OUP user menu

Chronic obstructive pulmonary disease and sudden cardiac death: the Rotterdam study

Lies Lahousse, Maartje N. Niemeijer, Marten E. van den Berg, Peter R. Rijnbeek, Guy F. Joos, Albert Hofman, Oscar H. Franco, Jaap W. Deckers, Mark Eijgelsheim, Bruno H. Stricker, Guy G. Brusselle
DOI: http://dx.doi.org/10.1093/eurheartj/ehv121 ehv121 First published online: 28 April 2015

Abstract

Aims Both sudden cardiac death (SCD) and chronic obstructive pulmonary disease (COPD) are common conditions in the elderly. Previous studies have identified an association between COPD and cardiovascular disease, and with SCD in specific patient groups. Our aim was to investigate whether there is an association between COPD and SCD in the general population.

Methods and results The Rotterdam study is a population-based cohort study among 14 926 subjects aged 45 years and older with up to 24 years of follow-up. Analyses were performed with a (time dependent) Cox proportional hazard model adjusted for age, sex, and smoking. Of the 13 471 persons included in the analysis; 1615 had a diagnosis of COPD and there were 551 cases of SCD. Chronic obstructive pulmonary disease was associated with an increased risk of SCD (age- and sex-adjusted hazard ratio, HR, 1.34, 95% CI 1.06–1.70). The risk particularly increased in the period 2000 days (5.48 years) after the diagnosis of COPD (age- and sex-adjusted HR 2.12, 95% CI 1.60–2.82) and increased further to a more than three-fold higher risk in COPD subjects with frequent exacerbations during this period (age- and sex-adjusted HR 3.58, 95% CI 2.35–5.44). Analyses restricted to persons without prevalent myocardial infarction or heart failure yielded similar results.

Conclusion Chronic obstructive pulmonary disease is associated with an increased risk for SCD. The risk especially increases in persons with frequent exacerbations 5 years after the diagnosis of COPD. This risk indicator could provide new directions for better-targeted actions to prevent SCD.

  • Comorbidity
  • Cardiovascular mortality
  • General population
  • Sudden cardiac death

Translational perspective

Sudden cardiac death (SCD) is a major health problem; however, risk stratification remains difficult and probably not all risk indicators have been identified. Chronic obstructive pulmonary disease (COPD) has been associated with an increased risk of cardiovascular disease and with SCD in specific high-risk patient populations. This study shows that COPD is a risk indicator for SCD in the general population and that the risk increases with COPD severity. This provides directions for further measures to prevent SCD.

Introduction

Sudden cardiac death (SCD) forms a substantial part of cardiovascular mortality, which is the leading cause of death globally,1 with a currently estimated incidence of 4–5 million cases worldwide per year.2 Sudden cardiac death is primarily caused by ventricular arrhythmias3 and risk factors for SCD include male sex, increasing age,2 a history of heart failure or cardiac ischaemia, cardiomyopathies, channelopathies, and alcohol and drug use.4 However, SCD is a heterogeneous outcome with multiple aetiological pathways of which many have not been established. This complicates an adequate risk assessment and prevention plan for SCD.

Chronic obstructive pulmonary disease (COPD) is the third leading cause of death.5 Chronic obstructive pulmonary disease is characterized by a progressive airflow limitation and is associated with a chronic inflammatory response in the airways and lungs.6 Besides these local pulmonary manifestations, systemic effects and comorbidities are frequently present and determine the prognosis of patients with COPD.7,8 Patients with COPD have a two- to three-fold higher risk of developing cardiovascular disease and half of COPD deaths can be attributed to cardiovascular disease.9 Although the underlying mechanisms for this high cardiovascular morbidity and mortality in patients with COPD are not fully established to date, a role for the accompanied hypoxia and hypoxaemia,6 higher heart rate,10 systemic inflammation,6 or recurrent exacerbations has been suggested. Moreover, patients with COPD suffer more frequently from ventricular arrhythmias than controls.11,12 In patients hospitalized for an exacerbation of COPD, atrial fibrillation and ventricular arrhythmias are independent predictors of death at 1 year.13 Furthermore, COPD has also been shown to be an independent predictor of SCD in high-risk patients after a percutaneous coronary intervention and coronary artery bypass graft13,14 and in patients with myocardial infarction.15 However, this association has not yet been assessed in the general population. Therefore, our objective was to assess whether COPD is an independent risk factor for SCD in the general population, and whether COPD, exacerbations, and systemic inflammation modify the association between COPD and SCD.

