Depressive symptoms and the risk of arthritis: A survival analysis using data from the osteoarthritis initiative.

BACKGROUND: Some studies investigated the association between depressive symptoms and arthritis; however, no longitudinal studies have documented the relationship between developing depressive symptoms and the risk of arthritis. Therefore, this study evaluated whether the development of depressive symptoms was associated with an elevated risk of arthritis.
MATERIALS AND METHODS: A survival analysis using Cox regression models was applied to osteoarthritis initiative data obtained over 6 years from adults (n = 3,662) aged ≥45 years at baseline. Developing depressive symptoms was defined using the 20-item Center for Epidemiologic Studies Depression scale (cutoff 16 points) between baseline and 1 year. Arthritis was defined answering "yes" to the following self-reported question: "Did the doctor say you developed arthritis since the last clinic visit about 1 year ago?" over the 6-year follow-up period.
RESULTS: The hazard ratios for developing arthritis were 3.51 (95% confidence interval [CI] = 2.32-5.29) and 2.03 (95% CI = 1.45-2.85) for men and women, respectively, as compared to those who did not develop depressive symptoms. There was a significantly (χ 2 = 73.672, P < 0.0001) lower survival probability at each time point throughout the study among men and women who developed depressive symptoms.
CONCLUSION: In both men and women, developing depressive symptoms increased the risk of arthritis, and the survival probability decreased at each time point. Copyright:

INTRODUCTION

Depression is a leading global cause of disability,[12] affecting 350 million people worldwide.[2] Depression might develop because of environmental (e.g., medical illness, obesity, stress, and poor sleep) or genetic factors that alter the immune and inflammatory process.[3] Many studies have provided evidence of the association between several genes related to immunoinflammation and depression.[45] These symptoms are strongly associated with cardiovascular disease, diabetes, chronic obstructive pulmonary disease, and chronic musculoskeletal disorders like arthritis;[67] however, the associations remain heterogeneous.[8] Moreover, most studies examining depression and arthritis comprised small sample sizes.[9]

Arthritis is a general medical condition; it comprises over 100 distinct diseases and conditions.[10] Arthritis is a major and growing public health problem with a sizeable impact on individuals' functional capacity and the ability to perform activities of daily living.[11121314] It is also associated with severe pain,[15] psychological distress,[16] and depression.[1718] There are approximately 750,000 hospitalizations[19] and 36 million outpatient visits annually attributed to arthritis.[20] It has been estimated that about 67 million adults will have self-reported doctor-diagnosed arthritis by the year 2030.[10]

There is an ample body of research on arthritis and its associated factors.[182122232425] Some of these studies demonstrated a relationship between depressive symptoms and arthritis.[182425] However, this relationship remains unclear. For example, Rathbun et al.[25] examined the transient impact of depressive symptoms on the longitudinal change of rheumatoid arthritis (RA) disease activity. The results demonstrated that depressive symptoms temporally influence the evolution of RA disease activity, and the magnitude was dependent on the time of symptoms' onset. Further, a cross-sectional general population study estimated the association between depressive symptoms and physical diseases in Switzerland.[7] The results showed that those with depressive symptoms had a higher chance of developing a physical illness like arthritis than did those who did not. Moreover, a recent study reported that higher scores for depressive mood and lower scores for social support were longitudinally not associated with the development of arthritis in patients with seropositive arthralgia, although they predict increased musculoskeletal symptoms, such as pain, aches, and stiffness.[26] Another longitudinal study suggested that chronic arthritis-related pain and depressive symptoms frequently co-occur in late life.[27] Notably, these studies differ from that of the present study in methodology, study population, terms, and definitions of depressive symptoms and arthritis. Thus, the goal of this study was to assess this relationship. It was hypothesized that developing depressive symptoms would increase the risk of developing arthritis.

MATERIALS AND METHODS

The current study used data from the osteoarthritis initiative (OAI), a publicly and privately funded multicenter, longitudinal, observational study examining the onset and progression of knee osteoarthritis. The OAI study enrolled men and women between February 2004 and May 2006 from four clinical sites in the United States (Baltimore, MD; Pittsburgh, PA; Pawtucket, RI; and Columbus, OH). An overview of this study is described elsewhere.[28] This study received ethical approval from the Committee on Human Research of the Institutional Review Board at the University of California, San Francisco and its affiliates (approval number: FWA00000068). All participants provided written, informed consent before participating.

