A sample of clinicians within the INHALE trial completed brief questionnaires within 24 hours of arm-randomization and prescribing decision. In addition to identifying factors that predicted prescribing, the investigators determined the proportion of antibiotic prescribing decisions that were consistent with the pneumonia panel test results.
What You Need to Know
While about 65% of antibiotic decisions aligned with pneumonia panel results, clinicians still rely on multiple clinical factors rather than strictly following test guidance.
Clinicians were far more likely to prescribe in line with guidelines when results were positive, quickly delivered, and supported by other signs of infection—highlighting the importance of both confirmation and timeliness.
Suspicion of other infection sources and discomfort with withholding antibiotics (especially after negative tests) can lead to deviations from protocol, reflecting a persistent “better safe than sorry” mindset in ICU settings.
The questionnaire elicited agreement on a 3-point scale (disagree, uncertain, agree) with the following statements: I believed the pneumonia panel results; I found the pneumonia panel results easy to interpret and understand; the patient was clinically deteriorating; the patient had laboratory or radiological evidence of deterioration; the patient likely had other sources of infection besides the lung; initial antibiotic prescription for respiratory tract infection or other infection was appropriate and did not need to be changed.
Of 481 questionnaires issued for eligible prescribing instances (for intervention and control arms), 295 (61%) responses were received; with 282 completed adequately for inclusion, and 159 of these relating to the intervention–arm cases for this analysis.104 (65.4%) of the prescribing decisions were consistent with pneumonia panel results; either meeting algorithm guidance as active against the identified pathogen, or no antibiotic prescribed with negative test.
Stewart and colleagues reported that prescribing consistent with protocol was most often predicted by: receiving a positive result; believing the result; receiving the result promptly; perceiving low likelihood of a non-respiratory source of infection; and no current antibiotics for a non-respiratory infection.
Receiving a positive rather than negative test result increased the odds of a prescription being consistent with protocol by a factor of 138.6. Perceiving the results as quick, and the patient having other evidence of infection increased the odds of protocol consistent prescribing by factors of 12.15 and 4.70 respectively. If, however, the clinician suspected another, non-respiratory source of infection, the odds of the protocol-consistent prescribing decreased by five-fold.
"Positive results may act in a 'confirmatory' manner, whereas negative results may generate cognitive dissonance—namely, between what clinicians expected to find, and what the tests revealed—especially when the test revealed nothing," the investigators suggest.
The investigators emphasize that multiple factors underlie the decision to prescribe an antibiotic, including the clinicians' beliefs about the test itself.They also note that there is particular hesitation to not prescribe, and a tendency to prescribe "just in case."
"'Correct' decisions do not automatically flow from diagnostic technology," they observe. "Pneumonia panel tests do not establish the 'ground truth' of the presence of infection somewhere in the body, and therefore it would be expected that in some cases, antibiotics would be started and continued despite negative results."
Reference
1. Stewart S-J F, Enne VI, Pandolfo AM, et al. Machine says go, doctor says no: an ecological momentary assessment analysis examining clinicians' perceptions of, and their antibiotic prescribing behaviour when using rapid molecular diagnostic tests in intensive care. Antimicrob Resist Infect Control. 2026 March 24; 15(1):42. doi: 10.1186/s13756-025-01690-8.
