Evidence based case report: Twenty year cough in a non-smoker

Publisher Website
Google Scholar
Abstract
Searching for evidence The first step was to formulate an answerable clinical question1—one of aetiology, differential diagnosis, diagnostic test accuracy, prognosis, treatment, or prevention. Here the initial clinical question was one of differential diagnosis: what are the possible causes, and frequency, of a chronic cough? My favourite ambulatory care text, Primary Care Medicine,2 listed 20 possible causes of chronic cough. These included environmental irritants such as smoking; lower respiratory tract problems such as lung cancer and asthma; upper respiratory tract problems such as chronic rhinitis or sinusitis; extrinsic compressive lesions such as aortic aneurysm; and miscellaneous causes such as psychogenic cough and reflux oesophagitis. Progressive conditions such as cancer or bronchiectasis were unlikely as Mrs V's condition had not changed over the 20 year course. The further information I needed was a differential diagnosis that included the frequency of possible causes. The chapter in Primary Care Medicine did not give the relative frequency for all diagnoses but did discuss two studies. Was this the best evidence on which to base my management? Preferably these studies would have investigated a large, consecutive, and representative series of patients with sufficient follow up to make any misdiagnosis apparent. Unfortunately, most relevant studies are not coded as such in Medline. The search requires two components—firstly, chronic cough or synonyms, and secondly, a “methodological filter”—to confine it to appropriate studies of differential diagnosis. A Medline search using “chronic near cough” (the special term near means that the “chronic” and “cough” need to be close together but allows for terms such as “chronic non-productive cough”) yielded 343 references in just the past five years. This was too many, so the search needed to be confined to articles that were potentially most relevant. I felt that a reasonable methodological filter might be to try to restrict references to those with an appropriate sample—that is, a random or consecutive set of cases, plus an adequate yardstick test or tests, and an appropriate follow up (to catch missed or mistaken diagnoses). Putting this together, I tried the following search: “chronic near cough,” “investigat* or diagnos* or cause*,” and “consecutive or follow up”. This search produced the four studies described in the table,3-6 including the two that had been discussed in Primary Care Medicine. These studies had used somewhat different diagnostic yardsticks, but, following Irwin,4 all required a response to specific treatment as a criterion for diagnosis. Given this need to establish response, the follow up in most studies was quite acceptable. Of particular interest was the structured approach to empirical treatment described by Pratter (I now keep a copy of this useful algorithm in my clinic desk).6 View this table: In this window In a new window Studies found by a Medline search on causes of chronic cough