Recent trends in COPD prevalence in Spain: a repeated cross-sectional survey 1997–2007

Review of previous literature

Numerous indirect assessments concur that the population burden due to COPD, both worldwide and in Spain, are set to increase in the near future. The global estimates of mortality and morbidity for given diseases made in 1990 were recently updated and confirmed a significant upward trend for COPD 5. In Europe, the COPD mortality rates range from <25 to >75 per 100,000 inhabitants within the various European countries with data, and its prevalence ranges from <2% to >10%, also with an expected increase 6. Both PLATINO (Latin American Project for the Investigation of Obstructive Lung Diseases) and BOLD (Burden of Obstructive Lung Disease) results identified substantial variability of COPD prevalence within centres. This evidence, together with the instability of results by time reported in here, points to the difficulties of applying current spirometric definitions of COPD 23. Both the PLATINO and BOLD surveys have identified a significant burden and undetected COPD in most areas, but their cross-sectional nature prevents any conclusion regarding temporal changes regionally or locally 12, 13. In asthma, repeated cross-sectional studies conducted in the same areas by the same authors, and using an identical/similar protocol, have reported the transition to the current, expanding asthma epidemic 24–27. Such evidence is scarce in adults from the population with post-bronchodilator forced spirometry and COPD, with repeated surveys only available from Finland 28 and Sweden 29. Both countries identified increases in the population burden of COPD 30 yrs after initiation, but did not use post-bronchodilator spirometry, a factor that can modify any final COPD assessment 30.

Limitations

Apart from the advantages of similar researchers and areas, and closely similar protocols whose differences were taken into account in our analysis, there are some limitations of our research that deserve further discussion. As mentioned above, protocols were not exactly the same and COPD guidelines and forced spirometry recommendations have changed during this relatively short period. We have ensured that re-calculation of individual data from both surveys was conducted consistently, and that similar definitions, thresholds and exclusions were applied. However, the subtle effect of differences in the sampling frame and recruitment, and slight technical changes in spirometry recommendations and/or tools cannot be ruled out to explain, totally or partially, our findings. Without an intention to be cumbersome, a long list of methodological issues are presented as follows. Spirometers used in both surveys differed (table 1). In IBERPOC it was a turbine spirometer, while in EPI-SCAN a pneumotacograph spirometer with high sensitivity was used. It has been reported that turbine spirometers create greater internal resistance to flow, sometimes failing the standards recommended by the American Thoracic Society at high flows 31. The manufacturer of the pneumotacograph spirometer Datospir-200 reports a maximum resistance of 0.12 kPa·L⁻¹·s⁻¹ and the turbine MasterScope 0.05 kPa·L⁻¹·s⁻¹. Buess et al. 32 suggested that the relationship between the internal resistance of a spirometer and the resistance of the respiratory system should be between 5–10%. For a range of resistance of the respiratory system of 0.5–1 kPa·L⁻¹·s⁻¹ and according to data provided by manufacturers, the pneumotacograph MasterScope meets that criterion, while the turbine spirometer would exceed it by 12–24%. An increase in internal resistance of pneumotacographs can produce an underestimation of expiratory volumes, detecting more COPD, which could have occurred in the IBERPOC study. As this effect occurs early in the expiratory manoeuvre, since most resistance occurs at high flows, it appears that the measure should affect FEV₁ more than the FVC, the latter being more dependent on the completion of the manoeuvre. Consequently, an increase of internal resistance of the spirometer could result in an underestimation of the FEV₁/FVC ratio. To our knowledge, there is no data in the literature comparing the Datospir 200 spirometer versus pneumotacographs. Accordingly, we re-analysed data considering an underestimation by 6% in the FEV₁/FVC due to turbine spirometers, producing a prevalence of COPD according to the old ERS criteria in EPI-SCAN of 10.6% (95% CI 9.6–11.5%). In the subgroup of individuals aged 40–69 yrs, the estimated prevalence of COPD would then be 9.2% (95% CI 8.2–10.2%), virtually identical to the 9.1% reported in the IBERPOC study.

