INTRODUCTION
In September 2025, the U.S. Food and Drug Administration (FDA) 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 1). 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 U.S. adults without diagnosed hypertension.
Figure 1. Illustration of the Apple Watch Hypertension Notification Feature Performance among Individuals (A) with and (B) without Hypertension.
This figure, as recommended by Zikmund-Fisher et al. in JAMA 2025, depicts expected alert patterns using the reported sensitivity and specificity of the Apple Watch Hypertension Notification Feature. Panel A shows outcomes among individuals with hypertension, where blue squares represent appropriate alerts (i.e., true positives) and gray squares represent inappropriate absence of alerts (i.e., false negatives). Panel B shows outcomes among individuals without hypertension, where red squares represent inappropriate alerts (i.e., false positives) and white squares represent appropriate absence of alerts (i.e., true negatives). 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 U.S. population. NHANES obtained institutional review board approval and written informed consent from all participants.2 This report follows the STROBE guideline for cross-sectional studies.
Eligibility to use the Apple Watch Notification Feature was determined based on the intended use population: non-pregnant, aged 22 years or older, unaware of having hypertension, and not taking antihypertensive medication.1 We estimated the prevalence (i.e. pre-test probability) of undiagnosed hypertension in the overall eligible population and select subgroups; see Supplemental Methods for additional details. Test performance metrics were applied to estimate the percent of true positives, false negatives, true negatives, false positives, positive and negative likelihood ratios, positive and negative predictive values, and post-test probabilities.3,4 All analyses were performed using R v.4.5.1.
RESULTS
There were 3,983 NHANES participants aged 22 years or older, not pregnant, unaware of a prior hypertension diagnosis, and not taking antihypertensive medication, representative of 127 million U.S. 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%). Table 1 presents pre- and post-test probabilities across different subgroups which vary in hypertension prevalence. For example, among eligible individuals aged 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 aged 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 1.
Performance of the Apple Watch Hypertension Notification Feature based on observed prevalence of hypertension (i.e., pre-test probability) overall and among subgroups of the eligible U.S. population.
| Characteristic | Observed prevalence of hypertension (i.e., pre-test probability) | + Post-test probability | − Post-test probability |
|---|---|---|---|
| Overall | 0.30 (0.27, 0.32) | 0.69 (0.63, 0.75) | 0.21 (0.19, 0.23) |
| Age, years | |||
| < 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/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 Other | 0.36 (0.27, 0.44) | 0.75 (0.67, 0.83) | 0.26 (0.19, 0.33) |
| Body mass index, kg/m2 | |||
| ≤ 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 | |||
| Less than high school | 0.33 (0.28, 0.38) | 0.73 (0.66, 0.79) | 0.24 (0.20, 0.28) |
| High school or greater | 0.29 (0.26, 0.32) | 0.69 (0.63, 0.75) | 0.21 (0.18, 0.23) |
| Poverty index ratio | |||
| < 1 | 0.26 (0.21, 0.30) | 0.65 (0.57, 0.72) | 0.18 (0.15, 0.21) |
| 1 ≤ and < 4 | 0.30 (0.28, 0.33) | 0.70 (0.64, 0.76) | 0.22 (0.19, 0.24) |
| ≥ 4 | 0.29 (0.23, 0.35) | 0.68 (0.60, 0.76) | 0.21 (0.16, 0.25) |
Estimates (95% confidence intervals) were calculated using NHANES 2017-2020. Eligibility was defined as NHANES participants aged 22 years or older, not pregnant, unaware of a prior hypertension diagnosis, and not taking antihypertensive medication. The positive likelihood ratio was 5.35 (95% CI 4.05-6.65) and the negative likelihood ratio was 0.64 (95% CI 0.59-0.68), based on Apple’s reported sensitivity and specificity.1 We estimated pre-test odds (i.e. prevalence of NHANES-defined hypertension divided by 1 minus the prevalence) and post-test odds (i.e. pre-test odds multiplied by the positive or negative likelihood ratio) to determine the post-test probability (i.e. post-test odds divided by 1 plus the post-test odds) of hypertension. + Post-test probability represents the probability of having hypertension among individuals who receive an alert. − Post-test probability represents the probability of having hypertension among individuals who do not receive an alert.
