To assess the determinants of patients' (PTGL) and physicians' (MDGL) global assessment of rheumatoid arthritis (RA) activity and factors associated with discordance among them.
To assess the determinants of patients' (PTGL) and physicians' (MDGL) global assessment of rheumatoid arthritis (RA) activity and factors associated with discordance among them.
A total of 7,028 patients in the Quantitative Standard Monitoring of Patients with RA study had PTGL and MDGL assessed at the same clinic visit on a 0–10-cm visual analog scale (VAS). Three patient groups were defined: concordant rating group (PTGL and MDGL within ±2 cm), higher patient rating group (PTGL exceeding MDGL by >2 cm), and lower patient rating group (PTGL less than MDGL by >2 cm). Multivariable regression analysis was used to identify determinants of PTGL and MDGL and their discordance.
The mean ± SD VAS scores for PTGL and MDGL were 4.01 ± 2.70 and 2.91 ± 2.37, respectively. Pain was overwhelmingly the single most important determinant of PTGL, followed by fatigue. In contrast, MDGL was most influenced by swollen joint count (SJC), followed by erythrocyte sedimentation rate (ESR) and tender joint count (TJC). A total of 4,454 (63.4%), 2,106 (30%), and 468 (6.6%) patients were in the concordant, higher, and lower patient rating groups, respectively. Odds of higher patient rating increased with higher pain, fatigue, psychological distress, age, and morning stiffness, and decreased with higher SJC, TJC, and ESR. Lower patient rating odds increased with higher SJC, TJC, and ESR, and decreased with lower fatigue levels.
Nearly 36% of patients had discordance in RA activity assessment from their physicians. Sensitivity to the “disease experience” of patients, particularly pain and fatigue, is warranted for effective care of RA.
There is increased emphasis on “patient-centered care” that gives due importance to a patient's perception of health and considers their priorities and preferences in making therapeutic decisions (1). This patient-centered care approach increases trust in the physician and enables patients to be more confident in expressing their concerns and thoughts, in prioritizing their problems, and in discussing their expectations or goals from health care, leading to shared goals and understanding of roles and responsibilities (2, 3). Patients who engage in collaborative care, shared decision making, and chronic disease self-management have improved health outcomes (4). However, discordance among patients and physicians in the assessed health status can result in patient dissatisfaction and can negatively affect patient care, treatment compliance, disease outcome, and consequent cost to society (5–9).
Discordance in patient- and physician-assessed level of well-being has been described in several chronic diseases (5, 10, 11). In patients with musculoskeletal disorders, discordance between patients and physicians has been reported in the rating of pain and overall health, and in willingness to take risks to improve health (12). Regarding rheumatoid arthritis (RA), physicians have been reported to differ from patients in the assessment of patients' physical and mental function (13, 14) and relevance of disability to the patient (15). In a single-center cross-sectional study of 207 patients that evaluated patient satisfaction with their RA activity (from excellent to unsatisfactory), patients had higher ratings than physicians at each satisfaction level (16). Moreover, only 60% congruence in assessments of RA activity fluctuation was noted among patients and physicians (17).
