Introduction: A diagnosis of hypertension should be based on multiple blood pressure (BP) measurements, taken on at least two separate occasions. We aimed to assess the impact of considering different criteria for a definition of hypertension, based on number of visits for blood pressure measurement, on estimates of hypertension prevalence, awareness, treatment and control, and on its association with two variables definitely related to hypertension: body mass index and left ventricular hypertrophy.
Methods: We used data from a cross-sectional study of 739 participants, aged > or = 45 years, randomly selected from a non-institutionalized Portuguese population, from January 2001 to December 2003. Main outcome measures were prevalence of hypertension (systolic BP > or = 140 mmHg and/or diastolic BP > or = 90 mmHg or current antihypertensive drug therapy) based on BP measurements on one visit, on BP measurements on a second visit or on fulfilling the same criteria on the two different visits.
Results: Estimated hypertension prevalence was 63.4% (95% CI: 59.8-66.9) using BP measurements from the first visit (HTN1) and 60.2% (95% CI: 56.6-63.8) using BP measurements from the second visit (HTN2). If both visits are used as criteria the estimated hypertension prevalence (HTN(Final)) was 56.3% (95% CI: 52.7-60.0), p (McNemar test) < 0.001, between HTN1 and HTN(Final) and between HTN2 and HTN(Final). Awareness, treatment and control changed from 60.2% to 64.4%, 53.1% to 59.8% and 24.9% to 22.0%, respectively, when using information from the first or both visits. All three different estimates of hypertension prevalence have a similar strong and independent association with body mass index (OR = 2.71 for body mass index > or = 30 with HTN(Final)) and with left ventricular hypertrophy (OR = 3.21 for HTN(Final) with left ventricular hypertrophy).
Discussion and conclusions: In many individuals labeled as hypertensive on a single evaluation, hypertension was not confirmed on reassessment, leading to a significant overestimation of 12.6% of the true prevalence. For this reason BP should be measured on at least two office visits both for clinical purposes and in epidemiological studies. On the other hand, this did not reflect on the association between hypertension and body mass index or left ventricular hypertrophy, suggesting that unconfirmed cases do not necessarily imply misclassification.