BACKGROUND

Uncontrolled hypertension is the largest single contributor to all-cause and cardiovascular mortality. Although most adults with hypertension are treated, more than 1 in 3 may be unaware they have high blood pressure (BP). The US Preventive Services Task Force strongly recommends screening all adults aged ≥18 years for hypertension, with follow-up testing using 24-hour ambulatory BP monitoring (ABPM) or home BP monitoring (HBPM) before a new hypertension diagnosis. Clinicians rarely order ABPM, however; they sometimes suggest HBPM but not according to recommended protocols. Whether BP kiosks (often in pharmacies) could be used to diagnose hypertension is unknown.

OBJECTIVES

BP Checks for Diagnosing Hypertension (BP-CHECK) was a randomized controlled study to evaluate performance, patient acceptability, and patient attitudes about clinic, HBPM, and kiosk hypertension diagnostic testing compared with ABPM (reference or gold standard) and longer-term outcomes after hypertension diagnostic testing. We also assessed provider knowledge, beliefs, and preferences about hypertension diagnostic tests.

METHODS

We enrolled individuals aged 18 to 85 years who had no hypertension diagnosis, were taking no antihypertensive medications, and had a systolic BP (SBP) ≥138 mm Hg or a diastolic BP (DBP) ≥88 mm Hg in the electronic health record (EHR) and an SBP ≥140 mm Hg or DBP ≥90 mm Hg at a screening visit. Participants were randomly assigned to 1 of 3 regimens for diagnosing hypertension: (1) clinic BP (follow-up clinic appointment), (2) HBPM (duplicate measurements twice daily for 5 days), or (3) kiosk BP in clinic or at a nearby pharmacy (triplicate measurements over 3 days). Participants were asked to complete their assigned regimen within 3 weeks of randomization. Participants returned for ABPM (the reference standard for diagnosing hypertension) at 3 weeks, with a final research visit 6 months after study enrollment. The primary outcome was the mean difference in SBP between the clinic, HBPM, and kiosk diagnostic regimens and the reference test (daytime ABPM). Secondary outcomes included mean differences in DBP and sensitivity and specificity (as well as positive and negative likelihood ratios) of the diagnostic regimens relative to the reference test. Additional secondary outcomes included adherence to diagnostic regimens, patient-reported outcomes after diagnostic testing and at 6 months, and BP outcomes at 6 months by randomization group. Qualitative interviews with patients and providers assessed their views on the BP measurement regimens as well as their interpretations of BP measurement and hypertension diagnostic testing.

RESULTS

Of 510 individuals randomly assigned, the mean age was 59 years; 48% were female, 80% were White, 7% were African American, 13% were other races, and 4% were Hispanic White. Mean baseline BP was similar across groups: 150/88 mm Hg. Among the 510 participants, 76 (14.9%) were excluded from primary analyses because of fewer than 14 daytime ABPM measurements and/or no clinic, HBPM, or kiosk BP measurements. Among those randomly assigned, 85.1% (434/510) had sufficient data for primary analyses (clinic, 82.6%; HBPM, 89.4%; kiosk, 83.3%). Using a diagnostic threshold of ABPM daytime mean SBP ≥135 mm Hg or DBP ≥85 mmHg as the reference standard, 71.7% of participants tested positive for hypertension. Compared with the reference standard, clinic, HBPM, and kiosk mean differences in SBP were 4.7 mm Hg lower, 0.1 mm Hg lower, and 9.5 mm Hg higher, respectively ( < .001 clinic and kiosk). Mean differences in DBP were 7.2 mm Hg lower, 0.4 mm Hg lower, and 5.0 mm Hg higher than mean daytime ABPM, respectively ( < .001 clinic and kiosk). The sensitivities of clinic, HBPM, and kiosk for detecting hypertension were 31.1%, 82.2%, and 96.0%, respectively. Specificities were 79.5% (clinic), 53.3% (HBPM), and 28.2% (kiosk); false positives were 5.6%, 13.8%, and 20.0%; and false negatives 50.0%, 12.5%, and 2.9%, respectively. Positive likelihood ratios of clinic, HBPM, and kiosk were 1.51, 1.76, and 1.34 respectively. Negative likelihood ratios were 0.87, 0.33, and 0.14, respectively. Adherence to BP data collection, defined as completing the minimum number of required BP checks, was 87.2% for clinic, 90.6% for HBPM, and 67.9% for kiosk, with 91.6% of all participants completing ABPM. HBPM was preferred. ABPM was more likely to disturb sleep and be uncomfortable, but participants perceived it as accurate. Six months after randomization, all groups had lower BP levels, with no significant differences in BP reduction, lifestyle behavior, or quality of life by randomization group. The proportion who were treated, defined as 1 or more prescription fills for an antihypertensive medication between randomization and 6-month follow-up, was similar across randomization groups (clinic, 25.6%; HBPM, 24.7%; and kiosk, 24.4%). Among individuals with hypertension based on ABPM testing, 40.1% (125/312) had a hypertension diagnosis recorded in the EHR by their provider between baseline and 6-month follow-up. Individuals with high BP measurements on ABPM and an EHR hypertension diagnosis had significantly greater declines from baseline to 6 months for both SBP and DBP compared with individuals with high BP measurements on ABPM and no EHR hypertension diagnosis (mean difference in SBP, −4.9 mm Hg [range, −7.8 to −2.1], < .001; mean difference in DBP, −2.3 mm Hg [range, −4.3 to −0.4], = .02). Providers believed that measuring BP manually using a stethoscope was more accurate than using automated, home, or kiosk BP monitors. Most physicians did not know standard protocols or out-of-office BP thresholds for diagnosing hypertension.

CONCLUSIONS

HBPM had better diagnostic performance, adherence, and acceptability than clinic or kiosk methods for making a new hypertension diagnosis. Most participants with high BP measurements on screening and ABPM diagnostic testing did not receive a hypertension diagnosis. New strategies are needed to improve hypertension diagnosis.

LIMITATIONS

Participants had to have high BP measurements in the EHR and at a baseline visit to be eligible, and most were confirmed to have hypertension on ABPM. Thus, our population may not be generalizable to those with lower BP measurements (and lower hypertension prevalence). Diverse racial and ethnic groups were underrepresented.