The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
There has been limited characterization of the burden of anxiety and depression, especially the former, among US adults with arthritis in the general population. The study objective was to estimate the prevalence and correlates of anxiety and depression among US adults with doctor-diagnosed arthritis.
The study sample comprised US adults ages ≥45 years with doctor-diagnosed arthritis (n = 1,793) from the Arthritis Conditions Health Effects Survey (a cross-sectional, population-based, random-digit–dialed telephone interview survey). Anxiety and depression were measured using separate and validated subscales of the Arthritis Impact Measurement Scales. Prevalence was estimated for the sample overall and stratified by subgroups. Associations between correlates and each condition were estimated with prevalence ratios and 95% confidence intervals using logistic regression models.
Anxiety was more common than depression (31% and 18%, respectively); overall, one-third of respondents reported at least 1 of the 2 conditions. Most (84%) of those with depression also had anxiety. Multivariable logistic regression modeling failed to identify a distinct profile of characteristics of those with anxiety and/or depression. Only half of the respondents with anxiety and/or depression had sought help for their mental health condition in the past year.
Despite the clinical focus on depression among people with arthritis, anxiety was almost twice as common as depression. Given their high prevalence, their profound impact on quality of life, and the range of effective treatments available, we encourage health care providers to screen all people with arthritis for both anxiety and depression.
Depression, a well-documented comorbidity among people with chronic diseases, including arthritis (1–6), can exacerbate functional disabilities (1), affect adherence to treatment (2), and be a barrier to self-care and self-management behaviors (3, 4). Despite its high prevalence in the general population and equal or stronger incapacitating effects on physical function (5), anxiety is often underrecognized and undertreated (6). Until recently, anxiety has been regarded largely as a comorbidity of depression, but its independent effects, including its role as a potential risk factor for depression, are increasingly recognized (7). Anxiety and depression are generally more common among people with arthritis than in the general population (8, 9), and interplay independently and synergistically with clinical outcomes such as pain and disability (10, 11).
Many studies examining the occurrence of these conditions among people with arthritis have studied depression only, have studied people with one type of arthritis (e.g., rheumatoid arthritis), or were clinic-based, not population-based, samples (4, 12–17). International population-based studies identifying major depression using the World Health Organization Composite International Diagnostic Interview (WHO-CIDI) indicate prevalences ranging from 2.2% (Japan) to 19% (Ukraine; in the US, age ≥18 years = 7–9% and age 54–65 years = 11%) (9, 18, 19). The prevalence of anxiety disorders among people with arthritis also varies internationally; a survey of 18 countries in the early 2000s found that people with arthritis were consistently more likely than those without arthritis to have anxiety disorders. Additionally, across the countries surveyed, US adults with arthritis had the first or second highest prevalence of each of the 4 specific anxiety disorders examined (generalized anxiety disorder [6%], social phobia [8%], agoraphobia/panic disorder [3%], and posttraumatic stress disorder [5%] ). The 2001–2003 US National Comorbidity Study Replication found that each of the 6 anxiety disorders measured was more common among people with arthritis (the prevalence among people with arthritis ranged from 1% for agoraphobia to 6% for social phobia; 10% reported a specific phobia) (9). All anxiety disorders were measured using the WHO-CIDI. A study of Australians ages ≥18 years with arthritis indicated that one-quarter had experienced an anxiety disorder in the past 12 months (19). A comparable estimate for the US is lacking.
To better characterize the burden of anxiety and depression among US adults with arthritis, we estimated the prevalence of each in a national sample of adults ages ≥45 years with self-reported doctor-diagnosed arthritis. As depression and anxiety can be highly responsive to clinical treatment (20), better recognition and detection of these conditions is a necessary first step to reducing the burden of these mental health conditions among people with arthritis. Therefore, we also examined the correlates of each condition to identify the sociodemographic, clinical, and other characteristics that can help health care providers (HCPs) identify those who are likely to have anxiety and/or depression.
Significance & Innovations
One-third of US adults with arthritis ages ≥45 years reported having at least one of anxiety and/or depression.
Although there is considerable clinical and research focus on depression among people with arthritis, anxiety was more common than depression (31% and 18%, respectively).
