Predictive Values of Routine Blood Pressure Measurements in Screening for Hypertension

There presently is no consensus regarding criteria for hypertension screening or case finding. Blood pressure variability results in differential ascertainment, depending upon the number of screening visits, the number of measurements per visit, and the method used in reducing the data obtained to an underlying measure of blood pressure status. To develop screening rules, the authors have computed the predictive values of blood pressure measurements for particular age-, race-, and sex-specific groups. Predictive value computations require estimates of between- and within-visit variability and of prevalence. Estimates of variability were calculated from data obtained at a worksite or a community blood pressure program from 991 persons, aged 30-69 years and not currently on antihypertensive medications, who were screened over two to five visits, one week apart. Estimates of prevalence were obtained from the screening of 158,955 adults by the Hypertension Detection and Follow-Up Program. Predictive values are presented for particular age-, race-, and sex-specific groups over a wide range of diastolic blood pressures, and are used to identify appropriate screening rules that minimize misclassification with the fewest possible blood pressure measurements. The results of a questionnaire sent to 30 hypertension experts indicated that the median acceptable predictive values for making screening decisions were 80% for predictive value positive and 77.5% for predictive value negative. When these criteria were adopted, 81%, 90%, and 93% of the subgroup of 901 persons with at least three visits from the above screened population were identified as having or not having true diastolic blood pressure greater than or equal to 90 mmHg after one, two, and three visits, respectively. The status of the remaining 7% remained uncertain after three visits.