Methods

Setting

The Rotterdam study is a prospective population-based cohort study in 14 926 people aged ≥45 years, which started in 1990.16 In addition to regular follow-up examinations at the research facility, participants are continuously monitored for morbidity and mortality through general practitioners and municipality records. The Rotterdam study was approved by the medical ethics committee and all participants gave written informed consent. This study complies with the Declaration of Helsinki.

Chronic obstructive pulmonary disease

Chronic obstructive pulmonary disease was diagnosed by an obstructive spirometry (FEV1/FVC < 70%) performed at the research centre visit or, if not available, by a physician based on the combination of clinical history, physical examination, and spirometry.17 The severity of COPD was determined by the forced expiratory volume in 1 s (FEV1), expressed as percentage predicted, according to the modified global initiative for chronic obstructive lung disease (GOLD) criteria (2007): mild COPD (GOLD I, FEV1 ≥ 80%), moderate COPD (GOLD II, 50% ≤ FEV1<80%), and severe COPD (GOLD III, FEV1 < 50%).18 Asthma patients were excluded. The date of incident COPD was defined as the date of obstructive lung function measurement, the date of COPD diagnosis in the medical records or the date of a first COPD medication prescription in someone with established COPD, whichever came first. Medication use was obtained through pharmacy-filled prescription data from seven pharmacies in the region which are all on one computer network. Moderate exacerbations were defined as needing a course of steroids and/or antibiotics and severe exacerbations as needing hospitalization. Frequent exacerbators were defined as COPD participants who had during follow-up on average two or more moderate or severe exacerbations a year.19

Sudden cardiac death

Sudden cardiac death cases were validated independently by two researchers and confirmed by a cardiologist through details from the medical records. In witnessed deaths, SCD was diagnosed according to Myerburg's definition endorsed by the European Society of Cardiology: ‘a natural death due to cardiac causes, heralded by abrupt loss of consciousness within one hour from onset of acute symptoms; preexisting heart disease may have been known to be present, but the time and mode of death are unexpected’.20,21 Unwitnessed deaths were coded as SCD if death was unexpected in persons found death, while they were in a stable medical condition 24 h before they were found and without evidence of a non-cardiac cause.22

Statistical analysis

Differences between subjects with and without COPD were studied using Mann–Whitney U and χ2 tests. In SPSS (IBM Corp., Somers, NY, USA), time-dependent cox proportional hazard models were performed to assess the association between COPD and SCD. Follow-up time was determined as the time between study entry or incident COPD date and death or end of study (1 January 2011). We also performed a sensitivity analysis adding the time between cohort entry and incident COPD to the non-COPD follow-up time. All models were adjusted for age and sex, and additionally for all biologically plausible covariates according to the previous literature which changed the risk estimate by >10%. Potential confounders were age (years), sex, height (cm), weight (kg), body mass index (kg/m²), smoking behaviour, pack-years, myocardial infarction, heart failure, coronary revascularisation procedures, systolic blood pressure (mmHg), diastolic blood pressure (mmHg), hypertension, total serum cholesterol (mmol/L), diabetes mellitus, atrial fibrillation, and heart rate corrected QT interval (ms) according to Bazett's formula. As high-sensitivity C-reactive protein (hsCRP) might be an important effect modifier, we investigated this variable in a separate analysis. High-sensitivity C-reactive protein serum levels (mg/L) were categorized as high vs. moderate/low, based on the American Heart Association classification.23 We constructed cumulative survival curves using the Kaplan–Meier method. With regard to the competing risk of non-SCD, we performed a sensitivity analysis using the data augmentation method and unstratified model described by Lunn & McNeil to estimate the hazard ratio (HR) of SCD in comparison with non-SCD.24 More details on the methods including the performed sensitivity analyses are provided in Supplementary material online.

Results

General characteristics

Table 1 shows the baseline characteristics of the study population (n = 13 471) with a median age of 64 years (interquartile range, IQR = 16) at the start of follow-up. Chronic obstructive pulmonary disease subjects were older, more often male, more (current) smokers, had a slightly lower body mass index, higher levels of hsCRP at baseline and had more frequently a medical history of myocardial infarction and coronary artery bypass graft. During a median follow-up of 3229 days (8.84 years; IQR = 3651 days; total of 123 024 person years of follow-up), 5197 (39%) of the participants died of whom 551 died suddenly. Eighty-two (5%) COPD subjects and 469 (4%) subjects without COPD died due to SCD (Figure 1). The SCDs were witnessed in about half of the cases (54%) and the ratio of witnessed vs. unwitnessed SCD was not different for COPD subjects vs. subjects without COPD (P = 0.975). The time of SCD was divided into three equal parts of 8 h and registered in 380 (69%) of 551 SCDs. The data show that SCD subjects with COPD died more frequently during night time (0–8 h; 53%) than SCD subjects without COPD (36%; P = 0.023) (Supplementary material online, Figure S1).