A distinct profile of those with anxiety and/or depression did not emerge in the multivariable models, indicating that all people with arthritis should be screened for anxiety and depression.
Only half of the respondents with anxiety and/or depression had sought help for their mental health condition in the past year, suggesting there is an unmet need for treatment of mental health conditions among people with arthritis.
MATERIALS AND METHODS
We analyzed data from the Arthritis Conditions Health Effects Survey (ACHES), a cross-sectional, random-digit–dialed telephone survey. ACHES, conducted by the Centers for Disease Control and Prevention (CDC), was designed to be representative of the civilian noninstitutionalized US population of adults ages ≥45 years with arthritis and/or chronic joint symptoms (21).
Telephone numbers were selected from a proprietary list linking phone numbers to US Census blocks. These numbers were first partitioned into 7 strata based on census-estimated percentages of Hispanics and non-Hispanic blacks associated with each block. To ensure geographic representation and reduce sample variation, the numbers were then sorted by census division and metropolitan status (i.e., urban versus rural counties) in each stratum. Numbers were then selected with equal probability within each of the 7 strata, with oversampling in those strata with high percentages of Hispanics and non-Hispanic blacks (21).
To maximize response rates, letters were mailed to the addresses associated with potential residential phone numbers at least 2 weeks prior to the first call. Trained interviewers called each number to identify 1) residential numbers and 2) household members who were ages ≥45 years and had doctor-diagnosed arthritis or chronic joint symptoms. We restricted our analysis to respondents with doctor-diagnosed arthritis (n = 1,793), who were identified with a response of “yes” to: “Have you ever been told by a doctor or other health professional that you have some form of arthritis, rheumatoid arthritis, gout, lupus, or fibromyalgia?” Given this method of case ascertainment, in this article, arthritis refers to people with arthritis and other rheumatic conditions. Interviews were conducted in English (or Spanish as needed) from June 2005 to April 2006. All residents in each household who met the inclusion criteria were eligible. Participants were compensated with a 100-minute prepaid long-distance phone card or a $5 donation to the Arthritis Foundation. Among eligible households, Council on American Survey Research Organizations response and completion rates were 51% and 86%, respectively (i.e., those with at least 1 age-eligible resident). Response and completion rates for eligible people in the household were 31% and 75% for the first household participant identified as eligible and 16% and 80% among other eligible respondents in the same household, respectively (21). The ACHES protocol was approved by the CDC Institutional Review Board.
Using a standardized questionnaire, interviewers collected information on sociodemographic characteristics and medical and psychosocial aspects of arthritis, including physical functioning and limitations, work effects, knowledge and attitudes about arthritis, self-management and self-care behaviors, and mental health. ACHES methods are described in detail elsewhere (21–23).
Anxiety and depression were assessed using the Arthritis Impact Measurement Scales (AIMS). Originally developed for use in longitudinal trials of rheumatoid arthritis to detect changes over time (24), AIMS was subsequently validated for use in studies of other arthritis types (25).
The AIMS anxiety and depression module comprises 12 questions (6 for anxiety and 6 for depression) and measures the frequency of symptoms (rating of 0–6; see Supplementary Appendix A, available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131a) in the past month. Following a validation study that reported that an AIMS depression subscale score of ≥4 was comparable to the Center for Epidemiologic Studies Depression Scale (CES-D) score cutoff for probable depression (16), multiple subsequent studies using AIMS to quantify the occurrence of anxiety and depression among people with arthritis have used ≥4 as the threshold for both conditions (4, 14, 16). Consistent with this, for each condition, we calculated the average subscale value and defined the presence of the condition as a mean value of ≥4.
Following the series of AIMS questions, respondents reported help-seeking behaviors (“During the past 12 months, have you sought help for stress, depression, or problems with emotions?”) and from whom this help was sought (i.e., “Did you seek help from any of the following: Family or friends? A self-help group or support group? A priest, minister, rabbi, or other religious counselor? A therapist or counselor? A physician?”).
We examined variables representing 3 domains of interest: 1) sociodemographic characteristics (i.e., age, sex, race/ethnicity, highest educational attainment, and current employment status) to develop a profile of affected individuals, 2) arthritis symptoms and physical function that can increase the likelihood of arthritis and depression, and 3) potentially modifiable health and self-management behaviors that are associated with arthritis symptoms and mental and physical function.
