Your First Blood Pressure Reading Isn’t a Verdict

You sit down, rest, and watch the blood-pressure monitor settle on 132/82. Under current US categories, that falls into Stage 1 hypertension. But the next two readings, taken one minute apart, are 126/76 and 126/76. Their average is 128/78, which belongs to a different category: Elevated.

Nothing in that little sequence proves the device was faulty. And nothing in it proves a medical diagnosis. It reveals a more interesting fact: blood pressure is a measurement; a blood-pressure category is a rule applied to it.

And the surprise is not simply that the first reading is too high. In our reanalysis, the label moved up for 5.5% of adults when three readings were averaged. The same instrument, the same person, only minutes apart. What changed: the pressure, the rule, or both?

Even under a careful protocol, the category can change

The US National Health and Nutrition Examination Survey (NHANES) measured blood pressure in a research centre, not in a rushed clinic queue. Participants first rested quietly for five minutes. An automatic monitor then took three readings about 60 seconds apart. We independently reanalysed 6,100 adults with three complete readings from August 2021 to August 2023, using the survey weights and sampling design.

Two different things are changing

Blood pressure is not an unchanging personal number. Breathing, posture, vascular tone, movement and emotion can alter it. But another part of the story is mathematical. Pressure is measured continuously; medical categories are separated by cut-offs. Two very similar readings can sit on opposite sides of a line without implying a similarly abrupt change in cardiovascular risk.

Our example makes the distinction visible. The first systolic reading is 132 mmHg, just above a US threshold at 130. Averaging the three readings gives 128. That four-point difference moves the label from Stage 1 to Elevated, even though neither result tells us the person’s usual blood pressure.

The percentages in our analysis also depend on which question we ask. Comparing two individual readings produces more disagreement than comparing the first with an average that already includes it. Restricting the question to crossing a single hypertension threshold produces a smaller percentage still. That is not a contradiction in the data. It is the consequence of changing the definition of disagreement.

The US classification uses boundaries at 120, 130 and 140 mmHg for systolic pressure, and 80 and 90 for diastolic pressure. In our sample, roughly 94% of changes occurred near boundaries that were reachable by a shift of five mmHg or less. That does not demonstrate a defective monitor. It demonstrates why a small shift in a measurement can produce a change in its label.

Three distinct events are easy to confuse: a change in a measured number; a change in a category assigned by a guideline; and a change in a clinician’s diagnosis or treatment decision. We measured the first two. NHANES did not adjudicate diagnoses, so our figures cannot tell us how often treatment would change. Even a small numerical shift can change a label near a threshold without creating a sudden change in personal cardiovascular risk.

Is one in six being misdiagnosed?

No. Our 16.4% is not a false-diagnosis rate. We compared two ways to summarise readings collected during one appointment. Neither is an independent diagnostic gold standard. Confirming hypertension generally requires repeated measurements on more than one occasion, and home or 24-hour monitoring is often used to resolve uncertainty.

Nor should all category changes be treated alike. The weighted share crossing the specific 130/80 threshold was 6.85% (95% CI 6.05–7.65%). The share crossing 140/90 was 4.65% (95% CI 3.89–5.41%). These figures answer different questions from the 16.4% four-category result.

What if your first reading was already high? Among adults initially in the US Elevated category, 23.2% moved to Normal when all three readings were averaged. For those initially at Stage 1, 19.7% moved to a lower category; for Stage 2, 18.9% did. These are survey-weighted conditional shares derived from our transition table, not probabilities of a diagnosis being overturned. A lower category still may not mean normal pressure.

Does the first reading really run higher?