A final limitation to consider is the often arbitrary decision when applying reference equations to estimate predicted lung function. We used those defined by Roca et al. 18, as they are preferred to other available equations for being locally produced and therefore more representative, as recommended elsewhere 21. However, they were obtained in the mid 1980s. As the Spanish population has grown taller and leaner 21 yrs later, their ongoing validity might be debatable, as their application produced an observed mean FEV₁ of 87.8% pred in 1997 and 100.8% in 2007 (table 1). Therefore, additional caution in the interpretation of changes in results of any spirometric survey should be granted when repeated, as changing lung reference equations might have a major effect. Only a re-analysis of raw data, as applied here, would identify this problem.

The use of lower limit of normal (LLN) to define and stage COPD has been postulated as more advantageous than previous and current recommendations, all based on fixed spirometry ratios 33, which may misdiagnose large segments of the very young and the very old 34, 35. In a way, the old ERS definition of COPD used in this study, with its variable post-bronchodilator ratio FEV₁/FVC percentage lower than a predicted value by sex, is already a type of LLN. However, to compare LLN to current thresholds was not the aim of our study, and we doubt our data will help to settle this controversy.

Should the findings be true, some factors can be considered that might explain, albeit partly, this unexpected finding of a COPD prevalence decrease. Changes in tobacco consumption in Spain have been well documented. The prevalence of smokers in adults in Spain significantly dropped in males from 39.9% in 1999 to 32.1% in 2007, while only from 24.6% to 22.1% in females in the same period 36. The actual point prevalence of smoking at the time of conducting the fieldwork of each study, IBERPOC in 1999 and EPI-SCAN in 2007, should not explain the COPD burden, but the cumulative exposure in this population, so it is hard to understand our finding of a 50.4% decrease in COPD prevalence due to changes in smoking. Although the comparison in pack-years of smoking exposure has to be made with care (table 2), the distribution in smoking exposure among individuals with and without airflow obstruction was 47.2±28.7 versus 26.7±19.4 pack-yrs in IBERPOC in 1997 37, a finding that was largely maintained in EPI-SCAN in 2007: 36.5±18.6 versus 25.0±17.2 pack-yrs (data not shown).

Of interest, the prevalences of respiratory symptoms in both periods (table 2) were similar for cough or sputum, and they are only significant for minor decreases in dyspnoea and wheezing; therefore, it is unlikely they help to explain the magnitude of the differences in spirometric values.

The birth cohort of IBERPOC participants in 1996–1997 suffered the consequences of the Spanish Civil War, from 1936 to 1939. During the 1940s, living conditions in Spain were extremely hard, including starvation and virtually universal infections. As the average year of birth of IBERPOC participants was 1942, the Barker hypothesis on the influence of pre-conception and early childhood factors on attaining full lung development 38, with a population shift on weight and other factors to influence later spirometry findings 39, and even the effect of tuberculosis 40, common in Spain at that time, might explain a significant unhealthy “cohort effect” in the IBERPOC participants. Conditions gradually improved in subsequent years, so EPI-SCAN participants (1952 being their average year of birth) were actually taller, leaner, and better educated (table 2). A final, relevant factor might be the existence of an outlier centre in the IBERPOC 1997 study; Manlleu participants had a COPD prevalence of 18.0% (95% CI 14.8–21.2), which was four-fold higher than the lowest participating area, mostly explained at that time by high occupational exposures in actually nonsmoking females, who had the mildest COPD 41. 10 yrs later in Vic, another rural village in the outskirts of Barcelona, only 10 km (∼6 miles) from Manlleu, but without any major local industry, we obtained a 4% COPD prevalence.

Finally, of public health interest, we have to underline from table 4 the nonsignificant drop in COPD underdiagnosis, but the substantial decrease in undertreatment, especially in those with severe COPD. To further reduce underdiagnosis, the implementation and wider use of spirometry screening in all settings, including quality spirometry in primary care 42, pharmacies 43, and elsewhere, require further research and resources 44.

Conclusion

To conclude, we report a substantial reduction of 50.4% in the prevalence of COPD in Spain from 1997 to 2007 in subjects aged 40–69 yrs, with also an unexpected shift to a milder severity in the population distribution. These findings remain unexplained, but highlight the difficulty found in comparing population findings of forced spirometry by time and place. Undertreatment (but not underdiagnosis) was significantly reduced during the same period.