There is a relative paucity of literature on the determinants of discordant assessment of RA activity between patients and physicians. A single-center study of 80 patients comparing physician and patient ratings of RA activity on a visual analog scale (VAS) found that levels of education and C-reactive protein were associated with higher physician ratings, whereas higher pain score, Health Assessment Questionnaire (HAQ) score, and tender joint count (TJC) were associated with a higher patient rating of RA activity (18). Multivariable analysis was not performed to establish independent determinants of the discordance. A recent study of 223 RA patients from 2 centers found higher levels of depressive symptoms to be the strongest independent predictor for discordance, whereas swollen joint count (SJC) and Cantonese/Mandarin language were associated with lower odds of discordance. However, this study did not consider the impact of patient-reported outcomes such as pain and fatigue nor of comorbidities on discordance, and did not have enough subjects to assess predictors of patient ratings lower than those by their physicians (19). Another recent single-center study of 110 patients found health literacy to be independently associated with the extent of discrepancy between patients' and physicians' RA activity assessment after adjusting for sociodemographic and treatment factors (sex, age, years of education, and biologic agent use), but did not adjust for the effect of patient-reported symptoms, physician-assessed joint counts, acute-phase reactant, psychological distress, and comorbidities (20). Moreover, other yet poorly defined patient-based factors such as quantity and quality of coping strategies used by RA patients to bear the burden of the disease may account for the discrepancies between patients' and physicians' assessments of disease (21). The purpose of this study in a large multinational cohort of RA patients was to assess the determinants of patient global assessment of RA activity (PTGL) and physician global assessment of RA activity (MDGL) scores, the level of concordance between patients' and physicians' assessment of RA disease activity, and the determinants of discordant assessment.
The Quantitative Standard Monitoring of Patients with RA (QUEST-RA) is a database of RA patients who received usual care from rheumatologists in ≥3 rheumatology clinics in several countries (QUEST-RA Investigators are listed in Supplementary Appendix A, available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658). One hundred nonselected consecutive RA patients were recruited from each participating clinic (22). The patients were assessed by a standard protocol to evaluate RA (23).
Demographic information and clinical characteristics of RA were obtained from the database. Patient demographic variables available included age, sex, education, and ethnicity. PTGL was assessed by the question, “In terms of joint tenderness (i.e., joint pain associated with light touch) and joint swelling (i.e., joint enlargement due to inflammation), how active would you say your rheumatic condition is TODAY?” on a 0–10-cm VAS with “not active at all” and “extremely active” as the anchors. This question is one of the items of the Rheumatoid Arthritis Disease Activity Index (RADAI) and was originally denoted as “arthritis now” (24). In a validation study of the RADAI in 484 RA patients, the authors reported “arthritis now” to be equivalent to PTGL (25). Both had identical and high correlation (Spearman's correlation 0.81) with pain. MDGL was assessed by the statement, “Please mark below your assessment of the patient's current disease activity” on a 0–10-cm VAS with “no activity” and “very active” as the anchors.
Pain and fatigue were assessed by 0–10-cm VAS, physical function was assessed by the HAQ (26), and morning stiffness duration in minutes was assessed by patient self-report questionnaire. The psychological HAQ was used to assess psychological distress by asking about the ability to deal with the usual stresses of daily life, feelings of anxiety and depression, and getting a good night's sleep. The responses are calculated in the HAQ format and scored from 0–3, with 0 denoting no psychological distress and 3 denoting very severe psychological distress. Psychological HAQ scores strongly correlate with the Arthritis Impact Measurement Scales (AIMS) anxiety scale, AIMS depression scale, Beck Depression Inventory, and Center for Epidemiologic Studies Depression Scale (27).
Physicians assessed 28 joints for tenderness (TJC) and swelling (SJC). Erythrocyte sedimentation rate (ESR) was measured in the local laboratory. Physicians completed information on comorbidities. Comorbidity burden was assessed by a composite comorbidity score (range 0–9) that comprised 11 comorbid conditions, including pulmonary disorders, myocardial infarction, other cardiovascular disorders, stroke, diabetes mellitus, hypertension, depression, fracture, gastrointestinal ulcer, other gastrointestinal disorders, and cancer (28). The comorbidity index was modified to exclude depression, resulting in a score range of 0–8 (29). Additionally, the impact of the presence or absence of 3 other painful comorbidities (osteoarthritis, fibromyalgia, and chronic back pain) was evaluated. Body mass index (BMI) was calculated from the weight and height using the standard formula.
