Individualized Perioperative Blood Pressure Management

Every anesthesiologist has been taught some version of the same rule: a patient's "normal" blood pressure is not 65 mmHg — it's whatever their body runs at. Push a chronically hypertensive 70-year-old down to a MAP of 65 during a four-hour laparotomy and their kidneys may be perfusing below what their brain and kidneys have adapted to. The fix seems obvious: find each patient's own baseline, and defend that number in theatre. IMPROVE-multi is the most rigorous test that idea has ever received, and the result is less comfortable than the theory.

The backstory

The logic behind individualized blood pressure targets has always made intuitive sense. Intraoperative hypotension — MAP under roughly 60–65 mmHg for any meaningful stretch — keeps showing up in observational data linked to kidney injury, cardiac injury, and death after noncardiac surgery. So guidelines settled on keeping MAP above 60–65 as a floor. But that floor is built for an average patient, and a lot of patients aren't average — someone whose resting pressure normally sits at 95 might genuinely be under-perfused at 65, even while technically "within guideline."

The trial that really put this on the map was INPRESS, back in 2017: keeping systolic pressure within 10% of baseline, rather than a flat threshold, dropped a composite of inflammation and organ dysfunction from 52% to 38%. Small trial, one hospital, 292 patients — but it was persuasive enough that "treat the patient's own number" became something close to received wisdom. IMPROVE-multi is essentially the bigger, better-designed sequel, asking the same question with a sharper definition of what "baseline" actually means.

One notable feature of the design: instead of a single preop blood pressure reading (which INPRESS used, and which is easily thrown off by nerves or a busy holding bay), this trial measured overnight blood pressure every 30 minutes before surgery and used the average nighttime MAP — the number a patient settles into during real sleep — as their individual target. It's a genuinely better baseline, which means this was a fair shot at the hypothesis rather than a strawman version of it.

What they did

1,142 patients having major abdominal, urologic, gynecologic, or liver surgery, randomized 1:1. One group had MAP kept at or above their own overnight baseline throughout surgery and for two hours after (capped between 65 and 110 mmHg, so nobody's chasing a reverse-dipper's freak overnight reading of 130). The other group just got routine care: MAP ≥65, managed however the anesthesiologist normally would.

The groups were well matched going in — similar age, similar comorbidities, about half with a history of hypertension in each. And the individualized group really did run at higher pressures during surgery: more time above target, more norepinephrine used to get there. So this wasn't a trial that failed to deliver its own intervention. It did what it set out to do.

And then — nothing

That's really the headline. Nothing happened.

*AKI, myocardial injury, nonfatal cardiac arrest, or death within 7 days. †Kidney replacement therapy, MI, nonfatal cardiac arrest, or death within 90 days. None of the 22 secondary outcomes reached significance.

What stands out most is that every single point estimate trends the wrong way — AKI, myocardial injury, the 90-day composite, even MI at 90 days. All numerically worse in the individualized group, none of it statistically significant, all of it consistent with plain noise around a true null. And this wasn't an underpowered trial that quietly missed something real — it hit its planned sample size with 90% power to detect an 8% absolute risk reduction. The trial looked hard for an effect, and there was nothing there to find. If anything there's a faint suggestion of harm from chasing higher pressures with more norepinephrine, though that signal shouldn't carry much weight given the confidence intervals.

How this sits with everything else

IMPROVE-multi isn't landing in a vacuum — the field had already been drifting this way. A smaller trial in 2021 found a fixed MAP target of 75 was no better than 60. POISE-3, with over 7,000 patients, found a hypotension-avoidance strategy no better than a permissive one for cardiovascular outcomes. What this trial adds is the individualization piece specifically, tested with about as good a baseline measurement as anyone's used — and it still came back null.

Worth saying plainly, because the authors do: hypotension in the routine-care group was rare and brief to begin with, since everyone was already being kept above 65 per guideline. So this trial isn't telling us what happens if MAP actually drops to 45 for twenty minutes — almost nobody in either arm experienced that. It's telling us that once patients are already being kept reasonably safe at the existing floor, pushing a subset higher because of their personal baseline doesn't seem to buy anything measurable. That's a narrower claim than "individualizing doesn't matter," and it's the one the data actually supports.

One limitation worth flagging: anesthesiologists obviously knew which group each patient was in, since they had to hit the target. Impossible to blind, but it does mean subtly different attention beyond just the number on the monitor can't be fully ruled out for the individualized group. The authors acknowledge this limitation themselves.