Pain as a mediator of sleep problems in arthritis and other chronic conditions

Authors

  • J. Denise Power,

    Corresponding author
    1. University of Toronto, and Toronto Western Research Institute, University Health Network, Toronto, Ontario, Canada
    • Arthritis Community Research and Evaluation Unit, Toronto Western Research Institute, 399 Bathurst Street MP10-322, Toronto, Ontario M5T 2S8
    Search for more papers by this author
  • Anthony V. Perruccio,

    1. University of Toronto, and Toronto Western Research Institute, University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Elizabeth M. Badley

    1. University of Toronto, and Toronto Western Research Institute, University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author

Abstract

Objective

To examine the associations between arthritis and insomnia symptoms and unrefreshing sleep, as well as the role of pain as a mediator of these relationships.

Methods

Analyses were conducted on the cross-sectional, nationally representative, weighted sample of adults ≥18 years of age (n = 118,336) in the 2000/2001 Canadian Community Health Survey. Four logistic regression models were estimated for each sleep problem (model 1: arthritis only; model 2: model 1 + sociodemographic characteristics, lifestyle factors, and other chronic conditions; model 3: model 2 + mental health [stress, depression]; and model 4: model 3 + pain). Mediation by pain was quantified by the percentage change in the effect of arthritis on a particular sleep problem by comparing models 3 and 4.

Results

The prevalence of insomnia symptoms and unrefreshing sleep in persons with arthritis was 24.8% and 11.9%, respectively. These estimates are twice as high as those for persons without arthritis. In multivariate regression analyses, the addition of pain decreased the effect of arthritis by 53% (insomnia symptoms) and 64% (unrefreshing sleep). The effect of arthritis was still statistically significant in these models, suggesting that pain is a partial mediator of these relationships.

Conclusion

Insomnia symptoms and unrefreshing sleep affect a considerable proportion of individuals with arthritis. Pain mediates a substantial amount of the relationship between arthritis and sleep problems. Better pain management could significantly improve sleep in individuals with arthritis.

INTRODUCTION

Difficulties with sleep have far reaching consequences. They compromise quality of life, psychosocial wellbeing, and occupational and educational performance, and result in increased morbidity, mortality, and health care utilization (1–6). There is growing evidence that sleep is affected as a consequence of physical health, with poor health precipitating or contributing to existing sleep difficulties (7–10). Arthritis has been found to be associated with sleep problems in a number of studies, although these studies have mainly been conducted on small patient populations and have tended to focus on inflammatory forms of the disease (8, 11–21).

There is some evidence to suggest that sleep problems may also be of concern in individuals with osteoarthritis. Leigh and colleagues (20) found that when compared with age- and sex-matched controls, male patients with osteoarthritis had more stage 1 sleep (very light stage of sleep similar to drowsiness) and less stage 2 sleep (light sleep). Weekly problems with sleep onset, sleep maintenance, and early morning awakenings were reported by 31%, 81%, and 51% of participants with knee pain (with and without radiographic evidence of osteoarthritis), respectively, in the Observational Arthritis Study in Seniors (22). Furthermore, one-third of individuals with self-reported arthritis, the vast majority of whom are likely to have osteoarthritis, reported that they had experienced arthritis-related sleep disruptions in the National Survey of Self-Care and Aging (23). This sleep disruption was associated with the most use of self-care, complementary, and medical treatments relative to other daily living disruptions due to arthritis (23).

Pain is the cardinal symptom of arthritis and is a strong correlate of sleep problems in community-based samples (8, 9, 15, 24, 25). A Gallup survey conducted in the 1990s found that 1 in 3 adult Americans experiences nighttime pain and sleeplessness, and that one-third of respondents with nighttime pain reported arthritic pain (26). In a study of hospitalized patients with rheumatic disease, pain was the most frequently cited cause of sleep disturbance (27). Furthermore, patients with osteoarthritis who undergo total hip arthroplasty have been found to experience significant improvements in sleep quality (28). This is likely due to the reduction in pain experienced after the procedure.

The majority of our knowledge about the relationships between arthritis, pain, and sleep problems has come from small clinical studies. Few of these studies have used multivariate analyses or have attempted to assess how much of the relationship between arthritis and sleep problems is in fact mediated by or due to pain. The size of this relationship has important clinical significance in terms of treatment strategies for patients with arthritis who experience sleep problems. Additionally, the consequences of arthritis, including sleep difficulties, are of increasing relevance to public health given the aging of the population and the increases in arthritis prevalence that have been projected (29, 30).