View this table:
Table 1

Baseline characteristics of the population at cohort entry (n = 13 471)

COPD (n = 1615)No COPD (n = 11 856)P-value
Age (years)70 (13)63 (15)<0.001
Males (%)909 (56)4727 (40)<0.001
Smoking status
 Never smoker (%)251 (16)4361 (38)<0.001
 Former smoker (%)689 (44)4903 (43)
 Current smoker (%)632 (40)2209 (19)
Pack-years cigarette smoking26 (37)3 (23)<0.001
Height (cm)170 (13)167 (14)<0.001
Weight (kg)74.9 (17.0)74.6 (17.8)0.773
BMI (kg/m²)25.8 (4.7)26.4 (4.9)<0.001
hsCRP (mg/L)1.9 (3.0)1.5 (2.5)<0.001
Myocardial infarction (%)118 (7)581 (5)<0.001
Heart failure (%)52 (3)235 (2)0.002
Coronary revascularization57 (4)334 (3)0.141
 CABGa (%)41 (3)212 (2)0.048
 PTCAa (%)19 (1)155 (1)0.608
Systolic blood pressure (mmHg)137 (29)137 (29)0.399
Diastolic blood pressure (mmHg)75 (16)77 (16)<0.001
Total serum cholesterol (mmol/L)6.2 (1.6)6.1 (1.6)0.036
Hypertension (%)741 (55)5344 (56)0.499
Atrial fibrillation (%)71 (5)438 (4)0.268
QTc interval (ms)430 (31)429 (30)0.618
Diabetes mellitus (%)154 (10)1196 (10)0.488
Potassium (mmol/L)4.2 (0.4)4.2 (0.4)0.948
GFR (sex specific ml/min/1.73 m²)74.8 (19.5)73.6 (19.2)<0.001
Diuretic use511 (31.6%)1966 (16.6%)<0.001
  • Categorical variables are expressed as numbers (percentage). Values of continuous variables are expressed as median (IQR).There were missings in baseline smoking status (n = 426), pack-years (n = 728), height (n = 1447), weight (n = 1438), BMI (n = 1455), hsCRP (n = 1922), myocardial infarction (n = 161), heart failure (n = 297), coronary revascularization (n = 583), systolic/ diastolic blood pressure (n = 1317), cholesterol (n = 1439), hypertension (n = 2547), atrial fibrillation (n = 1876), QTc interval (n = 1659), potassium (n = 3282), and GFR (n = 3325).

  • BMI, body mass index; CABG, coronary artery bypass graft; COPD, chronic obstructive pulmonary disease; GFR, glomerular filtration rate; hsCRP, high-sensitivity C-Reactive Protein; QTc, heart rate corrected QT interval according to Bazett's formula; PTCA, percutaneous transluminal coronary angioplasty.

  • aNumbers overlap.

Figure 1

Flow chart of the study population.

Association between chronic obstructive pulmonary disease and sudden cardiac death

Figure 2 illustrates the higher probability of SCD for COPD subjects compared with subjects without COPD. Adjusted for age and sex, COPD was significantly associated with an increased risk of dying due to SCD (HR 1.34, 95% CI 1.06–1.70). A stratified analysis in patients without prevalent myocardial infarction nor heart failure nor CABG nor PCI resulted in a comparable age- and sex-adjusted HR of 1.38 (95% CI 1.03–1.85). An additional analysis only including subjects with a study acquired spirometry resulted in an age- and sex-adjusted HR of 3.32 (95% CI 1.77–6.22) of SCD for COPD subjects with an obstructive spirometry compared with subjects with normal spirometry. Supplementary material online, Table S1 demonstrates the risk of SCD according to the severity of airflow limitation. A sensitivity analysis using the stratified Lunn–McNeil method to model the competing risk effect of other causes of death, resulted in an age- and sex-adjusted HR of COPD on SCD of 1.43; 95% CI 1.19–1.67. The cumulative incidence function (CIF) in Supplementary material online represents the higher incidence of SCD and other causes of death for subjects with COPD compared with subjects without COPD (see Supplementary material online, Figure S2). The CIFs for COPD vs. no COPD were statistically significant for SCD (Gray's test; P = 0.0006) and for non-SCD (Gray's test; P < 0.0001). Another sensitivity analysis adding the time between cohort entry and incident COPD to the control follow-up time did not substantially change the estimate.

