The Relationship Between ECG and Blood Pressure

An electrocardiogram and a blood pressure reading capture fundamentally different things: an ECG traces the electrical impulses that trigger each heartbeat, while blood pressure measures the physical force of blood pushing against artery walls. Yet these two measurements are deeply intertwined. Chronic high blood pressure reshapes the heart in ways that alter its electrical signature, and the timing of ECG signals can, under certain conditions, be used to estimate blood pressure without a cuff. The relationship runs in both directions, and understanding it has real consequences for how doctors assess cardiovascular risk and how wearable technology is evolving.

What an ECG Actually Shows About Blood Pressure

If you hand a cardiologist a standard 12-lead ECG printout and ask them to tell you the patient’s blood pressure, they cannot give you a number. The ECG does not measure pressure. What it can do is reveal the downstream effects of sustained high blood pressure on the heart’s structure and electrical behavior. Years of pumping against elevated pressure force the left ventricle to thicken, a condition called left ventricular hypertrophy, or LVH. That thickening changes the amplitude and timing of electrical signals the ECG picks up. So while the ECG cannot tell you the reading is 150/95, it can tell you the heart looks like it has been working against high pressure for a long time.

This distinction matters because it shapes how clinicians use the two measurements together. Blood pressure gives you the current state of the plumbing. The ECG gives you a window into whether that plumbing has already caused structural damage to the pump. A patient with borderline blood pressure but clear ECG signs of LVH is in more danger than the numbers alone suggest.

Left Ventricular Hypertrophy on the ECG

The most studied connection between ECG and blood pressure is the detection of LVH. Several scoring systems exist, the oldest being the Sokolow-Lyon voltage criteria developed in 1949 and the Cornell voltage and Cornell product criteria that followed decades later. All of them look at the size of specific waveforms to gauge whether the left ventricle has grown abnormally thick. A systematic review found that none of the newer, more sophisticated criteria clearly outperformed the original Sokolow-Lyon index, and that regardless of which scoring method was used, the ECG was a poor screening tool for ruling out LVH in people with hypertension.¹ Specificity was generally reasonable, meaning that when the ECG said LVH was present, it was usually right. But sensitivity was low, so many cases of true LVH were missed.

One newer set of criteria using the sum of deepest S-wave and tallest V4 R-wave showed a sensitivity of about 62%, significantly higher than the 35% achieved by Cornell voltage alone, while maintaining specificity above 90%.² Still, even the best single criterion misses roughly a third or more of LVH cases. Combining two different criteria improves matters. When both Cornell product and Sokolow-Lyon voltage indicated LVH, patients faced more than triple the risk of heart attack, stroke, cardiovascular death, and all-cause mortality compared to those without LVH by either criterion.³

The takeaway for patients is straightforward: the ECG is not a great way to screen for LVH because it misses too many cases. But when it does flag LVH, that finding carries real prognostic weight, especially when multiple criteria agree. An echocardiogram remains the standard tool for actually measuring wall thickness.

ECG Strain and Its Prognostic Power

Beyond the voltage measurements used to detect LVH, a pattern known as “ECG strain” adds another layer of information. Strain shows up as a characteristic depression of the ST segment and inversion of the T wave, typically in leads that face the thickened left ventricle. It has long been considered a worrisome sign, and recent research helps explain why. In people with hypertension, ECG strain was associated with higher heart mass and greater expansion of the space between heart muscle cells, pointing to diffuse scarring or fibrosis in the heart wall.⁴ Even when compared to hypertensive patients whose hearts were equally thick but who lacked the strain pattern, the strain group still had more fibrosis.

Data from the large LIFE trial reinforced the clinical importance of this pattern. Patients with ECG strain at baseline had a roughly 33% higher risk of major cardiovascular events after adjusting for treatment, blood pressure levels, and LVH severity. Those who developed new strain over the first year of the trial saw their risk double. Encouragingly, when strain regressed with treatment, the excess risk faded.⁵ This means tracking ECG strain over time gives doctors a marker of whether treatment is actually reversing heart damage, not just lowering the number on the cuff.

Can Treatment Reverse ECG Changes?