The 2000/2001 Canadian Community Health Survey (CCHS) is the first major Canadian survey in 10 years to incorporate questions on sleep difficulties. This study focuses on 2 types of sleep problems: insomnia symptoms and unrefreshing sleep. Insomnia is generally defined as an impression of inadequate sleep associated with deficits in initiating or maintaining sleep (31) and is the most common sleep difficulty in the general population (32, 33). Unrefreshing or nonrestorative sleep refers to lack of satisfaction with the sleep experience and is associated with a sense of being unrefreshed after sleep (34). This specific sleep problem has not generally been the focus of much research, particularly relative to insomnia. Furthermore, the 1991 General Social Survey found that although insomnia and unrefreshing sleep may intuitively seem to be highly related, they are in fact not (35). The objective of the current study was to examine the associations between self-reported arthritis and these 2 sleep problems, as well as the role of pain as a mediator of these relationships, using data from the CCHS and adjusting for a comprehensive list of potentially relevant factors, including other chronic conditions. It was necessary to statistically adjust for these conditions because of their co-occurrence with arthritis, but this also allowed us to consider the relationships of these conditions with sleep problems as well.

SUBJECTS AND METHODS

Sample.

The target population of the 2000/2001 CCHS (cycle 1.1), a cross-sectional, general population health survey, included all persons ages ≥12 years living in private dwellings in the 10 provinces and 3 territories. Persons living on Indian Reserves or Crown Lands, institutional residents, full-time members of the Canadian Armed Forces, and residents of certain remote areas were excluded from the survey. The CCHS covered ∼98% of the Canadian population age ≥12 years. The survey used a stratified 2-stage cluster design. In stage 1, separate strata were formed in each province, and then independent clusters were drawn from each stratum. In stage 2, dwelling lists were prepared for each cluster, and dwellings were selected from those lists. Generally, there was 1 randomly selected person per household. The overall response rate was 84.7%, with 130,880 individuals participating. All data from the CCHS are weighted to be representative at the regional and national level, accounting for sample design, adjustments for nonresponse, and poststratification. Further information on the design of the survey can be obtained from Statistics Canada (36). The analyses presented in this report were based on the sample of 118,336 adults ages ≥18 years.

Arthritis.

Subjects were asked if they had specific chronic conditions that had lasted or were expected to last ≥6 months and that were diagnosed by a health professional. Conditions included arthritis or rheumatism, excluding fibromyalgia. Responses were coded as the absence versus the presence of arthritis.

Pain.

Subjects were asked if they were usually free of pain or discomfort. Those who responded that they were not were asked to report the usual intensity of their pain or discomfort. Response options to this question were mild, moderate, or severe. A 4-level variable was created from the responses to these questions, classifying the usual amount of pain experienced as none, mild, moderate, or severe.

Insomnia symptoms.

Subjects were asked the following question: “How often do you have trouble going to sleep or staying asleep?” Response options were “most of the time,” “sometimes,” or “never.” Responses of “most of the time” were considered to represent the presence of insomnia symptoms. Responses of “sometimes” or “never” were grouped to represent the absence of such symptoms.

Unrefreshing sleep.

Subjects were asked how often they find sleep refreshing. Responses of “never” were considered to indicate unrefreshing sleep and responses of “most of the time” and “sometimes” were grouped to indicate the absence of this sleep problem.

Control variables.

Sociodemographic characteristics (33, 37), lifestyle factors (25, 38), other chronic conditions (8, 9), and mental health (39, 40) have all been linked to sleep disorders, and as such, an a priori decision was made to include these factors in multivariate analyses.

Sociodemographic characteristics.

Sociodemographic characteristics included sex and age grouped as 18–24 years, 25–34 years, 35–44 years, 45–54 years, 55–64 years, 65–74, and ≥75 years. Marital status (married/common law, single, widow/separated/divorced) and highest level of education (less than secondary school graduation, secondary school graduation, some postsecondary, postsecondary graduation) were also included. Statistics Canada's derived 4-level income variable was used for household income (41). This variable considers both the total household income and the number of individuals in the household. Household income is categorized as lowest, lower middle, upper middle, and highest. Because 10% of the weighted sample was missing data on income, a fifth category of “not stated” was added.

Lifestyle factors.

