Author + information
- Received May 19, 2015
- Accepted June 17, 2015
- Published online October 1, 2015.
- Kumar Narayanan, MD∗,
- Kyndaron Reinier, PhD∗,
- Audrey Uy-Evanado, MD∗,
- Carmen Teodorescu, MD, PhD∗,
- Lin Zhang, MD∗,
- Harpriya Chugh, BS∗,
- Gregory A. Nichols, PhD†,
- Karen Gunson, MD‡,
- Jonathan Jui, MD, MPH§ and
- Sumeet S. Chugh, MD∗∗ ()
- ∗The Heart Institute, Cedars-Sinai Medical Center, Los Angeles, California
- †Center for Health Research, Kaiser Permanente, Portland, Oregon
- ‡Department of Pathology, Oregon Health and Science University, Portland, Oregon
- §Department of Emergency Medicine, Oregon Health and Science University, Portland, Oregon
- ↵∗Reprint requests and correspondence:
Dr. Sumeet S. Chugh, The Heart Institute, Advanced Health Sciences Pavilion Suite A3100, Cedars-Sinai Medical Center, 127 S. San Vicente Boulevard, Los Angeles, California 90048.
Objectives The purpose of this study was to determine whether chronic obstructive pulmonary disease (COPD) is associated with sudden cardiac death (SCD) in the community.
Background COPD is linked to cardiovascular mortality; an association with SCD has not been systematically investigated in the general population.
Methods In the Oregon Sudden Unexpected Death Study (approximately 1 million population), adult SCD case subjects were compared with geographic control subjects with coronary artery disease. Detailed clinical and electrocardiographic risk marker information was obtained from medical records. The association of COPD with SCD in the overall population and in a propensity score–matched dataset was assessed with logistic models.
Results SCD case subjects (n = 728; age 69.9 ± 13.7 years) were more likely than control subjects (n = 548; age 67.2 ± 11.3 years) to have left ventricular ejection fraction ≤35% (27.5% vs. 12.0%; p < 0.0001), COPD (30.8% vs. 12.8%, p < 0.0001), diabetes mellitus (47.7% vs. 31.8%; p < 0.0001), use short-acting beta-2 agonist agents (SBAs) (22.3% vs. 12.6%; p < 0.0001), and less likely to use beta-blockers (60.6% vs. 66.4%; p = 0.03). In multivariable analysis, COPD was significantly associated with SCD (odds ratio [OR]: 2.2; 95% confidence interval [CI]: 1.4 to 3.5; p < 0.001). There was no significant interaction between COPD and medications, but an interaction was identified between SBAs and beta-blockers (p = 0.04); SBAs were strongly associated with SCD in subjects not taking beta-blockers (OR: 3.3; 95% CI: 1.4 to 7.7; p = 0.005) but not in those taking beta-blockers (OR: 1.3; 95% CI: 0.7 to 2.3; p = 0.39). The COPD-SCD association was maintained in a propensity score–matched analysis.
Conclusions COPD is associated with SCD risk in the community independent of medications, electrocardiographic risk markers, and left ventricular ejection fraction. Among other mechanisms, pro-arrhythmogenic right ventricular remodeling and systemic inflammation warrant further investigation.
Chronic obstructive pulmonary disease (COPD) is a major health problem that globally contributes to substantial morbidity and mortality (1). As a chronic condition, is associated with a large number of hospitalizations (2) and imposes a major financial burden on the health care system (3). Cardiovascular disease accounts for a sizeable proportion of morbidity and mortality in the COPD population (4), and studies have reported that COPD is an independent risk factor for cardiovascular mortality (5,6). Indeed, it has been suggested that the burden imposed by cardiovascular pathology in COPD patients may in fact be greater than that directly related to COPD itself (4). The absolute value (6) and the rate of decline (7) of the forced expiratory volume in the first second has been linked to risk of coronary artery disease (CAD).
