Predicting time to hospital discharge for extremely preterm infants

Pediatrics. 2010 Jan;125(1):e146-54. doi: 10.1542/peds.2009-0810. Epub 2009 Dec 14.

Abstract

Background: As extremely preterm infant mortality rates have decreased, concerns regarding resource use have intensified. Accurate models for predicting time to hospital discharge could aid in resource planning, family counseling, and stimulate quality-improvement initiatives.

Objectives: To develop, validate, and compare several models for predicting the time to hospital discharge for infants <27 weeks' estimated gestational age, on the basis of time-dependent covariates as well as the presence of 5 key risk factors as predictors.

Patients and methods: We conducted a retrospective analysis of infants <27 weeks' estimated gestational age who were born between July 2002 and December 2005 and survived to discharge from a Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network site. Time to discharge was modeled as continuous (postmenstrual age at discharge) and categorical (early and late discharge) variables. Three linear and logistic regression models with time-dependent covariate inclusion were developed (perinatal factors only, perinatal + early-neonatal factors, and perinatal + early-neonatal + later factors). Models for early and late discharge that used the cumulative presence of 5 key risk factors as predictors were also evaluated. Predictive capabilities were compared by using the coefficient of determination (R(2)) for the linear models and the area under the curve (AUC) of the receiver operating characteristic curve for the logistic models.

Results: Data from 2254 infants were included. Prediction of postmenstrual age at discharge was poor. However, models that incorporated later clinical characteristics were more accurate in predicting early or late discharge (AUC: 0.76-0.83 [full models] vs 0.56-0.69 [perinatal factor models]). In simplified key-risk-factors models, the predicted probabilities for early and late discharge compared favorably with the observed rates. Furthermore, the AUC (0.75-0.77) was similar to those of the models that included the full factor set.

Conclusions: Prediction of early or late discharge is poor if only perinatal factors are considered, but it improves substantially with knowledge of later-occurring morbidities. Predictive models that use a few key risk factors are comparable to the full models and may offer a clinically applicable strategy.

Publication types

  • Research Support, N.I.H., Extramural

MeSH terms

  • Cohort Studies
  • Female
  • Follow-Up Studies
  • Humans
  • Infant, Extremely Low Birth Weight*
  • Infant, Newborn
  • Infant, Premature, Diseases / diagnosis
  • Infant, Premature, Diseases / mortality*
  • Infant, Premature, Diseases / therapy*
  • Intensive Care Units, Neonatal
  • Length of Stay
  • Linear Models
  • Male
  • Patient Discharge / trends*
  • Predictive Value of Tests
  • Pregnancy
  • Probability
  • Retrospective Studies
  • Risk Assessment
  • Survival Analysis
  • Time Factors

Grants and funding