Respondents reported the severity of each of 3 symptoms (i.e., joint pain or aching, stiffness, and fatigue) in the past 7 days using a 0–10 scale (where 0 = no symptoms and 10 = most severe) and the number of days that they had experienced joint pain or aching in the past 7 days.
The survey included the Short Form 36 (SF-36) physical functioning subscales and a series of questions about the degree to which arthritis interfered with routine activities (e.g., spending time with family and friends, errands/shopping, and household chores) (21, 26). The SF-36 and interference variables were measured using Likert-style response scales (a lot, a little, or not at all), which we dichotomized (a lot versus a little/not at all). We analyzed individual items rather than the SF-36 score, since these individual items correspond to questions clinicians might use when asking about patients' physical function. Initial analysis showed moderate to strong correlations among the function (r = 0.4–0.8) and interference variables (r = 0.5–0.8). Therefore, we restricted analyses to 3 function variables (difficulty in walking several hundred feet, washing or bathing, and bending, kneeling, or stooping) and 1 interference variable (difficulty with errands and shopping). These physical function and interference variables have been associated with loss of independence among people with arthritis (27) and loss of independence has been correlated with depression in at least one previous study (4).
Physical activity was measured with 6 validated questions on frequency and duration of participation in leisure-time activities of moderate or vigorous intensity (28). Categories were based on the total number of minutes of physical activity each week, where 1 minute of vigorous exercise was equivalent to 2 minutes of moderate activity: recommended (≥150 minutes), insufficient (10–149 minutes), or inactive (<10 minutes) (29). Using a 0–10 scale (where 0 = no confidence and 10 = highest confidence), the participants rated their confidence on 3 aspects of self-management: belief that self-management education (SME) courses would help to manage arthritis or joint symptoms, ability to manage arthritis or joint symptoms, and ability to engage in moderate physical activity at least 3 times/week.
We examined the prevalence of anxiety and depression (to estimate the public health burden) and then stratified by independent variables to identify potential correlates for logistic regression models. For each outcome, we estimated the associations with independent variables with unadjusted and multivariable adjusted prevalence ratios (PRs) and 95% confidence intervals (95% CIs) (30). Last, we determined the proportion of respondents with anxiety and/or depression who had sought help for mental health conditions in the past 12 months and estimated the likelihood, with PRs and 95% CIs, of help seeking for anxiety, depression, and both conditions.
Sampling weights, based on the distribution of US adults ages ≥45 years with arthritis in the 2003–2005 National Health Interview Survey (NHIS) (21), were applied in all analyses to infer estimates to the national population of civilian noninstitutionalized adults ages ≥45 years with doctor-diagnosed arthritis. Statistical significance was defined using 2 criteria: nonoverlapping 95% CIs and a Wald's test (test of statistical significance of variable overall in the model) P value of less than or equal to 0.05. Analyses conducted in SAS, version 9.1 and SUDAAN, version 10 (Research Triangle Institute) accounted for the complex survey design.
Among adults with arthritis, 30.5% (11.5 million) reported anxiety, 17.5% (6.6 million) reported depression, and 14.7% (5.5 million) reported both. Most respondents with depression also had anxiety (84%), whereas half of those with anxiety also had depression (49.5%) (Figure 1).
Prevalence and correlates of anxiety.
At least half of the people in the following 6 subgroups reported anxiety: unemployed, unable to work, or disabled (62%); respondents who reported “a lot” of difficulty with bathing or dressing (63%), “a lot” of interference with errands or household chores in the past 7 days (51%), or that their arthritis or joint symptoms affected whether they worked for pay (52%); severe fatigue in the past 7 days (50.2%); and no confidence in their ability to engage in moderate physical activity at least 3 times/week (56%) (Table 1).
Table 1. Associations of sociodemographic, disease, and physical function and health behaviors, self-management, and self-efficacy with each of anxiety and depression: prevalence and unadjusted PRs*
Prevalence (95% CI)
Unadjusted PR (95% CI)
Prevalence (95% CI)
Unadjusted PR (95% CI)
PR = prevalence ratio; 95% CI = 95% confidence interval; HCP = health care provider.
