Reporting of pain by people with chronic obstructive pulmonary disease (COPD): comparative results from the HUNT3 population-based survey

Data source

In this study, we analyzed cross-sectional data obtained from the HUNT Study in Norway, which with 50,807 participants, is one of the largest health studies ever performed. The third health survey (HUNT3) was carried out in 2006 to 2008 and invited all inhabitants of the county of Nord-Trøndelag aged 20 years and older to participate. The health study included questions about socioeconomic conditions, health-related behaviors, symptoms, illnesses, and diseases. The general design and details of the HUNT3 study are given elsewhere [22, 23]. The county of Nord-Trøndelag is representative of Norway in terms of geography, economy, industry, sources of income, age distribution, and the morbidity and mortality rates of its inhabitants [24, 25].

Participants and sample

Of the 90,860 people invited, 50,807 participated, yielding a 54.1% participation rate. For the purpose of the present study, we included participants with a non-communicable chronic diseases (COPD, arthritis, heart disease, or diabetes), and healthy people who reported no disease (n = 18,811). In the questionnaire, medical conditions were assessed by self-report, with respondents indicating whether they have or ever had any of the 20 listed medical conditions. Cancer was not included as a specific disease group in this study because, due to the wording of the disease questions, respondents with a current diagnosis of cancer could not be distinguished from those who were successfully treated in the past and did not currently have chronic disease.

Because respondents could indicate more than one disease, we grouped them into mutually exclusive categories using a new variable named “disease”. The variable “disease” was coded as follows: All respondents who indicated chronic bronchitis, emphysema, or COPD were defined as having COPD (n = 1199) and were coded as 0. Those who indicated Rheumatoid arthritis, osteoarthritis, Bechterew’s disease, or fibromyalgia, but not COPD, were defined as having arthritis (n = 8582) and were coded as 1. Those who indicated myocardial infarction, angina pectoris, cardiac insufficiency, or unspecific heart disease, but not COPD or arthritis, were defined as having heart disease (n = 4109) and were coded as 2. Those who indicated diabetes, but not COPD, arthritis, or heart disease, were defined as having diabetes (n = 1254) and were coded as 3. Those who did not indicate any of the medical diseases listed in the questionnaire composed the non-disease group (n = 18,811) and were coded as 4. These five groups, totaling 33,955 respondents, constitute the sample in the present study.

Measures

Statistical analyses

Descriptive summary of the variables included in the variables were presented as percentages, mean with standard deviation (SD). Bivariate analysis of disease group and pain was performed with cross-tabs using chi-square test. The association between chronic pain and independent variables was analyzed using logistic regression, with odds ratios (ORs) and 95% confidence intervals (CIs). Chronic pain was treated as a categorical variable coded as 1 if the person reported pain lasting more than 6 months and 0 otherwise. The second dependent variable, pain intensity, was analyzed using ordinary linear regression, with regression coefficients, standard errors, and p-values. Pain intensity was rated on a six-point pain intensity scale with 0 representing no pain and 5 representing very severe pain. The models for both dependent variables (chronic pain and pain intensity) included the same independent variables. There were some missing values in the dataset. Excluding missing cases or including the missing cases as a separate category could have led to biased estimates. Therefore, multiple imputation was used to generate 5 imputed datasets using the mi impute chained command in Stata® 13. The procedure replaced each missing value with a set of plausible values based on all other variables included in the statistical model. We also ran the models without multiple imputation. The results of the analyses without imputed data (available upon request) were similar to those presented in Tables 2 and 3, in which we used multiple imputation. For further details, see White et al. [30].

Descriptive summary

Chronic pain in different disease groups

Participants with COPD were less likely to have chronic pain than those with arthritis, but more likely to have chronic pain than those with diabetes or no disease.

As shown in Table 1, the prevalence of chronic pain in the four disease groups ranged from a low of 43.0% in the diabetes group to a high of 75.4% in the arthritis group, with the non-disease group having a chronic pain prevalence of 25.9%. Chi-square test showed significant differences in chronic pain between the groups in the analyses (p < 0.001).

