Prediction model and assessment of probability of incident hypertension: the Rural Chinese Cohort Study

We aimed to develop a hypertension risk-prediction model among rural Chinese people. We included data for 9034 participants aged 18–70 years without baseline hypertension, diabetes, myocardial infarction, stroke, or heart failure in a rural Chinese cohort. The sample was randomly divided into a training set (60%) and testing set (40%). We used shrinkage estimates by the least absolute shrinkage and selection operator method in fitting a logistic model to explore the possibility of predicting the risk of hypertension in the training set. On multivariable analysis, age, parental hypertension, systolic and diastolic blood pressure, body mass index (BMI), and age by BMI were significant predictors of hypertension. After bootstrap validation, the corrected C-index, calibration intercept, and calibration slope were 0.7932, −0.0041, and 0.9938, respectively for the training set. Our model also had good discrimination (C-index, 0.7914 [95% CI 0.773–0.809]) and calibration (Hosmer–Lemeshow χ2 = 14.366, P = 0.073) for the testing set. Nomograms and score-based models were used to favor the clinical implementation and workability of the risk model. According to the risk score based on these factors, the cumulative risk for hypertension was <20% for 57.62% of participants, 20–40% risk for 27.24%, 40–60% for 12.19%, and >60% for 2.96% during the 6-year follow-up. The score-based area under the receiver operating characteristic curve for the present model and the Framingham risk-score model were similar (P = 0.282). The hypertension risk-prediction system we developed provides convenient approaches to identify individuals at high risk of hypertension.