SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Objective

To examine the evolution of psychosocial aspects of health-related quality of life in rheumatoid arthritis (RA) patients, and to identify their predictors.

Methods

All patients within a Swiss RA cohort and a US RA cohort who completed a Short Form 36 (SF-36) scale at least twice within a 4-year period were included. The primary outcome was psychosocial health as measured by the mental component summary (MCS) score of the SF-36. The evolution of this outcome over time was analyzed using structural equation models, which distinguish between the stable, the variable, and the measurement error components of the outcome's variance.

Results

A total of 15,282 patients (48,323 observations) were included. MCS scores were mostly stable over time (between 69% and 75% of the variance was not due to measurement error). The variable component of the SF-36 was mostly due to fluctuations at the moment of measurement and not to a global time trend of psychosocial health. Pain was the most important predictor of both the stable and variable components of psychosocial health, explaining ∼44% of the observed psychosocial health variance.

Conclusion

This large cohort study demonstrates that pain is the most important predictor of a patient's psychosocial health in RA patients. This suggests that physicians should place greater emphasis on pain management.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Rheumatoid arthritis (RA) is a chronic inflammatory disease characterized by progressive joint destruction that has a major impact on health-related quality of life (HRQOL) (1). Indeed, patients with RA experience joint pain, extraarticular manifestations, and functional limitations that often lead to permanent disability. Moreover, they are at greater risk of experiencing emotional disturbances (2), such as poor subjective well-being, low self-esteem, and sleep problems; they are also more likely to experience anxiety and depression (3, 4), resulting in an increased risk of suicide (5).

Although disease-modifying antirheumatic drugs (DMARDs) have been associated with a significant reduction of disease progression and functional disability (6–12), the evolution and variability of patients' HRQOL have been less studied (13, 14). Most studies assessing treatment effects have measured quality of life using disease-specific instruments centered on functional disability (7–12), such as the Health Assessment Questionnaire (HAQ) (15–17). Other aspects of HRQOL, such as mental and emotional health, vitality, and social functioning, all representing psychosocial health, are also important, but have been less studied than physical or disease-related outcomes. The lack of focus on patients' psychosocial health may result from the emphasis that current RA management guidelines place on preventing permanent joint damage and physical disability (18, 19).

The objectives of this study were 1) to examine the evolution of psychosocial health in RA patients, 2) to determine its variability over time, and 3) to identify patient- and disease-related predictors of psychosocial health. As pain is a central symptom of RA patients, it may play an important role in HRQOL, including its psychosocial aspects. The American College of Rheumatology has recently appointed a Pain Management Task Force, which stated that insufficient efforts have been devoted to pain management (20). Therefore, disease-related predictors of psychosocial health include disease activity and functional disability, but also self-reported pain. Understanding the relative impact of various aspects of a patient's experience on psychosocial aspects of HRQOL may help physicians to better understand patients' concerns and help patients not only to “do well,” but also to “feel well.”

Significance & Innovations

  • Psychosocial aspects of health-related quality of life (HRQOL) are mostly stable over a 4-year period (approximately 75% of the true variance).

  • Pain was the most important predictor of psychosocial aspects of HRQOL in treated rheumatoid arthritis patients, explaining approximately 44% of the observed variance.

  • Disease activity and functional disability each explained only approximately 23% of the observed psychosocial aspects of HRQOL.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Study population and measurements.

We used 2 longitudinal cohorts of RA patients: the Swiss Clinical Quality Management Program for Rheumatoid Arthritis database (SCQM) and the National Data Bank for Rheumatic Diseases (NDB), which have been described in detail elsewhere (21–23). Patients in both cohorts are enrolled on a continuous basis and assessed at regular intervals for disease activity, RA symptoms, and quality of life. The inclusion criteria for this analysis were a diagnosis of RA and at least 2 Short Form 36 (SF-36) assessments. We included all patients enrolled until the end of March 2009, with a maximum followup of 4 years per patient. Of the 22,995 patients of the combined cohorts, 15,282 fulfilled the inclusion criteria. In both cohorts, patients are typically treated by rheumatologists, either in private practice or in hospital outpatient clinics, with conventional (∼50%) or biologic antirheumatic therapies (∼50%), often in association with low-dose glucocorticoids in 36–41% at baseline (24, 25). Mean age at baseline was 53 and 59 years and 69% and 78% are women (in the SCQM and NDB, respectively) (25). Socioeconomic status (28% and 26% of college graduates, respectively) and baseline levels of patient-reported HRQOL (EuroQol 5-domain, SF-36 mental component summary [MCS], and SF-36 physical component summary [PCS]) were also very comparable between the 2 cohorts (25).