Figure 2

Kaplan–Meier curve of sudden cardiac death according to chronic obstructive pulmonary disease status (log-rank P < 0.001).

Figure 2 further illustrates that the distinction of SCD risk according to COPD status occurred ∼2000 days (5.48 years) after the study start (incident COPD date or cohort entry for participants without incident COPD), while the difference for other causes of death was seen much earlier (Supplementary material online, Figure S2). Since the proportional hazard assumption was violated, we evaluated the effect of COPD on SCD using a time-dependent Cox regression analysis stratifying time in more or <2000 days after COPD diagnosis. Within these strata, the proportional hazards assumption was satisfied and adjusted for age and sex, the HR for the period before 2000 days was non-significant (HR 0.66, 95% CI 0.43–1.02), whereas the HR for the period after 2000 days was significantly increased (HR 2.12, 95% CI 1.60–2.82; Table 2). We additionally adjusted the model for covariables which changed the risk estimate by >10%, i.e. only pack-years of cigarette smoking. Chronic obstructive pulmonary disease was associated with an almost two-fold increased risk to develop SCD, independent of age, sex, and pack-years of cigarette smoking (HR 1.93, 95%CI 1.44-2.59; Table 2). The other potential confounders (i.e. height, weight, body mass index, myocardial infarction, heart failure, coronary revascularisation procedure, blood pressure, hypertension, total cholesterol, diabetes mellitus, atrial fibrillation, and heart rate corrected QT interval) did not significantly influence the effect. We did not observe any significant interactions with sympathomimetic respiratory drugs or with cardiac treatment.

View this table:
Table 2

Chronic obstructive pulmonary disease and the hazard on sudden cardiac death for the period >2000 days (5.48 years) of follow-up

Categorical, all vs. no COPDModel 1Model 2
HR95% CIHR95% CI
COPD2.121.60–2.821.931.44–2.59
COPD without frequent exacerbations1.661.16–2.371.521.06–2.19
COPD with frequent exacerbations3.582.35–5.443.212.08–4.95
  • Model 1: age and sex adjusted.

  • Model 2: adjusted for age, sex, and pack-years of cigarette smoking. The other potential confounders (i.e. height, weight, body mass index, myocardial infarction, heart failure, coronary revascularisation procedure, blood pressure, hypertension, total cholesterol, diabetes mellitus, atrial fibrillation, and heart rate corrected QT interval) did not significantly influence the association.

  • Time-dependent Cox regression analysis for the period >2000 days of follow-up.

  • Frequent exacerbations were defined as having at least two moderate or severe exacerbations a year, averaged over the total years of follow-up.

  • CI, confidence interval; COPD, chronic obstructive pulmonary disease; HR, hazard ratio.

Influence of exacerbations

Regarding the influence of COPD exacerbations, 366 (23%) of 1615 COPD subjects had frequent exacerbations. Thirty-three (9%) frequent exacerbators compared with 49 (4%) COPD subjects without frequent exacerbations died due to SCD. The Kaplan–Meier curve for SCD in Figure 3 illustrates the poorer survival for COPD subjects with frequent exacerbations (log-rank P < 0.001). Adjusted for age, sex, and pack-years of cigarette smoking, the COPD frequent exacerbator phenotype was significantly associated with a more than three-fold increased risk of developing SCD (HR 3.21; 95% CI: 2.08–4.95; Table 2). A sensitivity analysis using the stratified Lunn–McNeil method to model the competing risk effect of other causes of death, resulted in an age- and sex-adjusted HR of the frequent exacerbators on SCD of 2.51 (95% CI: 2.15–2.86). Finally, the cumulative incidence of SCD was significantly higher when COPD subjects had frequent exacerbations (Supplementary material online, Figure S3).

Figure 3

Kaplan–Meier curve of sudden cardiac death according to chronic obstructive pulmonary disease status with or without frequent exacerbations (log-rank P < 0.001).

