Association between lung function decline and obstructive sleep apnoea: the ALEC study

Purpose To study changes in lung function among individuals with a risk of obstructive sleep apnoea (OSA), and if asthma affected this relationship. Methods We used data from the European Community Respiratory Health Survey II and III, a multicentre general population study. Participants answered questionnaires and performed spirometry at baseline and 10-year follow-up (n = 4,329 attended both visits). Subjects with high risk for OSA were identified from the multivariable apnoea prediction (MAP) index, calculated from BMI, age, gender, and OSA symptoms at follow-up. Asthma was defined as having doctor’s diagnosed asthma at follow-up. Primary outcomes were changes in forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC) from baseline to follow-up. Results Among 5108 participants at follow-up, 991 (19%) had a high risk of OSA based on the MAP index. Participants with high OSA risk more often had wheeze, cough, chest tightness, and breathlessness at follow-up than those with low OSA risk. Lung function declined more rapidly in subjects with high OSA risk (low vs high OSA risk [mean ± SD]: FEV1 = − 41.3 ± 24.3 ml/year vs − 50.8 ± 30.1 ml/year; FVC = − 30.5 ± 31.2 ml/year vs − 45.2 ± 36.3 ml/year). Lung function decline was primarily associated with higher BMI and OSA symptoms. OSA symptoms had a stronger association with lung function decline among asthmatics, compared to non-asthmatics. Conclusion In the general population, a high probability of obstructive sleep apnoea was related to faster lung function decline in the previous decade. This was driven by a higher BMI and more OSA symptoms among these subjects. The association between OSA symptoms and lung function decline was stronger among asthmatics. Electronic supplementary material The online version of this article (10.1007/s11325-020-02086-1) contains supplementary material, which is available to authorized users.


Sleep study -Icelandic subcohort
Among the 455 Icelandic participants in ECRSH III, altogether 416 underwent a whole night sleep study, with a T3 device (Nox Medical, Reykjavik, Iceland). A detailed description of the measurements has been published. 2 In short, two trained sleep technologists scored the sleep studies. Studies had to have ≥4 h of scorable oxygen saturation and more than two out of three respiratory traces: cannula flow, thorax and respiratory inductive plethysmography belts. Hypopneas were classified as a ≥30% drop in respiratory flow for ≥10 s with ≥4% oxygen desaturation.
Complete sleep study data was available on 401 participants. Thereof, 334 participants had data for MAP index calculation. Additionally, 9 participants with data for MAP index (5 with a MAP index > 0.5) had ongoing treatment for obstructive sleep apnea, and did therefore not undergo a sleep study. We included these participants in the group of participants with AHI > 15.
According to the Icelandic general population data, the sensitivity of the MAP index is 34/59=58%, and the specificity is 239/284=84% (figure A1). Figure A1. Venn diagram of the relationship between a positive MAP index and AHI (including those with previously diagnosed OSA as "AHI > 15").

AHI, ODI, OSA symptom index, and change in spirometry
Icelandic participants with sleep study in ECRSH III and spirometry at baseline and follow-up were 295. Thereof, 277 had answered the OSA symptom index questions.

OSA symptom index and change in lung function
The obstructive sleep apnea (OSA) symptom index had a linear relationship to changes in FEV1 and FVC (figures A2 and A3) (n=3419). Figure A2. Change in FEV1 between study visits, by obstructive sleep apnea symptom index. Adjusted for age, gender, BMI, smoking, centre, and baseline FEV1 value. Figure A3. Change in FVC between study visits, by obstructive sleep apnea symptom index. Adjusted for age, gender, BMI, smoking, centre, and baseline FVC value.
As changes in BMI had a significant effect on changes in lung function, we redid the analysis by only analyzing those with a change in BMI of more or less than 2 kg/m 2 ( figure A5, n=1940). The association of high OSA risk to difference in percent predicted FEV1 became non-significant, while the association to difference in percent predicted FVC remained significant (Table A3).
The association of the OSA symptom index to changes in FEV1 and FVC became non-significant among those without asthma, but remained significant among those with asthma ( Figure A4), and a significant interaction was found between the OSA symptom score and asthma. Table A3. Lung function decline between ECRHS II and III by OSA risk (MAP index >0.5), among participants with change in BMI ± 2 kg/m 2 between visits.

Spirometry data (mean ± SD)
Low OSA risk ( Figure A4. Multivariate mixed regression model on change in percent predicted of FEV1 and FVC between visits, only among participants with a change in BMI of less than ± 2 kg/m 2 . The graph shows regression coefficients with 50 and 95% confidence intervals. Adjusted for pack-years, baseline spirometry value, and centre. Subgraphs by doctor's diagnosed asthma. Note that variables differ in scale (categorical: gender; ordinal: OSA symptom index; continuous: BMI, age), affecting the interpretation of each coefficient. OSA: Obstructive sleep apnea; f-u: follow-up; % pred: percent predicted.

Current asthma -sensitivity analysis
Analysis by "current asthma" (compared to "doctor's diagnosed asthma" as in the main article) did not change the pattern of the results on changes in lung function in relation to the MAP factors, albeit some indices became non-significant (Figure 2 in main article and Figure A5). Those with doctor's diagnosed asthma but without current asthma (n = 429) were excluded from the analysis. Figure A5. Multivariate mixed regression model on change in percent predicted of FEV1 and FVC between visits, showing coefficients with 50 and 95% confidence intervals. Adjusted for pack-years, baseline spirometry value, and centre. Subgraphs by current asthma. Note that variables differ in scale (categorical: gender; ordinal: OSA symptom index; continuous: BMI, age), affecting the interpretation of each coefficient. OSA: Obstructive sleep apnea; f-u: follow-up; % pred: percent predicted.