The study's primary outcome was an aspect of HRQOL, psychosocial health, as measured by the MCS score of the SF-36 (26). Both the MCS and PCS were computed using the US population norms, which in a healthy population yield a mean ± SD of 50 ± 10. The SF-36 has been shown to be a reliable and valid generic HRQOL measure for RA patients (13–15). We chose to focus on the MCS in order to best capture the emotional, psychological, and social aspects of patients' experiences. The physical and disease-specific aspects are better assessed by instruments such as the Rheumatoid Arthritis Disease Activity Index (RADAI) for RA disease activity and the HAQ disability index for functional disability. Patients' pain level was assessed using a 10-cm visual analog scale (VAS). Sociodemographic variables (sex, age, education, and living arrangement, i.e., living with someone or alone) as well as disease duration were also recorded.

Analysis.

We used a specific type of structural equation model, the latent state-trait (LST) model (27, 28), to examine the longitudinal evolution and variability of psychosocial health. LST models offer the advantage over other longitudinal models such as the latent growth curve model of allowing the ability to analyze data that show no time trend but only random fluctuations. LST models permit the ability to distinguish various components of psychosocial health, as follows: psychosocial health = stable part (“trait”) + variable part (“state”) + error of measurement, where variable part = global time trend + fluctuations due to moment of measurement.

With LST models, it is possible to decompose the variance of each observed variable into a proportion due to the “true psychosocial health” (i.e., the reliability of the measure) and a proportion due to error of measurement. Similarly, the variance of the “true psychosocial health” can be then separated into a stable part that does not fluctuate over time (the “trait”) and a variable part (the “state”). Finally, the variable part can be further broken down into a part due to the influence of the preceding measure (global time trend) and a part due to the situation at the moment of measurement (fluctuations).

Figure 1 shows the specific model applied to our data. The stable part of psychosocial health is represented by one latent trait variable (Psy-soc health). The variable part due to the moment of measurement (fluctuations) is represented by a second set of latent variables (Year1 to Year4). Since there are 4 MCS measures (one for each year), there are 4 latent occasion-specific variables measuring the same construct (psychosocial health). To take into account the differences in formulation of the 4 subscales, we included a latent variable representing the deviation of the second observed variable (mcs2) from the trait (Psy-soc health dev) (29). With 1 score per year, the influence due to the moment of measurement (e.g., decrease in psychosocial health) and measurement error cannot be distinguished. On the contrary, when the same construct (psychosocial health) is measured at the same time with 2 or more instruments, one may distinguish the random influences of measurement errors and the systematic influences due to the time of measurement, as the yearly influences will impact all scores similarly, whereas measurement error will impact each score differently. Therefore, SF-36 subscales were aggregated in 2 test-half scores for each occasion of measurement (mcs1 and mcs2) (30). In this parceling procedure, each of these scores is comprised of 4 of 8 subscales of the SF-36 (mcs1: mental health, vitality, global health, physical functioning; mcs2: role emotional, social functioning, bodily pain, role physical). The 4 subscales were chosen in order to make each score equally informative (i.e., the sum of the MCS weights is equal across scores) and computed according to the usual MCS weights (26) (for a detailed explanation, see Supplementary Appendix A, available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658). To allow the assessment of the global time trend of psychosocial health, we added an autoregressive structure (i.e., each latent “Year1” to “Year4” variable was regressed on the preceding year) (31, 32).

thumbnail image

Figure 1. Diagram of the structural equation model (latent state trait). Isolated arrows show measurement errors. For better readability, the mcs variables' error was pictured only for mcs1-year1, and the set of variability covariates (i.e., RADAI1, HAQ1 and Pain1) was pictured only for the first year. Psy-soc health = latent-trait variable of psychosocial health; Psy-soc health dev = latent deviation of the second indicator of psychosocial health compared to the first indicator; Year = latent deviation of each year compared to the mean trait of psychosocial health (Psy-soc health); RADAI = Rheumatoid Arthritis Disease Activity Index; HAQ = Health Assessment Questionnaire.

