Pattern of variables describing desaturator COPD patients, as revealed by cluster analysis.

Abstract

STUDY OBJECTIVES: The aims of this study were to define, by cluster analysis, a pattern of clinical variables that differentiate desaturator (D) from nondesaturator (ND) patients affected by COPD, and to identify daytime variables that are predictive of nocturnal desaturation. PATIENTS: Fifty-one random, consecutive COPD outpatients (20 women; mean [+/- SD] age, 69.6 +/- 4.0 years) with mild daytime hypoxemia (Pao(2), 60 to 70 mm Hg) were enrolled into the study. Obstructive sleep apnea syndrome patients were excluded. MEASUREMENTS AND RESULTS: Lung volumes, arterial blood gas levels, and mean pulmonary artery pressure (MPAP) were measured, and nocturnal desaturation was evaluated with nighttime polygraphy. With least squares simple linear regression, the percentage of total recording time was highly correlated with a total nocturnal recording time of arterial oxygen saturation of < 90 mm Hg (T90) and MPAP (R = 0.84; R2= 71.20%); T90 was also highly correlated with daytime Paco2 (R = 0.70; R2= 48.96%). Multiple regression showed that T90 was highly correlated with both MPAP and Paco2 (R2= 97.75%). Hierarchical cluster analysis conducted with these three variables showed that D and ND patients differed in both nocturnal and daytime variables. The mean T90 was 30 +/- 3.5% in 19.2% and 8%, respectively, of the D and ND groups. Moreover, two D subgroups differing in MPAP and two ND subgroups differing in Paco2 were identified. CONCLUSIONS: D patients may be identified by a pattern of T90, MPAP, and Paco2 values, rather than by T90 alone, with the latter two variables being predictors of nocturnal desaturation severity.