Mixed-effects analyses of covariance (ANCOVAs) models were used to model PTGL and MDGL measures as functions of demographic and medical characteristics. Demographic variables included age, race (white, other races), sex, and postsecondary education/education >12 years (yes, no), whereas medical characteristics included duration of RA, patient's pain score (0–10-cm VAS), patient's fatigue score (0–10-cm VAS), HAQ (range 0–3), psychological HAQ (range 0–3), TJC (range 0–28), SJC (range 0–28), ESR, morning stiffness (0, 1–60, and >60 minutes), comorbidity index score (0, 1–2, and >2), BMI, fibromyalgia (yes, no), osteoarthritis (yes, no), and chronic back pain (yes, no). To adjust for correlations among subjects from the same reporting institution, the reporting institution was included as a random-effect variable in the models. The mixed procedure in SAS was used to implement these models (30).
The mean ± SD difference between the PTGL and MDGL VAS was calculated. Lin's concordance coefficient, which takes into account both the correlation as well as the precision of agreement between 2 continuous variables, was used to quantify the level of concordance between PTGL and MDGL ratings (31). A difference in rating between MDGL and PTGL of >2 cm was considered a relevant discordance for the purpose of analyses to understand the determinants of discordant RA activity rating by patients and physicians. Although there is no accepted definition of relevant discordance, we chose our cutoff based on a 2-cm change being considered the most frequently chosen value for minimum clinically important improvement (MCII) in PTGL by a group of patients, clinicians, and researchers (32). A 2-cm change was also found to be the MCII value in a group of nearly 5,000 RA patients using the 80% specificity cut point approach (33). Based on the difference in PTGL and MDGL, 3 groups were identified: concordant rating group (PTGL and MDGL within ±2 cm), higher patient rating group (PTGL exceeding MDGL by >2 cm), and lower patient rating group (PTGL less than MDGL by >2 cm). A sensitivity analysis was also done considering a 1-cm difference between PTGL and MDGL as a relevant discordance.
Demographic and medical characteristic variables were summarized using means and SDs for continuous data or proportions and counts for categorical data for each of these groups. Comparisons of the discordant rating groups to the concordant rating group were performed using t-tests for continuous variables or Pearson's chi-square tests for categorical variables. Two multivariable logistic regression models with the same demographic and medical characteristics variables as described above for the ANCOVA model were examined. One included only subjects in the higher patient rating and concordant rating groups to identify variables associated with PTGL exceeding MDGL by >2 cm. Similarly, the second model included only subjects in the lower patient rating and concordant rating groups in order to identify variables associated with MDGL exceeding PTGL by >2 cm. A random effect representing the reporting institution was included in these models to adjust for correlations between subjects seen at the same institution. We report adjusted odds ratios and 95% confidence intervals (95% CIs) for each significant predictor variable identified in these models.
Multiple imputations of missing data were performed using the Multivariate Imputation by Chained Equations library (34) as implemented in the statistical software R (35). The imputation process was repeated 5 times, producing 5 “complete” data sets. These data sets were analyzed as described above, resulting in 5 sets of results. These results were combined using the Mianalyze procedure in SAS (30).
The QUEST-RA database contained data on 7,568 patients recruited from 83 sites in 30 countries at the time of this analysis. Five hundred forty patients (7.13%) were excluded from the analysis (113 because of missing information on both PTGL and MDGL and 427 because of missing information on 5 or more study variables). Missing data were imputed as described above for 3,024 subjects (39.96%) who had missing information on less than 5 study variables. Therefore, our study analysis is based on information obtained from 7,028 patients (4,004 patients with complete data).
