Why almost half of hypertensive people don't know they have the disease

Systemic arterial hypertension is a chronic condition that causes sustained, silent, and progressively damaging elevation of blood pressure in the arteries, affecting target organs such as the heart, kidneys, and brain. The central epidemiological problem is not the lack of treatment: it is late or undiagnosed cases.

Data from Vigitel 2025 shows that almost 30% of Brazilian adults are hypertensive, compared to 22.6% in 2006. Globally, the World Health Organization estimates that 1.4 billion adults live with the condition. About half of this total are unaware of their diagnosis.

There are three main reasons for this global underdiagnosis. First, hypertension rarely causes symptoms in its early stages. Second, a significant portion of the population does not have regular check-ups. Third, and less intuitive for those outside the field: a single measurement in the doctor's office can miss hypertension due to two well-documented phenomena. White coat hypertension artificially inflates readings in a clinical setting. Masked hypertension does the opposite, showing normal pressure in the office but elevated readings in daily life. International guidelines recommend repeated ambulatory or home measurements precisely for this reason.

What the Apple Watch does, accurately: it does not measure blood pressure

The Apple Watch hypertension notification feature does not measure blood pressure and does not display values in mmHg. It analyzes patterns of the optical heart signal over 30 days and issues an alert if it detects signs consistent with chronic hypertension.

This distinction is fundamental for doctors who will receive patients with the notification on their phone screen. The Apple Watch does not provide a diagnosis, does not display a systolic or diastolic reading, and does not replace a sphygmomanometer. What it does is classify the user as having or not having signs compatible with hypertension after 30 days of passive use, and recommend that the patient measure their blood pressure with a conventional device for 7 days before consulting a healthcare professional.

• What the feature does

  Analyzes the PPG signal passively, accumulates evidence for 30 days, and alerts if it detects patterns compatible with chronic hypertension. Indicates the next step: measure blood pressure with a conventional device for 7 days.
• What the feature doesn't do

  Does not directly measure blood pressure. Does not display values in mmHg. Does not detect all cases of hypertension (sensitivity of 41.2% in the clinical study). Not recommended for those under 22, pregnant women, or those already diagnosed.

Photoplethysmography: the technology behind detection

Photoplethysmography (PPG) is a non-invasive optical technique that measures variations in blood volume in peripheral vessels using light. In the Apple Watch, green LEDs on the back of the watch illuminate the wrist skin; photodiodes capture reflected light, which varies with each heartbeat as blood flow increases and decreases.

The raw PPG signal contains much more information than just heart rate. The shape of the wave, specifically the amplitude and rise time of the pulse, reflects arterial wall stiffness. In people with hypertension, arteries become progressively stiffer, altering pulse morphology in characteristic ways: the wave rises faster, the reflected component arrives earlier, and pulse pressure widens. These patterns are not visible to the naked eye in the raw waveform but are detectable by deep learning algorithms.

It is worth noting: PPG is not equivalent to sphygmomanometric measurement. A cuff directly measures the force blood exerts on arterial walls by compressing and releasing the artery. PPG measures an indirect correlate, the vascular response to the heartbeat. This is why the Apple feature does not display values in mmHg and cannot replace a formal diagnosis.

The algorithm: three stacked stages of artificial intelligence

Apple's hypertension notification system uses a three-stage pipeline: feature extraction via deep learning, risk scoring by linear machine learning models, and notification evaluation by threshold over 30 days.

The watch collects 60-second segments of the PPG signal approximately every 2 hours during waking hours. Before processing any data, the accelerometer filters out moments when the user is moving, as movement degrades the PPG signal: only samples collected when the user is stationary are used.

The algorithm is static: once published, it does not learn from new user data. This is an intentional feature for regulatory safety.

The pivotal clinical study: methodology and actual results

Apple conducted a decentralized clinical study with 2,229 adult participants without a prior diagnosis of hypertension to validate the feature's performance before regulatory submissions to the FDA and other agencies, including Anvisa.

The reference methodology was robust: participants used two Omron Evolv home blood pressure monitors (clinically validated arm cuff) for 30 days, twice daily, following American Heart Association guidelines. The 30-day averages were used to classify each participant into categories: normal, elevated blood pressure, stage 1 hypertension (systolic 130 to 134 mmHg or diastolic 80 to 84 mmHg), and stage 2 hypertension (systolic 135 mmHg or higher or diastolic 85 mmHg or higher). Simultaneously, the Apple Watch passively collected PPG data.

