Article Text

Download PDFPDF

Extended report
The prospective association between psychological distress and disease activity in rheumatoid arthritis: a multilevel regression analysis
  1. Cécile L Overman1,
  2. Ercolie R Bossema1,
  3. Henriët van Middendorp1,
  4. Leoniek Wijngaards-de Meij2,
  5. Suzanne MM Verstappen3,
  6. Marcia Bulder1,
  7. Johannes WG Jacobs4,
  8. Johannes WJ Bijlsma4,
  9. Rinie Geenen1,4
  1. 1Department of Clinical and Health Psychology, Utrecht University, Utrecht, The Netherlands
  2. 2Department of Methodology and Statistics, Utrecht University, Utrecht, The Netherlands
  3. 3Arthritis Research UK Epidemiology Unit, The University of Manchester, Manchester, UK
  4. 4Department of Rheumatology and Clinical Immunology, University Medical Center Utrecht, Utrecht, The Netherlands
  1. Correspondence to Cécile L Overman, Department of Clinical and Health Psychology, Utrecht University, Heidelberglaan 1, PO Box 80.140, 3508 TC Utrecht, The Netherlands; c.l.overman{at}uu.nl

Abstract

Background Cross-sectional associations suggest a mutual impact of disease activity and psychological distress in rheumatoid arthritis (RA), but a prospective association has not been established.

Objective To examine concurrent and prospective associations between psychological distress and disease activity.

Methods Patients with RA (N=545, disease duration ≤1 year, age 18–83 years, 69% female, 64% rheumatoid factor (RF) positive) were monitored for 5 years. The Thompson joint score and erythrocyte sedimentation rate were assessed every 6 months. Depressed mood and anxiety were measured every 12 months. Multilevel regression analysis was used. RF positivity, age and female sex were included as covariates.

Results Concurrent levels of psychological distress and disease activity were positively associated (p≤0.04). Prospectively, depressed mood was associated with disease activity levels 6 months later (p≤0.04). The Thompson joint score was associated with psychological distress levels 6 months later (p≤0.03) and also with an increase in depressed mood over the subsequent 6 months (p=0.02). No other significant prospective associations were found (p≥0.07).

Conclusions Psychological distress and disease activity are positively associated when measured at the same time as well as when measured 6 months apart. While some support was found for the idea that a higher level of disease activity is a risk factor for an increase in psychological distress, the results do not support the notion that psychological distress is a risk factor for future exacerbation of disease activity.

Statistics from Altmetric.com

Psychological distress, comprising depressed mood and anxiety,1 is common among patients with rheumatoid arthritis (RA). Depressed mood is reported among 13–20% of patients with RA,2,,6 with some studies reporting rates even up to 42%.7,,10 These percentages are approximately twice as high as in the general population.11 Anxiety is also prevalent in patients with RA, often accompanied by comorbid depression.9 12 The proportion of patients with RA scoring above the clinical threshold for anxiety is two to four times greater9 12 than in the general population.11 13 As similar physiological stress systems are involved in both psychological distress and inflammation, a mutual impact of psychological distress and disease activity in RA could be possible.14

Psychological distress may affect disease activity by altering the functioning of the immune, endocrine and central nervous systems.15 Psychological stress and distress have been associated with overactivation of the hypothalamic–pituitary–adrenal axis and the sympathetic–adrenal–medullary axis16,,20 and with increased levels of pro-inflammatory cytokines,15 21,,25 possibly fuelling the activation of these axes.18 The enduring activation of stress axes can cause the immune system to downregulate its glucocorticoid receptors, thereby becoming desensitised to the inhibitory actions of the axes' hormonal products.16 19 26 27 This may allow an increase in pro-inflammatory cytokines and inflammation. Furthermore, psychological distress can indirectly affect physiological mediators and disease activity by influencing health behaviours, such as smoking, physical exercise, doctor visits, adherence and sleep quality.28 29 Psychological distress is thus expected to increase disease activity in patients with RA.

In addition to the negative effects psychological distress may have on disease activity, disease activity may also cause or increase psychological distress. A direct causal link between inflammatory processes and depressed mood is suggested by research showing an increase in depressive-like behaviour after the administration of pro-inflammatory cytokines in animals,30,,32 and of depression during immune treatment of cancer and hepatitis in humans.33,,35 It has been suggested that this negative effect can be abolished by blocking receptors for these cytokines,36 or by pretreatment with antidepressants.32 33 RA disease activity may also indirectly contribute to psychological distress because of pain and disability10 37 and the unpredictable and sometimes uncontrollable course of the disease.29 37 A period of more severe disease activity may thus increase the level of psychological distress in patients with RA.

