- Compared with single blood pressure measurement, using the average of past multiple measurements of systolic blood pressure (SBP), as an indicator of long-term exposure to elevated blood pressure, demonstrates a stronger association with incident cardiovascular disease (CVD).
- However, including information on long-term SBP that included current and other risk factors only minimally improved the performance of the multivariate models to predict future risk of CVD.
Why this matters
- Patients’ previously recorded BP measurements may be an indicator of their “usual” or long-term BP level, which may help to understand, communicate and put into context CVD risk associated with increased BP.
- 80,964 participants (age, 50 years) with BP measurement and without CVD and antihypertensive or lipid-lowering prescriptions were included using electronic health records (EHRs) from the UK Clinical Practice Research Datalink (CPRD).
- Funding: The Oxford Martin School and the National Institute for Health Research Oxford Biomedical Research Centre.
- Increase in current SBP (per 20-mmHg increase) was strongly associated with increased risk for incident CVD (adjusted HR [aHR], 1.22; 95% credible interval [CrI], 1.18-1.30), but associations were stronger for past (mean and time-weighted mean) and usual SBP (aHR ranging from 1.39 [95% CrI, 1.31-1.46] to 1.45 [95% CrI, 1.36-1.62]).
- In a model which included current SBP, sex, smoking, deprivation, diabetes mellitus, and lipid profile, the area under the receiver operating characteristic curve (AUC) was 0.747 (95% CrI, 0.722-0.811).
- The addition of past SBP mean, time-weighted mean or variability to this model increased the AUC to 0.750 (95% CrI, 0.727-0.811), 0.750 (95% CrI, 0.726-0.811) and 0.748 (95% CrI, 0.723-0.811), respectively, with all models showing good calibration.
- Routine EHRs did not include data on other risk factors or confounders.