Procalcitonin-guided stopping cut antibiotic exposure by just under a day and met its prespecified safety margin. A widely shared summary graphic says the opposite. The real debate is about the margin, not the result.
The circulating graphic concludes that procalcitonin is "an unreliable indicator for antibiotic cessation" and that mortality "was higher" with PCT. The trial concluded the reverse: PCT-guided care reduced antibiotic duration safely, meeting non-inferiority for 28-day mortality. The numbers in the graphic's own table are correct. The bottom line drawn from them is not.
Both strategies had point estimates slightly above zero. The PCT interval stays entirely left of the 5.4% margin, so non-inferiority was met. The CRP interval crosses it, so the CRP mortality result is inconclusive, not harmful.
PCT remains an unreliable indicator for antibiotic cessation.The trial's conclusion is that PCT-guided care reduces antibiotic duration safely compared with standard care. That is the finding the graphic was summarizing.
PCT stopped antibiotics one day earlier, but mortality was higher.Mortality was 20.9% with PCT versus 19.4% with standard care, a difference of 1.57 points with a 95% CI of −2.18 to 5.32. A confidence interval spanning zero is not a mortality increase. The trial was built to test non-inferiority, and PCT passed.
Strong antibiotic stop advice was given earlier in the PCT group.Accurate as a description, but it is the mechanism of the intervention rather than a finding. Clinicians in the PCT arm received protocolized stop advice; that is what the trial randomized.
CRP guidance did not meet non-inferiority.Correct. CRP did not shorten antibiotic duration (10.6 vs 10.7 days, P = .79), and its mortality comparison was inconclusive.
Look at the overall clinical picture.Always true, and the protocol agrees: PCT advice was advisory, and the treating clinician made the final call.
| Design element | ADAPT-Sepsis |
|---|---|
| Population | Adults in 41 NHS ICUs, within 24 h of IV antibiotics for suspected sepsis, expected to need ≥72 h of antibiotics |
| Arms | Daily PCT-guided advice, daily CRP-guided advice, or standard care; intervention-concealed |
| Co-primary outcomes | Total antibiotic duration to day 28 (effectiveness); 28-day all-cause mortality (safety, non-inferiority) |
| Non-inferiority margin | 5.4 percentage points |
There is a legitimate critique of ADAPT-Sepsis, and it is not the one in the graphic. A non-inferiority margin of 5.4 percentage points means the trial was designed to accept that PCT guidance might be up to about 5 points worse for mortality, in exchange for roughly one fewer day of antibiotics. The upper confidence bound landed at 5.32.
Put plainly: the trial excludes a mortality harm larger than about 5 points. It does not exclude a smaller one. Whether a margin that wide is acceptable for a stewardship intervention is a values question, and critics have argued it is too generous.
That critique cuts against the trial's design, not its reading. Within the margin it set, the result is clear.