Importance: There are concerns with translating results from the Systolic Blood Pressure Intervention Trial (SPRINT) into clinical practice because the standardized protocol used to measure blood pressure (BP) may not be consistently applied in routine clinical practice.
Objectives: To evaluate the concordance between BPs obtained in routine clinical practice and those obtained using the SPRINT protocol and whether concordance varied by target trial BP.
Design, setting, and participants: This observational prognostic study linking outpatient vital sign information from electronic health records (EHRs) with data from 49 of the 102 SPRINT sites was conducted from November 8, 2010, to August 20, 2015, among 3074 adults 50 years or older with hypertension without diabetes or a history of stroke. Statistical analysis was performed from May 21, 2019, to March 20, 2020.
Main outcomes and measures: Blood pressures measured in routine clinical practice and SPRINT.
Results: Participant-level EHR data was obtained for 3074 participants (2482 men [80.7%]; mean [SD] age, 68.5 [9.1] years) with 3 or more outpatient and trial BP measurements. In the period from the 6-month study visit to the end of the study intervention, the mean systolic BP (SBP) in the intensive treatment group from outpatient BP recorded in the EHR was 7.3 mm Hg higher (95% CI, 7.0-7.6 mm Hg) than BP measured at trial visits; the mean difference between BP recorded in the outpatient EHR and trial SBP was smaller for participants in the standard treatment group (4.6 mm Hg [95% CI, 4.4-4.9 mm Hg]). Bland-Altman analyses demonstrated low agreement between outpatient BP recorded in the EHR and trial BP, with wide agreement intervals ranging from approximately -30 mm Hg to 45 mm Hg in both treatment groups. In addition, the difference between BP recorded in the EHR and trial BP varied widely by site.
Conclusions and relevance: Outpatient BPs measured in routine clinical practice were generally higher than BP measurements taken in SPRINT, with greater mean SBP differences apparent in the intensive treatment group. There was a consistent high degree of heterogeneity between the BPs recorded in the EHR and trial BPs, with significant variability over time, between and within the participants, and across clinic sites. These results highlight the importance of proper BP measurement technique and an inability to apply 1 common correction factor (ie, approximately 10 mm Hg) to approximate research-quality BP estimates when BP is not measured appropriately in routine clinical practice.
Trial registration: SPRINT ClinicalTrials.gov Identifier: NCT01206062.