© 2007 American Thoracic Society
Avoiding Bias in the Annualized Rate of Change of FEV1To the Editor:Dr. Mannino and colleagues in their recent article reported that the most rapid decline in lung function in the Atherosclerosis Risk in Communities (ARIC) study was associated with increased risk of death (1). The rate of change was defined as the percentage change in FEV1 at 3 years from baseline in participants aged 44 to 66 years at baseline. Overall, participants with the most rapid declining lung function had a modestly increased risk of death. However, the method of defining rapid declining lung function introduces a bias that puts to question some of the conclusions. This can be demonstrated as follows. For healthy males, standing height 178 cm, predicted FEV1 was computed at an average age for the age groups quoted in Mannino and coworkers' article (Table 1) (2). In Mannino's approach, FEV1 should decline proportionately; thus, a 0.75% annualized change from baseline at age 47 years comes to 30 ml, approximating cross-sectional findings in their healthy reference populations. In the oldest person, a 0.75% annualized change is only 26.4 ml. Assuming a uniform rate of change in perfectly healthy subjects of 30 ml/year (last column of Table 1), expressing change as a percentage of the initial value inflates the observed percentage rates of decline in older subjects. This is exacerbated by the fact that longitudinal studies (3, 4) (and references therein) have revealed an age-related accelerating decline in FEV1.
This implies an age-related bias in the annualized rate of change of FEV1 (i.e., the number of elderly persons will be artificially increased in the group of rapid decliners); using age as a covariate does not resolve this bias. If the standard deviation of repeated measurements in the same person does not decline across age categories, this will exacerbate the problem because expressing change as a function of initial value is then influenced by regression to the mean (5). One wonders whether an approach like that of Burrows and colleagues (3), standardizing FEV1 for height cubed, or that of Chinn and coworkers (6), standardizing FEV1 for height squared, would not be more appropriate. An advantage of these approaches is that an apparently accelerated decline in FEV1 due to the decline in standing height with age is thus taken into account, removing another source of age (and height)-related bias. Even so, a 3-year interval is too short to accurately estimate longitudinal decline (3).
Falling Ratio Working Group FOOTNOTES Conflict of Interest Statement: P.H.Q. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. J.P.S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. M.R.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. G.R. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. REFERENCES
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
