The Apple Watch for Hypertension Screening

Introduction

The Apple Watch is the most widely adopted consumer health device globally. The recent US Food and Drug Administration (FDA) clearance of its Hypertension Notification Feature represents an important juncture in digital health. Novel cuffless blood pressure (BP) devices have the potential to overhaul how we screen and manage hypertension, with the proviso that they provide accurate and useful readings. In many ways, Apple’s release, intended to screen for hypertension, sidesteps debates on the provision of accurate BP numbers by focusing on hypertension classification. By shifting from BP values to classification, the feature avoids some challenges of cuffless measurement, though accuracy concerns remain and carry the risk of misclassification at scale.

What are the risks?

For many consumers, regulatory clearance conveys a sense of accuracy and safety. The Apple Watch does not actually measure BP, but instead it analyzes photoplethysmography signals over 30 days and provides a probabilistic notification that a user may have hypertension (without cuff calibration).¹ Hence, the Apple Watch does not provide an estimate of BP in mmHg.

For clinicians and patients alike, it is important to distinguish the Apple Watch’s signal from cuff-based BP measurement.² While the FDA 510(k) paperwork acknowledges that the Hypertension Notification “is not intended to replace traditional methods of diagnosis,” this has not been emphasized in media coverage, nor has the limited indication for “those who have not been previously diagnosed with hypertension” or the caveat that “the absence of a notification does not indicate the absence of hypertension”.¹ This gap between media coverage and clinical understanding creates an asymmetry that can contribute to public misperception and misplaced trust in the feature’s reliability. Although the device is explicitly not intended for individuals with known hypertension or during pregnancy,¹ in reality, adoption will very likely extend well-beyond these indications.

The FDA 510(k) summary data demonstrate the limitations of the Apple Watch Notification Feature. Compared to 15–30 days of home BP monitoring, the sensitivity of the device at detecting hypertension was 41% and the specificity was 92%.¹ In contrast, among adults without a known diagnosis of hypertension, the mean of high-quality office BP measurements obtained across two visits has a sensitivity of 71% and a specificity of 59%.³ Screening tools are generally expected to prioritize the ability to rule out a disease (typically due to high sensitivity, often at the expense of lower specificity) in order to minimize missed cases, particularly when the risk (i.e., burden/cost of further testing) from a false positive reading is low.⁴ For patients, a watch notification may prompt timely evaluation for hypertension, which could increase awareness and diagnosis. Even if a patient does not end up having hypertension, this encounter with a health care provider would be an opportunity to educate them on cardiovascular risk reduction and appropriate home BP monitoring technique using a validated, cuffed device. The main caveat of the Hypertension Notification Feature’s intention to screen for hypertension is that 59% of individuals with undiagnosed hypertension wearing the Apple Watch will not be alerted. Given that screening tests are expected to be good at ruling out a disease, absence of a notification could lead to false reassurance, deferred care, and delayed diagnosis. Thus, the Hypertension Notification Feature’s current performance is not suitable for largescale, reliable hypertension screening.

What lessons can we learn from the Apple Watch’s irregular heart rhythm notification? This feature increased awareness of atrial fibrillation and promoted innovation in remote arrhythmia monitoring. The Apple Heart Study screened over 400,000 participants; of the 2,064 participants with irregular pulse notifications, the positive predictive value for atrial fibrillation was 84%. However, the feature also created uncertainty about how to manage the resulting influx of data, resulting in high volumes of downstream testing (a portion of which was unnecessary and likely would not have occurred in the absence of the device alerts).⁵ Given the markedly higher prevalence of hypertension (almost 50% vs. <5%) and estimated hundreds of millions of Apple Watch users worldwide, the potential population-level impact is much larger, and even modest inaccuracy could have meaningful consequences at scale.

How did Apple evaluate the screening ability of the Hypertension Notification Feature?

Apple’s “validation study” enrolled 2,229 adults without prior hypertension and compared Apple Watch signals with home BP monitoring over 30 days, requiring a minimum of 15 days of readings to be included in the analysis.¹ Since the Apple Watch does not provide BP readings in mmHg, the feature cannot be tested in a formal study according to International Organization for Standardization (ISO) or Institute of Electrical and Electronics Engineers (IEEE) validation protocols. Instead, Apple performed an internal study comparing the feature against home, cuffed BP devices. The Hypertension Notification Feature was cleared under the FDA’s 510(k) pathway as “substantially equivalent” to another machine learning-based cardiovascular notification device which detects hypertrophic cardiomyopathy using electrocardiography.¹ This predicate device neither uses photoplethysmography nor screens for hypertension. As described in more detail in the AHA Scientific Statement on Cuffless Devices, this underscores that FDA clearance does not require evidence of validation using an international protocol.² This type of reliance on predicate devices for FDA clearance further illustrates structural limitations in existing regulatory pathways and highlights the urgent need for a distinct, standardized framework and testing protocol for machine learning-based, probabilistic diagnostic technologies like the Hypertension Notification Feature. Furthermore, currently available international validation protocols are not yet well-suited for intermittent cuffless BP devices and real-world use.^2,6 Further, photoplethysmography signals are influenced by skin tone, contact pressure, body habitus, hydrostatic pressure, vasoconstriction, and activity, conditions not addressed in available protocols.^2,7 FDA clearance should not be mistaken for accuracy or clinical readiness.

Cuffless device algorithms that rely on factors such as pulse arrival time or waveform morphology cannot directly measure BP; they can only infer change. Many technologies require calibration against cuff-based devices, while others use demographic or anthropometric inputs to provide BP estimation, which may embed bias and reduce generalizability. The Apple Watch does not use cuff-based calibration, and Apple has not disclosed an alternative calibration method.

Benefits versus risks of population screening

Accessible screening for hypertension remains a public health priority. Hypertension is the leading preventable risk factor for cardiovascular disease worldwide, resulting in over 10 million deaths each year. With one in three adults over 30 years worldwide having hypertension, and with nearly half unaware of their diagnosis,⁸ screening and detection are key. However, barriers to accurate BP measurement persist, particularly in underserved communities.² Consumer devices integrated into daily life could help overcome some of these barriers, although their adoption remains highest among affluent, health-literate populations, raising concern that underuse in low-resource settings could widen disparities if such tools are promoted as population screening strategies. However, clinical studies are needed to provide evidence that notification systems improve awareness and accurate detection of elevated BP and to define practical triage and confirmatory pathways for interpreting notifications in varied care settings. More robust scientific evidence is needed to inform the value of cuffless hypertension classification devices with respect to clinical decision-making and outcomes and to understand how device performance characteristics (e.g., sensitivity, specificity) affect this. Broadly, this exemplifies the tension between the promise of cuffless technologies and the risks of premature implementation.

Conclusion

The release of the Apple Watch’s Hypertension Notification Feature represents an important development in consumer digital health. Despite substantial consumer trust in Apple and the perceived legitimacy conferred by regulatory clearance, there is uncertainty regarding the Watch’s screening notifications, with low sensitivity and unclear performance across populations. Unless the above issues are addressed, this Apple Watch feature has limited prospects for moving beyond a consumer grade wellness perk. Addressing these uncertainties will require rigorous, transparent evaluation across diverse populations and contexts, along with research to determine how best to integrate such technologies into healthcare systems before they can be recommended for widespread hypertension screening.