Development and Internal Validation of a Risk Assessment Tool to Identify Neonates at Risk for 60-Day Hospital Readmission in Dar es Salaam, Tanzania, and Monrovia, Liberia.

Abstract

Hospital readmissions among neonates are common and may reflect ongoing illness. We conducted a prospective observational cohort study of neonates discharged from two hospitals, one in Dar es Salaam, Tanzania, and one in Monrovia, Liberia, to develop and internally validate a risk assessment tool to identify neonates at risk for unplanned readmission within 60 days of discharge. One hundred and fifteen candidate variables were collected. The outcome of unplanned readmission was identified through phone calls to caregivers. We constructed a multivariable logistic regression model with best subset selection to identify the optimal combination of variables to identify neonates at risk for readmission. We used bootstrap validation with 500 repetitions to internally validate the tool. Of the 2,344 neonates discharged, 98.5% were enrolled and had 60-day outcomes. Of these, 3.6% were readmitted within 60 days of discharge, with 41.7% of readmissions occurring within 14 days of discharge. The risk assessment tool included eight variables that were predictive of readmissions. Neonates who had documented abnormal posturing during hospital admission (adjusted odds ratio [aOR] 7.29, 95% CI 1.51-35.12), hydrocephalus (aOR 7.52, 95% CI 1.21-46.95), and low birth weight (aOR 3.16, 95% CI 1.69-5.92) had the greatest risk of readmission. The overall discriminatory value of the risk assessment tool was 0.77 (95% CI 0.76-0.79). The risk assessment tool demonstrated excellent calibration for predicting readmissions at low scores. However, this tool requires external validation before it can be used in sub-Saharan Africa to direct resources for follow-up of high-risk neonates.