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 new 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. Apple 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 healthcare system, resulting in missed opportunities for early detection and treatment. Moving forward, 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 Apple’s sensitivity and specificity estimates, which may not fully capture device performance. Although Apple reported that there was lower sensitivity in those aged younger than 60 years and who were not obese,1 these data could not be extrapolated to the NHANES population due to their modeling approach. Lastly, our subgroup-specific estimates assume that test performance is independent across subgroups.
Supplementary Material
Acknowledgements:
DKA and JAJ had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All analyses were independently double coded.
Disclosures:
Dr. Cohen is supported by R01HL153646, R01HL157108, R01HL155599, R01HL157264, U01HL160277, U24DK060990, and R01AG074989 from the National Institutes of Health (Bethesda, MD). Dr. Bress is supported by R01AG074989, K24AG080168, and R01AG065805 from the National Institute on Aging (Bethesda, MD) and R01HL139837 from the National Heart, Lung, and Blood Institute (Bethesda, MD). Additional support was provided by the University of Utah Study Design and Biostatistics Center, with funding in part from the Public Health Services research grant numbers UL1-RR025764 and C06-RR11234 from the National Center for Research Resources. Dr Picone is supported by a National Health and Medical Research Council Investigator Grant (GNT2018077) and is an Honorary Future Leader Fellow of the Heart Foundation of Australia (106618). All other authors have nothing to disclose.
Sources of Funding:
R01HL153646 from the National Heart, Lung, and Blood Institute. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Role of Funder:
The funder was not involved in the design or conduct of this study. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Non-standard abbreviations and acronyms
- BP
Blood pressure
- FDA
Food and Drug Administration
- NHANES
National Health and Nutrition Examination Survey
REFERENCES
- 1.U.S. Food and Drug Administration. 510(k) Summary for Apple Hypertension Notification Feature (K250507). Updated 12 Sep 2025. Accessed 15 Sep 2025, https://www.accessdata.fda.gov/cdrh_docs/pdf25/K250507.pdf
- 2.National Center for Health Statistics. NHANES Survey Methods and Analytic Guidelines. Centers for Disease Control and Prevention. Accessed 23 Sep 2025, https://wwwn.cdc.gov/nchs/nhanes/analyticguidelines.aspx [Google Scholar]
- 3.Richardson WS, Wilson MC, Guyatt GH, Cook DJ, Nishikawa J. Users’ guides to the medical literature: XV. How to use an article about disease probability for differential diagnosis. Evidence-Based Medicine Working Group. JAMA. Apr 7 1999;281(13):1214–9. doi: 10.1001/jama.281.13.1214 [DOI] [PubMed] [Google Scholar]
- 4.Walsh S, Choi E, Fang C, et al. The Diagnostic Accuracy of Using Borderline High Office Blood Pressure Thresholds to Diagnose Masked Hypertension According to the 2017 American College of Cardiology/American Heart Association Blood Pressure Guideline. Am J Hypertens. Apr 15 2025;38(5):288–294. doi: 10.1093/ajh/hpaf017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Jones DW, Ferdinand KC, Taler SJ, et al. 2025 AHA/ACC/AANP/AAPA/ABC/ACCP/ACPM/AGS/AMA/ASPC/NMA/PCNA/SGIM Guideline for the Prevention, Detection, Evaluation and Management of High Blood Pressure in Adults: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Hypertension. Oct 2025;82(10):e212–e316. doi: 10.1161/HYP.0000000000000249 [DOI] [PubMed] [Google Scholar]
- 6.Frommer D A decade with the Apple Watch: Thoughts, observations, and exclusive new data, after 3,675 days on the wrist. Accessed 15 Sep 2025, https://substack.newconsumer.com/p/a-decade-with-the-apple-watch
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.