RÉSUMÉ
BACKGROUND AND OBJECTIVES: Because blood pressure (BP) tracks from childhood to adulthood, some have recommended screening to identify children at high risk for the development of essential hypertension as adults. Others, however, have argued against this strategy because they believe that correlations between childhood and adulthood BP levels are too low. To address these issues, we considered prediction of adult BP from childhood levels. DESIGN: Cohort study with follow-up of participants from childhood to early adulthood. SETTING AND PARTICIPANTS: We ascertained BP in 337 schoolchildren from East Boston, Mass., and reexamined 317 (94%) of them 8 to 12 years later, at ages 18 to 26 years. MEASUREMENTS AND DATA ANALYSIS: On each of several visits, 1 week apart (four for children, three for adults), we obtained three BP readings with a random-zero sphygmomanometer. To calculate correlation coefficients (tracking correlations) between childhood and adult BP, we assumed a multivariate normal distribution and used an iterative maximal likelihood approach in a longitudinal model. We then used these correlations in expressions for sensitivity and specificity of childhood BP as a screening test and for positive predictive value for adult BP above specified cutoff points. RESULTS: During the 8- to 12-year interval, tracking correlations, corrected for within-person variability and adjusted for age, sex, smoking, and medication and alcohol use, were 0.55 for systolic BP and 0.44 for diastolic BP. The corresponding positive predictive values indicated that, for example, the probability that a 20-year-old man's true systolic BP will be > 139 mm Hg (> 90th percentile) was 0.44, given that his observed average systolic BP at age 10 years was > 95th percentile (> 117 mm Hg). The sensitivity of a 10-year old boy's systolic BP > 95th percentile to detect systolic BP > 139 mm Hg 10 years later was 0.17. The specificity of his BP < 95th percentile at age 10 years to detect systolic BP < 139 mm Hg at age 20 years was 0.97. For diastolic BP, predictive values and sensitivities were somewhat lower. CONCLUSIONS: After correction for within-person variability, tracking correlations from childhood to early adulthood are higher than previously reported. However, the resulting sensitivities and predictive values for childhood BP as a screening test for adult BP are of only modest magnitude. These data call into question the usefulness of routine BP measurement to identify children at high risk for the development of essential hypertension.