Another Piece of the Inhaled Corticosteroids-in-COPD Puzzle

Jørgen Vestbo, M.D.

Hvidovre University Hospital, Hvidovre, Denmark

	ARTICLE

TOP ARTICLE REFERENCES

In this issue of AJRCCM, Sin and Tu (pp. 580-584) publish their exciting pharmacoepidemiologic study of inhaled corticosteroids (ICS) in the elderly (1). Using Canadian register data, they studied 22,620 patients with COPD 65 yr of age and older after hospital admission and looked at 1-yr prognosis according to whether or not they had received a prescription of ICS within 3 mo of their admission. The outcome was death or readmission, two relevant outcome measures in any COPD study. Prescription of ICS seemed more frequent among patients who already received a lot of medication, which is consistent with these patients having more severe disease. The patients prescribed ICS were also slightly younger than those who did not get a prescription of ICS, less likely to have comorbidities, but more likely to have visited an emergency room for COPD in the preceding year. Prescription of ICS was associated with a significantly reduced risk of death and readmission; the crude 1-yr reduction in "death and/or readmission" was 10% (95% confidence interval, 6 to 15%). After adjusting for differences in age, sex, comorbidities, number of ER or office visits for COPD within 1 yr before the study, and concomitant medication, the reduction in mortality was 29% (22 to 35%) and the reduction in risk of readmission was 24% (20 to 29%).

The study was not a randomized controlled trial, and it is relevant to ask if this study can be viewed in context with the controlled trials that normally form the basis for our decisions on treatment options. However, before dismissing the study because of its design, lack of randomization, and the notion that only controlled trials reflect the truth, it seems worthwhile to reflect in general on pharmacoepidemiology and the similarities and differences between observational studies and controlled trials. Views on the utility of observational studies differ (2). Generally, there is agreement that the presence of invalidating bias is more likely to occur in the observational setting than in a controlled trial. In the study by Sin and Tu, however, several advantages are likely to reduce the risk of bias. Both recall bias and detection bias have been avoided through the use of a drug benefit database on prescriptions for medication and a hospital discharge database and a mortality register for outcomes. Confounding by indication is a potential bias arising from the lack of randomization. It refers to the fact that a given treatment may be chosen for the particular patient because of presence or absence of specific features of the disease. Most commonly, severity of disease can affect both choice of treatment and prognosis, but other disease characteristics may also play a role. For COPD, any feature of asthma would increase the likelihood of prescription of ICS and of a significant effect of the drug. Whereas the latter would lead to overestimation of an effect of ICS in COPD, an increased use of ICS with increasing severity would lead to an underestimation. In the study by Sin and Tu, both issues have been addressed, although differently. Asthma was considered unlikely in this age-group where the majority of obstructive lung disease is due to COPD. As the study includes only patients discharged with a diagnosis of COPD, I would tend to agree; asthma is rarely labeled as COPD, whereas COPD may be termed asthma in a discharge register. Confounding by severity can be looked at more directly as a fair number of descriptors of disease severity and comorbidities were used, and, in fact, the group prescribed ICS seemed the more severely ill. Nevertheless, previous experiences have taught us that treatment advantages can be exaggerated in an observational study (4), although several examples can be found where this has not been the case (2, 3). How do the findings of the present study then relate to the results from the clinical trials?

In fact, the study by Sin and Tu fits in very nicely with the results of the controlled trials published within the last few years. Whereas ICS do not affect the decline in FEV₁ (5), they seem to have an effect on exacerbations depending on severity of the disease. In the two recent studies including mild and mild-moderate disease (6, 7), no effect on exacerbation rate was found. The Lung Health Study II did not have exacerbations as an effect parameter, but, compared with placebo, inhaled triamcinolone led to both fewer unscheduled visits to a physician and fewer emergency room visits for respiratory problems (8). The two British studies including patients with severe COPD both showed that fluticasone propionate had an effect on exacerbations; in the study by Paggiaro and colleagues (9) through shifting exacerbations to the milder end of the scale, and in the ISOLDE study by reducing the actual number of exacerbations (10). As COPD progresses, the risk of an exacerbation resulting in death increases. The present Canadian study included only elderly patients who had already been hospitalized at least once and in this setting the effect on mortality of preventing exacerbations can be studied. It seems likely that this pharmacoepidemiologic study in a sense reflects the "real-life" impact of the findings from the controlled trials, including patients with moderate-severe COPD. To me and to other clinicians using ICS in patients with severe COPD who have experienced repeated exacerbations requiring oral corticosteroids, the results of the study are indeed reassuring.

