Translating Ambulatory Cuffless Wrist Blood Pressure Estimates to Cuffed Blood Pressure Equivalents for Clinical Decision-Making: The Continuum Bp Study

Objective: Current hypertension guidelines do not recommend cuffless wrist blood pressure (BP) devices for clinical use. Yet their prevalent use continues to rise. Clinicians are therefore confronted with cuffless BP data without a valid evidence-based framework for interpretation and decision making. In the Continuum BP study, we aimed to develop equations to translate BP estimates from two wrist BP devices employing distinct cuffless technologies—photoplethysmography (PPG; Aktiia/Hilo) and applanation tonometry (AT; Healthstats BPro Evo)—into corresponding 24-hour ambulatory BP monitoring (ABPM) values to support clinical use.Design and method: In a randomized crossover trial, 48 participants wore each wrist BP device in random order with concurrent 24-hour ABPM within a two-week interval, to complete two non-overlapping periods of BP recordings per participant. We quantified mean differences in average daytime (7am-11pm) systolic BP (SBP) between each wrist BP device and ABPM. Device-specific linear regression models predicted average daytime cuffed ambulatory SBP using cuffless SBP and covariates, selected via backward elimination from demographic, clinical, and lifestyle variables. Results: Sufficient data (14 or more daytime readings from all devices) was available from 34 participants (mean age 52 years; 50% female; 35% non-White; 29% hypertensive). The PPG device exhibited mean bias of -3.51 mmHg (SD 12.57). Device estimated average daytime cuffless SBP alone explained 41.8% of the variance in average daytime cuffed SBP (Table 1). A prediction model additionally incorporating Black ethnicity, body mass index, oscillometric SBP, and stress score improved explained variance to 52.6% with root mean square error (RMSE) 11 mmHg. The AT device showed mean bias of -10.98 mmHg (SD 12.45). Device estimated average daytime cuffless SBP alone explained 37.6% of the variance in average daytime cuffed SBP (Table 1). A model, additionally including age, body mass index, oscillometric SBP, stress score, and alcohol intake, improved explained variance to 71.7% with RMSE 8 mmHg. Conclusions: Device-specific prediction equations incorporating relevant individual characteristics enable translation of cuffless BP estimates to cuffed ABPM equivalents. Incorporating physical activity data may improve prediction accuracy. External validation is required to establish generalisability and clinical utility.