Introduction: There is growing interest in embedding research protocols into clinical practice, including use of electronic health records (EHR) for measures of blood pressure (BP). However, the accuracy and reliability of EHR-derived BP compared to research-based BP are unknown.
Hypothesis: We hypothesized EHR-derived measurements would overestimate systolic BP (SBP) and have higher variance compared to research measurements.
Methods: We analyzed EHR-derived BP and research BP measurements in participants from a randomized trial, Five+ Nuts and Beans. EHR-derived measurements were made following the implementation of a standardized BP protocol, including the use of automated devices, clinical training, and systems support, to enhance the quality of clinic BP measurements. EHR-derived SBP measurements and change in SBP, calculated from closest clinic visits prior to the start and after the end of the trial period, were compared with measures obtained in the research setting.
Results: Participants were 87% female (107 of 123); mean age was 58.6±9.7 yrs. Of the 123 participants, 118 had at least one primary care visit within 6 months of the research period. EHR-derived SBP were higher and more variable than research measurements, though these differences were attenuated for visits occurring within 7 days of research measurements (Table 1). EHR-based change in SBP was also more variable (SD=19.6) than research-based change (SD=14.0); using these SD’s, we estimate that a 2-arm trial conducted with clinic measurements would need to enroll 95% more participants (324 vs 166) to detect a 5 mmHg difference in SBP at a power of 0.9.
Conclusions: Despite implementation of a standardized BP protocol, clinic measurements of SBP were higher and more variable than research protocol-based measurements. These results suggest that, in clinical trials, use of BP measurements obtained in clinical care, even if standardized, will reduce power and require greater sample size compared to measurements obtained in research settings.