Of all manifestations of cardiovascular disease, sudden cardiac death (SCD) is the most lethal, with <10% survival on average (8). Whether COPD is linked to SCD is important both from a mechanistic risk perspective and from the point of view of primary prevention of SCD, since COPD is often encountered as a comorbidity. Although some studies have suggested that COPD may be a risk factor for SCD (9), robust population-based data are scarce. It is also important to consider this question in the context of associated medications such as short-acting beta-2 agonists (SBAs) that may be independently proarrhythmic (10). Furthermore, because COPD prevalence increases with age, cardiovascular risk factors and CAD often coexist, and these need to be accounted for as well. One of the difficulties in studying SCD in the community is that pre-arrest clinical details are often sparse, and information is mostly limited to that gathered by emergency medical services (EMS). Through the Oregon Sudden Unexpected Death Study (Oregon-SUDS), a prospective, population-based study of SCD in the Portland, Oregon, metropolitan area, we have established a mechanism to systematically obtain lifetime clinical history for all SCD victims. We therefore performed a comprehensive evaluation of the association between COPD and SCD in the community.
The Oregon-SUDS is an ongoing, community-based study of SCD using prospective multiple-source case ascertainment. Detailed methods have been published previously (11,12). Briefly, cases of out-of-hospital-cardiac arrest are ascertained from the Portland metropolitan area (catchment population of approximately 1 million) by use of multiple sources, namely, first responders (Portland fire department and local ambulance service), the county medical examiner’s office, and the emergency departments of participating local hospitals. Detailed information on circumstances, clinical history, and autopsy data (where available) is gathered, and SCD cases of cardiac pathogenesis are identified through a 3-physician adjudication process. SCD is defined as a sudden, pulseless collapse of a likely cardiac cause, occurring rapidly after symptom onset when witnessed or within 24 h of the subject being last seen in the usual state of health if unwitnessed. Cases of trauma, drowning, drug abuse, known terminal illness, or malignancy not in remission and extracardiac causes (such as pulmonary embolism) are excluded. In parallel, control subjects with known CAD are recruited from the same geographic area. Since the majority of SCD cases are found to have significant CAD (13), the control group enables the identification of risk markers that are specific to SCD. The control consists of subjects transported by the region’s EMS for symptoms of acute coronary ischemia, those undergoing angiography at one of the participating hospitals, patients with CAD seen in a clinic, or members of a health maintenance organization who have CAD. CAD is defined as at least 50% stenosis in a coronary artery, history of myocardial infarction (MI), or coronary revascularization. The present study was a community-based case-control analysis. Adult SCD case subjects (2002 to 2013) who had physician records and pre-arrest left ventricular ejection fraction (LVEF) information available were compared with CAD control subjects from the same geographic region over the same time period. No further specific matching was performed between case and control subjects.
COPD and other clinical information
Detailed demographic information and a lifetime clinical history for all case subjects were obtained from medical records (prior and unrelated to the SCD event). COPD was defined as a documented physician diagnosis of COPD in the medical records. Patients using home oxygen were excluded on the basis of the a priori definition of SCD for this study. Information on medications, including SBAs, long-acting beta-2 agonists, and beta blocker use, was obtained from physician records. The LVEF was obtained from the echocardiogram, left ventricular angiogram, or multigated acquisition scan performed closest but unrelated to the SCD event. Similarly, electrocardiographic (ECG) parameters, including the corrected QT interval (QTc, by the Bazett formula) and heart rate, were measured from archived ECGs closest and unrelated to SCD.