Alaska Native/American Indian, Asian, and Native Hawaiian or other Pacific Islander.
Disabled and unable to work were combined because of small sample sizes.
Homemakers and students were combined because of small sample sizes.
Categories were: recommended (≥150 minutes), insufficient (10–149 minutes), or inactive (<10 minutes), where 1 minute of vigorous exercise was equivalent to 2 minutes of moderate activity.
Have you ever taken a self-management education course?
Confidence that a self-management education course would help manage symptoms
Confidence in ability to manage arthritis or joint symptoms
Confidence in ability to engage in moderate physical activity at least 3 times/week
Currently being treated by doctor or HCP for arthritis or chronic joint symptoms?
Almost all of the independent variables were significantly associated with anxiety in unadjusted models (Table 1). In the multivariable model, anxiety was significantly higher among respondents who were ages 45–64 years (PR 1.7; referent: age ≥65 years), reported severe joint pain in the past week (PR 1.9; referent: no pain), and reported good (PR 1.4) or poor/fair self-rated health (PR 1.6; referent for self-rated health: excellent/very good) (Table 2). Anxiety was also higher among respondents who had no or moderate confidence in their ability to engage in moderate physical activity at least 3 times/week (PRs 1.5 and 1.3, respectively; referent: high confidence) (Table 2). After multivariable adjustment, respondents who were overweight or obese were 20% less probable to report anxiety (PR 0.8) (Table 2).
Table 2. Sociodemographic, physical, and psychosocial characteristics associated with anxiety and depression: statistically significant multivariable adjusted PRs*
Anxiety, PR (95% CI)
Depression, PR (95% CI)
The multivariable adjusted model comprised all but 1 variable examined in unadjusted models. This table shows statistically significant associations only; PRs for all variables examined in this multivariable model are shown in Supplementary Table 1 (available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658). PR = prevalence ratio; 95% CI = 95% confidence interval.
Severity of each of fatigue, joint stiffness, and joint pain was highly correlated (r = 0.7). To reduce collinearity, only the latter was included in the multivariable models; joint pain was selected because it is generally the most modifiable symptom among people with arthritis, and was also strongly associated with both anxiety and depression (see Table 1). Because severity of pain was highly correlated with number of days in the past week with pain, aching, or stiffness (r = 0.6), only severity of pain was included in the multivariable models.
Arthritis symptoms and physical health and function
Depression prevalence was highest among those who reported “a lot” of difficulties bathing or dressing themselves (48%) (Table 1). At least one-third of people in the following subgroups reported depression: unemployed, unable to work, or disabled (45%); Hispanics (37%); severe fatigue in the past 7 days (36.3%); and respondents whose arthritis or joint symptoms affected whether they worked for pay (33%), who had “a lot” of interference with errands or household chores in the past 7 days (36%), who had no or a low level of confidence in their ability to manage their arthritis or joint symptoms (34% and 45%, respectively), or who had no confidence in their ability to engage in moderate physical activity at least 3 times/week (42%).
Several correlates of depression were observed in unadjusted models (Table 1). In multivariable models, depression was significantly more common among those who were ages 45–64 years (PR 1.6; referent: age ≥65 years), reported low confidence in their ability to manage their arthritis or joint symptoms (PR 2.3), and had only moderate confidence in their ability to engage in moderate physical activity at least 3 times/week (PR 1.5) (Table 2).
Prevalence of help seeking for anxiety and depression in the past year.
Help seeking was highest among people with both conditions (57.1%) and lowest among those with anxiety only (45.1%; people with depression = 51.3%) (data not shown). Respondents were most likely to have sought help from their doctor (82–83%), followed by family and friends (45–46%); therapist/counselor (43–46%); priest, minister, rabbi, or other religious counselor (15–16%); and self-help or support groups (11–13%; sum exceeds 100% because respondents sought help from multiple sources).
