Hypertension Notification Feature (HTNF)

Performance Metrics

Clinical Evidence

The pivotal evidence is Apple's clinical validation, reported in the 510(k) summary for K250507 (cleared 11 September 2025) and in Apple's September 2025 validation paper. 2,229 adults with no prior hypertension diagnosis, recruited as an all-comers cohort and an at-risk cohort, wore an Apple Watch for 30 days and measured blood pressure twice daily with an FDA-cleared home cuff (Omron Evolv BP7000) as the reference. Hypertension was defined as a 30-day home average of 130/80 mmHg or above, the 2017 ACC/AHA threshold; stage 2 meant 135/85 mmHg or above. 1,863 participants provided at least 15 days of usable data and entered the primary analysis: 585 with hypertension and 1,278 without. Sensitivity was 41.2% (241 of 585, 95% CI 37.2–45.3) and specificity 92.3% (1,179 of 1,278, 95% CI 90.6–93.7). At the study prevalence of 31.4% the positive predictive value was 70.9% (95% CI 65.7–75.7). The split by stage matters. Sensitivity for stage 2 hypertension was 53.7% (95% CI 47.7–59.7), but for stage 1 it was 29.6% (95% CI 24.5–35.1), so seven in ten stage 1 cases went unnotified. Specificity was 95.3% (95% CI 93.7–96.5) among participants with normal blood pressure and 82.8% (95% CI 78.1–86.8) among those with elevated blood pressure. The stage 1 and elevated figures appear in the validation paper, not in the 510(k) summary. The feature met all pre-specified primary endpoints. The pre-specified subgroup analysis, adjusted for age, sex, BMI, race and blood pressure, found lower sensitivity in participants under 60 (risk ratio 0.69, 95% CI 0.55–0.85) and in those with a BMI of 30 or below (0.67, 95% CI 0.55–0.81). Both younger and lower-BMI participants also had slightly higher specificity (ratios 1.09 and 1.06). Hispanic participants had a higher adjusted sensitivity (1.41, 95% CI 1.07–1.88). The sensitivity ratios for sex, race and Fitzpatrick skin tone included 1.0. The subgroups were pre-specified, but the validation paper states that no statistical hypotheses were attached to them. A two-year longitudinal study of six non-overlapping 30-day windows put long-term specificity at 86.4% (95% CI 80.2–92.5) in 187 non-hypertensive participants, and 92.5% (95% CI 86.8–98.3) in the 121 with normal blood pressure, with most false notifications arriving in the first month. Independent work so far models the feature rather than re-testing it. A research letter in JAMA (Cohen et al., 2026) applied the 510(k) sensitivity and specificity to 3,983 NHANES 2017–2020 participants aged 22 and over who were unaware of any hypertension, representing 127 million US adults. A notification raised the probability of hypertension from 0.30 to 0.69; no notification lowered it only to 0.21. In adults under 30 the pre-test probability of 0.14 moved to 0.47 with a notification and 0.10 without; in adults 60 and over it moved from 0.45 to 0.81 and 0.34. The authors' central caution is false reassurance in the 58.8% of undiagnosed hypertensives who receive no alert. A commentary in Hypertension (Cohen et al., 2026) sets the 41% and 92% figures against a two-visit office blood-pressure mean (sensitivity 71%, specificity 59%) and concludes the feature is not suitable for large-scale, reliable hypertension screening in its current form. No independent prospective validation of the shipped feature and no outcome study (diagnosis rates, treatment uptake, cardiovascular events) has been located. Evidence grade is prospective: the pivotal study enrolled participants forward in time against a pre-defined cuff reference, but it is a manufacturer study reviewed by a regulator, not a peer-reviewed publication. AurisMark ◆◆ rests on the FDA clearance, the EUDAMED Class IIa registration and the Health Canada licence, all read from the registers, prospective-grade validation evidence, and no safety signal. A third mark would need peer-reviewed RCT-grade or outcome evidence and 24 months in market, neither of which exists.

Inside the Auris+ Listing

Regulatory Approvals

Safety Record

No recall, enforcement action or adverse-event report attributable to the Hypertension Notification Feature was found in the openFDA recall, enforcement and MAUDE endpoints as of 7 September 2026 (searches on the product name, product code SFR and the firm). The clinical risk is the one the labelling names: a wearer who receives no notification may conclude they have been screened clear, when the pivotal study shows the feature misses about six in ten undiagnosed hypertensives, seven in ten stage 1 cases and about half of stage 2 cases. The EUDAMED record carries two critical warnings: the feature cannot detect a heart attack, and it cannot detect pre-eclampsia or other hypertensive conditions of pregnancy. Performance outside the cleared population (diagnosed hypertensives, pregnancy, under-22s) is untested, and the Hypertension commentary expects real-world use to spread beyond it. The feature does not report a blood-pressure value and is not a substitute for cuff measurement.

Intended Use & Indications

The Hypertension Notification Feature is a software-only mobile medical application. It analyses photoplethysmography (PPG) data collected opportunistically by Apple Watch to identify patterns suggestive of hypertension and provides a notification to the user. The feature is for over-the-counter use by adults aged 22 and over who have not been previously diagnosed with hypertension. It is not intended to replace traditional methods of diagnosis, to monitor hypertension treatment effect, or to be used as a method of blood-pressure surveillance. It is not intended for use during pregnancy. The absence of a notification does not indicate the absence of hypertension (indications for use, FDA 510(k) K250507).