Impact of inter-arm blood pressure difference on cardiovascular risk estimation in primary care
The majority of cardiovascular events occur in people at low to intermediate estimated risk. Prediction of cardiovascular risk may be refined by taking account of novel risks markers. We have found that an inter-arm blood pressure difference (IAD) is an independent risk marker for cardiovascular events, after adjustment for QRISK2, Atherosclerotic Cardiovascular Disease (ASCVD), or Framingham cardiovascular risk scores. The National Institute for Health and Care Excellence currently recommends QRISK2 thresholds of >10% and >20% for initiation of statin and/or blood pressure lowering treatment as primary prevention of cardiovascular disease. Taking account of IAD therefore offers the potential for improved individual cardiovascular risk prediction, by reclassifying people across these 10-year risk thresholds. This study modelled the impact of adjusting cardiovascular risk prediction scores for IAD in a primary care population free of existing disease.
People aged 45-75 years, free of cardiovascular disease, had bilateral blood pressure measured simultaneously, three times, with a Microlife Watch BP Office device during National Health Service (NHS) Health Checks in one rural general practice in Devon, England. Systolic IAD was defined as mean right minus mean left systolic blood pressure. QRISK2, ASCVD and Framingham risk scores were calculated and adjusted using hazard ratios for IAD derived from our INTERPRESS-IPD Collaboration analyses, based on data from over 57,000 records. Unadjusted and adjusted risk scores were classified according to 10% and 20% risk thresholds and the proportions reclassified were reported. Analyses were carried out using Stata v15.0.
Data existed for 334 participants [mean (standard deviation): age 57.4 (9.3) years, blood pressure 132/79 (14/8.5) mmHg]. Systolic IAD was ≥10 mmHg for 31 (9.3%) participants. Mean risk scores were: QRISK2 8.0 (6.7), ASCVD 6.9 (6.5) and Framingham 10.7 (8.1) before adjustment for IAD, and 8.9 (7.7), 7.1 (6.7) and 11.2 (8.5), respectively, following adjustment. Overall, 18 (5%) participants were reclassified from below to above the 10% or 20% QRISK2 risk thresholds; corresponding figures for ASCVD and Framingham were 3 (1%) and 10 (3%), respectively. For individuals with initial risk scores between 8% and 9.9% only, numbers reclassified above 10% for QRISK2 were 13/35 (37%), ASCVD 3/29 (10%) and Framingham 7/38 (18%).
Our findings confirm that, by taking account of systolic IAD, individual cardiovascular risk estimates for people attending NHS Health Checks can be refined. Using this approach individual evidence-based decisions on interventions for primary prevention of cardiovascular disease can be discussed. This could facilitate targeting of treatment to those at greater than expected cardiovascular risk, particularly when unadjusted risk scores are close to treatment thresholds. This study also confirms the known tendency of Framingham scores to over-estimate risk in an English cohort in comparison to the UK derived QRISK2 score.