The association between spirometry measurement quality, cognitive function, and mortality.

BACKGROUND

Population studies that assess lung function usually exclude results of individuals with poor-quality measurements, which often means excluding many subjects. Impaired cognition is frequently associated with poor-quality spirometry; excluding such subjects may introduce a selection bias in studies with lung function as either outcome or exposure. We investigated the association between poor-quality spirometry and impaired cognitive function and whether poor-quality spirometry is associated with future mortality risk independently of cognitive function.

METHODS

We used data from a prospective cohort in three Central and Eastern European countries; 12,087 individuals aged 45-75 years (54% females) with complete information on variables of interest were included. Standard memory, verbal fluency, and executive cognitive domain tests were converted into latent variable z-scores and divided into quartiles. Spirometry tests were classified into two categories based on repeatability criteria: good- (71%) vs. poor-quality spirometry (29% of participants). Those with good-quality spirometry were further classified, using forced vital capacity (FVC) and forced expiratory volume in the first second (FEV), as healthy spirometry (63%) or impaired spirometry (8%). Multinomial logistic regression was used to assess the association between poor-quality spirometry and cognitive function, and a Cox proportional regression was used to analyze the risk of total mortality over a 17-year follow-up.

RESULTS

After controlling for a range of covariates, higher cognitive function predicted lower odds of poor-quality spirometry. In the highest cognitive function quartile, compared with the lowest quartile, the odds ratio of poor-quality spirometry was 0.82 (95%CI: 0.72-0.92). Impaired spirometry was associated with higher mortality risk even after adjusting for cognition (adjusted hazard ratio 1.63, 95%CI: 1.45-1.84), but mortality risk was similar in subjects with poor- vs. good-quality (HR 1.02, 95%CI: 0.93-1.10).

CONCLUSION

Higher cognitive function was associated with a lower risk of poor-quality spirometry. The lack of independent association of poor-quality spirometry with mortality suggests that excluding poor-quality spirometry measurements from analyses is unlikely to introduce a major bias. However, discarding poor-quality spirometry from epidemiological analyses might imply the exclusion of vulnerable subjects. These findings should be confirmed in future studies representing other populations.