Introduction COPD is a disorder characterised by high morbidity and mortality. Although several parameters have been used to predict survival among COPD patients, most of the information on the prognostic value of pulmonary function comes from studies, either conducted in selected COPD populations or where only simple spirometry was measured. Few studies have comprehensively assessed lung function parameters and investigated their impact on survival; a prior smaller study from our group suggested carbon monoxide gas transfer may have prognostic value.1
Objective The study aimed to identify potential predictors of survival in a cohort of stable COPD outpatients.
Methods Data from patients, who had their first full lung function tests including blood gas analysis between February 1996 and May 2010 were extracted from the hospital's clinical COPD database. Patients with major co morbidities, such as malignancy, chronic renal failure and chronic heart failure were excluded. Survival data were available for all patients, until May 2011. Demographic data, PaO2 and PaCO2, transfer factor, and plethysmographic lung volumes were initially entered in a univariate regression model. Age, Body Mass Index (BMI), FEV1% predicted, FEV1/FVC, TLC% predicted, TLCOc% predicted, KCOc% predicted, RV% predicted, IC/TLC, PaCO2 and PaO2, were found to be univariately associated with survival and then entered in a stepwise Cox regression analysis model. Corresponding HRs and 95% CI were calculated for each independent predictor.
Results Data were available for 641 patients (62.2% male); mean age 61.9±10.2 years, FEV1 38.4±19.7% and BMI 24.3±5.3 kg/m2. Median survival was 92.9 months. Survival rates at 3 and 5 years (all cause mortality) were 0.88 and 0.62. In the total population, age (HR 1.05, 95% CI 1.03 to 1.07), PaO2 (HR 0.843, 95% CI 0.76 to 0.934) and TLCOc% (HR 0.975, 95%CI to 0.965 to 0.986) independently predicted survival. Abstract P220 Figure 1 presents the Kaplan–Meier survival curves, adjusted for age and PO2, for the two population groups, separated using the TLCOc% median value as a cut-off point (>38.0 and =38.0% predicted).
Conclusions Gas transfer measurement provides additional prognostic information compared to spirometry.