A risk score predicting new incidence of hypertension in Japan

Abstract

The prevention of hypertension starts with the awareness of risk. Our aim was to construct a simple and well-validated risk model for nonhypertensive people in Japan consisting of basic clinical variables, using a dataset for two areas derived from the Japan Multi-Institutional Collaborative Cohort Study. We constructed a continuous-value model using data on 5105 subjects participating in both the baseline survey and a second survey conducted after 5 years. The area under the receiver operating characteristic curve (AUC) and the Hosmer-Lemeshow χ² statistic for the entire cohort were 0.826 and 7.06, respectively. For validation, the entire cohort was randomly divided 100 times into derivation and validation sets at a ratio of 6:4. The summarized median AUC and the Hosmer-Lemeshow χ² statistic were 0.83 and 12.2, respectively. The AUC of a point-based model consisting of integer scores assigned to each variable was 0.826 and showed no difference, compared with the continuous-value model. This simple risk model may help the general population to assess their risks of new-onset hypertension.