Categorical variables (expressed as number and percentage) were compared with the chi-square or Fisher exact test, and continuous variables (expressed as mean and standard deviation) were compared with the Student t test or Mann-Whitney U test. Odds ratios (ORs) for the association of COPD and SBA use with SCD were obtained by use of multivariable logistic regression, with adjustment for LVEF and clinical and ECG markers significant in univariate analysis. We assessed for possible interaction between COPD and medications (SBA, beta-blocker use), as well as between SBA and beta-blockers. Stratified analysis was performed in case significant interaction was detected. A logistic model with a dummy variable for COPD and SBA use was employed to evaluate the potential additive risk of presence of both COPD and SBA use. We further performed sensitivity analysis using propensity score matching. The propensity matching was performed with PROC LOGISTIC (SAS version 9.3 SAS Institute, Cary, North Carolina) and a macro for 1-to-1 case-control propensity score matching, according to methods described by Parsons (14). First, we calculated each subject’s propensity score in a logistic regression model with COPD as the outcome variable, using the following covariates: age, sex, smoking status, diabetes mellitus, hypertension, LVEF, and medications, including angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, beta-blockers, and SBAs. On the basis of the predicted probability of COPD (propensity score) from the logistic regression model, case subjects were then matched to control subjects with the closest propensity score by use of nearest-neighbor matching using a 2-digit match to achieve matching while preserving sample size (i.e., subjects were matched if both the case and control subject had propensity scores that were identical in the first significant digit [the tenths], with the hundredths allowed to range from 0.01 to 0.09 for a matched pair), which resulted in a dataset with 330 case and control subjects each. We also performed additional sensitivity analysis with stronger matching (3-digit matching, with propensity score tenths and hundredths required to be identical); the resultant dataset had 418 patients (209 case and control subjects respectively). The association of COPD with SCD in the propensity score–matched dataset was analyzed by binary logistic regression. A 2-tailed p value of ≤0.05 was considered significant. Analyses were performed with SPSS software version 21.0 (IBM Corporation Inc., Armonk, New York) and SAS software version 9.3 (SAS Institute Inc.).
Initial analysis was performed in 728 SCD case subjects and 548 control subjects with CAD. Clinical and demographic characteristics of case and control subjects are outlined in Table 1. Case subjects were older than control subjects (age 69.9 ± 13.7 years vs. 67.2 ± 11.3 years; p < 0.0001), with similar sex distribution. Case and control subjects did not differ significantly with respect to mean body mass index, hypertension, mean cholesterol level, presence of obstructive sleep apnea (OSA), or smoking status. Case subjects were significantly more likely than control subjects to have diabetes mellitus (47.7% vs. 31.8%; p < 0.0001), LVEF ≤35% (27.5% vs. 12.0%; p < 0.0001), higher heart rate (78.5 ± 18.8 beats/min vs. 70.0 ± 16.8 beats/min; p < 0.0001), and higher QTc (463.3 ± 45.3 ms vs. 434.5 ± 36.4 ms; p < 0.0001). Implantable cardioverter-defibrillators (ICDs) were present only in a minority of SCD case subjects (3.6%) and control subjects (0.5%).
COPD and medications
COPD was significantly more prevalent in case subjects than in control subjects (30.8% vs. 12.8%; p < 0.0001) (Figure 1). The prevalence of asthma was not significantly different (9.8% vs. 9.1%; p = 0. 71). Case subjects were significantly more likely to be taking SBAs (22.3% vs. 12.6%; p < 0.0001) (Figure 1) and were less likely to be taking beta-blockers (60.6% vs. 66.4%; p = 0.03). Only a small percentage of case and control subjects were taking long-acting beta-2 agonists, which was again more likely to be seen in case subjects, with borderline significance (6.0% vs. 3.6%; p = 0.052). All subjects taking beta-2 agonists were prescribed the medication in the inhaled form; no subject was taking oral beta-2 agonists. Case subjects were also less likely to be on angiotensin receptor blockers (9.2% vs. 13.1%; p = 0.03) and statins (48.5% vs. 67.3%; p < 0.0001); however the proportion taking angiotensin-converting enzyme inhibitors was similar.
