The 120 mmHg systolic target

A trace: from how the blood pressure was measured, up to the threshold a clinician acts on.

Paulina Del Mundo

2026-06-09

Almost everyone in American medicine now repeats some version of the same sentence: 130 is the new high, and for patients at risk you treat toward something under it. That recommendation did not fall out of the sky. It sits at the end of a pathway that runs all the way back to a question almost no one asks at the bedside: how was the blood pressure actually measured?

This is a trace. It starts at the bottom of that pathway, with the raw datapoint, and builds back up, appraising the statistics at each rung, to ask whether the recommendation at the top can bear the weight a clinician puts on it. The aim is not to tell anyone how to treat a patient. It is to show what the number underneath the bedside is made of, and where it is sturdier or softer than its confident phrasing suggests.

From the datapoint up

01 — Measurement

How the pressure was taken

Here is the foundation the whole pathway rests on. SPRINT measured blood pressure with an automated device (an Omron 907-family monitor) as the average of three readings taken a minute apart after five minutes of quiet seated rest. This is automated office blood pressure, AOBP: rested, repeated, machine-averaged. It is not how a typical busy clinic takes a pressure, which is often a single cuff reading on a patient who just walked in.

A popular version of the critique says SPRINT was “unattended,” with no staff in the room, so its 120 is not a real-world 120. That version is mostly a myth, and worth retiring. The SPRINT protocol did not actually specify whether staff were present, and sites varied widely: a survey of 88 sites and 8,645 participants found some where the patient was always alone, some never alone, and several in between. Within the trial, attended versus unattended readings differed by only about 1.5 mmHg systolic. Observer presence is not the issue.

The better-founded point is separate and survives scrutiny. Automated, rested, averaged office readings run lower than the ordinary manual office reading, independent of who is in the room. How much lower is genuinely contested: single studies have reported gaps as large as roughly 16 mmHg, while pooled meta-analyses land much smaller, on the order of 3 to 9 mmHg, with high heterogeneity across devices and populations. Carry that uncertainty up the pathway. Every number above this rung was generated by a measurement method most clinics do not reproduce.

02 — Model

What the trial actually tested

On that measurement, SPRINT built a clean randomized comparison. It assigned participants to an intensive systolic target of under 120 or a standard target of under 140, achieved a separation between the arms of roughly 12 to 14 mmHg, and summarized the result with a Cox proportional-hazards model. On its own terms this is about as credible as clinical evidence gets: randomization, a hard composite endpoint, a pre-specified analysis.

One feature of the design has to travel up with us. The contrast tested was under-120 versus under-140. Nobody in SPRINT was treated to a goal of under-130, which, as we will see at the top of the pathway, is the number the guideline actually recommends.

03 — Estimate

The numbers it produced

SPRINT reported a 25 percent lower rate of its primary composite outcome (myocardial infarction, other acute coronary syndromes, stroke, acute decompensated heart failure, or cardiovascular death) in the intensive arm: hazard ratio 0.75, 95% confidence interval 0.64 to 0.89, p < 0.001. All-cause mortality was also lower, hazard ratio 0.73, 95% confidence interval 0.60 to 0.90, p = 0.003. The trial was stopped early, at a median of about 3.3 years against a planned five, when the benefit crossed a pre-specified boundary.

Those relative numbers are striking. The absolute numbers are the ones a patient lives in. Over that 3.3-year window the trial reported a number-needed-to-treat of 61 to prevent one primary-outcome event and 90 to prevent one death. Two cautions travel with these figures. Trials stopped early for benefit tend, as a class, to overstate the effect size, because they stop on a favorable random excursion. And the intensive arm did not come free: SPRINT recorded significantly more hypotension, syncope, electrolyte abnormalities, and acute kidney injury in the intensively treated group. The benefit is real. It is also a net of benefit and harm, summarized as a single hazard ratio, resting on the measurement at the bottom of this pathway.

04 — Synthesis

From one trial to “the evidence”

A single estimate becomes a recommendation only once it is treated as the body of evidence. SPRINT (the SPRINT Research Group, NEJM 2015) randomized 9,361 adults aged 50 and older who were at increased cardiovascular risk but who did not have diabetes and had not had a stroke. That population is the synthesis step’s fine print. The trial is strong evidence about high-risk, non-diabetic, stroke-free adults, and the guideline leans on it heavily for the lowered systolic target. Whether it licenses a target for everyone else is a question the trial cannot answer on its own, and the trial that ran the same comparison in the excluded population reached a different answer (see the contrast below).

05 — Decision rule

The cutoffs at the point of care

Synthesis becomes a set of numbers a clinician can act on. A reading at or above 130/80 is now hypertension. For a patient with known cardiovascular disease, diabetes, chronic kidney disease, a 10-year risk of at least 10 percent, or age 65 and up, drug treatment is recommended starting in the 130–139 / 80–89 band, aiming below 130. For lower-risk adults the drug-initiation threshold stays nearer the old 140/90.

There is one more clause in the rule, and it is the bottom of the pathway resurfacing at the top. The guideline does not just say what number to hit. It specifies how to take the reading: patient seated and rested, arm supported, cuff sized correctly, result based on the average of multiple readings. That clause is the guideline’s attempt to keep a clinic’s number on the same scale as SPRINT’s. It is also the clause most often skipped.

