Prospective external validation of the predicting out-of-office blood pressure (PROOF-BP) algorithm: an observational cohort study

Problem

The PROOF-BP algorithm, which combines change in clinic blood pressure over multiple readings on a single occasion with patient characteristics, has been shown to predict a patient’s out-of-office blood pressure (BP) level and hence determine whether subsequent ambulatory blood pressure monitoring (ABPM) is likely to affect decision making in the diagnosis of hypertension. The present study aimed to collect diagnostic data from routine practice to prospectively validate this new approach.

Approach

This study used a Prospective observational cohort design. Ten Primary Care practices and one hospital in the UK enrolled consecutive patients aged >18 years, referred for ABPM. All participants underwent clinic BP monitoring, with the PROOF-BP algorithm applied, as well as daytime ABPM. Pre-specified outcomes included the proportion of true/false positive/negative results for detecting hypertension using the PROOF-BP strategy (algorithm + ABPM) compared to the reference ABPM. The sensitivity and specificity of the PROOF-ABPM strategy for detecting hypertension was compared to existing approaches including those recommended in the UK and US. Area under the receiver operator characteristic curve (AUROC) statistics were used to examine model performance.

Findings

A total of 887 patients (mean age 52.8±16.2 years) were enrolled during the study period across all sites. PROOF-BP had a sensitivity of 97% which was better than existing approaches recommended in the UK (78%) and US (86%). However, PROOF-BP had a specificity of 76% which was worse than both UK (97%) and US (100%) approaches. AUROCs were highest in those strategies utilising ABPM. Overall, PROOF-BP resulted in more patients having their hypertensive status correctly classified (difference vs. UK 6%, p<0.001; difference vs. US -0.3%, p=0.802) and less utilisation of ambulatory monitoring (difference vs. UK -12%, p<0.001; difference vs. US -28%, p<0.001).

Consequences

Prospective validation of the PROOF-BP approach suggests it is possible to accurately classify an individual’s hypertensive status with comparable or better accuracy and less utilisation of ABPM compared with existing strategies. The PROOF-BP strategy can be recommended for use in routine practice for patients where ambulatory blood pressure monitoring is being considered.