Clinical Factors May Predict Length of Hospital Stay for Extremely Preterm Infants

Laurie Barclay, MD

December 17, 2009

December 17, 2009 — Clinical factors may predict length of hospital stay for extremely preterm infants, according to the results of a retrospective analysis reported online in the December 14 issue of Pediatrics.

"The take-home message is that we can't predict time to discharge very accurately from day one of hospitalization," lead author Susan R. Hintz, MD, MS, from Lucile Packard Children's Hospital and Stanford University School of Medicine in Stanford, California, said in a news release. "Really important things happen during the course of hospitalization that affect when an extremely premature baby will be discharged."

The goal of this study was to develop, validate, and compare the ability of several models to predict the time to hospital discharge for infants younger than 27 weeks' estimated gestational age. Time-dependent covariates as well as 5 key risk factors were considered as potential predictors. The study sample consisted of 2254 infants younger than 27 weeks' estimated gestational age who were born at a Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network site between July 2002 and December 2005 and who survived to discharge.

Using postmenstrual age at discharge, the investigators modelled time to discharge as a continuous variable and as a categoric variable (early vs late discharge). They then developed 3 linear and logistic regression models with time-dependent covariate inclusion: perinatal factors only, perinatal plus early-neonatal factors, and perinatal plus early-neonatal plus later factors. They also assessed models for early and late discharge using the cumulative presence of 5 key risk factors as predictors, and they compared predictive capabilities 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.

Although prediction of postmenstrual age at discharge was poor, models including later clinical characteristics more accurately predicted early or late discharge. The AUC for full models was 0.76 to 0.83 vs 0.56 to 0.69 for perinatal factor models. Simplified key-risk-factors models showed that predicted probabilities for early and late discharge compared favorably with the observed rates and that the AUC (0.75 - 0.77) was similar to those of the models that included the full factor set. The 5 key factors or groups of factors predicting later-than-usual hospital discharge were birth weight less than 750 g, the need for surgery during hospitalization, sepsis or gastrointestinal tract infections, chronic lung problems, and severe problems with retinal development.

"It was encouraging that this very streamlined, five-factor model was as good as the much more complicated statistical model that we used to predict if a baby would be discharged early or late," Dr. Hintz said.

Limitations of this analysis include lack of generalizability to all institutions and differences in care approach between sites.

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

The National Institutes of Health and the Eunice Kennedy Shriver National Institute of Child Health and Human Development supported this study. The study authors have disclosed no relevant financial relationships.

Pediatrics. Published online December 14, 2009.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.

processing....