Every day thousands of physicians worldwide are facing the dilemma whether “to treat or not to treat this coughing patient with antibiotics”. A test that safely discriminates a bacterial from a non-bacterial cause is the holy grail, and some feel that procalcitonin (PCT) is just that. Results from multiple – mostly European – studies and meta-analyses pushed them in that direction. Yet, we are not that often sure of the causative pathogen, and diagnostic research in the absence of a gold standard is tricky business. And now there is this marvelous study that perfectly addresses that uncertainty and questions all prior PCT evidence, and that was discussed in our PhD journal club.
This was a randomized controlled trial in 14 emergency departments that compared usual care to care plus PCT and an antibiotic-prescription guideline. The single most important words in the paper: “We enrolled patients in the emergency department for whom the treating clinician had given an initial diagnosis of acute lower respiratory tract infection (LRTI), but had not yet decided to give or withhold antibiotics and about whom there was uncertainty regarding the need for antibiotics, such that PCT data could influence the prescribing decision.” After enrolling 1,656 such patients (826 with PCT) median durations of antibiotic treatment were 4.2 and 4.3 days with PCT and usual care. Adverse events attributable to withholding antibiotics occurred with equal frequency in both groups, as did all secondary endpoints. Although antibiotic prescription occurred more frequently when PCT was higher (0.1 to 0.25 and 0.25 to 0.5 μg/L), these groups were relatively small as most patients had undetectable PCT levels.
The beauty of trial was that it really mimicked clinical practice (i.e., the test was used for the right patients) and that all personnel was trained and educated before the start of the study (i.e, they had been told what to do with a PCT result). Yet, as in real life, clinicians retained complete autonomy over their patients and were allowed to defer from the guideline.
The decision to only enroll those patients for whom there was doubt whether to start antibiotics, is really unique in diagnostic studies for LRTI. Previous studies frequently included patients for which the diagnostic test is not intended (antibiotics prescribed (or not) regardless of test result), or had antibiotic prescription tightly protocolized, thereby excluding the (most unpredictable) real life variable; the physician. Even more studies were observational with the test performed afterwards and then compared to a final diagnosis (bacterial or non-bacterial) based on expert panel opinion, sometimes with exclusion on undetermined episodes. This inflates the test performance, and the unfortunate physician cannot decide at the spot who will be in that undetermined group.
Motivated by their supervisors Tessa Mulder and the other PhD students tried to find a fatal flaw, but failed. It cannot be excluded that prior improvements in antibiotic stewardship in these centers prevented PCT to make a difference, questioning the representativeness of the study setting and the generalizability of the results. But even if so, that was then achieved without PCT. If the fatal flaw remains unidentified this study may well serve as a new (gold?) standard on how to determine the real-life added value of a new diagnostic test to optimize antibiotic use for infections in which a gold standard is missing.
Tessa Mulder does a PhD on the prevention of surgical site infections after colorectal surgery.