Although many patients with RA attribute their disease flares to psychological stress,14 38 research so far has not conclusively demonstrated a relation between patients' psychological status and disease activity in prospective studies. Some studies did find a concurrent association between depression or anxiety and objective measures of disease activity (eg, erythrocyte sedimentation rate; ESR),2 39 40 but others did not.10 41 In other patient groups or in the general population, some studies were able to link depression or depressed mood to higher levels of pro-inflammatory cytokines such as interleukin 6,15 21 22 24 25 42 interleukin 123 and tumour necrosis factor alpha,24 but others did not find such an association.19 43 44 Due to these inconsistent results and because of the cross-sectional nature of most of the studies, the prospective link between psychological distress and immune or disease activity remains uncertain. As yet, only one study employed a longitudinal mixed model design to examine the possible determinants of depression and anxiety in patients with RA over a period of 10 years.40 Depression was concurrently associated with high disease activity and anxiety was, unexpectedly, associated with low disease activity, which emphasises the need to examine depressed mood and anxiety as separate dimensions of psychological distress. This hallmark study did not examine the prospective associations between psychological distress and disease activity.

The aim of the current study was to examine both the concurrent and prospective associations between psychological distress and disease activity in patients with a recent diagnosis of RA. To achieve this goal, our study used multilevel regression analysis to examine clinical and psychological data of a large cohort of RA patients over a period of 5 years after diagnosis. We expected to find concurrent and also prospective associations between psychological distress and disease activity levels. Moreover, we hypothesised that higher than average levels of psychological distress would be associated with a subsequent increase in disease activity. Also, higher than average levels of disease activity were hypothesised to be associated with a subsequent increase in psychological distress.

Methods

Participants

The study population comprised 545 patients with a recent diagnosis of RA, recruited between 1990 and 2002 in six rheumatology outpatient clinics of the Utrecht Foundation for Rheumatism Research in the Utrecht region, The Netherlands, who were enrolled in a prospective trial comparing the effectiveness of several drug treatment strategies. The study was approved by the ethical review boards of all participating hospitals. Details of the study have been described.45 Briefly, inclusion criteria were a diagnosis of RA as defined by the revised criteria of the American Rheumatism Association,46 disease duration of 1 year or less and age 17 years or greater. Patients diagnosed with psychiatric disorders, co-morbid diseases or drug usage, possibly interfering with one of the therapeutic strategies, and fertile female patients not taking adequate contraceptive measures or who were breastfeeding were excluded from the study.

Psychological and clinical assessments

Psychological distress was assessed at baseline, before randomised treatment with one of four medication strategies and then annually until 5 years after inclusion.45 Psychological distress was measured with the anxiety and depressed mood-scales of the impact of rheumatic diseases on general health and lifestyle (IRGL) questionnaire.47 The anxiety scale consists of 10 items (range 10–40) derived from the Spielberger state–trait anxiety inventory.48 49 The depressed mood scale consists of six items (range 0–24), derived from a questionnaire by Zwart and Spooren.50 These items of the IRGL are scored on a four-point Likert scale with higher scores representing more psychological distress. Scores of 6 or greater for depressed mood and 23 or greater for anxiety are considered to indicate distress according to a norm group of 362 RA patients.47 The reliability and validity of the IRGL scales are satisfactory.47

Disease activity was assessed at baseline and afterwards every 3 months in the first 2 years and once every 6 months in the remaining 3 years. The used measures of disease activity were ESR in mm/h (Westergren) and the Thompson articular index, a weighted score including both swollen and painful joints (range 0–534).51

Statistical analysis

Multilevel regression analysis was applied to examine concurrent and prospective associations between disease activity and psychological distress variables. This method of analysis accommodates missing values obviating the need to eliminate all data from a particular case, and takes account of both within and between-subject variance of the hierarchical data (ie, repeated measures within RA patients). Data were analysed using HLM 6.0852 and SPSS 16.0.2.

The score distributions of both the disease activity variables and the psychological distress variables were positively skewed. After logarithmic transformation the score distributions of ESR and anxiety and their residuals reached normality. The score distributions of the Thompson joint score and depressed mood, both with one third of the scores being zero and their residuals remained positively skewed after transformation. To examine if the score distribution of these variables would affect the results, we performed the multilevel analyses twice, once with the original continuous scores as dependent variables and once with categorised scores of the Thompson joint score and depressed mood, and compared the results. The results did not differ substantially, allowing us to leave the Thompson joint score and depressed mood as continuous dependent variables in the final analyses. However, it was expected that the Thompson joint scores and depressed mood scores would indicate a dichotomous distinction between having and not having joint problems or a depressed mood and a linear effect of the scale when having these problems. Therefore, when the Thompson joint score or depressed mood acted as independent variables, a dichotomous variant (0 is for zero and 1 is for scores exceeding zero) of these variables was added to the model to examine both effects.