Among those with anxiety, depression, or both, more than half (55%) of all respondents had not sought help in the past year. This was only slightly improved when limited to those who were currently seeing a doctor or HCP for their arthritis or joint symptoms (46%) (Table 3). Across all sociodemographic groups, 36–66% of respondents had not sought help (Table 3), and among those who were currently seeing a doctor or HCP for their arthritis or joint symptoms, the range was 21–61%. Similar patterns were observed among those with each of anxiety, depression, and both conditions (data not shown).
Table 3. Percentage with anxiety, depression, or both in the past month who have not sought help in the past 12 months for “stress, depression, or problems with emotions”
Currently being treated for arthritis or joint symptoms
Alaska Native/American Indian, Asian, and Native Hawaiian or other Pacific Islander.
Disabled and unable to work were combined because of insufficient sample sizes.
Homemakers and students were combined because of insufficient sample sizes.
One-third of respondents with arthritis had anxiety, depression, or both. Anxiety was almost twice as common as depression (31% and 18%, respectively), and virtually all respondents with depression also had anxiety. Approximately half of the respondents with anxiety and/or depression had sought help for their mental health condition in the past year. Most of the statistically significant associations observed in the multivariable analysis were moderately strong; nevertheless, a distinct profile of characteristics of those with anxiety and depression did not emerge.
We found that anxiety was more common than depression in this population-based sample of people with arthritis, a pattern that has been observed in clinic-based samples (4, 12–14, 16). Anxiety can elicit independent and at least equally debilitating effects as depression (5, 31, 32). Despite this and the high prevalence of anxiety in previous studies, few influential rheumatology texts (33) mention anxiety, suggesting that the magnitude and impact of this problem among people with arthritis are underrecognized (31).
The US Preventive Services Task Force recommends screening of all adults for depression when systems are in place to ensure accurate diagnosis, effective treatment, and appropriate followup; there is insufficient evidence supporting universal screening when effective treatment and followup are unavailable (34). HCPs do not appear to routinely and systematically screen for anxiety or depression (35). We believe that screening of all people with arthritis for anxiety and depression is indicated when the same conditions (e.g., effective treatment) are met. Although there are differences in the treatment for depression and anxiety, pharmacotherapy and cognitive–behavioral therapy are considered effective methods of treating depression and many forms of anxiety (20). Furthermore, the appropriate treatment of depression among people with arthritis can lead to clinically significant reductions in pain, improved functional outcomes, and continued compliance with antidepressant use for at least 1 year following treatment (36). For this reason, treating existent mental health conditions should be regarded as a fundamental part of managing arthritis symptoms. Both anxiety and depression were common among respondents who were currently being seen by an HCP for their arthritis and joint symptoms, but approximately half of those reporting anxiety or depression had not sought help for their mental health conditions in the past year. HCP visits for management of arthritis symptoms may be an opportunity to screen for and treat anxiety and depression.
The relationship across anxiety, depression, and pain is complex, with evidence that each condition acts indepen dently and synergistically as a risk factor and outcome for each other (10, 11, 37). Furthermore, each is an independent determinant of disability, further complicating this interrelationship. Similar to depression, anxiety can persist and worsen if untreated (7); the importance of addressing anxiety is emerging only now (18). Kessler et al propose that anxiety may be underrecognized and undertreated because it can be an appropriate response to stressful life events and circumstances, and therefore treatment may not seem indicated (7). In fact, psychosocial distress among people with arthritis may signal the presence of other threats to their well-being, such as economic insecurity (the prevalence of each of anxiety and depression in our study was higher among those who were unemployed, disabled, or unable to work) (38). In at least one study, anxiety was an even stronger predictor of functional limitations than depression among people with arthritis (5), and it can be an obstacle to the behavioral changes associated with reducing pain and depression, such as physical activity. Minor and Brown examined the efficacy of an exercise program for people with arthritis, and both high baseline anxiety and depression scores were independently associated with an increased risk of not exercising at 3, 9, and 18 months postintervention (32).