While patients and physicians shared most of the variables (Table 1) that were independently associated with their assessment of RA disease activity (pain, fatigue, HAQ, psychological HAQ, morning stiffness, SJC, TJC, and ESR), some variables were significant only for PTGL (female sex) or for MDGL (age, obesity, and fibromyalgia). However, there were remarkable differences in the relative importance of various individual variables in PTGL and MDGL ratings as shown by the partial R2 plots (Figure 1). These plots depict the contribution of each variable in explaining the observed variance in the PTGL or MDGL ratings after controlling for the effects of the remaining variables. Pain was overwhelmingly the single most important determinant of PTGL rating, followed by fatigue. In contrast, MDGL ratings were most influenced by the physical examination (SJC and TJC) and laboratory markers of inflammation (ESR).
|Variable||PTGL model||MDGL model|
|Estimate (95% CI)||P||Estimate (95% CI)||P|
|Intercept||0.9356 (0.4547, 1.4166)||0.0001||1.6572 (1.2211, 2.0934)||< 0.0001|
|Age||−0.0001 (−0.0037, 0.0034)||0.9432||−0.0077 (−0.0107, −0.0047)||< 0.0001|
|Sex (female)||−0.1184 (−0.2234, −0.0134)||0.0271||−0.0423 (−0.1307, 0.0461)||0.3480|
|Race (nonwhite)||0.0026 (−0.1361, 0.1413)||0.9709||−0.0026 (−0.1404, 0.1353)||0.9706|
|Education (>12 years)||0.0197 (−0.0747, 0.114)||0.6823||−0.0035 (−0.0868, 0.0799)||0.9345|
|RA duration||0.0027 (−0.002, 0.0074)||0.2561||−0.0023 (−0.0062, 0.0016)||0.2557|
|Pain||0.49 (0.4674, 0.5125)||< 0.0001||0.1621 (0.1431, 0.1811)||< 0.0001|
|Fatigue||0.1587 (0.1394, 0.178)||< 0.0001||0.0378 (0.0215, 0.0542)||< 0.0001|
|Morning stiffness duration, minutes|
|1–60||0.1637 (0.0498, 0.2776)||0.0051||0.093 (0.0001, 0.186)||0.0497|
|>60||0.2993 (0.1699, 0.4286)||< 0.0001||0.2653 (0.168, 0.3627)||< 0.0001|
|HAQ||0.3069 (0.2239, 0.3899)||< 0.0001||0.3226 (0.2517, 0.3936)||< 0.0001|
|Psychological HAQ||0.0872 (0.0105, 0.164)||0.0259||−0.0877 (−0.152, −0.0234)||0.0075|
|TJC28||0.0211 (0.0134, 0.0288)||< 0.0001||0.0612 (0.0545, 0.068)||< 0.0001|
|SJC28||0.0395 (0.0295, 0.0496)||< 0.0001||0.1199 (0.1111, 0.1288)||< 0.0001|
|ESR||0.0024 (0.0005, 0.0043)||0.0139||0.0197 (0.018, 0.0214)||< 0.0001|
|BMI||−0.0018 (−0.0111, 0.0074)||0.6969||−0.0127 (−0.0204, −0.0049)||0.0015|
|1–2||−0.1052 (−0.2003, −0.01)||0.0303||−0.0333 (−0.1129, 0.0463)||0.4122|
|>2||−0.1094 (−0.2598, 0.0409)||0.1536||−0.1638 (−0.2888, −0.0388)||0.0102|
|Chronic back pain||−0.0948 (−0.2164, 0.0267)||0.1261||−0.0825 (−0.186, 0.021)||0.1180|
|Osteoarthritis||−0.0303 (−0.1529, 0.0922)||0.6275||−0.0196 (−0.1237, 0.0845)||0.7116|
|Fibromyalgia||−0.2125 (−0.4572, 0.0322)||0.0887||−0.3404 (−0.5443, −0.1365)||0.0011|
The mean ± SD VAS scores for PTGL and MDGL were 4.00 ± 2.70 and 2.91 ± 2.37, respectively. Patients rated their RA activity higher than physicians across all countries with a mean ± SD difference of 1.09 ± 2.38. Lin's concordance correlation coefficient value was 0.51 (95% CI 0.49, 0.53), indicating moderate concordance between PTGL and MDGL ratings. The concordance between PTGL and MDGL ratings varied widely across different countries (Figure 2). Most patients and their physicians did not differ too greatly in their scores, as 4,454 patients (63.4%) had an RA activity rating within 2 cm of their physician's ratings. A total of 2,106 patients (30%) scored their RA activity >2 cm higher than their physicians (higher patient rating group) with a mean ± SD difference of 3.97 ± 1.57, whereas 468 patients (6.6%) scored their RA activity 2 cm lower than their physicians (lower patient rating group) with a mean ± SD difference of −3.4 ± 1.26.