In this longitudinal study, the approach taken to analyze the datasets is illustrated in Figure 1. Specific datasets were used: 0.2.2, 1.2.1, 3.2.1, 5.2.1, 6.2.1, and 8.2.1. Participants (n = 3,662) who were aged ≥45 years at baseline and did not already have depressive symptoms were followed over 6 years. The participants were divided into two groups: those (n = 153) who had depressive symptoms but did not have arthritis between baseline and 1-year follow-up and those (n = 3509) who had neither depressive symptoms nor arthritis. Participants who were diagnosed with arthritis between the 1- and 6-year follow-up periods were compared to determine whether having depressive symptoms was associated with an elevated risk of developing arthritis.

Figure 1

Analysis of the risk of arthritis over time in participants aged ≥45 years who did not already have depressive symptoms

Depressive symptoms' status at baseline and during the 1-year follow-up period was assessed using the self-reported, 20-item Center for Epidemiologic Studies Depression Scale (CES-D). This scale captures the frequency of feelings and behaviors over the past 7 days and is measured on a four-point scale (0 = rarely or none of the time to 3 = most or all the time). The scale items were summed; total scores ranged from 0 to 60, with higher scores indicating greater depressive symptoms.[29] A cutoff score of 16 was used: depressive symptoms if the score ≥16 and no depressive symptoms if the score was <16.[3031] The scale demonstrated excellent validity in the current sample.[32]

Arthritis was defined answering “yes” to the following self-reported question over the 6-year follow-up period: “Did the doctor say you developed arthritis since the last clinic visit about 1 year ago?” Similar questions were used in another national survey[33] and epidemiological studies[34] with adequate sensitivity and specificity.[35]

According to the available literature, the following potential covariates were included in the analyses: age (years), sex (female/male), race (non-Caucasian/Caucasian), education (≤ primary school/≥ high school), marital status (married/unmarried), annual income (36] and body mass index (BMI; kg/m2) was used to determine obesity.[37]

A flowchart of the current study sample is shown in Figure 2. There were 4,796 participants in the OAI study. Participants who already had depressive symptoms (n = 480) or arthritis (n = 654) at baseline were excluded from analyses; therefore, the final sample size was 3,662. Excluded participants (n = 1,134) were significantly more likely than included participants to be younger (59.0 years), unmarried (34%), live alone (22%), and obese (BMI = 30 kg/m2) than were those who were included in this study.

Figure 2

Flowchart of the study sample

As a part of the survival analysis, interval censoring was considered by undertaking periodic assessment (such as the annual interview of respondents in the OAI study), and the event of interest (in this case, arthritis) is known to occur in the interval between two assessments. Therefore, those who participated in every year (up to 6 years) were included in analyses. A Kaplan–Meier estimate of survival function (survival being determined as not developing arthritis) was initiated to investigate how the unadjusted survivor function for the two groups (those who had depressive symptoms between baseline and 1-year follow-up and those who did not have depressive symptoms) changed over time. Whether possible confounding variables such as age, sex, race, marital status, education, annual income, healthcare coverage, comorbidity, and BMI in the 1-year follow-up period were predictive for the risk of developing arthritis was examined by univariate analysis using Chi-square tests. The risk of arthritis for the two groups was computed using a Cox regression model to show the hazard function. The model was adjusted for age, sex, race, marital status, education, annual income, healthcare coverage, comorbidity, and BMI in the 1-year follow-up period. The analysis was performed separately in men and women to determine the sex differences in having depressive symptoms and the risk of arthritis. All data processing and analyses were performed with Statistical Analysis Software version 9.3 (SAS Institute Inc., NC, USA) for Windows.

RESULTS

Between baseline and the 1-year follow-up, 153 participants developed depressive symptoms. Participants' characteristics stratified by the group are shown in Table 1. Between baseline and the 1-year follow-up, those who developed depressive symptoms were significantly unmarried, lived alone, obese, and had less annual income than did those who had neither depressive symptoms nor arthritis [Table 1].