Ideally, treatment and management of anxiety and depression include simultaneous clinical and self-management interventions. There are multiple inexpensive, convenient, and evidence-based self-management interventions for anxiety and depression that complement clinical care. Aerobic exercise is an effective treatment for mild to moderate depression and is associated with reductions in anxiety (39); some strength training activities may also elicit the same effects for depression (29). For those wanting guidance on safely exercising, community-based physical activity programs (e.g., Walk With Ease, EnhanceFitness) teach people with arthritis strategies to reach recommended levels of physical activity without exacerbating symptoms or worsening disease (40). SME interventions (e.g., Chronic Disease Self-Management Program, Arthritis Self-Management Program) have been proven to lead to reductions in anxiety and depression (40). Therefore, another strategy for HCPs is recommending participation in physical activity and evidence-based SME interventions (http://www.cdc.gov/arthritis/interventions.htm). Recommendation from an HCP is key; ACHES respondents who had received a recommendation from their HCP to attend an SME class were 18.5 times more likely to report attending one than those without a recommendation (41).
AIMS anxiety and depression subscales were used to define these conditions. AIMS is one of the most commonly used and reported instruments for the study of anxiety and depression among people with rheumatic conditions (4, 13–16). As mentioned previously, although AIMS subscales have not been validated directly in population-based studies, the AIMS depression subscale is strongly correlated (r = 0.81) with the CES-D (42), a population measure of depression symptoms. Increasing levels of AIMS depression and anxiety scores are also associated with lower levels of physical function (measured with the Health Assessment Questionnaire [HAQ]) (4, 13), suggesting construct validity. To our knowledge, the AIMS anxiety subscale has not been validated against any other population-based measure of anxiety. We believe there is construct validity to the anxiety subscale because of the association between anxiety and HAQ physical function scores (4, 13). Also, we found that having sought help in the past 12 months for mental health conditions was strongly associated with both anxiety (PR 3.5, 95% CI 2.9–4.2) and depression (PR 3.4, 95% CI 2.9–4.0) (data not shown), suggesting that the subscales detect mental distress.
The types of anxiety disorders (e.g., generalized anxiety disorder, panic disorder) detected by the AIMS anxiety subscale have not been characterized. Also, because symptoms of anxiety can be a manifestation of depression (43), the proportion of anxiety among respondents attributable to depression is unknown.
Our estimates indicate that the population burden of anxiety and depression among adults with arthritis is substantial, but may be underestimated for several reasons. We used a conservative definition of depression (i.e., the AIMS depression subscale cutoff of “probable” rather than “probable and possible”) (16, 42) that would exclude what a provider might detect and treat. Further, there may be underreporting of the presence and frequency of symptoms if ACHES respondents did not disclose information about symptoms of anxiety and depression because of social stigma concerns.
This study has the following limitations. First, people with arthritis were ascertained by a question on self-reported doctor-diagnosed arthritis and not by examination. A clinic-based validation study found that this question had a high positive predictive value (44). Second, ACHES is a cross-sectional study and it is not known whether the correlates studied are predictors or sequelae of anxiety and depression. Third, several 95% CIs in both unadjusted and multivariable analyses bordered on statistical significance, especially for depression, for which there were fewer people affected. This suggests that there was insufficient power to detect modest statistically significant associations. Last, despite multiple strategies to maximize survey participation, response rates were low, particularly among blacks and Hispanics. A previous analysis indicated that the sociodemographic characteristics of ACHES respondents are similar to adults with arthritis ages ≥45 years in the nationally representative NHIS (22), suggesting that ACHES results are generalizable to the US population.
ACHES is the most comprehensive population-based national survey of US adults with arthritis to date. We found that both anxiety and depression are common among people with arthritis and the prevalence of anxiety was higher than the prevalence of depression. A distinct profile of people with these conditions was not evident because the prevalence of these conditions was relatively high across all of the subgroups. Approximately half of the affected respondents whose arthritis was being treated by an HCP had not sought treatment in the past year for their mental health condition, indicating a missed opportunity for HCP intervention. This is important because HCPs can have a significant impact on reducing the burden of anxiety and depression among people with arthritis through systematic screening for both conditions, treatment based on current standards of care, and their strong influence in recommending physical activity and SME programs to their patients.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Murphy had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Murphy, Sacks, Brady, Hootman, Chapman.
Acquisition of data. Sacks.
Analysis and interpretation of data. Murphy, Hootman, Chapman.
The authors would like to thank the Battelle staff for coordinating all aspects of ACHES, the ACHES steering group's expertise and time in development of the survey, and the ACHES respondents for their participation in this study.