COPD and use of SBAs: Effects on SCD risk
Table 2 shows the results of the multivariable logistic regression models. Model 1 was adjusted for age, diabetes mellitus, LVEF, and medications (including SBA use), whereas model 2 was adjusted for smoking and ECG markers in addition to the variables adjusted for in model 1. COPD was significantly associated with SCD in both models (model 1 OR: 2.6 [95% confidence interval (CI): 1.8 to 3.7]; model 2 OR: 2.2 [95% CI: 1.4 to 3.5]; both p < 0.0001). No significant interaction was observed between COPD and either SBA use (p = 0.19) or beta-blocker use (p = 0.59). However, significant interaction was noted between SBAs and beta-blockers (p = 0.04). Hence, subjects were stratified by beta-blocker status to assess the association between SBA use and SCD, by use of similar logistic models as before. A strong association was observed between SBA use and SCD status among subjects not taking beta-blockers (OR: 2.6 [95% CI: 1.3 to 5.2], p = 0.006 for model 1; OR: 3.3 [95% CI: 1.4 to 7.7], p = 0.005 for model 2), even after adjustment for COPD and other risk factors. However, such an association was not present among subjects taking beta-blockers (OR: 0.9 [95% CI: 0.6 to 1.4], p = 0.72 for model 1; OR: 1.3 [95% CI: 0.7 to 2.3], p = 0.39 for model 2). When patients with SBA use not taking beta-blockers were compared with those taking both SBA and beta-blockers, most cardiac risk factors were noted to be similar between the 2 groups, and the main difference was a significantly higher heart rate in the SBA-only group (85.7 ± 20.6 beats/min vs. 75.8 ± 15.0 beats/min; p < 0.0001). In a logistic regression model that used dummy variables and adjustment for all other risk factors, presence of both COPD and SBA use was associated with higher odds for SCD (OR: 3.4; 95% CI: 2.0 to 5.7; p < 0.001) than COPD alone (Figure 2).
COPD and OSA
There was no evidence of significant interaction between COPD and OSA (p = 0.81). Also, presence of both COPD and OSA did not further increase the OR for SCD compared with presence of COPD alone (OR: 2.6 vs. 2.5, respectively).
Effect of COPD on SCD risk evaluated by propensity score–matched analysis
Analysis was performed as described with a propensity score–matched dataset (330 case and control subjects each). COPD remained significantly associated with SCD in the propensity score–matched dataset (OR: 3.9; 95% CI: 2.2 to 6.7; p < 0.0001). When the propensity match criteria were made stronger (3-digit matching, with propensity score tenths and hundredths required to be identical), the COPD-SCD association remained similar (OR: 3.2; 95% CI: 1.6 to 6.3).
In this population-based study, COPD was found to be significantly associated with SCD after adjustment for LVEF and other cardiovascular risk factors. COPD and SBA were also each independently associated with SCD, which suggests that the COPD-SCD risk was unrelated to SBA use. Furthermore, the presence of both COPD and SBA use had a higher OR for SCD than COPD alone. Although no interaction was seen between COPD and use of SBAs or beta-blocking drugs, a significant interaction was observed between beta-blockers and SBAs, with beta-blockers appearing to reduce SBA-related SCD risk, likely by blunting the sympathetic activity caused by SBAs. This premise is supported by the fact that compared with patients taking both SBA and beta-blockers, patients taking SBAs only had a significantly higher resting heart rate but had no differences in the frequency of other risk factors. There was a borderline-significant association seen between long-acting beta-agonists and SCD; however, only a small proportion of cases overall were taking long-acting beta-2 agonists. Because COPD often coexists with other comorbidities, we performed a sensitivity analysis after propensity score matching based on COPD status. The COPD-SCD association was maintained in the results of the propensity-matched dataset. Furthermore, severely ill patients on home oxygen therapy were excluded in our study. Taken together, these results indicate that COPD was associated with SCD risk independent of clinical/ECG risk markers and medications, which also suggests that COPD may exert deleterious effects on the heart via other mechanisms.
