readmission rates varied significantly between hospitals and across geographic regions, ranging from 2.2–28.4% 7–14 days after discharge to 2.7–34.4% 365 days after discharge. Some of this variation was explained by hospital casemix. However, after risk adjustment, there remained a 7.9–11.5 fold difference in readmission rates between hospitals with the lowest and highest rates across the five time frames that did not change when complications of preterm birth were included in the risk-adjustment model.

Conclusions There is substantial variation in readmission rates of premature infants that is only partially explained by gestational age and social factors. Readmission rates may provide a measure of the quality of NICU care and the integration of services within a geographic area.