Statistics Canada's derived variables for type of smoker and type of drinker were used (41). Type of smoker was grouped as current smoker (daily, occasional former daily, occasional), former smoker (former daily, former occasional), and never smoker. Type of drinker was grouped as regular drinker, occasional drinker, and nondrinker (former, never). For physical activity, Statistics Canada's derived physical activity index (active, moderate, inactive) was used (41). This index is created based on the energy expenditure of leisure activities during the 3 months prior to the survey and considers both the duration and frequency of each activity. Body mass index (BMI) was calculated based on self-reported height (in meters) and weight (in kilograms) as weight/height2. BMI was grouped into 4 categories: <20 kg/m2 (insufficient weight), 20–24.9 kg/m2 (acceptable weight), 25–27 kg/m2 (some excess weight), and >27 kg/m2 (overweight).

Other chronic conditions.

A series of dichotomous variables (present versus absent) was created for the following conditions that subjects indicated had lasted or were expected to last 6 months or more and were diagnosed by a health professional: other musculoskeletal conditions (fibromyalgia or back problems), allergies, lung problems (asthma or chronic bronchitis), diabetes, cancer, digestive problems (stomach or intestinal ulcers or bowel disorders), cardiovascular problems (high blood pressure, heart disease, or stroke), migraine, neurologic problems (multiple sclerosis, epilepsy, or Parkinson's disease), and other chronic conditions (thyroid condition, chronic fatigue syndrome, multiple chemical sensitivities, or urinary incontinence).

Mental health.

The CCHS includes a series of questions on depression that are a subset of items from the Composite International Diagnostic Interview (CIDI), which measures major depressive episodes (41). The CIDI is a structured diagnostic instrument that was designed to produce diagnoses according to the definitions and the criteria of both the Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised and the Diagnostic Criteria for the Research of the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision. From these questions, a short-form score was calculated for each subject. A dichotomous variable for depression was created from this score as 4–8 (depressed) versus 0–3 (not depressed). A score of 4 was used as the cutoff because it represents an 80% probability of caseness depression. Subjects were asked about their level of stress using the following question: “Thinking about the amount of stress in your life, would you say that most days are (not at all stressful, not very stressful, a bit stressful, quite a bit stressful, extremely stressful)?” Responses were grouped into 3 categories: not at all/not very, a bit, quite a bit/extremely.

Statistical analysis.

Estimates of the prevalence of arthritis, insomnia symptoms, unrefreshing sleep, and pain (by level of severity) were generated for the study population and compared for persons with and without arthritis. The prevalence of insomnia symptoms and unrefreshing sleep was also determined by level of pain for both groups.

A series of 4 logistic regression models was estimated for each of the sleep problems. In the first of these models, arthritis was the only independent variable included. In the second model, all of the control variables (sociodemographic, lifestyle factors, other chronic conditions) with the exception of the mental health variables were added. These latter variables for stress and depression were added as a separate block based on previous research indicating that they have particularly strong independent relationships with sleep problems (7, 8, 10, 40, 42–44). In the final model, the pain variable was added. All variables were treated as categorical.

The extent of mediation was quantified as the percent change in the estimated effects (45):

equation image

where ORM− = odds ratio for the independent variable of interest attained from a regression model that does not consider the hypothesized mediator and ORM+ = odds ratio for the same variable attained from a regression model that does include the hypothesized mediator. This method quantifies the change in the OR using 1 as a reference point. For clarity, superscripts were used in the tables to indicate the magnitude of the percentage change in the estimated effects of arthritis and the other chronic conditions due to the addition of pain.

All analyses were performed on the weighted data set using SAS software version 8.2 (SAS Institute, Cary, NC). Due to the complexity of the design of the CCHS, variance for all estimates was determined using the bootstrap resampling method according to Statistics Canada guidelines (36). This method was used to compute 95% confidence intervals (95% CIs) for prevalence estimates and ORs, and was used to determine the significance of differences between proportions. For all analyses, a 2-tailed P value ≤0.05 was used to indicate statistical significance.

RESULTS

The prevalence of arthritis was 16.8% (95% CI 16.5–17.1%), affecting 3.9 million Canadians ages ≥18 years (Table 1). Slightly more than 17% of Canadians reported usually experiencing some level of pain, with 9.2% (95% CI 9.0–9.5%) and 2.9% (95% CI 2.7–3.0%) reporting that pain was moderate and severe, respectively. Prevalence estimates for insomnia symptoms and unrefreshing sleep were 14.1% (95% CI 13.8–14.3%) and 7.7% (95% CI 7.5–7.9%), respectively.

