Research Letter
Impact of a Smartwatch Hypertension Notification Feature for Population Screening
Jordana B. Cohen, Daniel K. Addo, Joshua A. Jacobs, et al
JAMA 2026;335;(11):1001-1003. doi:10.1001/jama.2025.26925
In September 2025, the US Food and Drug Administration cleared the Apple Watch hypertension notification feature, which uses photoplethysmography signals to notify users of possible hypertension without using a blood pressure (BP) cuff, prompting further evaluation.1 In a validation study comparing the notification feature to home BP monitoring over 30 days,1 41.2% of individuals with undiagnosed hypertension received an alert for potentially having hypertension and 58.8% did not, whereas among those without hypertension, 92.3% did not receive an alert and 7.7% received an inappropriate alert (Figure). These test performance metrics translate to a positive likelihood ratio of 5.35, indicating moderate evidence to rule in hypertension, and a negative likelihood ratio of 0.64, indicating weak evidence to rule out hypertension. We characterized the potential impact of this notification feature on hypertension screening among US adults without diagnosed hypertension.
Figure. Illustration of the Smartwatch Hypertension Notification Feature Performance Among Individuals With and Without Hypertension
This figure depicts visualization of risk based on expected alert patterns using the reported sensitivity and specificity of the Apple Watch hypertension notification feature. A, Blue dots represent true-positive or appropriate alerts and gray dots represent false-negative or inappropriate absence of alerts. B, Pink dots represent false-positive or inappropriate alerts, and light gray dots represent true-negative or appropriate absence of alerts. Each panel is scaled to 100 individuals to illustrate the relative frequency of correct and incorrect notifications.
Methods
We analyzed data from the National Health and Nutrition Examination Survey (NHANES) 2017-2020, a nationally representative, cross-sectional survey of the US population. NHANES obtained institutional review board approval and written informed consent from all participants.2 This report follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Eligibility to use the notification feature was determined based on the intended use population: nonpregnant, 22 years or older, unaware of having hypertension, and not taking antihypertensive medication.1 We estimated the prevalence (ie, pretest probability) of undiagnosed hypertension in the overall eligible population and select subgroups (eMethods in Supplement 1 for additional details). Test performance metrics were applied to estimate the percent of true-positive, false-negative, true-negative, false-positive results; positive and negative likelihood ratios; positive and negative predictive values; and posttest probabilities.3,4 All analyses were performed using R version 4.5.1.
Results
There were 3983 NHANES participants aged 22 years or older, not pregnant, unaware of a prior hypertension diagnosis, and not taking antihypertensive medication, representative of 127 million US adults. Using the estimated prevalence of hypertension in the overall eligible population, the positive predictive value was 69.1% (95% CI, 63.3%-74.9%) and the negative predictive value was 79.0% (95% CI, 76.6%-81.3%). The Table presents the probability of hypertension before and after testing across different subgroups that vary in hypertension prevalence. For example, among eligible individuals younger than 30 years, an alert increases the probability of hypertension from 0.14 (95% CI, 0.11-0.17) to 0.47 (95% CI, 0.39-0.56), wherein the absence of an alert lowers it to 0.10 (95% CI, 0.07-0.12). Among individuals 60 years or older, an alert increases the probability of hypertension from 0.45 (95% CI, 0.39-0.50) to 0.81 (95% CI, 0.76-0.86), and the absence of an alert lowers it to 0.34 (95% CI, 0.29-0.39).
Table. Performance of the Smartwatch Hypertension Notification Feature
| Characteristica | Estimated probability, 95% (CI)b | ||
|---|---|---|---|
| Overall observed prevalence of hypertension (ie, pretest probability)c | Positive posttest probabilityd | Negative posttest probabilitye | |
| Overall | 0.30 (0.27-0.32) | 0.69 (0.63-0.75) | 0.21 (0.19-0.23) |
| Age, y | |||
| <30 | 0.14 (0.11-0.17) | 0.47 (0.39-0.56) | 0.10 (0.07-0.12) |
| ≥30 and <45 | 0.26 (0.22-0.29) | 0.65 (0.58-0.72) | 0.18 (0.15-0.21) |
| ≥45 and <60 | 0.36 (0.32-0.41) | 0.75 (0.70-0.81) | 0.27 (0.23-0.31) |
| ≥60 | 0.45 (0.39-0.50) | 0.81 (0.76-0.86) | 0.34 (0.29-0.39) |
| Sex | |||
| Male | 0.33 (0.29-0.38) | 0.73 (0.67-0.79) | 0.24 (0.20-0.28) |
| Female | 0.26 (0.23-0.29) | 0.65 (0.59-0.72) | 0.18 (0.16-0.21) |
| Race and ethnicity | |||
| Hispanic | 0.24 (0.22-0.26) | 0.63 (0.56-0.69) | 0.17 (0.15-0.19) |
| Non-Hispanic Asian | 0.28 (0.24-0.31) | 0.67 (0.61-0.73) | 0.20 (0.17-0.22) |
| Non-Hispanic Black | 0.36 (0.32-0.40) | 0.75 (0.70-0.81) | 0.26 (0.23-0.30) |
| Non-Hispanic White | 0.30 (0.26-0.34) | 0.69 (0.63-0.76) | 0.21 (0.18-0.25) |
| Non-Hispanic otherf | 0.36 (0.27-0.44) | 0.75 (0.67-0.83) | 0.26 (0.19-0.33) |
| Body mass indexg | |||
| ≤30 | 0.24 (0.21-0.27) | 0.63 (0.56-0.70) | 0.17 (0.14-0.19) |
| >30 | 0.39 (0.36-0.43) | 0.78 (0.73-0.82) | 0.29 (0.26-0.32) |
| Education | |||
| <High school | 0.33 (0.28-0.38) | 0.73 (0.66-0.79) | 0.24 (0.20-0.28) |
| ≥High school | 0.29 (0.26-0.32) | 0.69 (0.63-0.75) | 0.21 (0.18-0.23) |
Discussion
Guidelines recommend population-wide hypertension screening for all adults using validated, cuff-based BP measurement.5 With more than 200 million Apple Watch users worldwide,6 the hypertension notification feature offers opportunities to identify adults with undiagnosed hypertension but with potential for hypertension misclassification as well. A large proportion of individuals unaware of their hypertension may be made aware. However, an even larger proportion of individuals with undiagnosed hypertension could receive no alert at all. The manufacturer indicates that the feature is not intended to diagnose hypertension and is designed to prompt users to seek clinical evaluation if hypertension is suspected.1 However, false reassurance may discourage some individuals with undiagnosed hypertension from obtaining appropriate screening or engaging with the health care system, resulting in missed opportunities for early detection and treatment. Rigorous validation, strategies to improve cuffless device accuracy, and thoughtful integration into population screening will be essential if cuffless technologies are to contribute meaningfully to hypertension detection.
Limitations include reliance on the manufacturer’s sensitivity and specificity estimates, which may not fully capture device performance. Although Apple reported that there was lower sensitivity in those younger than 60 years and those who were not obese,1 these data could not be extrapolated to the NHANES population due to their modeling approach. Lastly, the subgroup-specific estimates assume that test performance is independent across subgroups.
