From the authors
- S. Suissa
P. Barber correctly points out the difficulties in identifying chronic obstructive pulmonary disease (COPD) patients using administrative databases, and especially in distinguishing between COPD and asthma. This is particularly important when assessing the impact of inhaled corticosteroids because these drugs have been shown in randomised trials to be extremely effective in asthma but not in COPD. Thus, a study based on a mixed population that includes both asthma and COPD patients will result in an average effect for inhaled corticosteroids.
To maximally ensure the accuracy of a first-time COPD diagnosis for cohort entry, we used three criteria: 1) 55 yrs of age or over; 2) three or more prescriptions on at least two different dates for a bronchodilator within a 1-yr period; and 3) no other prescriptions for bronchodilators or other asthma drugs during the 5-yr period prior to the three defining prescriptions 1, 2. With these criteria, we were confident that the proportion of asthma patients would be kept to a negligible level.
In our previous research on asthma, we effectively used similar criteria to study asthma, with the difference that age at cohort entry (diagnosis) was restricted to between 5 and 44 yrs, thus reducing the possibility of including COPD patients 3–5.
While observational databases studies are challenging, they are indispensable to complement other types of studies into the effects of drugs. As P. Barber notes, scientific rigor must be introduced in their design and analysis, which can be best achieved by intense collaborations between clinicians and methodologists. We have and will continue to work in this direction.
- © ERS Journals Ltd