Download figure to PowerPoint

We imposed restrictions on the estimated parameters (e.g., constraining all the error variances of the observed mcs variables to be equal) and used the sample size–adjusted Bayesian information criterion to determine the best-fitting model (see Supplementary Appendix B, available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658). To examine the predictors of stable psychosocial health, we regressed the latent-trait variable on the sociodemographic variables, disease duration, and mean disease activity (RADAI), disability index (HAQ), and VAS pain scores. To examine the predictors of the variable psychosocial health, we regressed the latent occasion-specific variables on the year-specific RADAI, HAQ, and VAS pain scores. All parameters were estimated with Mplus 5.2 using the maximum likelihood estimator. Missing data were imputed by full information maximum likelihood. A sensitivity analysis estimating the parameters separately for each country of origin showed no evidence of effect modification (data not shown).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

The combined database included 15,282 RA patients (11,223 from NDB, 4,059 from SCQM), yielding a total of 48,323 observations (mean number of measures per patient 3.2). Table 1 shows the distribution of patient characteristics by country of origin.

Table 1. Patients' characteristics at baseline by cohort*
 SCQM (n = 4,059)NDB (n = 11,223)
  • *

    Values are the mean ± SD unless otherwise indicated. Percentages reflect the proportion based on nonmissing data. Due to the large sample size, all differences, except living alone (vs. not), were statistically significant. SCQM = Swiss Clinical Quality Management Program for Rheumatoid Arthritis; NDB = National Data Bank for Rheumatic Diseases; RA = rheumatoid arthritis; RADAI = Rheumatoid Arthritis Disease Activity Index; HAQ = Health Assessment Questionnaire; DI = disability index; VAS = visual analog scale; MCS = mental component summary; PCS = physical component summary.

  • Computed using US population norms (mean ± SD 50 ± 10).

Women, no. (%)1,752 (68.7)8,715 (77.8)
Age, years52.3 ± 14.059 ± 12.7
Education, years12.6 ± 2.313.2 ± 3.0
Living alone, no. (%)2,598 (81.2)8,186 (81.4)
RA disease duration, years9.9 ± 9.712.9 ± 10.9
RADAI3.92 ± 2.182.66 ± 1.56
HAQ DI1.06 ± 0.751.01 ± 0.71
Pain VAS6.17 ± 2.357.03 ± 2.26
MCS46.72 ± 12.0450.11 ± 11.07
PCS35.65 ± 10.5336.70 ± 10.77

Based on the LST model, approximately 85% of the total variability in MCS scores was attributable to true differences in psychosocial health (Table 2), and 15% was attributable to measurement error. The stable part of psychosocial health accounted for 69% up to 75% of true variance. This suggests that psychosocial health as measured by the SF-36 does not evolve much over a period of 4 years in RA patients (Table 2).

Table 2. Variance components of the psychosocial health measurement, as assessed by the Short Form 36 MCS score*
MeasureReliabilityStable partVariable part
TotalDue to time trendDue to moment of measurement
  • *

    Stable and variable (total) parts sum to 1. MCS = mental component summary.

MCS year 10.860.690.31
MCS year 20.850.750.250.010.24
MCS year 30.850.720.280.040.24
MCS year 40.850.730.270.030.24

The variable part of psychosocial health represented 25–31% of true variance over the 4 years (Table 2). Most of the variability was attributable to fluctuations due to the moment of measurement (Table 2), and not to a global time trend in psychosocial health. Indeed, the longitudinal evolution of psychosocial health over the years predicted only 1–4% of the psychosocial health assessment 1 year later (Table 2). Expressed differently, this corresponded to autoregression coefficients of 0.18 from the first to second year, 0.31 from the second to third year, and 0.30 from the third to fourth year. These coefficients were not significantly different from 0.

Prediction of the stable part of psychosocial health.