Lahousse et al. (15) recently reported an increased SCD risk in patients with COPD in the Rotterdam study, with an OR of 1.34. They found this risk to be higher in the early years after COPD diagnosis, as well as in those with frequent exacerbations. Similar to our study, they did not find a significant interaction between COPD and sympathomimetic drugs (15). The other community-based study that directly addressed the question of whether COPD contributes to SCD risk evaluated SCD cases from the ARREST (Amsterdam Resuscitation Study) registry and compared these with non-SCD case subjects from a pharmacological database. The study concluded that COPD was linked to higher SCD risk, and this was highest in those also taking SBAs (9). However, that study relied on drug information to ascertain the presence of COPD and other risk factors, unlike the present study in which physician researchers obtained information directly from medical records. Other studies have suggested that COPD contributes independently to cardiovascular mortality and potentially to SCD. In the UPLIFT (Understanding the Potential Long-term Impacts on Function With Tiotropium) COPD trial, SCD constituted 4.4% of all COPD deaths (16). VALIANT (Valsartan in Acute Myocardial Infarction Trial) among patients with acute MI showed that COPD was associated with higher risk of sudden death but not MI or stroke (17). OSA has been reported to be a risk factor for SCD, and it has been suggested that the presence of both OSA and COPD may result in even higher cardiovascular risk (18,19). In the present study, we did not find evidence of a significant interaction between OSA and COPD; this needs further focused evaluation in larger analyses.
It could be argued that COPD may merely be a coexisting condition or a “marker” for the type of person likely to have cardiac disease as well; however, growing evidence suggests that there may be a true link. The exact mechanism underlying the COPD-SCD relationship remains speculative; however, several possibilities can be advanced. Abnormalities of cardiac repolarization including increased QTc and QT dispersion have been identified in COPD patients and shown to be related to SCD (20,21), although in the present study, the COPD-SCD association was independent of QTc. A direct relation between hypoxemia and ventricular irritability has been suggested (22) but is likely to be operative only in a small subset of severe cases. However, the increased work of breathing in COPD and resultant increased oxygen demand (23) are additional stressors for a person with heart disease and therefore potential contributors to risk. Another area of increasing focus has been the inflammation that accompanies COPD (24). Considering the proximity of the lungs to the heart, inflammatory mediators generated in the lung can potentially have adverse consequences for the heart. A milieu of chronic inflammation could aid progression of atherosclerosis and increase arterial stiffness (25). COPD has been linked to greater coronary artery calcium scores (26). Cytokines can recruit leukocytes, contributing to inflammatory plaque rupture (27). Inflammatory mediators can also predispose to hypercoagulability and coronary thrombosis (28). Autonomic instability that leads to proarrhythmia may play a role as well. Reduced heart rate variability, which improved with a physical rehabilitation program, has been demonstrated in COPD (29). Finally, right ventricular (RV) remodeling may play a role in the proarrhythmic effects of COPD, because pulmonary hypertension results in dilation, hypertrophy, and eventually, failure of the RV (30). Exercise, which can be a potential trigger for arrhythmia, also causes worsening pulmonary hypertension, with the potential to increase cardiac vulnerability in the patient with COPD (31). RV failure can also lead to elevated levels of natriuretic peptides (32,33), which are known to be associated with increased risk of ventricular arrhythmia (34).
The COPD-SCD risk association is especially important in the context of primary prevention ICDs, since COPD may contribute to a greater comorbidity burden, and ICD implantation may be less likely in such candidates (35). However, it has been demonstrated that ICDs reduce mortality in patients with COPD (36), and the frequency of appropriate ICD shocks is actually higher than in non-COPD subjects (37). The findings of the current study suggest that the COPD-SCD association would also need to be considered in this decision-making process.
Finally, the interplay between COPD and associated medications in the context of SCD risk remains of considerable interest. Inhaled SBAs, which are often prescribed in COPD, may further increase risk of arrhythmia (38). Furthermore, studies of inhaled beta-2 agonists in patients with asthma suggest that regular use, in contrast to use on an as-needed basis, is associated with worse control of asthma and an increase in mortality (39,40). On the other hand, beta-blockers, which may reduce risk of arrhythmia, are often underprescribed in the COPD patient because of fear of worsening bronchospasm. It is noteworthy that a meta-analysis performed by Salpeter et al. (41) showed no worsening of lung function among COPD patients taking cardioselective beta-blockers. Especially for the COPD patient with concomitant heart disease, clinicians may need to consider minimizing SBA use and favor increased use of cardioselective beta-blockers.
