LAUSR

Objectives: There is a systematic question, “What is the optimal target for blood pressure (BP) lowering during antihypertensive therapy in adults?” Answers are usually on basis of clinical trials and their results, but their data are always limited. ROC curve analysis is a non-parametric method and can be used for this reference. Methods: Data in this paper come from clinical trials FEVER (The Felodipine Event Reduction Study), CHIEF (Chinese Hypertension Intervention Efficacy Trial) and an epidemiologic study APCSC (Asia Pacific Cohort Studies, Chinese Studies). ROC (receiver operating characteristic)curves were calculated and drown with aid of software SAS 9.2. In this paper high BP is a risk factor (for stroke and all CVD events) and also a marker variable (for ROC). The stroke and CVD are state variables of the ROC curves. These two kinds of variables are on basis of ROC curves. Results: Clearly, relation between CVD (yes/no) and BP is a proper function corresponds to the standard logistic distribution. Some Cut-off points of the curves are used as estimates of BP norms. In the Table,MAX YD: maximum of Youden's index; CROSS: the cross-over point of TPR and TNR; 80%TPR: 80% true positive rate. Conclusion: The cut-off points of ROC curves suggested the optimal target for SBP which varying between 127/76 and 149/88, according different data source and cut-off points.