Table 2 shows the sociodemographic and clinical characteristics of the 2 patient rating groups and the univariable analyses comparing the concordant rating group with each of the 2 discordant patient rating groups. Among the sociodemographic characteristics, a higher patient rating was associated with increasing age and a lower proportion of subjects with education >12 years, whereas lower patient ratings were associated with female sex and nonwhite race. In general, patients in the concordant rating group had lower symptom severity, as well as better physical and psychological functional health status, compared to the 2 discordant patient rating groups, whereas physician- and laboratory-assessed RA activity parameters were lower (with the exception of TJC) in the higher patient rating group, and were higher in the lower patient rating group. The higher patient rating group had higher comorbidity burden and BMI and a higher prevalence of fibromyalgia and osteoarthritis. The lower patient rating group had a significantly lower prevalence of fibromyalgia.
|Patient rating groups||P†|
|Higher patient rating group (n = 2,106)||Concordant rating group (n = 4,454)||Lower patient rating group (n = 468)||Higher vs. concordant||Lower vs. concordant|
|Age, years||56.66 ± 13.76||54.67 ± 13.84||54.47 ± 13.82||< 0.001||0.774|
|Nonwhite race, %||27.29||29.22||34.76||0.104||0.017|
|Education >12 years, %||31.18||35.14||31.72||0.001||0.133|
|RA duration, years||11.91 ± 10.38||10.31 ± 8.98||10.96 ± 9.64||< 0.001||0.161|
|Pain (0–10-cm VAS)||5.4 ± 2.35||3.44 ± 2.62||4 ± 2.71||< 0.001||< 0.001|
|Fatigue (0–10-cm VAS)||5.49 ± 2.6||3.86 ± 2.83||4.1 ± 2.92||< 0.001||0.09|
|Morning stiffness duration, minutes||64.57 ± 79.01||45.68 ± 67.83||61.66 ± 76.45||< 0.001||< 0.001|
|HAQ||1.22 ± 0.73||0.9 ± 0.77||1.09 ± 0.74||< 0.001||< 0.001|
|Psychological HAQ||2.03 ± 0.7||1.77 ± 0.67||1.87 ± 0.7||< 0.001||0.002|
|SJC28||3.63 ± 4.51||4.31 ± 5.45||7.84 ± 6.63||< 0.001||< 0.001|
|TJC28||6.36 ± 6.83||6.61 ± 7.7||10.09 ± 8.35||0.17||< 0.001|
|ESR, mm/hour||28.63 ± 22.67||29.87 ± 24.13||42.58 ± 29.22||0.042||< 0.001|
|PTGL (0–10-cm VAS)||6.15 ± 1.97||3.15 ± 1.7||2.41 ± 1.42||< 0.001||< 0.001|
|MDGL (0–10-cm VAS)||2.19 ± 2.49||2.95 ± 2.45||5.81 ± 1.69||< 0.001||< 0.001|
|Comorbidity index||1.09 ± 1.28||0.91 ± 1.15||0.91 ± 1.2||< 0.001||0.977|
|Fibromyalgia, %||4.55||2.5||1.33||< 0.001||0.043|
|Chronic back pain, %||15.75||14.43||12.44||0.166||0.22|
|BMI, kg/m2||25.86 ± 5||25.52 ± 4.7||25.41 ± 4.67||0.008||0.645|
|DAS||4.13 ± 1.68||4.13 ± 1.84||5.21 ± 1.37||0.861||< 0.001|
Multivariable logistic analyses (Table 3) showed that the odds of higher patient rating increased with increase in age, patient-reported symptoms (pain, fatigue, and presence of morning stiffness), and psychological HAQ scores. An increase in physician-assessed variables (SJC and TJC) and laboratory markers of inflammation (ESR) independently decreased the odds of higher patient rating. In contrast, an increase in SJC, TJC, and ESR increased the odds of lower patient rating. Lower levels of fatigue were the only patient-reported predictor independently associated with decreased odds of lower patient rating. Sex, ethnicity, duration of RA, education, HAQ score, BMI, and comorbidities were not independently associated with discordant assessment of RA activity. Sensitivity analysis using an MCII of 1 cm showed almost identical results (see Supplementary Table 1, available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658). However, longer RA duration increased the odds of a higher patient rating, whereas psychological HAQ score was no longer associated with higher patient rating. A higher psychological HAQ score decreased the odds of lower patient rating, but fatigue was no longer associated with lower patient rating.
