Development of cardiovascular disease risk prediction model for Sri Lankans

Abstract

INTRODUCTION AND OBJECTIVES: There are no Cardiovascular (CV) risk prediction models derived from Sri Lankans. Therefore, we aimed to develop a model to predict the risk of cardiovascular diseases (CVD) among Sri Lankans. METHODS: We developed a model to predict the risk of developing CVDs among Sri Lankans by comparing risk factors of patients who have had and haven’t had acute CVDs. Risk factors were selected depending on the odds ratios of each risk predictor and the feasibility of using those in clinical practice. Two separate models were developed for diabetics and non-diabetics. A scoring system [diabetics; 0-23 and non-diabetics 0-14] was designed based on weighted scores of each risk predictor. Predictive validity of the model was tested by calibration and discrimination. Receiver Operator Characteristic (ROC) curve was used to determine the cut-off value. RESULTS: The model consisted of five predictors; sex, current-smoking status, premorbid systolic blood pressure > 140 mmHg, antihypertensive medication usage and high-density-lipoprotein(HDL) < 45 mg/dL. Discrimination of the model was measured by the area under the ROC curve (diabetics; 0.76, 95% Confidence Interval: 0.68-0.84, non-diabetic; 0.91, 0.86-0.96). Calibration with goodness of fit by Hosmer and Lemeshow test (diabetics; p=0.75, non-diabetics; 0.66) was satisfactory. The tool demonstrated a good predictive ability with sensitivity and specificity of 71.1% (95%CI: 61.3% - 80.8%) and 68.4%(65.3% - 80.5%) in diabetics and 82.2% (95%CI: 72.7% - 91.7%) and 90.9% (95%CI: 84.9% - 96.9%) in non-diabetics. CONCLUSION: The model demonstrated good discrimination and well calibration. It can be used in screening high-risk Sri Lankans for developing cardiovascular diseases.