Possibly, but NHANES did not measure anxiety or identify the cause of a change. In an additional exploratory analysis using CDC-derived CSV copies, the survey-weighted difference between the third and first readings averaged just −0.22 mmHg systolic (95% CI −0.47 to +0.03) and −0.81 mmHg diastolic (−0.96 to −0.67). These small population means do not describe any one person’s change. When we reassigned the label “first” uniformly among each person’s three observed readings, the population gap between downward and upward category moves fell from 5.39 to 3.69 percentage points. Most of the directional imbalance therefore survives without a privileged first measurement. This is an exploratory mathematical check, not an estimate of the physiological cause; it is not included in the archived v1.0 report.

Try your own three readings

The calculator shows how a rule changes the label for a set of numbers. It cannot diagnose hypertension, identify the reason for a reading’s variation or replace confirmation over multiple days. For home measurements, follow the appropriate home-monitoring guidance rather than treating office thresholds as a personal diagnostic test.

Use this educational calculator to explore the first reading and the average of all three under US or European office categories. It is not a diagnosis or treatment recommendation.

What should you do with a high reading?

Use a validated upper-arm monitor with the correct cuff size. Rest quietly for at least five minutes, keep your back supported, feet flat and your supported arm at heart level, and avoid talking during measurement. Take repeat readings about a minute apart. Record them and look for a pattern over several days rather than choosing the lowest result.

Which number should you believe? Do not cherry-pick the lowest one, or treat a single high result as your lifelong baseline. Keep the individual readings and the context: time of day, body position, cuff, activity and medications. A clinician can interpret the pattern across visits, and use home or 24-hour monitoring when needed. A single series cannot identify white-coat hypertension or determine which measurement best predicts future cardiovascular events.

When not to wait for an average: If a reading is higher than 180 systolic and/or 120 diastolic mmHg, follow AHA guidance: repeat after at least a minute and contact a healthcare professional promptly if still that high. With chest pain, breathing difficulty, weakness, new vision changes, difficulty speaking or other concerning symptoms, seek emergency help immediately. Do not use a lower average to dismiss the warning.

How we know, and what we still cannot know

We analysed 6,100 of the 6,337 examined adults in NHANES 2021–2023 who had three complete blood-pressure readings. Reported percentages are weighted estimates for this complete-case survey sample, not a count of diagnosed patients. The confidence intervals reflect sampling uncertainty, but not possible systematic errors or bias from missing readings. This independent public-data reanalysis has not been peer-reviewed.

Earlier studies already documented reclassification after repeated measurements, notably Handler, Zhao and Egan (2012), and Tang et al. (2021) examined streamlined blood-pressure screening protocols. Our contribution is an updated, openly described calculation using recent US observations and current classification rules, not the discovery of a new physiological mechanism. The exploratory order-symmetrisation analysis above tests an additional mathematical explanation; it does not establish its clinical cause.

Read the source research: The Null Institute: NHANES 2021–2023 blood-pressure reclassification analysis. It includes weighted estimates, the complete category-transition matrix, source documents, methodology and statistical limitations. Archived report and reproducibility files (v1.0): DOI 10.5281/zenodo.23342882 ↗.

Independent calculation check (14 October 2026). Reanalysis of 6,100 individual NHANES records reproduced the survey-weighted 16.4087% category-change estimate (95% CI 15.36–17.45%). This check used publicly available CDC-derived CSV copies; the bytes of freshly downloaded original CDC XPT files were not independently authenticated. This is not a hypertension misdiagnosis rate. Methods and verification details at The Null Institute.

References

1. CDC NHANES 2021–2023, oscillometric blood pressure protocol.
2. CDC NHANES 2021–2023, demographics and weights.
3. AHA/ACC, 2025 high blood pressure guideline.
4. European Society of Cardiology, 2024 guideline.
5. American Heart Association, home blood-pressure monitoring.
6. Handler, Zhao & Egan, 2012, prior NHANES investigation of repeated measurements.
7. Tang et al., 2021, simplified blood-pressure screening using different measurement protocols in NHANES and other populations.
8. Nikitin, I. (2026). NHANES blood-pressure reclassification analysis, version 1.0. The Null Institute. DOI: 10.5281/zenodo.23342882.