Background Debate continues about the predictive value of different spirometry based definitions of chronic obstructive pulmonary disease (COPD). We explored the hazard of death for three overlapping definitions, and any significant interactions with smoking.
Methods Data from the Health Survey for England (years 1995–1997, 2001) and Scottish Health Survey (1995, 1998, and 2003) were used. 30001 adults aged 45+ were interviewed, had spirometry data, and linked mortality records (HSE extracted in 2011; SHeS in 2008). Cox regression was used to estimate the hazard ratios of those with COPD/airflow obstruction, compared with those without, adjusted for age, socio-economic position, smoking, drinking, body mass index (BMI), and longstanding-illness. Sex-specific models, for all-cause and respiratory mortality, used three definitions: fixed threshold (FT): FEV1/FVC < 0.70 and FEV1 < 80% predicted; lower limit of normal (LLN): FEV1/FVC1; self-reported COPD as a longstanding-illness (COPD).
Results For men meeting FT criteria the hazard ratio (HR) for all-cause mortality was 1.49 (95% CI 1.37, 1.62), compared with those not meeting the criteria. Despite capturing a smaller group, the HR for LLN was not significantly different: 1.60 (1.44, 1.78), nor was the HR for those reporting COPD 1.73 (1.45, 2.05). Similar, overlapping ratios were found for women: FT 1.52 (1.37, 1.68); LLN 1.62 (1.42, 1.85) and COPD 1.76 (1.42, 2.18). The relationship between COPD/obstructed airflow and mortality was stronger for respiratory mortality, with greater differentiation between measures. Among men, FT HR 2.69 (2.16, 3.36) just overlapped with LLN 4.20 (3.30, 5.35), but was lower than reported COPD 5.19 (3.79, 7.10). Among women, the figures were: FT 3.46 (2.68, 4.47), LLN 4.24 (3.17, 5.66), COPD 6.36 (4.31, 9.37). Smoking was a significant risk factor for mortality in all models, independently of airway obstruction, but results showed a significant interaction between cigarette pack-years and COPD. Among adults without COPD, the HR gradually increased to 1.7 (all-cause mortality) for those with 50+ pack-years, relative to never smokers and around 2 for respiratory deaths. Among adults reporting COPD, the HR for pack-years ≥1 was around 3 for all-cause and >6 for respiratory deaths.
Conclusion Participants mentioning COPD as a longstanding illness had the highest respiratory mortality. The HR for those meeting LLN criteria was higher, but not significantly so, than using FT. Any history of smoking increased mortality dramatically among those reporting COPD, compared with a more graded increase of risk by number of pack-years for those not reporting COPD.