Abstract FR450: The Use of Automated Blood Pressure Measurement Devices is Associated with Decreased Threshold Bias for Systolic Blood Pressure in a Large Healthcare System in Central and Northeastern Pennsylvania

Introduction: Studies have shown that when office blood pressure (BP) is measured, a common error is terminal digit preference whereby specific terminal digits, such as “0”, are recorded. In addition, quality metric goals may inadvertently incentivize BP recording just below goal BP level (i.e. threshold bias). Limited data exist on the relationship between automated vs. manual BP device usage and threshold bias. In preparation for implementation of automated office BP devices across primary care in a large regional healthcare system, we examined threshold bias across clinics. We hypothesized that clinics using automated BP devices would show less threshold bias compared to clinics using manual devices. Methods: We identified the last outpatient BP measure in 2023 for patients on anti-hypertensive medications receiving care at Geisinger, a healthcare system in Pennsylvania. We included clinics with ≥200 patients and defined use of manual or automated BP devices based on the proportion of systolic BP (SBP) measurements with odd values. Clinics were defined as using automated devices if ≥40% of SBP measurements were odd (theoretical expectation ~50%), manual devices if <10% of their SBP values were odd, and a mix of automated/manual device usage if 10-<40% of values were odd. Threshold bias was assessed by calculating the ratios of clinic-level counts of SBP values 130-139/140-149 and 120-129/130-139 mm Hg, and diastolic BP (DBP) values 85-89/90-94 mm Hg. Higher ratios may indicate possible threshold bias. Results: Of 192 qualifying clinical departments, 105 were classified as using manual devices, 61 as using automatic devices, and 26 as using a mix of automated/manual devices. The mean (SD) ratio of SBP 130-139/140-149 mm Hg was 2.2 (0.94) in manual device clinics, 1.4 (0.41) in mixed device clinics, and 1.3 (0.29) in automated device clinics (ANOVA p<0.001). The mean (SD) ratios of SBP 120-129/130-139 mm Hg were 1.4 (0.37), 1.0 (0.21), and 0.87 (0.18) in manual, mixed, and automated device clinics, respectively (ANOVA p<0.001). The same trend was not observed for the ratio of DBP 85-89/90-94 mm Hg. Conclusions: Use of automated BP devices, based on recorded BP values, is associated with lower threshold bias for SBP. Ongoing efforts to implement automated BP devices are underway with expectations of improving BP accuracy and addressing therapeutic inertia to reduce cardiovascular risk.