One of the more encouraging findings in this area is that effective blood pressure treatment can actually shrink LVH as measured by the ECG, and that shrinkage translates into fewer heart attacks and strokes. The LIFE study compared losartan-based therapy to atenolol-based therapy in hypertensive patients with ECG-confirmed LVH. After six months, even after accounting for blood pressure levels, the losartan group showed substantially greater regression of both Cornell product and Sokolow-Lyon voltage, and this advantage persisted throughout the study.⁶

A separate large trial found that the ACE inhibitor ramipril caused regression of ECG-LVH independently of its blood pressure-lowering effect. Patients whose LVH regressed had lower rates of death, heart attack, stroke, and heart failure compared to those whose LVH persisted.⁷ The fact that LVH regression happened beyond what blood pressure reduction alone would explain suggests that certain medications have direct effects on the heart muscle, and that serial ECGs can track those benefits. For patients already being treated for hypertension, this means a follow-up ECG is not just a formality; it can show whether your heart is actually recovering.

Pulse Transit Time and the Dream of Cuffless Monitoring

The relationship between ECG and blood pressure extends beyond diagnosis into an entirely different domain: using the ECG signal as part of a system to estimate blood pressure continuously, without a cuff. The core idea relies on pulse transit time, the interval between the heart’s electrical R-wave on the ECG and the arrival of the resulting pulse at a peripheral site, typically measured by a fingertip or wrist optical sensor. When arteries are stiff or blood pressure is high, the pulse wave travels faster, shortening this interval. When pressure drops, the wave slows and the interval lengthens.

In controlled research settings, the correlation between these timing measurements and blood pressure can be strong. One study found that both pulse arrival time (which includes the heart’s internal processing delay) and the vascular-only transit time showed strong negative correlations with blood pressure, with correlation coefficients below −0.8 in individual subjects.¹³ However, the picture gets muddier in larger and more diverse populations. A large surgical database found that the average correlation between pulse arrival time and systolic blood pressure was only about −0.37, and even weaker for diastolic pressure.¹⁴ The gap between individual and population-level results is a core challenge: the relationship is real within a given person, but it varies so much between people that a one-size-fits-all model does not work well.

The Pre-Ejection Period Problem

A major complicating factor is the pre-ejection period, or PEP. This is the delay between the ECG’s R-wave and the moment the heart actually ejects blood into the aorta. It is not a vascular delay; it is an internal cardiac delay driven by how fast the heart muscle contracts. PEP accounts for a substantial and variable share of the total transit time, ranging from roughly 12% to 35% depending on what drugs a person is taking or what physiological state they are in.¹⁵ Because PEP changes with stress, medications, posture, and exercise in ways that do not mirror what blood pressure is doing, it introduces noise into the transit-time estimate.

At rest, PEP variability within a single individual tends to be small, roughly 1% to 1.5% of the total transit time, which makes resting measurements more reliable.¹⁶ But the moment a person stands up, exercises, or encounters stress, PEP becomes highly variable and strongly influences the blood pressure estimate. One research group described PEP as a “highly stress-dependent parameter” and showed that the blood pressure uncertainty it introduces can inflate to extreme values at high systolic pressures.¹⁷ Including PEP explicitly in the measurement, rather than just lumping it into transit time, appears to improve accuracy, particularly after exercise.¹⁸

Can AI Estimate Blood Pressure from an ECG Alone?

A separate and more ambitious line of research tries to skip the optical pulse sensor entirely and estimate blood pressure from the ECG waveform alone, using machine learning. The idea is that subtle features of the ECG shape encode information about vascular stiffness and cardiac output that a trained algorithm could extract. Some results look promising on paper. One deep-learning model reported estimation errors of roughly ±6 mmHg for diastolic and ±8 mmHg for mean arterial pressure, which met commonly cited accuracy standards.¹⁹

But a critical review of this literature struck a more cautionary note. Researchers found that heart rate and heart rate variability extracted from ECG offered no added value for cuffless blood pressure measurement, that frequency-domain ECG features yielded very poor results, and that waveform complexity features required age data and calibration to achieve acceptable results for systolic pressure while still failing for diastolic pressure.²⁰ The title of that paper captures the current scientific consensus bluntly: estimating blood pressure from ECG waveforms alone is not yet viable. The promising results from individual studies often depend on specific databases, calibration schemes, or patient populations that do not generalize well.

The combined approach, ECG plus a photoplethysmography sensor measuring the pulse optically, remains more credible. Systems using both signals together and incorporating techniques like signal preprocessing and regression models have reported errors in the range of ±5 to ±7 mmHg for systolic and ±3 to ±6 mmHg for diastolic pressure.²¹²² These numbers sound good, but they tend to come from controlled settings rather than daily wear.

Wearable Devices and the Validation Gap

Consumer smartwatches that claim to measure blood pressure using some combination of ECG and optical sensors are already on the market. Their accuracy in real-world use, however, remains questionable. A clinical evaluation of three wearable watch-type blood pressure devices found that while all three produced mean readings within 5 mmHg of a standard cuff measurement, none of them passed all the criteria required by the ISO 81060-2 validation protocol, primarily because the spread of individual readings was too wide.²³ Being close on average is not the same as being reliable for any given reading, and it is the individual reading that matters when you are trying to decide whether your blood pressure is controlled.