Table 1

Participants’ characteristics stratified by group

Characteristics	No depressive symptoms or arthritis, (n=3509; 96%)	Depressive symptoms only (n=153; 4%)	P
Age (years), mean±SD	63.4±9.2	63.7±9.4	0.72
Sex, n (%)
Male	1527 (43)	55 (36)	0.06
Female	1982 (57)	98 (64)
Race, n (%)
Caucasian	2871 (82)	132 (86)	0.16
African American/Asian/other	638 (18)	21 (14)
Marital status, n (%)
Married	2433 (69)	94 (61)	0.04
Unmarried/divorced/widowed	1076 (31)	59 (39)
Education, n (%)
≤Primary school	491 (14)	25 (16)	0.41
≥High school	3018 (86)	128 (84)
Annual income (US dollars), n (%)
<50000	1193 (34)	65 (42)	0.03
≥50,000	2216 (66)	88 (58)
Household composition, n (%)
Live alone	715 (20)	43 (28)	0.02
Live with others	2794 (80)	110 (72)
Healthcare coverage, n (%)
Yes	3456 (98)	146 (95)	0.03
No	53 (2)	13 (5)
Charlson Comorbidity Index, n (%)
0	2264 (75)	112 (73)	0.60
≥1	751 (25)	41 (27)
Obesity, n (%)
Obese	1212 (46)	66 (55)	0.05
Nonobese	1425 (54)	54 (45)

SD – Standard deviation

Results of the univariate analysis are shown in Table 2. Between the 1- and 6-year follow-up, there was no significant difference in the likelihood of men and women to develop arthritis. Compared to those who were married or lived with others, a higher percentage of the unmarried/divorced/widowed participants and those who lived alone had a significantly higher risk of developing arthritis. Participants with less annual income and healthcare coverage and those who were obese were significantly more likely to develop arthritis compared to their counterparts.

Table 2

Results of the univariate analysis

Characteristics	n (%)	Risk of developing arthritis over time (%)	Univariate χ2 (P)
Depressive symptoms
Had depressive symptoms between baseline and 1 year	153 (4)	22	30.75 (<0.001)
Did not had depressive symptoms	3509 (96)	18
Sex
Male	1582 (43)	20	3.42 (0.06)
Female	2080 (57)	24
Race
Caucasian	3003 (82)	19	1.97 (0.16)
African American/Asian/other	659 (18)	34
Marital status
Married	2527 (69)	21	4.27 (0.04)
Unmarried/divorced/widowed	1135 (31)	23
Education
≤Primary school	516 (14)	26	0.67 (0.41)
≥High school	3146 (86)	21
Income per year (US dollars)
<50,000	1258 (34)	24	4.68 (0.03)
≥50,000	2404 (66)	21
Household composition
Live alone	758 (21)	24	5.33 (0.02)
Live with others	2904 (79)	21
Healthcare coverage
Yes	3602 (98)	22	8.54 (0.03)
No	60 (2)	13
Charlson Comorbidity Index
0	2376 (75)	21	0.28 (0.60)
≥1	792 (25)	25
Obesity
Obese	1278 (46)	22	3.77 (0.05)
Nonobese	1479 (54)	18

The Kaplan–Meier survival curve for the risk of arthritis between 1- and 6-years follow-up for both groups is shown in Figure 3. There was a significantly (χ2 = 73.672, P < 0.0001) lower survival probability at each time point throughout the study period among men and women who developed depressive symptoms only between baseline and 1 year, as compared to those who had neither depressive symptoms nor arthritis.

Figure 3

Probability of arthritis over time in initially arthritis-free men and women who were having depressive symptoms at 1 year follow-up and in those who were not having depressive symptoms. Dep = 0 sex 1 = 1, men who did not have depressive symptoms; dep = 0 sex 1 = 2, a woman who did not have depressive symptoms; dep = 1 sex 1 = 1, men who had depressive symptoms; dep = 1 sex 1 = 2, a woman who had depressive symptoms

Cox regression models in Table 3 show the effect of developing depressive symptoms and the risk of developing arthritis between 1- and 6-years in men and women. Compared to both men and women who had neither depressive symptoms nor arthritis, men and women who had developed depressive symptoms between baseline and 1 year were significantly more likely to have arthritis. Table 3 also shows that being younger and African American/Asian/other decreased the hazard ratio for the risk of arthritis for women. In both sexes, obesity increased the hazard ratio of the risk of arthritis, whereas having no healthcare coverage increased the hazard ratio of the risk of arthritis in men only.

