To estimate the sensitivity and specificity of self-reported hypertension
(HTN) as compared with the clinical diagnosis in epidemiological studies.
We searched MEDLINE, Embase, Scopus, Web of Science, LILACS, Google Scholar, and ProQuest Dissertations & Theses Global: Health & Medicine databases. In addition, we screened the references’ lists of relevant reports to identify potentially eligible articles. There were no date or language restrictions. Studies were selected by two independent reviewers, who also extracted data and assessed methodological quality using the Quality Assessment of Diagnostic Accuracy Studies criteria. A meta-analysis
was performed to summarize sensitivity and specificity across studies and estimate heterogeneity.
Out of 2304 records, 22 were included, corresponding to a population of 112 517 adults (55% women). There was substantial variation in sensitivity and specificity across countries and age groups. Several different techniques, devices, and reference ranges were used to diagnose HTN, and self-reporting underestimated its prevalence in the majority of studies. The sensitivity was 42.1% (95% confidence interval 30.9–54.2) and the specificity was 89.5% (95% confidence interval 84.0–93.3), with high heterogeneity (I2
Less than half of patients with HTN would not be identified by self-reporting in epidemiological studies. Self-reported HTN has important limitations and may represent an important source of bias in research depending on regional, socioeconomic, and cultural differences.