|Higher patient rating||Lower patient rating|
|OR (95% CI)||P||OR (95% CI)||P|
|Age, Δ10 years||1.07 (1.012, 1.131)||0.0182||0.993 (0.908, 1.086)||0.8780|
|Female sex||0.921 (0.785, 1.081)||0.3153||1.297 (0.982, 1.713)||0.0672|
|Nonwhite race||0.956 (0.764, 1.197)||0.6942||1.009 (0.745, 1.366)||0.9557|
|Education >12 years||0.981 (0.851, 1.132)||0.7946||0.88 (0.693, 1.118)||0.2937|
|RA duration, Δ5 years||1.033 (0.998, 1.07)||0.0672||1.059 (1.001, 1.121)||0.0504|
|Pain (range 0–10)||1.437 (1.388, 1.488)||< 0.0001||0.943 (0.888, 1.001)||0.0545|
|Fatigue (range 0–10)||1.136 (1.103, 1.17)||< 0.0001||0.928 (0.882, 0.977)||0.0041|
|Morning stiffness duration, minutes|
|1–60||1.284 (1.091, 1.511)||0.0027||1.12 (0.824, 1.523)||0.4644|
|>60||1.285 (1.074, 1.537)||0.0062||1.343 (0.959, 1.882)||0.0855|
|HAQ (range 0–3)||0.954 (0.844, 1.08)||0.4577||0.944 (0.76, 1.174)||0.6057|
|Psychological HAQ (range 0–3)||1.137 (1.014, 1.273)||0.0273||1.047 (0.858, 1.277)||0.6537|
|SJC28 (range 0–28)||0.931 (0.915, 0.947)||< 0.0001||1.066 (1.045, 1.087)||< 0.0001|
|TJC28 (range 0–28)||0.968 (0.956, 0.98)||< 0.0001||1.022 (1.004, 1.041)||0.0141|
|ESR||0.987 (0.984, 0.99)||< 0.0001||1.015 (1.011, 1.02)||< 0.0001|
|1–2||0.906 (0.784, 1.048)||0.1838||0.92 (0.727, 1.163)||0.4844|
|>2||1.152 (0.928, 1.431)||0.1989||0.968 (0.658, 1.423)||0.8669|
|BMI||1.004 (0.991, 1.018)||0.5444||1 (0.978, 1.024)||0.9672|
|Chronic back pain||0.898 (0.747, 1.079)||0.2511||0.784 (0.564, 1.092)||0.1499|
|Osteoarthritis||1.04 (0.869, 1.246)||0.6680||0.988 (0.718, 1.361)||0.9411|
|Fibromyalgia||1.02 (0.724, 1.438)||0.9079||0.66 (0.281, 1.553)||0.3415|
Our data show that patients and physicians rate RA activity differently. Pain is overwhelmingly the single most important independent determinant of PTGL, followed by fatigue. While physicians do take into account patient-reported outcomes (particularly pain and physical functional status), they rely predominantly on joint counts and ESR for determining the MDGL score. Clearly, physicians are most focused on “RA-specific outcomes,” whereas patients are more focused on how the general health state is affected by RA. Similar findings of a particularly strong association of pain and fatigue with PTGL and joint counts with MDGL have been reported (31). However, unlike our finding of a strong association of ESR with MDGL, one study found this association to be weak (36). Similar observations have been made in other diseases such as systemic lupus erythematosus (SLE), where patients base their assessments of disease activity on its psychological and functional effects, and physicians rely on physical and laboratory abnormalities (37–39).