Most of these devices also require periodic calibration against a conventional cuff, which undercuts the “cuffless” promise somewhat. The calibration aligns the watch’s algorithm to your personal baseline, but that baseline drifts over time, and there is no consensus on how often you need to recalibrate.²⁴ For now, these devices are best thought of as trend monitors, useful for spotting changes over time but not replacements for a validated cuff when an actual number matters.

Heart Rate Variability as a Bridge Between ECG and Blood Pressure

Beyond waveform shapes and transit times, the beat-to-beat rhythm of the heart captured by the ECG provides another connection to blood pressure. Heart rate variability, the slight fluctuation in the interval between consecutive heartbeats, reflects the balance between the sympathetic (accelerator) and parasympathetic (brake) branches of the autonomic nervous system. In people with hypertension, this balance is shifted: sympathetic drive to the heart and blood vessels is heightened, while the vagal brake is weakened.²⁵

This shift shows up as reduced overall heart rate variability. People with established hypertension have been found to have lower beat-to-beat variability and diminished baroreflex modulation, the mechanism that fine-tunes heart rate in response to blood pressure changes, compared to those with normal pressure. Interestingly, those in the “high-normal” blood pressure range already show some changes, particularly increased low-frequency variability during rest, possibly reflecting greater blood pressure instability even before a formal hypertension diagnosis.²⁶ These findings suggest that the ECG-derived rhythm analysis might eventually serve as an early warning system, flagging autonomic imbalance before blood pressure readings cross diagnostic thresholds.

ECG and Blood Pressure in Pregnancy

The interplay between ECG and blood pressure takes on special significance during pregnancy. Preeclampsia and gestational hypertension are leading causes of maternal and fetal complications, and researchers have explored whether ECG changes might predict these conditions before blood pressure spikes. Left atrial abnormalities detected on a standard ECG have been associated with a fourfold increased risk of developing hypertensive disorders during pregnancy.²⁷ If validated further, routine ECG screening early in pregnancy could help identify women who need closer monitoring.

The relationship also extends to the next generation. In the Copenhagen Baby Heart Study, newborns exposed to maternal preeclampsia had slightly prolonged QRS durations and lower R-wave amplitudes compared to unexposed infants. Babies of mothers with gestational hypertension showed modestly prolonged QT intervals.²⁸ The differences were small and their long-term significance remains unclear, but the findings suggest that the hemodynamic stress of maternal hypertension leaves a measurable electrical imprint on the fetal heart.

Why ECG-Based LVH Screening Falls Short in Adolescents

In adults, even an imperfect ECG screening test for LVH carries useful prognostic information. In teenagers, the picture is murkier. A study comparing hypertensive and normotensive adolescents found that similar numbers in both groups had ECG evidence of LVH, making the ECG essentially useless as a discriminator in that age range.²⁹ The likely explanation is that normal cardiac growth during adolescence produces voltage patterns that overlap heavily with early hypertrophy. For young patients with suspected hypertension, echocardiography remains the preferred tool for evaluating heart structure, and clinicians should be cautious about reading too much into ECG voltage criteria that were developed and validated in adults.

Exercise Testing and Blood Pressure Responses

Exercise stress testing brings ECG and blood pressure monitoring together in a controlled clinical scenario. During graded exercise, both heart rate and blood pressure rise. An exaggerated blood pressure response to exercise, where systolic pressure climbs disproportionately high for the level of exertion, has been linked to future cardiovascular events. In one large study following nearly 5,000 initially normotensive men for an average of 17 years, the magnitude of exercise-induced systolic blood pressure elevation was significantly associated with cardiovascular events and death from any cause, even after adjusting for confounders including LVH.³⁰ The ECG portion of the stress test is typically used to detect ischemia (reduced blood flow to the heart), but combining it with the blood pressure trajectory during exercise creates a richer risk profile than either measurement provides alone.

Blood pressure behavior during tilt-table testing, where a patient is strapped to a table and tilted upright, also produces ECG changes. T-wave amplitude shifts during tilting have been found to correlate with orthostatic outcomes like drops in blood pressure upon standing.³¹ For patients who faint or feel dizzy when they stand, these combined ECG and blood pressure observations during tilt testing can help sort out whether the problem is in the heart’s electrical system, the blood vessels’ ability to constrict, or the nervous system’s control over both.