Influence of baseline systemic inflammation

Of the 13 471 subjects, 11 549 (85.7%) had a baseline hsCRP measurement. A linear regression model adjusted for age and sex, demonstrated that the natural logarithm of hsCRP concentration was significantly increased in subjects with COPD and SCD compared with subjects with COPD without SCD (P < 0.05). When stratified according to the level of baseline systemic inflammation, the HR for SCD was significantly increased in COPD subjects with frequent exacerbations having an hsCRP level >3 mg/L (HR 3.67; 95% CI 1.97–6.85; Table 3). In contrast, in all subjects with a low-to-moderate degree of systemic inflammation, the HR for SCD was significantly increased in COPD subjects without frequent exacerbations (HR 1.84; 95% CI 1.16–2.91; Table 3).

View this table:
Table 3

Chronic obstructive pulmonary disease and the hazard on sudden cardiac death for the period >2000 days (5.48 years) of follow-up, stratified according to the serum level of high-sensitivity C-reactive protein

Model 1Model 2
HR95% CIHR95% CI
hsCRP ≤ 3 mg/LCOPD1.981.32–2.981.761.15–2.68
COPD without frequent exacerbations2.091.34–3.261.841.16–2.91
COPD with frequent exacerbations1.600.65–3.901.480.60–3.64
hsCRP > 3 mg/LCOPD1.881.16–3.051.931.18–3.16
COPD without frequent exacerbations1.210.62–2.351.240.63–2.42
COPD with frequent exacerbations3.511.90–6.493.671.97–6.85
  • Model 1: age and sex adjusted.

  • Model 2: adjusted for age, sex, and pack-years of cigarette smoking. The other potential confounders at baseline (i.e. height, weight, body mass index, myocardial infarction, heart failure, coronary revascularisation procedure, blood pressure, hypertension, total cholesterol, diabetes mellitus, atrial fibrillation, and heart rate corrected QT interval) did not significantly influence the association.

  • Time-dependent Cox regression analysis for the period >2000 days of follow-up.

  • Frequent exacerbations were defined as having at least two moderate or severe exacerbations a year, averaged over the total years of follow-up.

  • CI, confidence interval; COPD, chronic obstructive pulmonary disease; hsCRP, high-sensitivity C-reactive protein; HR, hazard ratio.

Discussion

In this population-based study in community-dwelling middle-aged and elderly persons, we showed an ∼30% increased risk of SCD in subjects with COPD. The risk almost doubled in the period 2000 days (5.48 years) after the diagnosis of COPD, a trend which was not observed for other causes of death. Among subjects with a higher degree of baseline systemic inflammation, COPD patients with frequent exacerbations had a 3.7-fold increase in the risk of SCD.

An increased risk of SCD in patients with COPD has been previously described in specific patient populations at high risk of SCD.13 Our prospective cohort study demonstrated for the first time that COPD is associated with an increased risk of SCD in the general population, even when the already increased risk of COPD on all-cause mortality was taken into account. There are several factors in persons with COPD that could play a role in the mechanism which ultimately leads to SCD. Chronic obstructive pulmonary disease has been associated with an increased risk of ventricular arrhythmias11,12 and a prolongation of the QTc interval,25 currently the most well-known electrocardiogram-derived risk indicator for SCD.26 QTc prolongation in COPD patients can originate from different mechanisms. First, COPD is associated with autonomic neuropathy, which is associated with QTc prolongation.25 Also, common comorbidities in COPD patients, such as coronary heart disease and heart failure are associated with a prolonged QTc.27 Other factors that are present in persons with COPD are a higher resting heart rate,10 hypoxia and hypoxaemia,6 cardiac ischaemia,4,9 and heart failure4,28 which are all known risk factors for SCD. Although information on time of SCD was incomplete, our data suggest that patients with COPD die more frequently during night time, potentially due to the reduced ventilation and blunted responses to hypercapnia resulting in more ventricular ectopic episodes while they are sleeping.29,30 Since baseline cardiovascular diseases did not have a substantial impact on the association between COPD and SCD, our data suggest that preexisting cardiovascular disease in patients with COPD does not explain the observed association. Moreover, our data suggest that mechanisms through which COPD might raise the risk of SCD act in a severity- or duration-dependent manner since the estimate increased by duration of COPD. One mechanism could be that COPD leads to hyperinflation, followed by remodelling, pulmonary hypertension, cardiac arrhythmias, and subsequently SCD.

