Continuing the home beta-blocker in suspected infection was associated with lower mortality

Design

retrospective cohort analysed as a target trial emulation with inverse probability of treatment weighting, Cox regression, single US academic centre

Population

4,635 adults on chronic beta-blockers admitted from the emergency department with suspected infection, excluding shock and heart rate <40 or >120; 1,172 (25.3%) continued

Primary outcome

90-day all-cause mortality

Effect

hazard ratio 0.77 (95% CI 0.61–0.98), p=0.03; in-hospital mortality odds ratio 0.60 (0.30–1.20), p=0.15

Holding the beta-blocker is close to reflex when a patient arrives from the emergency department with a possible infection — the reasoning being that blunting the tachycardic response in impending sepsis is dangerous. This study used target trial emulation on 4,635 adults already on chronic beta-blockers admitted from one large US academic emergency department with suspected infection, defined by orders for blood cultures and broad-spectrum antibiotics. Patients in shock, with heart rate below 40 or above 120, or needing a clinician-directed intravenous beta- or calcium-channel blocker were excluded, with inverse probability of treatment weighting for baseline differences.

Only 1,172 patients (25.3%) had an oral beta-blocker continued within 48 hours. Continuation was associated with lower 90-day all-cause mortality (hazard ratio 0.77, 95% CI 0.61–0.98, p=0.03) and shorter hospital stay. In-hospital mortality did not differ significantly (odds ratio 0.60, 95% CI 0.30–1.20).

The length-of-stay result is where the reasoning has to slow down. An incidence rate ratio of 0.39 means stays about 60% shorter, which is an implausibly large effect for continuing an oral tablet, and it points at what the propensity score could not weight away: patients who are well enough to take oral medication are well enough to go home sooner. That same mechanism could produce part of the mortality signal. What survives is the direction and the absence of harm — in a non-shocked patient who was taking a beta-blocker yesterday, there is no evidence here that stopping it helps, and some that it does not.

• This applies to patients not in shock, with heart rate between 40 and 120 — the exclusions define the population, and shock was excluded.
• Abrupt beta-blocker withdrawal has its own risks in ischaemic heart disease and is the harm this practice may be trading for.
• Three-quarters of eligible patients did not have theirs continued; check the drug chart on admission rather than letting omission happen by default.
• Where the patient cannot take oral medication, that is a separate decision — do not read this as a reason to give an intravenous substitute.
• This is one US academic centre; the direction is plausible, the magnitude is not transferable.

Why it matters

The default is to hold the drug, and the only evidence available points the other way.

Don't overread it

This is an observational single-centre analysis — patients whose beta-blocker was continued were probably less unwell in ways the weighting could not capture.

The statistics, in plain English

A hazard ratio of 0.77 with an upper bound of 0.98 only just excludes no effect, and the in-hospital mortality estimate (0.60, 0.30–1.20) does not exclude it at all — so the mortality benefit rests on events accumulating after discharge, which is where confounding by how well the patient was doing has most room to operate. The length-of-stay ratio of 0.39 is the tell: no oral tablet shortens admissions by 60%, so that result is measuring patient condition, not drug effect, and the same bias is present in the mortality estimate to an unknown degree. Target trial emulation reduces confounding by indication; it does not remove it.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.