Population-based design, robust adjudication of SCD cases, and detailed lifetime clinical history for each subject are some of the strengths of the present study. We also chose control subjects with CAD so as to focus specifically on SCD risk beyond CAD. However, there are some limitations to consider. Although both multivariable models and propensity matching analyses were performed, owing to the observational nature of the study, we cannot exclude an influence of unmeasured confounders with certainty. In community-based studies of this nature, it is challenging to obtain all aspects of data uniformly for all subjects. Thus, precise data on measures that reflect overall status, such as hospitalizations, were not available. We chose those subjects with LVEF information available for this analysis, and this could result in some bias; however, it is important to account for ejection fraction, which is the main marker presently used in risk stratification. Finally, our results would need to be confirmed in prospective studies and in other populations before they can be broadly applied.
In this population-based study, COPD was significantly associated with risk of SCD independent of the LVEF, medications used, and clinical and ECG risk markers. These results suggest that COPD may need to be considered as a separate cardiovascular risk factor, and novel mechanisms of risk related to inflammation and RV remodeling need to be further explored.
COMPETENCY IN MEDICAL KNOWLEDGE: COPD is associated with SCD in the general population even after accounting for associated medications, ECG markers, and the left ventricular ejection fraction, which suggests the need to evaluate novel mechanisms related to inflammation and right ventricular remodeling.
TRANSLATIONAL OUTLOOK: Clinicians may need to consider COPD as a cardiovascular risk factor. Further prospective studies are needed to study the interplay between COPD, medications, and SCD risk, which may have implications for the use of short-acting beta-agonists and beta blockers in routine clinical practice among COPD patients.
The authors acknowledge the important contribution of American Medical Response, Portland/Gresham fire departments, and the Oregon State Medical Examiner’s office.
This study was funded in part by National Heart, Lung, and Blood Institute grants R01HL105170 and R01HL122492 to Dr. Chugh. Dr. Nichols has received grant funding from Merck, Novartis, Boehringer-Ingelheim, and AstraZeneca. Dr. Chugh holds the Pauline and Harold Price Chair in Cardiac Electrophysiology Research at the Heart Institute, Cedars-Sinai Medical Center, Los Angeles, California. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- coronary artery disease
- confidence interval
- chronic obstructive pulmonary disease
- emergency medical services
- implantable cardioverter-defibrillator
- left ventricular ejection fraction
- myocardial infarction
- odds ratio
- obstructive sleep apnea
- corrected QT interval
- right ventricle
- short-acting beta-2 agonist
- sudden cardiac death
- Received May 19, 2015.
- Accepted June 17, 2015.
- American College of Cardiology Foundation
- Lozano R.,
- Naghavi M.,
- Foreman K.,
- et al.
- Ford E.S.
- Hole D.J.,
- Watt G.C.,
- Davey-Smith G.,
- Hart C.L.,
- Gillis C.R.,
- Hawthorne V.M.
- Chugh S.S.,
- Jui J.,
- Gunson K.,
- et al.
- ↵Parsons LS. SAS Users Group Paper 225-25, Using SAS® Software to Perform a Case-Control Match on Propensity Score in an Observational Study. Available at: http://www2.sas.com/proceedings/sugi25/25/po/25p225.pdf. Accessed May 1, 2015.
- Lahousse L.,
- Niemeijer M.N.,
- van den Berg M.E.,
- et al.
- Hawkins N.M.,
- Huang Z.,
- Pieper K.S.,
- et al.,
- Valsartan in Acute Myocardial Infarction Trial Investigators
- Gami A.S.,
- Olson E.J.,
- Shen W.K.,
- et al.
- Sin D.D.,
- Man S.F.
- Mills N.L.,
- Miller J.J.,
- Anand A.,
- et al.
- Newman A.B.,
- Naydeck B.L.,
- Sutton-Tyrrell K.,
- Feldman A.,
- Edmundowicz D.,
- Kuller L.H.
- Suwa T.,
- Hogg J.C.,
- Quinlan K.B.,
- Ohgami A.,
- Vincent R.,
- van Eeden S.F.
- Zupanic E.,
- Zivanovic I.,
- Kalisnik J.M.,
- Avbelj V.,
- Lainscak M.