Creating a Risk Score for Prolonged Length of Stay Following Pediatric Lung Transplants.

Abstract

Pediatric lung transplant (LT) outcomes correlate with prolonged length of stay (pLOS), yet a paucity of research exists predicting pLOS in this population. We aim to identify factors associated with pLOS following pediatric LT, and to create a risk score identifying individual patients at increased risk of pLOS. Using the OPTN database, we analyzed 733 pediatric patients who received an LT between January 2000-July 2022. We used LASSO regularization to identify factors predicting pLOS. A risk score was calculated using odds ratios for each variable in the model. LASSO was run on 51 factors and identified 13 to be included in the model. The variables with the highest negative impact on LOS were recipient ethnicity (Asian, Native American, Pacific Islanders, or Mixed Race) (OR = 3.187), recipient requiring life support (OR = 2.354), and recipient age 2-10 years old (OR = 2.203). In contrast, low cold ischemia time (OR = 0.562), time on waitlist 21-60 days (OR = 0.284), and diagnosis of primary pulmonary hypertension (OR = 0.307) had protective effects on LOS. Using the risk score, we stratified patients into three equal groups: low (mean LOS = 20.60 days), medium (mean LOS = 22.53 days), and high risk (mean LOS = 36.72 days) of pLOS. The c-statistic for the model was 0.7931 and 0.7757 for the risk score. Using this model and risk score, physicians and healthcare systems could identify patients at risk of pLOS and could intervene on preventable variables before transplantation.