To the Editor:
I read with interest the paper by Soriano et al. 1, in which the authors stated that fluticasone proprionate alone or in combination with salmeterol is associated with increased survival of chronic obstructive pulmonary disease (COPD) patients managed in primary care.
First, according to their data, the authors should have considered replacing fluticasone with inhaled steroids. More importantly, they did not discuss potential limitations that could have cast doubt on this statement.
Some of the data presented in table 1 of this paper pointed towards potential important differences between the three “active” groups and the reference group at baseline. Indeed, baseline treatment in the three groups receiving fluticasone and/or salmeterol could be described as fairly typical for moderate-to-severe COPD patients, with >90% of the patients receiving inhaled bronchodilators. In contrast, less than one-half of the reference group used inhaled bronchodilators at baseline, which is also clearly different from the reality of drug use in COPD in the UK 2. This is peculiar for a group of COPD patients that show a high mortality. The 25% mortality at 2 yrs in this group parallels the prognosis of COPD patients requiring long-term oxygen therapy 3, 4 or with a forced expiratory volume in one second (FEV1) of <30% predicted in the intermittent positive pressure breathing trial 5.
Moreover, patients in the reference group were significantly older, had significantly more comorbidities (the combination of the two pointing towards potentially more severe comorbidities), and were less likely to have severe COPD as defined by the authors (use of oxygen or nebulised therapy).
Finally, the cause of death was not reported.
These points lead to the following questions being raised. 1) Could the method of inclusion in the UK General Practice Research Database (GPRD) explain these differences? Neither the description by the authors nor previous papers related to the GPRD 6–8 clarify whether inclusion in the database as a COPD patient required contact with a general practitioner (GP) for primary diagnosis of COPD or whether the mention of COPD at a contact for another purpose (and therefore as a secondary diagnosis) was sufficient. Since COPD is prevalent, particularly so in an old population such as the reference group of Soriano et al. 1 (mean age±sd 72±10 yrs) 9, the detection of COPD as a secondary diagnosis when contact is made with the GP should not be a rare occurrence. Accordingly, the primary reason for the healthcare contact leading to the inclusion in the GPRD could have been a comorbid state with a worse prognosis. This could play a role in the higher all-cause mortality observed in the reference group. 2) The authors reported the sensitivity of a diagnosis of COPD in the GPRD but did not report the specificity, which is equally important in the interpretation of their data. As patients from the three active groups seem to be quite typical COPD populations, could patients misdiagnosed as COPD be overrepresented in the reference group? 3) Since other studies based on the GPRD were related to the cause of death recorded in the database 10, 11, could the authors have clarified whether the cause of death was available in the different groups?
- © ERS Journals Ltd