Introduction: The use of self-reported biometric values, such as blood pressure (BP), has been proposed as an efficient strategy for monitoring clinical care, evaluating health system performance, and conducting pragmatic randomized trials. Unfortunately, there is limited evidence about whether self-reported biometric readings are accurate and, if so, whether their accuracy is predicted by readily identifiable patient characteristics. Enrollment data from the ongoing Medication adherence Improvement Support App For Engagement - Blood Pressure (MedISAFE-BP) trial provide a unique opportunity to address these questions.
Methods: MedISAFE-BP is a randomized clinical trial evaluating the effect of the Medisafe smartphone application on BP among subjects with poorly controlled hypertension, defined as ≥140mmHg systolic per JNC8 guidelines. Subjects were recruited through online patient communities, social media, and targeted advertisements. Subjects who indicated that their BP was poorly controlled while on medication underwent further screening. After informed consent, subjects provided baseline information including demographics, comorbidities, the number of BP medications they were currently taking, hypertension knowledge, patient activation measured by the Consumer Health Activation Index, and self-reported adherence. Subjects were then mailed a home BP cuff to verify their self-reported blood pressure. We evaluated the positive predictive value of self-reported poorly controlled hypertension using the measured BP readings. We then used multivariable logistic regression to identify predictors of having a measured BP value that was actually poorly controlled.
Results: Our study cohort consisted of 1,142 individuals who self-reported as having poorly controlled BP. The positive predictive value of poorly-controlled BP by self-report was only 37%. In fact, 284 (24%) subjects had systolic BPs that were normal (systolic BP < 120 mmHg). Factors that were independently associated with accurate self-report included older age (odds ratio [OR] 1.3 per decade, 95% confidence interval [CI] 1.2-1.5), a history of prior stroke (OR 2.5, CI 1.2-5.2), diabetes mellitus (OR 1.5, CI 1.1-2.2), and a low level of activation (OR 1.63, CI 1.2-2.2). Hypertension knowledge, education, and self-reported adherence were not associated with accurately self-reporting BP.
Discussion: In this cohort of individuals who reported that their BP was poorly controlled, only one-third actually had elevated BP when measured with a home BP cuff. While this discrepancy may have many underlying causes, it suggests that the use of self-reported BPs is not an accurate method of monitoring hypertension control at the population-level. Reassuringly, several factors are independently associated with accurate self-reported BPs, and thus there may be some subgroups for whom self-report can be relied upon.