In univariate analyses, pain (measured by the VAS) and disease activity (RADAI) were strongly associated with the stable component of psychosocial health. In contrast, functional disability (HAQ), disease duration, age, education, and living alone (versus not) showed more modest associations with psychosocial health (Table 3). In multivariate analyses, pain and disease activity remained strongly associated with psychosocial health, and the association with functional disability increased. All of the other associations between the stable trait of psychosocial health and the demographic characteristics became weak. Pain was the strongest predictor of the stable part of psychosocial health, both in univariate and in multivariate analysis.

Table 3. Predictors of the stable part of psychosocial health*
CovariatesUnivariate analysisMultivariate analysis
Standardized coefficientPStandardized coefficientP
  • *

    RADAI = Rheumatoid Arthritis Disease Activity Index; HAQ = Health Assessment Questionnaire; DI = disability index; VAS = visual analog scale.

Age0.28< 0.0010.14< 0.001
Education0.24< 0.0010.10< 0.001
Sex0.010.300.000.60
Living alone or together0.04< 0.0010.000.85
Disease duration−0.13< 0.001−0.030.004
RADAI−0.53< 0.001−0.60< 0.001
HAQ DI−0.17< 0.001−0.59< 0.001
Pain VAS−0.78< 0.001−0.82< 0.001

Prediction of the variable part of psychosocial health.

In univariate analyses, yearly measures of disease activity, function, and pain were modestly, but significantly, associated with the variable part of psychosocial health (Table 4). In multivariate analyses, only pain remained a significant predictor of the variable part of psychosocial health, whereas yearly disease activity or function did not.

Table 4. Regression coefficients for predictors of the variable part of psychosocial health (i.e., year specific)*
CovariatesUnivariate analysisMultivariate analysis
Year 1Year 2Year 3Year 4Year 1Year 2Year 3Year 4
  • *

    RADAI = Rheumatoid Arthritis Disease Activity Index; HAQ = Health Assessment Questionnaire; DI = disability index; VAS = visual analog scale.

  • P < 0.001.

  • P > 0.05.

RADAI−0.28−0.21−0.19−0.21−0.08−0.06−0.06−0.06
HAQ DI−0.21−0.17−0.16−0.18−0.09−0.07−0.07−0.07
Pain VAS−0.30−0.27−0.25−0.27−0.24−0.22−0.21−0.22

Overall, pain explained more than 60% (e.g., −0.822 for year 2) of the stable part of psychosocial health and approximately 5% (e.g., −0.222 for year 2) of the variable part of psychosocial health (Figure 2). Altogether, this represents 44% of observed psychosocial health. In contrast, disease activity and functional disability explained 35% and 36%, respectively, of the stable part of psychosocial health, and only a negligible fraction of the variable part. Altogether, this represents 23% and 22%, respectively, of observed psychosocial health.

thumbnail image

Figure 2. Proportion of variance of psychosocial health explained by pain, disease activity (RADAI), and function (HAQ), separately for the stable part and for the variable part of psychosocial health. Because the 3 predictors (pain, RADAI, HAQ) are correlated, the variance explained by these predictors exceeds the variance of the stable part. * = because the proportion of variance explained by the RADAI and HAQ is small, the lined area is not visible; QoL = quality of life; RADAI = Rheumatoid Arthritis Disease Activity Index; HAQ = Health Assessment Questionnaire.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

The ultimate goal of treating a chronic disease is to preserve the patient's quality of life, and not primarily, at least according to patients, to prevent one additional erosion (33). In this study, we examined the longitudinal evolution of patients' psychosocial health, its variability, and the predictors of change in psychosocial health in a very large cohort of patients treated for RA. We found that psychosocial health, as measured by the MCS of the SF-36, is comparable to a healthy population. Moreover, it does not change much over 4 years. Indeed, psychosocial health is mostly stable over time and the variable component reflects essentially year-specific influences and not a consistent longitudinal trend (i.e., progressive deterioration). Longitudinal studies of RA patients have shown that the deterioration of functional disability and the progression of structural joint damage have become very slow in recent years, which may be explained by the widespread use of DMARDs (7, 10, 11, 24, 34, 35). This is consistent with our finding of the stability of psychosocial health over a followup of 4 years. Since treated RA patients have mostly stable psychosocial health, analyses with very long followup times are required to determine whether psychosocial health either improves (15) or worsens (16) gradually over longer periods of time, which was not demonstrable in this study.

