Rapid detection of viral causes of community-acquired pneumonia (CAP) is the holy grail for reducing unnecessary antibiotic use and controlling antimicrobial resistance. Already 25 years! In those years, multiple rapid diagnostic tests have been developed, only to end up in the “graveyard of promising tests” after clinical evaluation. Yet, most of these tests did not die from natural causes, but because of inappropriate study designs.
Recent Case Study
A recent example can be seen in last week’s Nature Communications. In four Spanish hospitals, 242 patients with CAP were randomised to receive either multiplex real-time PCR in non-invasive respiratory samples plus conventional microbiological testing (n=119), or conventional microbiological testing alone (n=123). The primary endpoint was days of antibiotic therapy (DOT). The study was conducted between February 2020 and April 2023, without patients suspected of COVID-19 as these were admitted to hospitals through other routes.
According to the abstract: “The median DOT was 10.04 (IQR 7.98, 12.94) in the multiplex PCR plus conventional microbiological testing group and 11.33 (IQR 8.15, 16.16) in the conventional microbiological testing alone group (difference −1.04; 95% CI, −2.42 to 0.17; p = 0.093). No differences were observed in adverse events and 30-day mortality. Our findings do not support the routine implementation of multiplex real-time PCR in the initial microbiological testing in hospitalised patients with CAP.”
What determines the success of rapid diagnostic testing in CAP, assuming the test result is accurate? And how can this be studied with maximum efficiency, i.e. the smallest sample size?
- The test result should always guide a treatment decision; and
- A large proportion of patients should have a viral infection.
This study, like many others, fell short of both requirements.
Patient Selection: A Critical Factor
Who should be enrolled when assessing the effects of such a test? A rapid diagnostic test will not change antibiotic prescriptions for every consecutive CAP patient. Based on the clinical severity of CAP in an individual patient, a physician may well decide to prescribe antibiotics (or not) regardless of the test result. Such patients should, preferably, not be included in the study (as the test does not guide treatment).
The same applies if test results do not reach physicians. The researchers of this trial stressed that “attending physicians received clinical interpretations of the PCR results, but the research team did not provide specific recommendations regarding antibiotic use based on the microbiological findings”. Again, the test may not guide treatment.
The Impact of Background Noise
In both scenarios, even if the test is accurate, the result may not change practice, and its potential benefits can be drowned in “background noise”; patients in the study in which the test cannot perform.
In this study, all patients were treated with antibiotics, including the eight (!) patients (6.7% of the population tested) with monomicrobial viral infection. So, the maximum potential benefit would have been zero, or a reduction in DOT, for these eight patients. The difference (max 6.7% reduction in DOT) is a far cry from the 27% reduction (from 11 to 8 days) that the sample size was based on.
Designing for Success
The optimal design would be to only randomise those patients, whose treating physician indicates that a test result will guide clinical management. In acute infections, this is operationally challenging, as treatment decisions cannot be postponed for too long. We tried this approach but failed and ultimately abandoned our efforts.
The next best design accepts that test results will not be used appropriately in all patients (for the reasons mentioned) but ensures a large enough sample size to demonstrate an effect, if present, despite background noise. This approach requires a much larger sample size and a significantly lower reduction in antibiotic use pursued but could lead to a meaningful and evidence-based strategy for rapid diagnostic testing in CAP patients.