In this paper, we describe the prevalence of clinically-relevant comorbid conditions in a nationally-representative sample of people with physician-diagnosed COPD. We found that 96.4% of adults with physician-diagnosed COPD have at least one condition that may complicate the treatment of COPD. Most notably, 51.8% of people with COPD 45 and older are taking more than 4 medications (polypharmacy), 55.6% report mobility difficulty, 60.4% have hypertension, and 54.6% have arthritis.
These prevalence values are relatively consistent with those found in previous studies of comorbidities in COPD. However, there is a large range of previously reported prevalence values. For example, estimates of arthritis in COPD range from 22% [5
] to 70% [23
]. Neither of these studies, nor any other recent studies investigating comorbidity in COPD, have examined nationally-representative data. This both limits the applicability of these prevalence estimates and helps account for the large ranges in these estimates. In fact, several papers have cited lack of national representation or specific population bias as a weakness [5
]. As such, our study both confirms the high prevalence of comorbidities in patients with COPD and provides specific prevalence values that are relevant on a national scale.
Another strength of our study is the range of clinically-relevant conditions assessed. While there is a lot of data on, for example, cardiovascular disease in COPD [8
], there are few studies that look at the wide variety of medical conditions and functional limitations we have assessed. This is important partially because comorbidity has been found to be an important aspect of quality of life in COPD [25
], as well as an independent risk factor for hospitalization [28
]. In addition, comorbidities increase the risk of hospitalization and mortality in patients with COPD [8
], and significantly increase the costs of treating COPD [29
]. These conditions are also highly relevant for clinical decision-making and self-management.
The classification of the conditions into disease, clinical factor, and health status factor domains highlights that a wide range of conditions relevant to the clinical management of people with COPD are quite prevalent, and that these relevant conditions extend beyond traditionally-defined diseases.
Physicians must be judicious when caring for patients with COPD. The high prevalence of comorbidity and polypharmacy means physicians must be cognizant of potential adverse drug events and nonadherence. People with COPD are often complex, and, thus, we will need to improve our ability to prioritize treatment recommendations based on relative benefits and harms and patient preferences. Current guidelines, and our evidence base, do not yet adequately inform this critical clinical decision-making [13
Clinical practice guidelines generally do not address how to treat COPD in the context of comorbid conditions [11
]. As such, these guidelines may be of little help when dealing with the majority of COPD patients [30
]. For example, β-blockers, which are indicated for cardiovascular disease, may worsen lung function in some patients with COPD; some studies, however, have shown that this is not a contraindication to the initiation of β-blockers [32
]. Conversely, bronchodilators, which are believed to be beneficial for pulmonary function, may worsen tachyarrhythmias [9
]. Guidance about these potential interactions, and the quality of evidence supporting any recommendations about them, would be very useful to clinicians. Our results can inform clinical practice guideline priority-setting processes to determine which comorbidities should be addressed in future COPD guidelines.
It is imperative that therapeutic trials be designed to reflect the true
population of people with COPD. Many studies exclude patients with significant comorbid conditions [33
]. Herland et al. found that only 17% of a group of COPD patients would be eligible for a “typical” clinical trial. 65.9% of COPD patients in this study were “excluded” due to significant comorbidities, including diabetes, depression, and ischemic heart disease [34
]. These exclusions may be troublesome given the high prevalence of these diseases in people with COPD, and the potential for interactions between the diseases and their treatments. Our study highlights the need for future clinical trials to evaluate safety and effectiveness in COPD patients with multiple comorbidities.
There is evidence that treating COPD may benefit the course of comorbid conditions, and, visa-versa [35
]. For example, several observational studies have shown improved outcomes in COPD patients treated with statins [36
], independent of whether patients have a comorbid diagnosis of ischemic heart disease [38
]. While a randomized controlled trial has shown an improvement in exercise tolerance in COPD patients treated with statins [39
], more prospective intervention trials are needed to look at the use of statins in COPD [40
Given these potential complications of treatment and the interactions between comorbid conditions and COPD, some researchers have started to advocate for an integrated-care approach to the management of patients with COPD. Sonetti et al. in a recent review advocate for a chronic care model approach to COPD management, with an approach that includes automatic screening for common comorbidities [41
]. In order to truly move to such a system, however, it is necessary to have a good understanding of how best to treat COPD in the context of comorbid conditions. First steps to accomplishing this are (a) determining common “patterns” of co-existing conditions (b) including patients with comorbid conditions in clinical trials with appropriate analytic strategies to understand heterogeneity of treatment effect [43
] (c) evaluating current treatment regimens in patients with different patterns of comorbid conditions and (d) continuing to study possible pathophysiologic connections between COPD and comorbidities. Further research should also continue to explore the effects comorbid conditions have on outcomes (health-related and other) in COPD.
As spirometry data is not available in NHANES 1999–2006, we were not able to look at comorbidities in the context of the severity of COPD. This is significant because a recent study showed that increased respiratory impairment was associated with a higher risk of having comorbid hypertension, CVD, and diabetes [8
Also due to the absence of spirometry data, we defined COPD via self-report. While this does not meet the gold standard definition for COPD (which is spirometric) [5
], we are aware of no other applicable datasets that are nationally-representative and as comprehensive and current as NHANES 1999–2008. For example, NHANES 1988–1994 has spirometry data on a subset of participants; but, given the changing demographics of the US, it is unlikely this data is entirely representative of the current US population.
We acknowledge that the prevalence of comorbidities in physician-diagnosed COPD may be different than that in spirometrically-defined COPD; for example, patients with a higher burden of disease and lower health status may be more likely to receive a physician diagnosis of COPD. There is also likely misclassification of some subjects - subjects with spirometrically-defined COPD in the group without physician-diagnosed COPD, and subjects who reported a physician-diagnosis of COPD, but who would not meet spirometric criteria. Given the available data, and the desire to assess comorbidities in a larger sample of adults across multiple waves of NHANES, we believe physician-diagnosed COPD is a relevant outcome.
A study by Barr et al. found that self-report-based surveys are an appropriate way to study respiratory disease in healthcare professionals [44
], and many studies of comorbidities in COPD have used self-report [20
]. We found little difference in the prevalence of comorbid conditions between COPD subjects with and without a history of smoking, which helps validate our definition of COPD. Our definitions of many of the comorbidities were similarly limited by the need to, in some cases, define conditions by self-report.
We were also limited by factors included in the NHANES data set. We would have liked to assess warfarin use in COPD patients (our sample size was too small), human immunodeficiency virus (NHANES only runs HIV tests on subjects between the ages of 18 and 49), insomnia/sleeping problems [27
] (only assessed in NHANES 2005–2008), gastroesophageal reflux disease, pulmonary embolism, and pneumonia (these conditions were not directly assessed in NHANES).