Because psychosocial health does not change much over time, treatment strategies should focus on factors that influence the stable part of psychosocial health. Our most interesting finding in this regard is that self-perceived pain was the strongest predictor of the stable part of psychosocial health, more important than disease activity and functional disability. Pain was also the only significant predictor of the variable part of psychosocial health, although to a smaller extent, which suggests that antirheumatic treatments have an insufficient effect on pain relief. Indeed, in this large cohort of patients all treated for their RA and followed by rheumatologists, contrary to psychosocial health, the reported level of pain was considerable, with a mean VAS above 6. These results match survey findings reporting that approximately two-thirds of RA patients have inadequate pain relief (36, 37), even when their disease is considered to be “well-controlled” (37, 38). Furthermore, many patients report that their physician focuses on disease control but is less concerned with pain relief (39). This is further corroborated in long-term followup studies in which patient report of insufficient pain relief remains surprisingly stable, despite parallel improvement in disease control and function (7, 10).

The reasons for high pain levels in RA patients are complex and intricate. On the one hand, some patients may be ambivalent regarding the chronic use of analgesics and wary of their potential side effects (39). On the other hand, rheumatologists' main focus in patients with RA has not been pain relief, but the control of inflammation and the prevention of permanent damage or disability (18, 19). Most rheumatologists would not consider themselves as “pain physicians” in their professional identity (18, 20). Other potential barriers to effective pain management include reluctance to prescribe opioids, inclination for immunologic research over pain research, or inadequate financial compensation (18). Recently, a newly appointed task force of the American College of Rheumatology has reviewed these issues, acknowledging insufficient efforts devoted to pain management and discussing new perspectives in education and research (20). Yet whether better pain management will improve quality of life remains uncertain and cannot be answered by our observational data. Nevertheless, our results are a reminder for clinicians to pay careful attention to patients' symptoms and not to forget to treat arthritis pain, as this could impact patients' well-being more than anything else.

In regard to sociodemographic predictors, previous studies have suggested that the quality of social support of RA patients may play an important role in their quality of life (40). We found that living alone (versus with someone) was weakly associated with psychosocial health, but the association was not confirmed in multivariate analysis. However, we did not measure the quality of social support, which seems to be a stronger determinant of psychosocial health than marital status or living arrangements (41–43).

The main limitation of our study is a followup period limited to 4 years. As previously discussed, given the stability of psychosocial health, a longer followup time would be needed to capture minor time trends. Another limitation we did not measure is the incidence of psychiatric comorbidities, such as anxiety and depression. However, although these specific diagnoses could be confounders of the relationship between RA activity and psychosocial health, they can also stand in the causal pathway between them. For example, chronic pain can cause depression, which will impact psychosocial health. Therefore, we chose to focus on disease-related predictors of psychosocial health, such as disease activity, functional disability, and self-reported pain. Finally, by restricting our sample to patients who answered the questionnaires at least twice, we may have selected individuals who are more compliant and have a higher quality of life.

The main strengths of our study are the large multicentered cohorts that enhance the generalizability of our results and the use of structural equation modeling, which helped explore and quantify all components of variability in psychosocial health measures. Moreover, we assessed psychosocial health using the MCS of the SF-36, an instrument that is well suited to capture the emotional, psychological, and social aspects of patients' well-being, and has been shown to be a reliable and valid generic HRQOL measure for RA patients (13–15).

Pain level was the most important predictor of both the stable part and the variable part of psychosocial aspects of HRQOL in treated RA patients. This suggests an overall insufficient pain relief, independent of the control of disease activity. Our results prompt greater focus and efforts on pain management in RA patients.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

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. Courvoisier 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. Courvoisier, Glauser, Wolfe.

Acquisition of data. Michaud, Wolfe, Finckh.

Analysis and interpretation of data. Courvoisier, Agoritsas, Glauser, Wolfe, Cantoni, Perneger, Finckh.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

We are grateful to the SCQM and the NDB staff for data management and support and to the participating physicians and patients who made this study possible. A list of rheumatology offices and hospitals that are contributing to the SCQM registries can be found at www.scqm.ch/institutions.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
ACR_20652_sm_s1.doc42KSupplementary Data

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.