Before multilevel analyses, the time trend of the psychological distress and disease activity variables across 5 years was examined. The testing of polynomial terms revealed that for every model the best fit was acquired with an equation specifying both a linear and quadratic trend. Covariates were determined with correlational analysis before multilevel analyses. Rheumatoid factor (RF) positivity, female sex and age at baseline were significantly correlated with most disease activity and psychological distress variables and were entered as covariates in the multilevel analyses. Type of medication used, radiographic damage score,53 education level and marital status were not significantly correlated with psychological distress and disease activity variables, and were not included in multilevel analyses.

In multilevel analyses, concurrent and prospective association models were tested for each of the four dependent variables: ESR, Thompson joint score, depressed mood and anxiety. In the concurrent association models, the dependent variable was predicted by time variables, covariates and the psychological distress or disease activity variables of interest, measured at the same time as the dependent variable. In the prospective association models, the dependent variable was predicted by time variables, covariates and the distress or disease activity variables of interest, measured 6 months before. Every model was tested both with and without the previous measurement (6 or 12 months before) of the dependent variable. With this addition of the previous measurement of the dependent variable to the model, the other variables in the prospective association models predict a change of the dependent variable, instead of its absolute value. The three types of models were built up step by step: from an initial model, including only time variables (linear and quadratic); to a second model, examining the effects of the covariates; to a final model, including all possible explanatory variables. The significance of the effects of included variables was determined using the Wald test. In all statistical tests, significance levels were set at p<0.05.

Results

Patient characteristics and missing data

Table 1 shows the patient characteristics. Of the 545 patients with RA, 426 (78%) completed 60–100% (>20) of the 34 psychological and clinical assessments. Of the 119 patients with less than 60% of the assessments completed, 86 patients completed between 40% and 60%, and 33 patients completed 18–40%. Patients who completed less than 60% did not significantly differ from the other patients with respect to baseline scores of psychological distress and disease activity (p≥0.14) or patient characteristics (p≥0.41) with the exception of a more frequent positive RF status (68% vs 54%, p=0.004) in the group with the fewest missing values.

Description of time trends

Table 1

Baseline characteristics of the 545 patients with RA

Both psychological distress and disease activity scores declined over the years, with a pronounced decline in the first year and a more gradual decline in the years thereafter (table 2). Psychological distress was common in our patient sample. At baseline, 45% of the depressed mood scores and 36% of the anxiety scores reflected distress. Approximately 25% of the patients still experienced psychological distress after 5 years (table 2).

Table 2

Trends for the psychological distress and disease activity variables during the first 5 years after being diagnosed with RA

Concurrent association analyses

First, we examined concurrent associations between psychological distress and disease activity variables. The final models are shown in table 3 for depressed mood and anxiety and in table 4 for the Thompson joint score and ESR. With respect to trend effects, scores on all dependent variables decreased across the 5-year interval, with most often a significant linear (time) and quadratic (time2) trend. With respect to the covariates, RF positivity was associated with anxiety and ESR, female sex was associated with more severe depressed mood and anxiety, and an older age was associated with a lower Thompson joint score and a higher ESR. After taking into account the time trends and covariates, higher scores on both disease activity variables were associated with more severe depressed mood and anxiety and vice versa.

Table 3

Concurrent association analyses for psychological distress

Table 4

Concurrent association analyses for disease activity

Prospective association analyses of levels

Subsequently, prospective (ie, time-lagged) associations between psychological distress and disease activity variables were examined. Previous scores on the dependent variable were not taken into account. Therefore, these analyses reflect associations between absolute levels of psychological distress and disease activity measured 6 months apart. Four associations between disease activity and psychological distress variables were significant. Scores exceeding zero on the depressed mood scale were associated with a higher Thompson joint score (B=10.7163 (SE 4.9783), p=0.03) and a higher ESR 6 months later (B=0.0353 (SE 0.0176), p=0.04). Furthermore, a higher Thompson joint score was associated with higher levels of depressed mood (B=0.0028 (SE 0.0013), p=0.03) and anxiety (B=0.0001 (SE 0.00004), p=0.02) 6 months later. The other associations in these analyses were not significant (p≥0.07).

