The Statistical Model for Prediction of Heat-Related Illnesses in Conscript Training Course

Abstract

Background: Heat-related illnesses (HRI) are a major health problem among conscripts. Risk assessment using statistical equations is one strategy to help prevent HRI at the individual level. Objective: To create and evaluate an appropriate statistical model to predict HRI in basic conscript training courses. Materials and Methods: The study employed a prognostic and prospective design, divided into two phases. The model was developed in the first phase while the second evaluated the model. In the model development phase, the sample comprised first and second turn conscripts. The model evaluation phase involved a sample of first and second turn conscripts not in the year of the model development phase. Data on personal and environmental factors were collected in the model development phase to adjust the score level to align with the risk level. In the evaluation phase, data were collected using variables obtained during model development by categorizing the risk groups into two levels, low and high, and sorting them according to their symbolic color. Data were analyzed in the development phase using binary logistic regression and clinical predictive rule. Scores in the model evaluation phase were analyzed using the Net Reclassification Index (NRI). Results: In the model development phase, 2,217 subjects took part in the study, with a 100% response rate. The incidence of HRI was 1.6 per 1,000 persons/day. The predictive factors included alcohol consumption within seven days of military service, fever, systolic blood pressure, body mass index, and urine color. In the model evaluation phase, 2,217 subjects participated in the study, with a 100% response rate. When compared with symbolic color classification, a traditional risk assessment, the NRI was equal to 61.4% and considered to be appropriate. Conclusion: The use of score scales based on factors in the statistical model proved to be a suitable additional method for predicting heat-related illnesses at the individual level. Keywords: Statistical model; Heat-related illnesses; Conscripts