The impact of rapid molecular diagnostic testing for respiratory viruses on outcomes for emergency department patients

In reply: Katelaris and colleagues raise concerns that uncontrolled before‐and‐after studies are vulnerable to bias and that other factors (eg, the severity of influenza) may have confounded our findings.1 I agree that uncontrolled before‐and‐after studies, being observational studies, are generally prone to different types of bias and that other residual (unmeasured) factors may have confounded our study outcomes.2

The potential impact of the differences in the prevalence and severity of influenza during the two study periods was adequately acknowledged in our article as a limitation. The objective of our study was to determine the association between the use of rapid polymerase chain reaction (PCR) for respiratory viruses in emergency departments and patient and laboratory outcomes.1 It is important to remember that, consistent with previous similar studies,3,4 our study evaluated the correlation (association) between rapid PCR use and outcomes (eg, hospital admission) and did not claim any cause‐and‐effect relationship. The causal inferences can only reliably be drawn using randomised controlled trials. As noted in our article, because of its inherent design limitations,2 our study could not fully account for all possible confounders or temporal trends in the severity and prevalence of influenza.1

I also acknowledge that a controlled before‐and‐after study or an interrupted time series analysis would offer a better estimation of the impact of an intervention. However, because of the nature of our data, we were not able to conduct such a study or analysis. I agree that even if our findings implied potential economic benefits, a formal economic evaluation is needed. The potential role of rapid PCR in antimicrobial stewardship programs should also be explored.