Table 1. Prevalence estimates of arthritis, sleep problems, and pain in the Canadian adult population, Canadian Community Health Survey 2000/2001
CharacteristicNumber (in thousands)Percentage (95% confidence interval)
All subjectsWith arthritis*Without arthritis
  • *

    P < 0.001 for all characteristics comparing subjects with arthritis and those without arthritis.

Arthritis3,91416.8 (16.5–17.1)
Insomnia symptoms3,03214.1 (13.8–14.3)24.8 (24.1–25.6)10.6 (10.3–10.9)
Unrefreshing sleep1,6607.7 (7.5–7.9)11.9 (11.3–12.5)6.1 (5.9–6.4)
Pain    
 Mild1,2425.3 (5.1–5.5)10.7 (10.1–11.3)4.2 (4.0–4.4)
 Moderate2,1559.2 (9.0–9.5)25.3 (24.5–26.0)6.0 (5.8–6.2)
 Severe6682.9 (2.7–3.0)9.6 (9.0–10.3)1.5 (1.4–1.6)

Significantly higher proportions of individuals with arthritis reported pain and both of the sleep problems than did subjects without arthritis (Table 1). Approximately one-quarter of persons with arthritis reported insomnia symptoms, more than twice the proportion of those without arthritis. Approximately 12% of individuals with arthritis reported experiencing unrefreshing sleep, again approximately twice the proportion of those without arthritis.

The prevalence of both insomnia symptoms and unrefreshing sleep increased with increasing pain severity in persons with and without arthritis (Figure 1). However, individuals with arthritis had a higher prevalence of sleep difficulties, particularly insomnia symptoms, at each pain level, with the exception of those with mild pain reporting unrefreshing sleep.

Figure 1.

Prevalence of sleep problems by arthritis status and pain severity in the Canadian adult population, Canadian Community Health Survey 2000/2001.

The results of multivariate analyses for insomnia symptoms are shown in Table 2. The strong bivariate association between arthritis and insomnia symptoms (OR 2.74, 95% CI 2.60–2.89) was reduced substantially by inclusion of the control variables for sociodemographic characteristics, lifestyle factors, and other chronic conditions in model 2. The addition of the mental health variables did not have a large impact on the arthritis OR; the effect of arthritis on insomnia symptoms decreased by 10% compared with model 2. However, both stress (quite a bit/extremely stressful versus not at all/not very stressful: OR 2.11, 95% CI 1.96–2.27) and depression (OR 2.25, 95% CI 2.08–2.44) did have strong independent effects on insomnia symptoms. Inclusion of the pain variables in model 4 further reduced the effect of arthritis on insomnia symptoms by 53%, although the effect of arthritis was still statistically significant (OR 1.26, 95% CI 1.18–1.35). This suggests that pain mediates a substantial amount of the relationship between arthritis and insomnia, although it does not fully account for the relationship.

Table 2. Logistic regression models: predictors of insomnia symptoms in the Canadian adult population, Canadian Community Health Survey 2000/2001*
CharacteristicOdds ratio (95% confidence interval)
Model 1Model 2Model 3Model 4
  • *

    Model 1: arthritis as the only independent variable; model 2: model 1 + sociodemographic characteristics, lifestyle factors, and other chronic conditions; model 3: model 2 + mental health; model 4: model 3 + pain.

  • 40–60% change in the estimated effect due to the addition of pain.

  • Less than 20% change in the estimated effect due to the addition of pain.

  • §

    20–40% change in the estimated effect due to the addition of pain.

  • Change in the estimated effect was not calculated because neurologic conditions were not significantly associated with unrefreshing sleep.

