Abstract
The primary objectives of this study are as follows: (1) Develop two primary risk prediction models: 1.1. for determining the likelihood of developing diabetes and hypertension in the normal population; and 1.2. for determining the risk of developing diabetes and hypertension-related complications and mortality. In addition, (2) estimates the health expenses and productivity loss due to diabetes and hypertension. This is a quantitative study utilizing Ministry of Public Health outpatient and inpatient data. The study consisted of 1.2 million samples and was followed over a period of 6 years, from January 1st, 2017, to February 28th, 2023. The primary statistic employed in model development is discrete-time survival analysis. For data classification, two machine learning techniques, namely Lasso Logistic Regression and XGBoost (Extreme Gradient Boosting), were utilized. Additionally, model accuracy was assessed using the area under the Receiver Operating Characteristic (ROC) Curve. The study found that the XGBoost method had higher prediction efficiency than Lasso Logistic Regression in all models. The diabetes and hypertension risk models for the general population that used the XGBoost method were accurate, with the area under the curve (AUC) values ranging from 0.86 to 0.88 for diabetes and 0.84 to 0.87 for hypertension. However, the predictive performance for the risk of complications in diabetes and hypertensive patients varied depending on the type of complications, with the AUC ranging from 0.63 to 0.84. Predicting chronic kidney disease yielded the best AUC, ranging from 0.82 to 0.84 and 0.80 to 0.82 in diabetic and hypertensive patients, respectively. Additionally, the present study examined the comprehensive health expenditure of individuals diagnosed with diabetes and hypertension. This expenditure comprised direct medical costs, direct non-medical costs (including food and travel), and opportunity costs incurred by both patients and their caregivers. The annual total expenditures per patient vary as follows: 6,114 baht, 25,322 baht, 18,124 baht, and 37,575 baht for patients without both diabetes and hypertension, patients with only diabetes, patients with only hypertension, and patients with both conditions. Caregiver opportunity cost is the determining factor for health expenditure differences between patients with diabetes and hypertension alone. It is recommended that the Ministry of Public Health incorporate the diabetes and hypertension risk models into nationwide diabetes and hypertension screening programs for individuals aged 35 years and older. These models offer a more comprehensive assessment of the risks involved in diabetes and hypertension than point-of-care glucose tests and blood pressure measurements alone. In addition, prior to implementing the risk model, criteria should be established to categorize patients according to their risk, and health recommendations should be tailored to each risk group. This will enable local agencies to efficiently disseminate information to the public.
