The Science Journal of the American Association for Respiratory Care
Recently, it has been reported that DLCO is a predictor of all-cause mortality within the general population independent of FEV1 and even in absence of apparent clinical respiratory disease (Am J Epidem 1998; 147:1001-18). We examined 126 (54M/72W) subjects from Barcelona study (ARRD 1990;141:1026-32) explored twice 10 years apart, to investigate: a)the longitudinal decay of DLCO during the 10-yr follow-up period, expressed either in absolute or percent change, was strongly dependent on the initial DLCO value (DLCOinitial). The latter explained up to 80% of the variance of the final DLCO (DLCOfinal). Different covariates such as changes throughout the study in FVC, FEV1, height, BMI and time of follow-up explained all together less than 2.5% of the variance of DLCOfinal. A linear equation model did not show homoscedasticity. In contrast a logarithmic transformation of DLCO presented: a) slightly improved predictability of DLCOfinal; b) homoscedascidity; and, c) the constant term of the equation was negligible. The formula proposed to predict the longitudinal decay of DLCO is:
DLCOfinal= (DLCOinitial)0.977. High reliability of DLCO measurements of DLCO was demonstrated by means of structural equation models with latent variables.
(Supported, in part, by FIS 98/0052-01 SGR-0086 and SEPAR 96)