Chronic pain was also associated with several other independent variables (Table 2). Men had lower odds of reporting chronic pain (OR 0.81, 95% CI: 0.77–0.86) than women. The odds of reporting chronic pain increased with increasing age (OR 1.07, 95% CI: 1.05–1.09) and was higher for those employed in occupations with low educational requirements (OR 1.17, 95% CI: 1.10–1.24) than for those employed in occupations with high educational requirements. Daily smoking was associated with much higher odds of reporting chronic pain (OR 1.30, 95% CI: 1.22–1.37) compared with people who did not smoke. Physical activity was negatively associated with the odds of reporting chronic pain. The odds of reporting chronic pain was lower in people who exercised nearly every day (OR 0.63, 95% CI: 0.56–0.72) than in those who never exercised. Those with comorbidity had higher odds of reporting chronic pain than those without comorbidity (OR 1.75, 95% CI: 1.48–2.07). The odds of reporting chronic pain also increased with increasing anxiety (OR 1.08, 95% CI: 1.06–1.09) and depression scores (OR 1.05, 95% CI: 1.04–1.06). Living arrangement was not significantly associated with the odds of reporting chronic pain.

Differences in pain intensity between different disease groups

Pain intensity was most commonly rated as moderate in the four disease groups, with 50.6% of the arthritis group and 39.3% of the COPD group reporting moderate pain in the last 4 weeks. Furthermore, 19.4% of the arthritis group and 15.8% for COPD reported severe pain, while very severe pain was reported by < 2% for all groups. In contrast, the most common pain intensity rating in the non-disease group was no pain (42.0%) (Table 1).

Pain intensity was also associated with several other independent variables. Women reported greater pain intensity than men did, and pain intensity increased with increasing age. People employed in occupations with low educational requirements reported greater pain intensity than did those employed in occupations with high educational requirements. Moreover, smoking was associated with greater pain intensity, and exercising was associated with lower pain intensity. People with comorbidity reported higher pain intensity than those without comorbidity, and pain intensity also increased with increasing anxiety and depression scores. Living arrangement was not significantly related to pain intensity.

Chronic pain in people with COPD and different medical conditions

The finding that people with arthritis had the highest probability of chronic pain is to be expected because pain is the primary symptom of arthritis [31]. A large European survey on pain that included in-depth interviews with 4839 respondents with chronic pain showed that arthritis was the most common cause of pain (42%) [17]. COPD and heart disease rated next, which generally correspond to findings in a large US cohort, where Roberts et al. found that, after arthritis, patients with COPD were the most likely group to report chronic pain [14]. While we found no significant differences between people with COPD and heart disease, Roberts et al. reported higher rates of chronic pain in COPD than heart disease. While Roberts et al. used specific International Classification of Disease (9th edition) Clinical Modification codes to define COPD and pain, but Centers for Medicare and Medicaid definitions were used to define heart disease, we included a wide range of self-reported heart conditions in the heart disease group. This methodological difference may explain the slight discrepancy between our results and the findings of Roberts et al.

The self-reported comorbidity rate in COPD of 80% was lower than the other disease groups, where heart disease and diabetes had 100% comorbidity. The literature concerning the role of comorbidities in the presence of pain in COPD is conflicting [6]. Nonetheless, COPD is a complex disease that has profound effects on cardiac function and gas exchange with systemic consequences [32]. According to Barnes and Celli (2009), systemic inflammation may also initiate or worsen common comorbid diseases in COPD, such as ischemic heart disease, heart failure, osteoporosis, lung cancer, depression and diabetes [32]. The finding in the present study suggests that comorbidity is highly related to the probability of chronic pain and higher pain intensity in all the non-communicable diseases included in this study.

Sex differences in reporting of pain

Chronic pain was more common in women than in men, and women also reported greater pain intensity. These findings are consistent with the research literature, which generally reports a higher prevalence of pain in women than in men [33–36]. Other clinical and epidemiological studies have shown that women also have a higher prevalence of many painful disorders, such as rheumatoid arthritis, compared to men [34, 37, 38]. In a study of people with COPD, Christiansen et al. also reported a higher frequency of pain in women than in men [39]. Many explanations for the differences in pain between women and men have been proposed and are supported by evidence; these include neuroimmunological [40], hormonal, genetic, and psychological factors [41]. Moreover, it has been questioned whether men and women construct different meanings of pain, which might influence their reports of the pain experience [42].