We found a moderate degree of concordance in MDGL and PTGL scores. The QUEST-RA study was performed in a routine clinical setting for RA patients. Patients completed the study questionnaire prior to the clinical encounter with their physician. Therefore, physicians were aware of patients' ratings of RA activity and patients' symptom burden while making the MDGL assessment. ESR results were also available for some but not for all patients (depending on the clinical care setup in the participating clinic) prior to MDGL assessment. Despite awareness of patients' perspectives, a substantial proportion of physicians (36.6%) had relevant discordance with their patients in RA activity assessment. Most of the discordance was associated with patients rating RA activity higher than physicians.
Moreover, when discordance was present the magnitudes of difference between the PTGL and MDGL scores on average were substantially more than the MCII (nearly 2 and 1.5 times the MCII of 2 cm in the higher and lower patient rating groups, respectively). This also explains why the prevalence of discordance in our study is similar to the data reported by Nicolau et al (36.5%) and Barton et al (35%), even though they used slightly larger differences in PTGL and MDGL VAS ratings of 3 cm and 2.5 cm, respectively, as denoting significant discordance (18, 19).
The discordance in PTGL and MDGL may originate from several sources, including patients' experiences of RA, dynamics of the patient and physician relationship, cultural factors, random factors, and other systematic differences between patients' and physicians' assessments of RA activity. While we could not adjust for all of these sources of variation, we did include the reporting institution as a random effect in the model. Pain was the single most important source of discordance between PTGL and MDGL. Pain perception in RA is not only affected by the degree of inflammatory burden from the ongoing RA activity, but also by the enhanced pain sensitivity (hypernociception) secondary to inflammatory cytokines, autonomic dysfunction, and psychological variables, particularly anxiety and depression (40). In addition, there is substantial variability in pain sensitivity in humans that is strongly affected by polymorphisms in genes involved in pain modulation (41, 42). Importantly, sensitivity and attention to pain are important determinants in meeting patients' expectations during visits for arthritis care. Patients with declining levels of pain are less likely to report unmet expectations (43).
Psychological distress as assessed by the psychological HAQ (which has a strong correlation with anxiety and depression) was independently associated with higher patient rating. A recent study reported depressive symptom severity as the strongest independent determinant of discordance in PTGL and MDGL (19). This study, however, did not adjust for severity of pain and fatigue, which are the dominant predictors of self-reported depression in RA patients (29). Patients with depression are also at greater risk of symptom underestimation by their physicians (44).
There is a small proportion of RA patients with relatively more severe joint involvement and higher ESR but rather low severity of self-report symptoms, leading them to underestimate disease activity. These patients either may not perceive the abnormal joints or be unaware of the importance of the presence of clinical manifestations such as swollen joints in strongly predicting structural damage from RA (45).
While a discordance in the assessment of physical function status among RA patients and their physicians has been consistently demonstrated (13, 14), there are conflicting data on its association with PTGL and MDGL discordance. One study has reported the HAQ to be associated with higher patient rating (18), whereas another did not show an association of HAQ scores with patient and physician discordance (19). We did not find HAQ scores to be an independent determinant of discordance between PTGL and MDGL. We also did not find comorbidities and coexistence of common painful conditions to be an independent determinant of discordance. This finding could partly be due to significant partial correlations between these variables and symptom experiences of the patients (pain, fatigue, and stiffness). Ethnicity was evaluated as white or nonwhite, and was not associated with discordant RA activity assessment. Similar findings were noted in RA and SLE patients (19, 39).