Arthritis2.74 (2.60–2.89)1.61 (1.51–1.71)1.55 (1.46–1.65)1.26 (1.18–1.35)
Age group, years (versus 18–24)    
 25–34 1.05 (0.93–1.18)1.02 (0.91–1.16)1.01 (0.89–1.14)
 35–44 1.11 (0.98–1.26)1.06 (0.94–1.21)1.02 (0.90–1.16)
 45–54 1.38 (1.21–1.56)1.39 (1.22–1.58)1.31 (1.15–1.50)
 55–64 1.29 (1.12–1.48)1.47 (1.27–1.70)1.40 (1.22–1.62)
 65–74 1.12 (0.96–1.31)1.44 (1.23–1.68)1.40 (1.20–1.64)
 ≥75 1.28 (1.08–1.51)1.73 (1.47–2.05)1.69 (1.42–2.00)
Female sex (versus male) 1.23 (1.16–1.31)1.19 (1.13–1.27)1.22 (1.15–1.30)
Education (versus postsecondary graduation)    
 Less than secondary graduation 1.34 (1.25–1.44)1.38 (1.28–1.48)1.34 (1.24–1.43)
 Secondary graduation 1.11 (1.02–1.20)1.15 (1.06–1.25)1.14 (1.05–1.23)
 Some postsecondary 1.14 (1.03–1.27)1.14 (1.03–1.26)1.13 (1.02–1.25)
Income (versus highest)    
 Lowest 1.39 (1.26–1.55)1.35 (1.21–1.50)1.26 (1.13–1.40)
 Lower middle 1.27 (1.16–1.39)1.26 (1.14–1.38)1.21 (1.10–1.33)
 Upper middle 1.09 (1.01–1.18)1.10 (1.02–1.19)1.08 (1.00–1.17)
 Not stated 1.11 (0.99–1.24)1.14 (1.01–1.27)1.10 (0.98–1.23)
Marital status (versus married/common-law)    
 Widowed/separated/divorced 1.22 (1.14–1.31)1.13 (1.05–1.21)1.13 (1.05–1.21)
 Single 1.16 (1.07–1.26)1.11 (1.02–1.22)1.11 (1.02–1.21)
Body mass index (versus acceptable weight)    
 Insufficient weight 1.00 (0.90–1.11)0.97 (0.87–1.08)0.96 (0.86–1.07)
 Some excess weight 0.98 (0.91–1.06)0.98 (0.91–1.06)0.98 (0.91–1.06)
 Overweight 1.08 (1.01–1.15)1.08 (1.01–1.15)1.07 (1.00–1.14)
Alcohol intake (versus nondrinker)    
 Regular drinker 1.00 (0.93–1.08)1.01 (0.94–1.08)1.05 (0.97–1.13)
 Occasional drinker 1.00 (0.93–1.08)1.01 (0.94–1.09)1.03 (0.95–1.11)
Smoking status (versus never smoker)    
 Current smoker 1.50 (1.41–1.60)1.39 (1.30–1.48)1.36 (1.27–1.45)
 Former smoker 1.12 (1.05–1.20)1.10 (1.03–1.18)1.11 (1.04–1.19)
Physical activity index (versus active)    
 Moderate 0.97 (0.89–1.04)0.95 (0.88–1.03)0.94 (0.87–1.02)
 Inactive 1.07 (1.00–1.14)1.02 (0.95–1.09)0.98 (0.91–1.05)
Chronic conditions (present versus absent for each condition)    
 Allergy 1.16 (1.10–1.23)1.12 (1.05–1.18)1.11 (1.05–1.18)
 Cardiovascular conditions 1.35 (1.26–1.45)1.31 (1.22–1.40)1.27 (1.18–1.36)
 Respiratory conditions 1.37 (1.27–0.49)1.31 (1.20–1.42)1.26 (1.16–1.37)
 Digestive conditions 1.61 (1.47–1.76)1.47 (1.34–1.61)1.33 (1.21–1.46)§
 Migraine 1.88 (1.74–2.03)1.67 (1.54–1.81)1.56 (1.44–1.69)
 Diabetes 1.35 (1.21–1.50)1.33 (1.19–1.48)1.29 (1.15–1.44)
 Cancer 1.61 (1.38–1.87)1.51 (1.29–1.76)1.43 (1.23–1.68)
 Neurologic conditions 1.07 (0.86–1.33)1.01 (0.80–1.27)0.96 (0.76–1.20)
 Other musculoskeletal 1.74 (1.65–1.84)1.60 (1.52–1.70)1.31 (1.23–1.39)
 Other chronic conditions 1.44 (1.34–1.55)1.37 (1.27–1.47)1.28 (1.18–1.38)§
Stress (versus not at all/not very stressful)    
 A bit stressful  1.20 (1.13–1.28)1.17 (1.10–1.25)
 Quite a bit/extremely stressful  2.11 (1.96–2.27)1.96 (1.82–2.10)
Depressed (versus not depressed)  2.25 (2.08–2.44)2.11 (1.95–2.29)
Pain (versus none)    
 Mild   1.76 (1.60–1.95)
 Moderate   2.22 (2.05–2.40)
 Severe   3.21 (2.83–3.64)