Anxiety and depression

We found higher probabilities of chronic pain and higher pain intensity with increasing anxiety and depression scores. These findings are consistent with prior research showing that higher levels of anxiety are associated with increased clinical pain and heightened experimental pain severity [34]. There is some evidence that the affective dimensions of dyspnea, the main symptom in COPD, such as distress and unpleasantness, are processed in the same areas of the brain that process fear, anxiety, and pain [43, 44], which suggests a possible pathway to explain this association. Our findings suggest that reducing anxiety might be beneficial by also helping patients to reduce their experience of pain.

Pain and socioeconomic and lifestyle factors

The regression analyses in this study indicate that lower socioeconomic status, which was assessed by the educational requirement for one’s occupation, increased the odds of reporting chronic pain and was associated with greater pain intensity. The research literature shows that the prevalence and intensity of pain are higher among people living in areas of low socioeconomic status [45, 46]. Furthermore, in a study on comorbidity that included 1.75 million patients treated at 341 health centers in Scotland, Barnett et al. [47] found that people living in deprived areas were more likely to have COPD, depression, and pain as comorbidities than other disorders. In that study, among people with COPD, 31% of those living in deprived areas reported having a painful condition, whereas this percentage was reduced by half (15%) for those living in more prosperous areas [47]. In the present study, we had no socioeconomic information other than educational level, but our findings support the relationships between socioeconomic factors and diseases and pain.

The percentage of smokers was high in all disease groups, particularly in those with COPD, in which 76% smoked daily or occasionally. Tobacco smoking has long been acknowledged as the primary cause of COPD in industrial countries [48], and smoking is a recognized risk factor for chronic pain [49]. There is also epidemiological evidence that smokers have higher rates of chronic pain and that they rate their pain as more intense than do nonsmokers [50]. International evidence shows that smoking is more common in groups with low socioeconomic status [51]. Moreover, the bi-directional relationships between smoking and depression in the general population are well established (e.g. [52]). The mechanisms responsible for the relationships between smoking, pain, and depression are complex [50]. In addition to its stimulating effect, nicotine may also have antidepressant effects that maintain smoking [53]. Consequently, the high prevalence of smokers in the present study may suggest that, in people with COPD and other chronic conditions, cigarette smoking represents one way of coping with pain and depression. Many smokers report that smoking improves their mood, suggesting that it may function as self-medication of their depression. The associations between smoking and pain underline the continuing need for tobacco-control policies, which should consider lifestyle and socioeconomic position. Since indoor smoking in public was prohibited in Norway in 2004, smoking habits have changed radically, and the overall percentage of daily smokers has halved [54].

The finding that exercising was inversely related to chronic pain and pain intensity support the available evidence suggesting that physical activity and exercise may improve pain severity [55]. On the other hand, people with pain may also lack the energy to exercise or may find exercise can exacerbate their pain.

Strengths and weaknesses of the study

The primary strength of the study is the large and unselected sample, which is representative of Norway’s general population. This differs from most previous studies on pain in people with COPD, which have been conducted in clinical, outpatient settings, or pulmonary rehabilitation settings. The participants with COPD may have had other diseases, but we had no information to determine whether COPD was their primary diagnosis. However, by including a variable measuring the presence of comorbidity, we controlled for comorbidity in the regression analyses. We acknowledge some limitations of this study. First, the medical diagnoses relied on self-report, and some morbidities were probably underreported [3]. In the general population, poor agreement between self-report and health administrative data has been reported, particularly in people with COPD [56]. Although the non-disease group probably included individuals with unreported medical conditions, the high number of participants in this group contributed to the high statistical power, which should have neutralized the impact of false negatives. Second, we used employment in a job requiring different levels of education as a proxy for socioeconomic status. Having more details data about educational level and income would have strengthened the validity of our findings. Despite these shortcomings, we believe that the epidemiological data included here will help to corroborate and extend the findings of previous studies on the relationships between diseases including COPD and the experience of pain.