1069 — Are We Underestimating Blood Pressure in the Clinic? Concerning Discrepancies between Manual and Automated Measurements

Robinaugh D (VA New York Harbor Healthcare System), Friedberg JP (VA New York Harbor HCS), Lipsitz S (Brigham and Women’s Hospital, Harvard Medical School, Boston, MA), Ulmer M (VA New York Harbor HCS), D'Orio V (VA New York Harbor HCS), Natarajan S (VA New York Harbor Healthcare System, New York University Medical School)

Objectives:
Despite the reliability of automated blood pressure (BP) measurement, manual measurement remains the primary means of assessment in the VA. We aimed to: 1) Evaluate whether there are discrepancies between BP taken manually versus those taken with an automated machine, 2) Estimate and quantify potential sources of variation, particularly zero-digit terminal digit preference (TDP), a common source of observer bias where measurements ending in zero occur at higher frequencies than other digits.

Methods:
We used data from 441 hypertensive veterans who had their BP measured manually as part of their clinic visit. Participants were then approached by a trained research assistant to participate in the study, at which point their BP was measured using the Omron HEM-907XL automated BP machine. To determine if there were differences between manual and automated BP, we: 1) Assessed if the paired difference (manual BP – automated BP) had median zero using Wilcoxon sign-rank tests, 2) Compared TDP between manual and automated BP measurement using McNemar's test, and 3) Evaluated the effect of TDP on measurement discrepancy by comparing relative bias [(manual BP–automated BP)/(automated BP)] between TDP and no TDP using Wilcoxon rank-sum tests.

Results:
Manual BP measurements were lower than automated BP (p < .0001) for both systolic (148 vs. 138 mm Hg) and diastolic BP (80 vs. 77 mm Hg). TDP was more frequent in manual BP (50% of measurements) than in automated measurements (11%; p < .001). The relative bias for manual BP measurements showing TDP was greater than measurements with no TDP (-7.1% vs. -5.0% for SBP; p < .05 and -4.0% vs. -2.5% for DBP; p=.27).

Implications:
Manual BP measurement may systematically underestimate actual BP. While TDP accounts for a portion of the relative bias between manual and automated BP, significant differences remain even in scores not showing TDP. Possible explanations for the remaining discrepancy between manual and automated BP measurements include expectation bias, insufficient training, and observer-participant interaction.

Impacts:
The data suggest that the continued use of manual BP could lead to misdiagnosis and inadequate care for individual patients in clinical settings and systematically underestimate blood pressure data in research settings.