Results similar to those for insomnia symptoms were found in analogous regression models for unrefreshing sleep (Table 3). After controlling for sociodemographic characteristics, lifestyle factors, and other chronic conditions, individuals with arthritis were 1.5 times (95% CI 1.38–1.64) more likely to report unrefreshing sleep than those without arthritis. The addition of the mental health variables in model 3 reduced the effect of arthritis on unrefreshing sleep by 12%, and both stress and depression had strong independent effects, with ORs of ∼2. The inclusion of the pain variable in the final model had a large impact on the arthritis/unrefreshing sleep relationship, reducing the effect of arthritis by 64% (OR 1.16, 95% CI 1.06–1.27) and suggesting that pain partially mediates the relationship.

Table 3. Logistic regression models: predictors of unrefreshing sleep in the Canadian adult population, Canadian Community Health Survey 2000/2001*
CharacteristicOdds ratio (95% confidence interval)
Model 1Model 2Model 3Model 4
  • *

    Model 1: arthritis as the only independent variable; model 2: model 1 + sociodemographic characteristics, lifestyle factors, and other chronic conditions; model 3: model 2 + mental health; model 4: model 3 + pain.

  • 60–85% change in the estimated effect due to the addition of pain.

  • Less than 20% change in the estimated effect due to the addition of pain.

  • §

    20–40% change in the estimated effect due to the addition of pain.

  • 40–60% change in the estimated effect due to the addition of pain.

  • #

    Change in the estimated effect was not calculated because neurologic conditions were not significantly associated with unrefreshing sleep.

Arthritis2.02 (1.88–2.18)1.50 (1.38–1.64)1.44 (1.32–1.57)1.16 (1.06–1.27)
Age group (versus 18–24)    
 25–34 0.93 (0.81–1.07)0.91 (0.79–1.05)0.89 (0.77–1.02)
 35–44 0.85 (0.73–0.99)0.81 (0.70–0.95)0.78 (0.66–0.91)
 45–54 0.80 (0.68–0.94)0.80 (0.68–0.94)0.75 (0.63–0.88)
 55–64 0.68 (0.57–0.80)0.75 (0.63–0.89)0.71 (0.59–0.84)
 65–74 0.51 (0.42–0.62)0.63 (0.52–0.78)0.61 (0.50–0.75)
 ≥75 0.51 (0.42–0.63)0.67 (0.55–0.83)0.65 (0.53–0.79)
Female sex (versus male) 1.03 (0.95–1.11)1.00 (0.92–1.08)1.02 (0.94–1.10)
Education (versus postsecondary graduation)    
 Less than secondary graduation 1.31 (1.19–1.44)1.35 (1.23–1.49)1.30 (1.18–1.43)
 Secondary graduation 1.16 (1.06–1.28)1.21 (1.10–1.33)1.19 (1.09–1.31)
 Some postsecondary 1.03 (0.91–1.18)1.03 (0.90–1.17)1.02 (0.90–1.17)
Income (versus highest)    
 Lowest 1.14 (1.00–1.30)1.09 (0.95–1.24)1.00 (0.88–1.14)
 Lower middle 1.09 (0.97–1.22)1.07 (0.96–1.21)1.03 (0.91–1.15)
 Upper middle 1.05 (0.95–1.16)1.06 (0.96–1.17)1.03 (0.93–1.14)
 Not stated 1.22 (1.07–1.41)1.25 (1.09–1.44)1.20 (1.05–1.38)
Marital status (versus married/common-law)    
 Widowed/separated/divorced 1.18 (1.07–1.29)1.09 (0.99–1.20)1.08 (0.99–1.19)
 Single 1.02 (0.92–1.14)0.99 (0.89–1.10)0.98 (0.88–1.10)
Body mass index (versus acceptable weight)    
 Insufficient weight 1.04 (0.91–1.18)1.01 (0.88–1.15)1.00 (0.87–1.14)
 Some excess weight 1.01 (0.90–1.12)1.00 (0.90–1.12)1.01 (0.90–1.13)
 Overweight 1.07 (0.98–1.16)1.07 (0.98–1.16)1.06 (0.97–1.15)
Alcohol intake (versus nondrinker)    
 Regular drinker 0.81 (0.73–0.88)0.81 (0.74–0.89)0.84 (0.76–0.92)
 Occasional drinker 0.95 (0.85–1.05)0.96 (0.86–1.06)0.97 (0.88–1.08)
Smoking status (versus never smoker)    
 Current smoker 1.48 (1.35–1.63)1.38 (1.25–1.52)1.35 (1.22–1.49)
 Former smoker 1.05 (0.96–1.14)1.03 (0.94–1.12)1.04 (0.95–1.14)
Physical activity index (versus active)    
 Moderate 0.96 (0.87–1.06)0.95 (0.85–1.05)0.94 (0.84–1.04)
 Inactive 1.35 (1.23–1.48)1.30 (1.18–1.42)1.24 (1.13–1.36)
Chronic conditions (present versus absent for each condition)    
 Allergy 1.15 (1.07–1.24)1.11 (1.03–1.19)1.10 (1.02–1.19)
 Cardiovascular conditions 1.29 (1.18–1.40)1.24 (1.14–1.35)1.19 (1.10–1.30)§
 Respiratory conditions 1.26 (1.14–1.39)1.20 (1.08–1.33)1.15 (1.04–1.27)§
 Digestive conditions 1.42 (1.26–1.60)1.30 (1.15–1.47)1.15 (1.02–1.31)
 Migraine 1.71 (1.56–1.88)1.52 (1.38–1.68)1.40 (1.27–1.55)§
 Diabetes 1.42 (1.25–1.61)1.40 (1.23–1.59)1.34 (1.18–1.53)
 Cancer 1.50 (1.18–1.90)1.41 (1.11–1.78)1.32 (1.04–1.68)§
 Neurologic conditions 1.15 (0.87–1.53)1.10 (0.81–1.48)1.04 (0.78–1.40)#
 Other musculoskeletal 1.64 (1.52–1.77)1.51 (1.39–1.63)1.22 (1.12–1.32)
 Other chronic conditions 1.47 (1.34–1.62)1.39 (1.27–1.53)1.29 (1.17–1.42)
Stress (versus not at all/not very stressful)    
 A bit stressful  1.16 (1.06–1.27)1.13 (1.03–1.24)
 Quite a bit/extremely stressful  2.07 (1.87–2.29)1.91 (1.72–2.11)
Depressed (versus not depressed)  1.90 (1.73–2.08)1.75 (1.60–1.93)
Pain (versus none)    
 Mild   1.66 (1.46–1.89)
 Moderate   2.17 (1.97–2.39)
 Severe   3.65 (3.18–4.20)

