Abstract
Introduction:
There are multiple readmission prediction models available in high-income countries (HIC) aimed at predicting the risk of readmission within 30 days of discharge after surgical intervention. The accurate prediction of a patient’s individual readmission risk allows for targeted outpatient follow-up and early recognition and management to decrease morbidity, mortality, and cost. There are limited risk prediction models available for low-middle income countries (LMIC).
Methods:
This is a weighted prediction model development study, using prospective electronically captured data including all patients undergoing surgical intervention at a large regional hospital in South Africa, from January 2014 to December 2017. The discriminative performance of the predictive model was assessed using receiver operating characteristic (ROC) analysis based on multivariate logistic analysis of previously identified risk factors for readmission, with temporal validation performed on the 2018 surgical patient cohort.
Results:
In total 5588 patients were included with 200 (3.6%) readmitted within 30 days. Univariate analysis identified the following risk factors for readmission: major operative magnitude, emergency operation, higher operative wound classification, and unplanned reoperation. Multivariate analysis revealed that operative wound classification III and IV (OR = 2.5, p < 0.001), unplanned reoperation (OR = 17.4, p < 0.001) and malignancy (OR = 1.9, p = 0.04) were significant predictors of readmission within 30 days. Comparing weighted multivariate analysis findings, a cut-off score of 25 demonstrated optimal sensitivity (>0.75) and 1-specificity (<0.25) for predicting readmission within 30 days. Findings using ROC analysis revealed that the derivation group had an area under the curve (AUC) of 0.82 compared to 0.78 in the temporal validation group.
Conclusion:
The predictive model named theSurgical Risk Assessment Tool for Unplanned Readmissions (STUR) has been developed and internally validated in a LMIC and can be easily applied at discharge.It enables accurate risk stratification and potential targeted follow up.