An interesting feature of this register linkage study is the timing of recruitment. So far, all the controlled trials on ICS have recruited participants in a stable state, often with a standard run-in period in which patients had to demonstrate stability. Indeed, in the trial including the most severely ill patients, the ISOLDE study (10), a significant number of participants were withdrawn from the run-in period because of exacerbation, the majority of them after stopping their usual ICS (11). As there is a growing need for large randomized studies in COPD with mortality as outcomeand, in fact, some are already ongoingthis approach to timing of recruitment could be useful. Future studies with admission and/or mortality as outcome ought to learn from this experience from clinical epidemiology and perhaps view admission for an exacerbation as an equivalent of the acute myocardial infarction in ischemic heart disease. Through this, the benefit of using both controlled trials and observational studies in clinical research on COPD may be further enhanced.

	References

TOP ARTICLE REFERENCES

1. Sin DD, Tu JV. Inhaled corticosteroids and the risk of mortality and readmission in elderly patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2001; 164: 580-584 [Abstract/Free Full Text].

2. Benson K, Hartz AJ. A comparison of observational studies and randomized, controlled trials. N Engl J Med 2000; 342: 1878-1886 [Abstract/Free Full Text].

3. Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies, and the hierarchy of research designs. N Engl J Med 2000; 342: 1887-1892 [Abstract/Free Full Text].

4. MacMahon S, Collins R. Reliable assessment of the effects of treatment on mortality and major morbidity: II. observational studies. Lancet 2001; 357: 455-462 [Medline].

5. Mapp CE. Inhaled glucocorticoids in chronic obstructive pulmonary disease. N Engl J Med 2000; 343: 1960-1961 [Free Full Text].

6. Vestbo J, Sørensen T, Lange P, Brix A, Torre P, Viskum K. Long-term effect of inhaled budesonide in mild and moderate chronic obstructive pulmonary disease: a randomised controlled trial. Lancet 1999; 353: 1819-1823 [Medline].

7. Pauwels RA, Löfdahl C-G, Laitinen LA, Schouten JP, Postma DS, Pride NB, Ohlsson SV. for the European Respiratory Society Study on Chronic Obstructive Pulmonary Disease. 1999. Long-term treatment with inhaled budesonide in persons with mild chronic obstructive pulmonary disease who continue smoking. N Engl J Med 1999; 340: 1948-1953 [Abstract/Free Full Text].

8. The Lung Health Study Research Group. Effect of inhaled triamcinolone on the decline in pulmonary function in chronic obstructive pulmonary disease. N Engl J Med 2000;343:1902-1909.

9. Paggiaro PL, Dahle R, Bakran I, Frith L, Hollingworth K, Efthimiou J. on behalf of the international COPD study group. Multicentre randomised placebo-controlled trial of inhaled fluticasone propionate in patients with chronic obstructive pulmonary disease. Lancet 1998; 351: 773-780 [Medline].

10. Burge PS, Calverley PM, Jones PW, Spencer S, Anderson JA, Maslen TK. Randomised, double blind, placebo controlled study of fluticasone propionate in patients with moderate to severe chronic obstructive pulmonary disease: the ISOLDE trial. BMJ 2000; 320: 1297-1303 [Abstract/Free Full Text].

11. Jarad NA, Wedzicha JA, Burge PS, Calverley PMA. An observational study of inhaled corticosteroid withdrawal in stable chronic obstructive pulmonary disease. Respir Med 1999; 93: 161-168 [Medline].