All of the other chronic conditions included in the analyses were also significantly related to insomnia symptoms (Table 2) and unrefreshing sleep (Table 3), with the exception of neurologic conditions. Migraine demonstrated the strongest associations. Pain appeared to play a role in mediating the effects of other musculoskeletal conditions similar to that of arthritis. It was also a substantial mediator of the effects of digestive conditions.

DISCUSSION

Arthritis and sleep difficulties are common problems in the adult population. Slightly more than 1 in 6 adult Canadians reported having arthritis in 2000/2001, whereas approximately 1 in 7 reported insomnia symptoms. There has been relatively little research focusing on sleep problems specifically in individuals with arthritis, particularly among those with noninflammatory forms of the disease. This study has demonstrated that Canadians with self-reported arthritis, the vast majority of whom have osteoarthritis, report insomnia symptoms and unrefreshing sleep at approximately twice the rate of the rest of the population. Our findings also suggest that much of these sleep difficulties are due to pain, the chief symptom of all types of arthritis.

Sutton et al (8) examined the association of insomnia with an array of potential explanatory factors using data from the 1991 Canadian General Social Survey and found significant independent associations with rheumatic disorders and pain. Our findings are consistent with the work of Sutton et al in that we similarly found that a significant association between arthritis and insomnia symptoms persisted after adjustment for pain, as it did for unrefreshing sleep. However, we also demonstrated that the addition of pain to regression models for both insomnia symptoms and unrefreshing sleep greatly reduced the estimated effect of arthritis. This finding indicates that a substantial amount of the relationship between arthritis and sleep problems is mediated by pain. Adjustment for pain reduced the effect of arthritis on unrefreshing sleep and insomnia symptoms by 64% and 53%, respectively. This suggests that considerable improvements in sleep could be achieved by better managing pain in persons with arthritis.