06 — Recommendation

“Treat to a goal of less than 130/80 mmHg”

In 2017 the American College of Cardiology and American Heart Association rewrote the definition of high blood pressure for the first time in fourteen years (Whelton et al., Hypertension 2018). Stage 1 hypertension now begins at 130–139 systolic or 80–89 diastolic, down from the old 140/90, and the treatment goal is less than 130/80.

The strength rating matters, and it is not uniform. The systolic target below 130 carries the guideline’s strongest endorsement, Class I with level of evidence B-R, for adults who already have cardiovascular disease or a 10-year predicted risk of at least 10 percent. For lower-risk adults the same target drops to Class IIb, the “may be reasonable” tier. The diastolic under-80 rests largely on expert opinion. So the headline number is really a family of recommendations of uneven evidentiary weight, and the part doing the heavy clinical lifting is the systolic target in higher-risk patients, grounded substantially in the one trial we just climbed.

The verdict

Now stand at the top and look back down the pathway we built.

The measurement is the rung that constrains everything above it, but the honest version of that constraint is narrower than the popular one. SPRINT was not uniformly unattended, and the attended-versus-unattended difference was trivial. What does hold is that AOBP reads lower than routine office measurement by an uncertain margin, so the guideline’s number and a typical clinic’s number are not interchangeable. The guideline anticipated this by specifying proper technique; the gap opens in the clinics that do not follow it.

The estimate is genuine. A well-conducted randomized trial showed a mortality benefit, which is rare and not to be waved away. It should be discounted modestly for early stopping, read on the absolute scale alongside its relative one, and weighed against a real increase in hypotension, syncope, electrolyte disturbance, and kidney injury.

The step from what was tested to what was recommended is where confidence and evidence diverge most. SPRINT tested under-120 in high-risk, non-diabetic, stroke-free adults. The guideline recommends under-130 for a broader population, with a Class I label on the systolic target. The direction is well supported. The specific cut point (under-130 was never a randomization target, but an interpolation between the two arms SPRINT compared), the breadth of the population it is applied to, and the assumption that a clinic 130 means a SPRINT 130 are all softer than the Class I phrasing conveys.

None of this argues against treating high blood pressure, or against lower targets for the patients SPRINT studied. It argues for knowing which rung your confidence is actually resting on.

For a high-risk adult whose pressure is measured properly, the pathway holds up well. For a low-risk or diabetic patient, or one whose pressure is taken on a single hurried cuff, more of the recommendation’s weight is hanging on extrapolation than the number suggests.

The contrast that bounds it

The cleanest check on how far SPRINT generalizes is the trial that ran the same comparison in the population SPRINT excluded. ACCORD-BP (the ACCORD Study Group, NEJM 2010) randomized 4,733 adults with type 2 diabetes to the identical intensive under-120 versus standard under-140 systolic targets. Its primary composite outcome (nonfatal myocardial infarction, nonfatal stroke, or cardiovascular death) showed no significant benefit: hazard ratio 0.88, 95% confidence interval 0.73 to 1.06, p = 0.20, over a mean 4.7 years. Stroke, a pre-specified secondary outcome, was reduced (hazard ratio 0.59, 95% confidence interval 0.39 to 0.89), but the headline result was null.

Two trials, the same blood-pressure targets, different populations, divergent answers. That is exactly why SPRINT’s exclusion of diabetics is not a footnote. It is a boundary on the recommendation, and reading SPRINT as if it applied uniformly is the kind of move this trace exists to catch.

This trace is methodological commentary, not clinical advice. It is an appraisal of what the statistics behind a recommendation can and cannot support, written for people who design studies, defend methods, and read the literature critically. It is not guidance for treating a patient, and it is not a substitute for the guidelines themselves or for clinical judgment.

If you have a recommendation, a guideline-development question, or a manuscript whose statistical basis you want traced this way, that is the kind of work I take on. Book a discovery call →

Sources

  • SPRINT Research Group. A Randomized Trial of Intensive versus Standard Blood-Pressure Control. N Engl J Med. 2015. doi:10.1056/NEJMoa1511939.
  • SPRINT Research Group. Final Report of a Trial of Intensive versus Standard Blood-Pressure Control. N Engl J Med. 2021;384(20):1921–1930. doi:10.1056/NEJMoa1901281.
  • Whelton PK, Carey RM, Aronow WS, et al. 2017 ACC/AHA Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults. Hypertension. 2018;71:e13–e115. doi:10.1161/HYP.0000000000000065.
  • Johnson KC, Whelton PK, Cushman WC, et al. Blood Pressure Measurement in SPRINT. Hypertension. 2018;71(5):848–857. doi:10.1161/HYPERTENSIONAHA.117.10479.
  • Bo Y, Kwok K-O, Chung VC-H, et al. Comparison between automated office blood pressure and conventional office blood pressure measurement: a systematic review. Curr Hypertens Rep. 2021;23:14. doi:10.1007/s11906-020-01118-1.
  • ACCORD Study Group; Cushman WC, Evans GW, Byington RP, et al. Effects of Intensive Blood-Pressure Control in Type 2 Diabetes Mellitus. N Engl J Med. 2010;362:1575–1585. doi:10.1056/NEJMoa1001286.