© 2004 American Thoracic Society
Inhaled Corticosteroids and Fracture Risk in COPDTo the Editor:Lee and colleagues (1) report that the use of more than 700 µg per day of beclomethasone or equivalent inhaled corticosteroid for at least 2 years was associated with an increase in the risk for nonvertebral fracture (adjusted odds ratio [equivalent to the rate ratio], 1.68; 95% confidence interval [CI], 1.10–2.57), but only with current use. This finding, based on a case–control study of 1,708 cases of nonvertebral fracture and 6,817 matched control subjects selected among patients with chronic obstructive pulmonary disease, is not entirely consistent with previous studies. Our recent study of 9,624 cases of hip or upper-extremity fracture and 191,622 matched control subjects found no such effect at this dose (2). The only increased risk we found was among the subset followed for more than 8 years and using daily doses of more than 2,000 µg of inhaled corticosteroids (rate ratio, 1.61; 95% CI, 1.04–2.50), but not necessarily with current use. We verified in our data whether current use of lower doses among subjects with at least 4 years of follow-up increases this risk (Table 1) . Our analysis found no increase in this risk with current use of even very high doses of inhaled corticosteroids.
A methodologic aspect of the study by Lee and colleagues (1) may perhaps explain this discrepancy. For each case, the date of fracture was taken as the index date to determine current use of inhaled corticosteroids. This index date was assigned to every control subject matched to the case. However, the article does not mention whether the control subjects also were "at risk" on the index date. In other words, was each control subject available to be a "current" user on that day, namely to have the opportunity to receive a prescription during the 30-day period before the index date? The increasing crude odds ratios (0.96, 1.18, 1.29) with narrowing exposure intervals before the index date (anytime, prior 90 days, prior 30 days) suggest that this may be so. For the case subjects, we can safely presume that they were available for receipt of a prescription during the 30-day period before the index date because they had a fracture on that date. Control subjects who were not at risk on the index date—for example, if they had died or were in hospital, and therefore were unavailable to receive a prescription for inhaled corticosteroids—would be considered unexposed by the analysis. This would result in an overestimation of the odds ratio. It would be useful if the authors could redo this analysis only with control subjects at risk on the index date.
McGill University Health Centre Montreal, Quebec, Canada FOOTNOTES Conflict of Interest Statement: S.S. has been reimbursed for attending several conferences and also has participated as a speaker in scientific meetings financed by various pharmaceutical companies (Schering-Plough, AstraZeneca, and GlaxoSmithKline), and has received funding for research grants from AstraZeneca ($89,000), Schering-Plough ($80,000), and GlaxoSmithKline ($159,000). P.E. has received speaker fees and has served on advisory boards for AstraZenca ($6,000/year), GlaxoSmithKline ($4,000/year), and Merck Frost ($4,000/year), and in 2002–2003 he rereceived a $50,000 research grant from GlaxoSmithKline. S.S. is the recipient of a Distinguished Investigator award from the Canadian Institutes of Health Research (CIHR). REFERENCES
From the Authors: Suissa and Ernst question whether in our study (1) of the association between inhaled corticosteroid use and nonvertebral fractures we included control subjects that were not "at risk" for exposure to inhaled corticosteroids at their index date and thus overestimate the odds ratio in patients that we classified as current users. When control subjects were individually matched to cases, they were required to be alive at the cases fracture date to be a control for that patient. However, we did not exclude control subjects that were hospitalized on the cases fracture date. Of the 6,817 control subjects included in the analysis, 68 (1.0%) were hospitalized on their index date. The number of days these patients were in the hospital before their index date ranged from 1 to 159 days. The median was 6 days and the interquartile range was 3 to 18 days; thus, most of the 68 control subjects were "at risk" for the majority of the 30-day period before their index date. However, 14 control subjects were hospitalized for the entire 30-day period before their index date and 1 was hospitalized for more than 90 days. These patients were classified as unexposed when, as noted by Suissa and Ernst, they were not at risk for exposure. Removing these patients from the analysis slightly decreased the estimated odds ratio in patients classified as current users of average daily doses greater than 700 µg from 1.68 (95% confidence interval [CI], 1.10–2.57) to 1.66 (95% CI, 1.09–2.54). Excluding these patients from the analysis does not change the interpretation of our findings: that is, current users of greater than 700 µg per day of inhaled corticosteroids had an increased risk for nonvertebral fractures.
Hines VA Hospital and Northwestern University Feinberg School of Medicine Chicago, Illinois Acknowledgments T.A.L. has received research grants from GlaxoSmithKline ($86,000), AstraZeneca ($63,000), a Boehringer Ingelheim–Pfizer partnership ($91,000), and Pfizer ($20,000). K.B.W. has received research grants from Pfizer, AstraZeneca, Boehringer Ingelheim, and GlaxoSmithKline, has participated as a member of the Merck Respiratory Advisory Board, and has consulted for AstraZeneca on several projects. REFERENCES
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