The main results were as follows. Overall sensitivity was 41.2% (95% CI: 37.2% to 45.3%), meaning the algorithm identified about 4 out of 10 people with confirmed hypertension. For stage 2 hypertension, sensitivity rose to 53.7%. Overall specificity was 92.3%, reaching 95.3% among participants with genuinely normal pressure. In practical terms: most alerts issued corresponded to people with some degree of actual blood pressure elevation.

Subgroup analyses showed that sensitivity is higher in older participants and those with a BMI over 30, which makes pathophysiological sense: in these groups, vascular disease is more advanced, and PPG signals are more pronounced. After adjustment for covariates, there was no clinically significant performance difference between sex, race, and skin tone, including Fitzpatrick skin types V and VI, which is relevant for a diverse population like Brazil's.

What this changes in clinical practice: the patient who arrives with the alert on their phone

The feature does not produce a diagnosis, but it can be the trigger that leads an asymptomatic patient to seek medical attention for the first time, or to start home monitoring that will support a formal diagnosis.

When a patient comes to the clinic with an Apple Watch notification, the recommended course of action is clear. Apple advises the user to measure their blood pressure with a conventional sphygmomanometer twice a day for 7 days and bring the records to the doctor. This protocol aligns with the guidelines of the Brazilian Society of Cardiology and international recommendations for diagnosing hypertension based on home monitoring.

The doctor should interpret the notification as additional data in clinical reasoning, not as a conclusive laboratory result. A sensitivity of 41% means that more than half of hypertensive individuals will not receive an alert: the absence of a notification does not rule out the disease. On the other hand, a specificity of 92% means that most of those alerted do indeed have some degree of blood pressure elevation.

The scale of the feature is the most impressive data from a public health perspective. Apple projects that the feature will alert more than 1 million people without a hypertension diagnosis in the first year of global availability alone. No traditional clinical screening program can operate at this scale with such marginal cost.

Feature limitations and regulatory context in Brazil

The feature was approved by the FDA in the United States and has been available in Brazil since January 2026 for compatible models. Despite its availability, clinical limitations are significant and need to be communicated to patients.

• Low sensitivity for stage 1

  The sensitivity for stage 1 hypertension was only 29.6%. The feature is better at detecting more severe cases, which already show greater arterial stiffness.
• Requires continuous use

  The algorithm needs 30 days of data. Users who do not wear their watch regularly may not accumulate enough samples to trigger the evaluation.
• Excluded populations

  The feature is not indicated for individuals under 22, pregnant women, or those already diagnosed with hypertension. The vascular pathophysiology in these groups is distinct, and the model was not trained to capture it.
• Absence of notification does not rule out the disease

  Given the sensitivity limit, the absence of notification should not be clinically interpreted as ruling out hypertension. Conventional screening remains indicated.

What this feature means for the future of preventive medicine

The Apple Watch hypertension notification feature represents a paradigm shift in chronic disease screening: for the first time, a mass-market consumer device operates as a passive and continuous cardiovascular screening tool.

From a technical perspective, what Apple has demonstrated is that a deep learning model with only 3.3 million parameters, running passively on the wrist, can extract relevant clinical information from a signal that was previously only used to count heartbeats. This paves the way for future versions of the algorithm to detect other cardiovascular markers, such as subclinical atherosclerosis, incipient heart failure, or blood pressure variability.

For Brazilian medicine, the context is particularly relevant. With almost 30% of the adult population hypertensive and a healthcare system that historically has not been able to conduct large-scale active screening, passive screening tools based on devices already present in the daily lives of a growing portion of the population have a real potential for epidemiological impact, provided they are well integrated into the primary care flow.

For doctors, residents, and students, understanding how these technologies work is no longer optional. Patients already arrive at the clinic with wearable data. Knowing how to interpret an Apple Watch notification, contextualize its limits, and integrate it into clinical reasoning is part of contemporary medical competence.

Frequently asked questions

Direct answers to the most common questions about the Apple Watch hypertension notification feature.