Background Chronic obstructive pulmonary disease (COPD) is a commonly reported cause of death and associated with smoking. However, COPD mortality is high in poor countries with low smoking rates. Spirometric restriction predicts mortality better than airflow obstruction, suggesting that the prevalence of restriction could explain mortality rates attributed to COPD. We have studied associations between mortality from COPD and low lung function, and between both lung function and death rates and cigarette consumption and gross national income per capita (GNI).

Methods National COPD mortality rates were regressed against the prevalence of airflow obstruction and spirometric restriction in 22 Burden of Obstructive Lung Disease (BOLD) study sites and against GNI, and national smoking prevalence. The prevalence of airflow obstruction and spirometric restriction in the BOLD sites were regressed against GNI and mean pack years smoked.

Results National COPD mortality rates were more strongly associated with spirometric restriction in the BOLD sites (<60 years: men r_s=0.73, p=0.0001; women r_s=0.90, p<0.0001; 60+ years: men r_s=0.63, p=0.0022; women r_s=0.37, p=0.1) than obstruction (<60 years: men r_s=0.28, p=0.20; women r_s=0.17, p<0.46; 60+ years: men r_s=0.28, p=0.23; women r_s=0.22, p=0.33). Obstruction increased with mean pack years smoked, but COPD mortality fell with increased cigarette consumption and rose rapidly as GNI fell below US$15 000. Prevalence of restriction was not associated with smoking but also increased rapidly as GNI fell below US$15 000.

Conclusions Smoking remains the single most important cause of obstruction but a high prevalence of restriction associated with poverty could explain the high ‘COPD’ mortality in poor countries.