Prospective association analyses of change

Finally, prospective associations between psychological distress and disease activity variables were examined, but now the previous score of the dependent variable was taken into account. Therefore, these analyses examined the association between psychological distress or disease activity variables and a change in the dependent variable during the subsequent 6 months (instead of its absolute value 6 months later). As was to be expected, the association between the two repeated measurements of the dependent variable was highly significant in all models (p<0.001). One model analysing the prospective association of change was significant: a higher Thompson joint score was associated with an increase in depressed mood during the subsequent 6 months (B=0.0034 (SE 0.0014), p=0.02). The other associations in these analyses were not significant (p≥0.21).

Discussion

The hypothesis of this study was that psychological distress and disease activity would be both concurrently and prospectively associated. The results partly supported our hypotheses. Psychological distress and disease activity were indeed concurrently associated. In addition, more severe depressed mood (but not anxiety) was associated with higher disease activity 6 months later, and a higher Thompson joint score (but not a higher ESR) was associated with more severe psychological distress 6 months later. Moreover, only a higher Thompson joint score was associated with an increase in depressed mood across 6 months. The other hypothesised associations of psychological distress or disease activity with subsequent change were not confirmed.

The prevalence of depressed mood (45%) and anxiety (36%) in our sample at baseline was largely similar to the prevalence found in previous studies.2 3 7,,10 After 5 years, the scores of a quarter of our sample still reflected depressed mood and anxiety. The high number of patients with scores indicative of psychological distress together with the approximately similar decline of disease activity and psychological distress in the first 5 years after diagnosis both suggest an effect of disease activity on psychological distress.

In multilevel regression analyses, we first took into account the influence of time trends and patient characteristics.54 Consistent with results of previous studies,55 56 a positive RF status, female sex and an older age were associated with several disease activity and psychological distress variables. Our findings support the notion that in patients with more severe depressed mood, disease activity is probably greater, not only at the same time but also several months later, and that in patients with more swollen and painful joints, psychological distress is probably greater at the same time and later on. These findings, however, do not support causal or predictive conclusions. Although a causal influence of the one variable on the other cannot be excluded, it is possible that extraneous variables explain the observed concurrent and prospective associations.

In the prospective analyses predicting change, a higher Thompson joint score was associated with an increase in depressed mood during the next 6 months, but the other associations between disease activity variables and a subsequent change in psychological distress were not significant. Notably, psychological distress variables were not associated with a change in disease activity. Therefore, our data do not support the notion that psychological stress may cause disease flares. Research so far did indicate some weak evidence that past stress may have a negative effect on RA disease activity,14 and may increase the inflammatory response in healthy people.57 However, based on our analyses, the popular14 38 and in physiological theory grounded15 belief that psychological distress may fuel inflammatory activity should be modified or should otherwise be confirmed in future prospective research.

A major strength of this study is that it used a longitudinal design like only one other study.40 Our study is innovative with respect to its prospective analyses of levels and change in psychological distress and disease activity. However, our study also has limitations. First, we used existing data with relatively long intervals between assessments. More frequent monitoring, for example, every 3 months during 5 years, will increase the chance of finding disease flares and substantial mood changes. Although costly and taxing for patients, such a design would be feasible because regular follow-up examinations every 1–3 months have been recommended recently for optimal treatment.58 Second, while the prospective analyses of absolute levels are very liberal in suggesting predictive associations, the prospective analyses of change are very conservative. This predictive association between absolute levels is found when anxiety is relatively high both during and 6 months after an episode of having painful and swollen joints. Then, although the Thompson joint score does not predict a still further increase in anxiety, the subsequent anxiety level in patients with a high Thompson joint score is still higher than normal. Third, because we only included the Thompson joint score and ESR as a reflection of disease activity, our findings do not generalise to cytokine and hypothalamic–pituitary–adrenal axis functioning.

To conclude, the results suggest a positive association between psychological distress and disease activity in RA patients when measured at the same time and when measured 6 months apart. There is some minor support that a higher level of disease activity forecasts a further increase in psychological distress 6 months later. However, our results do not support the notion that psychological distress is a risk factor for the future exacerbation of disease activity.

Acknowledgments

The authors would like to thank all participating rheumatologists and research nurses of the Utrecht Foundation for Rheumatism Research (SRU) for their contribution to this paper.

References

View Abstract

Footnotes

  • Funding This study was funded by a grant from the Faculty of Social and Behavioural Sciences of Utrecht University and by grants from the Dutch Arthritis Association.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.