Table 3

Cox regression model estimating the hazard function of developing arthritis over time, stratified by sex

Parameter	Men			Women
	Parameter estimate	Hazard ratio	P	Parameter estimate	Hazard ratio	P
Arthritis
Participants who had depressive symptoms at 1-year follow-up	1.25	3.51	<0.01	0.71	2.03	<0.01
Age	0.06	1.02	0.02	−0.01	0.99	0.04
African American/Asian/other	−0.43	0.65	0.06	−0.31	0.73	<0.01
Unmarried/divorced/widowed	0.35	1.43	0.11	−0.07	0.93	0.32
≤Primary school	0.04	1.05	0.75	−0.11	0.89	0.10
<50,000 US dollars	−0.23	0.79	0.34	−0.09	0.91	0.11
Live alone	−0.31	0.73	0.19	0.07	1.05	0.38
No healthcare coverage	1.51	4.53	0.02	0.26	1.31	0.15
Comorbidities, ≥1	0.12	1.23	0.12	0.12	1.16	0.03
Obese	0.02	1.03	<0.01	0.02	1.02	<0.01

DISCUSSION

To the best of our knowledge, this was the first study to evaluate whether developing depressive symptoms is associated with an increased risk of developing arthritis. The findings support the hypothesis that developing depressive symptoms elevated the risk of arthritis in both men and women as compared to those who did not have depressive symptoms. Both men and women who developed depressive symptoms between baseline and the 1-year follow-up had a lower probability of not having arthritis at each time point throughout the study period. Furthermore, obesity and having no healthcare coverage increased the risk of arthritis in both sexes if they had developed depressive symptoms between baseline and 1 year.

Previous findings showed that low education, disease activity, and disability were associated with developing depressive symptoms, and they were also associated with having arthritis.[3839] Moreover, previous studies reported that individuals who had depression and arthritis displayed more significant functional impairment and worse health-related quality of life (HRQoL) compared to those who did not have depression but had arthritis.[2140] Research also demonstrated a temporal association between depression and pain in patients with arthritis; however, the real relationship between developing depressive symptoms and the risk of arthritis might have been masked in the literature.[414243]

The present findings cannot be compared with previous literature directly because previous studies mainly focused on improved depression symptoms and arthritis.[44] These studies also focused on specific outcomes such as functional well-being, pain level, and disability and did not focus on the association between developing depressive symptoms and the risk of arthritis. As noted in the introduction, a cross-sectional general population study that was conducted in Switzerland reported that depressive symptoms were associated with arthritis after controlling for age, sex, education, occupation, and household income. However, the present study focused on the development of depressive symptoms; therefore, the current study findings may help guide clinicians on the effectiveness of depression treatment.

The current study has much strength including the use of longitudinal study design and the validated and widely used 20-item CES-D. Further, the survival analyses employed data from a publicly and privately funded, multicenter, ongoing, longitudinal prospective cohort study. However, some study limitations need to be acknowledged when interpreting the findings. Social desirability bias and recall bias are plausible, as doctor-diagnosed arthritis was self-reported by respondents. This study also did not assess if participants received treatment or medication for depression.

Despite these limitations, the current findings warrant that treating depressive symptoms may decrease the longitudinal risk of developing arthritis. For example, antidepressant treatment was found to reduce depressive symptoms in the general population,[4546] and it may also improve their HRQoL. Studies have also reported that the combination of psychotherapy and antidepressants is more effective in treating significant depressive symptoms.[4748] Moreover, the Improving Mood-Promoting Access to Collaborative Treatment trial reported that patients with arthritis receiving a combination of antidepressants and psychotherapy showed a reduction in depressive symptoms, pain, and improved HRQoL.[44] Hence, addressing the effects of depression treatment in individuals with coexisting depressive symptoms and arthritis is warranted in the community and population-based studies.

CONCLUSION

In both men and women, developing depressive symptoms increased the risk of arthritis. There was a significantly lower survival probability at each time point throughout the study among men and women who developed depressive symptoms. Future studies are needed to examine the factors that cause developing depressive symptoms and the mechanism behind the association between depressive symptoms and arthritis.

Financial support and sponsorship

The author extends his appreciation to the Research Centre, College of Applied Medical Sciences, and the Deanship of Scientific Research at King Saud University for supporting this research.

Conflicts of interest

There are no conflicts of interest.