Reply to “Correspondence to ‘Association between Chronic Rhinosinusitis and New Onset Asthma Implications for Prevention.’”

To the Editor:

We thank Cheng et al. for their careful consideration and comments on our manuscript entitled “Sinus inflammation and chronic rhinosinusitis are associated with a diagnosis of new onset asthma in the following year.” ¹ Here we respond to each of their comments.

Cheng et al. were concerned about confounding. We agree that consideration and control, through a variety of methods, for potential confounding variables is critical in observational epidemiologic research. We note that we matched or adjusted for a large set of confounding variables, including age, sex, race, ethnicity, medical assistance status (a surrogate for family socioeconomic status, as emphasized by Cheng et al.), number of outpatient encounters, body mass index, calendar year, tobacco use, allergic rhinitis, gastroesophageal reflux disease, chronic obstructive pulmonary disease, obstructive sleep apnea, and pneumonia. We agree that potential confounding by air pollutants, a set of important environmental exposures mentioned by Cheng et al., would be an interesting topic for future research.

Sinus surgery and eosinophilic inflammation were of particular interest to Cheng et al. In our analysis, we evaluated how surgery for chronic rhinosinusitis (CRS) and eosinophilia influenced the associations and reported these results in the published manuscript. We agree that more research on these questions would be useful. Our analysis required a very large sample size, with exclusion of over 43,000 patients with prevalent asthma (and thus many patients with prevalent CRS) and the comparison of over 35,000 patients with new onset asthma to almost 900,000 patients without new onset asthma. The potential benefits of CRS surgery in relation to new onset asthma would be most effectively studied by a randomized controlled trial but, given the estimated annual incidence of asthma of 7-10 per 1000 in our study, would require randomization of very large numbers of patients. This is a challenging study to do, so we hope the results we presented help to shed some light on this topic in the meantime.

It is unclear to us why Cheng et al. conclude that methodologic limitations may have led to overestimation of associations. If we misclassified some patients without CRS as having CRS because we were unable to be certain about the 12-week duration requirement, a concern highlighted by Cheng et al., this seems it could bias results away from or towards the null, depending on whether the misclassification was non-differential or not. Lack of information on CRS duration in the electronic health record (EHR) is why we used more than one case definition to find CRS. We discussed in the published manuscript why we believe that duration considerations were unlikely to be invalidating. While measurement error is always a concern in studies of this type using EHR data, we note that a positive sinus CT scan identified with analysis of text and validated with chart review, goes beyond prior approaches and is strong evidence of radiologic sinus inflammation without a high prevalence of misclassified scans.

We hope we have been appropriately responsive to the concerns of Cheng et al. and again thank them for their consideration of our manuscript.