Exacerbations could provide another mechanism through which SCD is caused. Suissa et al. recently described that with every exacerbation both the risk of subsequent COPD exacerbations and the risk of mortality increased.31 Elevated biomarkers of cardiovascular morbidity, such as troponine,32 midregional-pro atrial natriuretic peptide,33 and N-terminal-pro brain natriuretic peptide34 (the latter two are both indicators of heart failure) are associated with exacerbation mortality and a higher risk of myocardial infarction and stroke after a COPD exacerbation has been observed.35 Wedzicha et al. described that COPD exacerbations may lead to increased cardiovascular morbidity through systemic inflammation.36 This hypothesis is endorsed by another study that showed increased levels of C-reactive protein (CRP) and interleukin-6 following COPD exacerbations.37 Systemic inflammation in patients with COPD can be evaluated using CRP levels, an acute phase protein.38 C-reactive protein is strongly associated with cardiovascular disease in general,39,40 and SCD41 and COPD specifically.17,38 Our results demonstrate that the effect of frequent exacerbations was modified by the baseline degree of systemic inflammation. Similar to the subgroup of COPD patients with persistent inflammation which demonstrated high incidence of exacerbations and worse survival, the risk of SCD was minimally affected by the degree of airflow limitation.42 Possibly, COPD patients with substantial systemic inflammation develop more atherosclerosis, increasing the risk of coronary heart disease whereby each exacerbation further increases the risk of SCD.

The association between COPD and SCD provides various directions for further research. The first is to unravel the underlying mechanisms. Secondly, effective strategies to reduce the risk of SCD in COPD patients are needed. In addition to increased awareness of COPD as a risk factor for SCD, preventive options such as medical treatment (e.g. β-blockers), placement of implantable cardioverter defibrillator (ICD) or withdrawing QTc-prolonging drugs, should be investigated.

Our study has several strengths and some limitations. First, in previous studies, the association between COPD and SCD has been assessed only in specific patient populations who already had an increased risk of SCD, but not in a general population setting with adjustment for all known confounders. Second, SCD was validated by research physicians blinded to the COPD status of the subject. Another strength of this study is that we have prospectively gathered data on morbidity and mortality events, and on covariables, limiting the risk of information bias. However, some potential misclassification may have occurred by inclusion of unwitnessed deaths, though the percentage of unwitnessed deaths in our study was not different for COPD subjects compared with subjects without COPD. A first limitation is that autopsies and electrocardiography at time of death were rarely performed,2 hampering the evaluation of exact causes of death. However, we applied a universal definition, which is the best method for identification of SCD cases in a population-based study and indicates similar data validity.22 By exclusion of non-cardiac causes in all sudden deaths, we refined sudden deaths to only those of cardiac origin. Secondly, we do not have information on successful resuscitations. Importantly, numbers of subjects with ICD placement were very low, implicating that use of ICD could not distort our results. Another limitation is that lung function measurements were not performed in all study participants at baseline. However, a sensitivity analysis in subjects with spirometry, confirmed the increased risk of SCD in subjects with an obstructive lung function compared with subjects with a normal lung function. Finally, it has previously been suggested that participants in a prospective cohort study are more healthy than non-participants.43 The Rotterdam Study has an overall response rate of 72%. Persons with more severe COPD could have had lower participation rates. However, in our study this would have only resulted in an underestimation of the effect.

In conclusion, our study demonstrated that COPD is a risk factor for SCD in the general population, especially in the period more than 5 years after the diagnosis of COPD. The risk further increases in COPD patients with frequent exacerbations having a higher degree of baseline systemic inflammation. These data could facilitate accurate risk stratification and pave the path to targeted preventive actions for patients with COPD.

Funding

This work was supported by grants from the Fund for Scientific Research Flanders (FWO) project G035014N, the Netherlands Organisation for Health Research and Development (ZonMw) (Priority Medicines Elderly 113102005 to M.E. and P.R.R.; and DoelmatigheidsOnderzoek 80-82500-98-10208 to B.H.S.). L.L. is the recipient of a Belgian Thoracic Society Fellowship. O.H.F. works in ErasmusAGE, a centre for aging research across the life course funded by Nestlé Nutrition (Nestec Ltd.); Metagenics Inc.; and AXA. The Rotterdam Study is supported by the Erasmus MC and Erasmus University Rotterdam; the Netherlands Organisation for Scientific Research (NWO), the Netherlands Organisation for Health Research and Development (ZonMw), the Research Institute for Diseases in the Elderly (RIDE), the Netherlands Genomics Initiative (NGI), the Ministry of Education, Culture and Science, the Ministry of Health Welfare and Sports, the European Commission (DG XII), and the Municipality of Rotterdam.

Acknowledgements

The authors thank the study participants, the staff from the Rotterdam Study, and the participating pharmacists and general practitioners.

Conflict of interest: none declared.

References

View Abstract