In contrast to our findings, Wilcox et al (22) found that although statistically significant, pain explained little variance in sleep disturbance in their cohort of older adults with knee pain. However, they hypothesized that this was due to the fact that everyone in the study cohort had to report pain on most days of the week to meet the study's inclusion criteria. An advantage of using a large population-based health survey such as the CCHS is that it includes individuals with the full range of symptom severity.

We found that pain only partially mediated the effect of arthritis on sleep problems. Furthermore, even individuals with arthritis who had no pain reported insomnia symptoms in significantly higher proportions than those without arthritis (17.1% versus 9.4%). It is not clear what other symptoms or factors associated with arthritis may influence sleep. The regression analyses presented here controlled for sociodemographic characteristics, lifestyle factors, other chronic conditions, and mental health, including both stress and depression. ORs for arthritis in fully adjusted regression models were still significant at 1.26 for insomnia symptoms and 1.16 for unrefreshing sleep. It would be interesting to determine if medication use explains some of these remaining associations. Unfortunately, we did not have sufficient data to examine this.

Pain is a symptom of many of the conditions we included as control variables, and not surprisingly, pain also appeared to play a substantial mediating role in the effects of several of these conditions. One of the strongest mediating effects of pain was observed for other musculoskeletal conditions (fibromyalgia, back problems). Adjustment for pain reduced the effect of musculoskeletal conditions on unrefreshing sleep and insomnia symptoms by 57% and 48%, respectively. It should be noted that at least some of the individuals reporting back problems are likely to actually have arthritis. As for our findings for arthritis and other musculoskeletal conditions, pain generally appeared to play a somewhat larger role in mediating the effects of chronic conditions on unrefreshing sleep than on insomnia symptoms.

The independent effects of depression and the highest level of stress considered were larger than the effects of the physical conditions included in our analyses. However, it is reasonable to hypothesize that these factors may also be mediators of the effects of chronic conditions on sleep problems. In the case of arthritis, they appeared to be only minor mediators. Stress and depression actually appeared to play larger mediating roles for the control conditions, including other musculoskeletal disorders. In the case of migraine and particularly allergies, these mental health variables were larger mediators than pain. This finding is somewhat surprising in the case of migraine, but may be related to the episodic nature of the pain associated with this condition as opposed to one's usual experience as captured by the CCHS. Mental health also appeared to be about as large a contributing factor as pain for the associations of cancer, respiratory, and cardiovascular problems with sleep problems.

Our findings are consistent with research on sleep problems not specific to arthritis. This includes the strong independent associations found with pain (8, 9, 15, 24, 25) as well as the associations found with the other chronic conditions studied (7–10). Consistent with other studies, we also found that women were significantly more likely than men to report insomnia symptoms (8, 33, 42, 43). Our findings of strong independent effects for stress and depression are also in accordance with previous findings (7, 8, 10, 40, 42–44).

There are several limitations of this study. First, all of the data were based on self report, and we were unable to distinguish between types of arthritis. Unfortunately, the CCHS did not include questions on pain experienced at rest or questions on the cause(s) of pain. Pain assessed in this manner cannot be strictly attributed to a particular condition. Including other causes of pain would result in overestimating the role of arthritis pain in mediating the relationship between arthritis and sleep problems. Also, there is debate as to the proper method to quantify and test the statistical significance of mediation (46). Most of the work in this area has been based on continuous rather than binary outcomes, as considered in our study. Finally, the cross-sectional design of the CCHS precludes causal conclusions because it does not allow one to determine the direction of the relationship between pain and sleep problems. It is likely that the relationship is bidirectional to some extent, because sleep disturbances have been shown to decrease the pain threshold (47). A major strength of this study is the large and nationally representative sample of the CCHS and its high response rate of 84.7%, which enable generalization of study findings to the noninstitutionalized Canadian population ages ≥18 years.

The results of this study indicate that sleep difficulties affect a significant proportion of individuals with arthritis. The findings also suggest that treatment of the pain associated with arthritis could significantly improve sleep in these patients. Because many individuals never discuss their arthritis or sleep problems with a physician (48, 49), education about the availability of effective treatments for both of these disorders is warranted. Given the aging of the population and the increases in arthritis prevalence that are expected (29, 30), it is important that the public be encouraged to seek treatment for these conditions to reduce the impact on the population as a whole.

Acknowledgements

The authors thank Drs. Susan Bondy and Edward Adlaf for their input regarding data analysis and preparation of this manuscript.

Ancillary