Multiple studies in patients with chronic diseases have shown that patient–physician discordance in health status is associated with negative consequences such as underestimation of patients' symptom burden and functional status, undertreatment of patients' symptoms, patient dissatisfaction, noncompliance with prescribed treatment, decreased trust in the physician leading to adverse disease outcome, and consequent cost to society (3, 5–9). To the best of our knowledge, there is no direct evidence linking discordance in RA disease activity assessment to adverse outcomes in RA patients. A study that evaluated patient–physician concordance in assessment of physical function in RA patients showed less pain and depressive symptoms in patients who had concordant assessment of physical function with their physician (14). Further research is needed on how the discordance between MDGL and PTGL impacts clinical outcomes for RA patients and whether interventions to reduce discordance would improve outcomes.
Our study has some limitations. First, the questions used for PTGL and MDGL assessment were not identical. However, the “arthritis now” question of the RADAI used by us to assess “PTGL” has been considered equivalent to the “patient's global assessment of RA activity” question (25). Second, there is no standardized way to define discordance. We selected a 2-cm VAS difference to define concordance categories for our main analyses based on it being the most frequently chosen value for the MCII in PTGL by a group of patients, clinicians, and researchers (32), as well as the calculated value for nearly 5,000 RA patients using a robust methodologic approach (33). More research is needed on how to define discordance. Third, our study had a cross-sectional design. Patient expectations and perceptions vary according to whether there has been an improvement or worsening of health compared to the past (46). The perception of improvement of disease activity of patients with RA is considerably different depending on the disease activity level at which they start (47). We simply do not know relative activity direction and magnitude of relative change in RA activity of our patients over the recent past. Fourth, we do not have data on self-efficacy and learned helplessness, which have been shown to affect patients' perceptions of disease (46). Other limitations include the potential impact of unmeasured cultural factors on disease activity assessment and perception. The QUEST-RA study was conceived to promote objective RA activity assessment and establish a database of RA patients from several countries with differing health care systems and social, cultural, and economic backgrounds. Multiple dimensions related to culture such as ethnicity; belief about disease causation, course, and outcome; patient–physician relationship dynamics; and economics of health care may impact perception of disease (48, 49). To decrease variance from these non–RA-related factors, every patient and physician in the study completed a standard protocol. Moreover, validated translated versions of the study questionnaires in the patient's native language were used when available, or were developed using the standard method of translation and back translation. Second, we included the individual centers of patient enrollment as a random effect in statistical modeling to adjust for some of these factors. We did not have detailed data on socioeconomic factors, and having patients from such varied backgrounds would make it difficult to be comparative. Therefore, we used a broad category of education and ethnicity for our analyses. We also did not have the data on health literacy, which has been shown to be independently associated with the extent of discrepancy in PTGL and MDGL (20).
For effective care of any chronic disease such as RA, there needs to be a “collaborative definition of a problem” that takes into account both the patient-defined problems as well as clinical assessment by the physician (2). Physicians need to pay more attention to the “disease experience” of the patient. Conversely, physicians may need to educate patients with minimal symptoms about the importance of clinically meaningful changes such as swollen joints and/or elevated inflammatory markers.
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. Khan 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. Khan, Spencer, Sokka.
Acquisition of data. Abda, Aggarwal, Alten, Ancuta, Andersone, Bergman, Craig-Muller, Detert, Georgescu, Gossec, Hamoud, Jacobs, Laurindo, Majdan, Naranjo, Pandya, Pohl, Schett, Selim, Toloza, Yamanaka, Sokka.
Analysis and interpretation of data. Khan, Spencer, Sokka.
Abbott (Finland) provided funds for the study data collection. It had no role in study design, what and how data are collected, data analysis, and writing of the manuscript. It had no role in approval of the content of the manuscript.