Exploring Delayed Pediatric Diagnoses in Emergency Department – Reply

In Reply

We appreciate the criticisms raised by Tsai and Ma. We believe their concerns do not substantively impact our findings.¹ Higher volume is clearly associated with improved patient outcomes, but prior to our study, that association had not clearly been extended to diagnostic quality.² Emergency Department (ED) volume is a marker of clinical experience, which itself is related to expertise. Experience and expertise matter in diagnostic quality.³ Experience and expertise are difficult to measure directly, but both attributes are associated with patient volume; clinician background; patient population; setting; hospital resources; such as the presence of inpatient beds that support pediatric care; and more. That is, those nonvolume factors mediate, not confound (as proposed by Tsai and Ma), the volume-delayed diagnosis association we observed across nearly 59000 ED encounters. By definition, mediators should not be adjusted because they are the reason an association exists.⁴ For example, it would stand to reason that the staffing of pediatric specialists or the availability of 24-hour imaging would only exist in hospitals with higher pediatric volume and would tend to reduce the likelihood of delayed diagnosis. In that case, such resources would provide a reason for a volume-delay association, rather than causing such an association to be incorrect.

We agree that some revisits may be a marker of disease progression.⁵ We appreciate the mention of Kawasaki disease as an example of a disease that requires time to recognize a diagnosis. However, as we noted in our discussion, it is not clear why the relative balance of delayed diagnosis vs progression of disease would differ by volume. Such a difference would need to be present to confound our main findings.

We agree that clinicians’ experience and interactions with patients are important in quality care, but we are not aware of evidence demonstrating that more clinician time spent with patients is associated with either volume or diagnostic quality. The data cited by Tsai and Ma demonstrating higher rates of individuals who left without being seen and longer length of stay suggest that higher-volume EDs should have lower quality, but we found the opposite. We agree that referrals can have both positive and negative impacts on diagnostic quality at the receiving ED, but we conducted a sensitivity analysis excluding referrals that did not change the clear volume-delay association. Additionally, we are not aware of research evaluating the association of ED volume and clinician burnout, or clinician burnout and missed diagnoses, as ED volume and clinician workload are not automatically linked. However, regardless, we found the opposite association between pediatric volume and missed diagnoses.

Overall, we do not believe the criticisms impact the interpretation of our findings, and in some cases lend support. We appreciate Tsai and Ma calling attention to solutions that could help improve diagnosis in all EDs, and we are highly committed to testing and implementing those solutions in the future.