Arm Based on LEg blood pressures (ABLE-BP): can systolic ankle blood pressure measurements predict systolic arm blood pressure? An individual participant data meta-analysis from the INTERPRESS-IPD Collaboration

OBJECTIVES: To determine associations between arm and ankle systolic blood pressures (SBPs), develop and validate a multivariable model predicting arm SBP from ankle SBP, and investigate associations between ankle SBP, cardiovascular disease and mortality. DESIGN: Ankle-arm SBP differences were examined in two-stage individual participant data (IPD) meta-analyses using multivariable hierarchical linear regression models. Models were used to derive and validate a prediction model for arm SBP based on ankle SBP. Model performance was assessed using area under the receiver operating characteristic (AUROC) curve analyses. Prognostic associations of ankle SBP with outcomes were examined using Cox proportional hazards models. DATA SOURCES: Searches identified cohorts for the Inter-arm Blood Pressure Difference IPD (INTERPRESS-IPD) Collaboration from Medline, Old Medline, Medline in process, Embase and CINAHL databases from inception until January 2017; unpublished data were also sought. Required primary outcomes were all-cause mortality, cardiovascular mortality, and/or fatal and non-fatal cardiovascular events. ELIGIBILITY CRITERIA: Prospective studies from community, primary care or general clinic settings, without language restriction, that recorded SBP in both arms were eligible. Adults aged ≥18 years with SBP measured in all four limbs, in a supine position, were included in the current analyses. People with peripheral artery disease were excluded. DATA EXTRACTION AND SYNTHESIS: Anonymised datasets were individually cleaned and then combined into a single dataset for the INTERPRESS-IPD Collaboration. RESULTS: The current dataset included 33 710 participants from 14 studies; mean age 58 years, 45% female, mean baseline arm blood pressure 138/80 (SD: 20/12) mm Hg. Mean ankle SBP was 12.0 mm Hg (95% CI 8.8 to 15.2) higher than arm SBP. The multivariable model predicting arm SBP from ankle SBP demonstrated excellent performance (AUROC curves, sensitivities and specificities were >0.82, 0.80 and 0.82, respectively, at all BP thresholds from 130 to 160 mm Hg). Model performance was superior to existing arithmetic formulae.Ankle SBP was neither associated with all-cause nor cardiovascular mortality (HR 1.000 (0.997 to 1.002; p=0.682) and 1.001 (0.996 to 1.005; p=0.840), respectively). However, lower-reading ankle SBP was associated with fatal or non-fatal cardiovascular events (HR 1.005 (1.002 to 1.007; p<0.001). CONCLUSIONS: On average, ankle SBP is 12 mm Hg higher than arm SBP. Estimating individual arm SBP from ankle SBP measurements with a multivariable model is more accurate than existing fixed arithmetic formulae. This model, operationalised in an online calculator (https://ablebp. RESEARCH: exeter.ac.uk/), could facilitate hypertension management and cardiovascular care for people unable to have arm SBP measured. PROSPERO REGISTRATION NUMBER: CRD42015031227.