In this study we found that approximately one third of patients with RA have unacceptable pain up to 5 years after diagnosis, and that nearly two thirds of these patients have pain despite low inflammatory activity. Pain improvement was only significant between inclusion and the 6 month-visit, and average pain was thereafter essentially unchanged over time. Several baseline variables, including PROs (i.e. VAS pain, VAS PGA and HAQ), and clinical outcomes like low SJC and low grip force, as well as female sex and low age, predicted unacceptable pain. Out of these associations, those with VAS for pain and PGA were consistently significant over time, and both low SJC and high VAS PGA remained significantly associated with unacceptable pain in the multivariate analyses at 5 years. Unacceptable pain with low inflammatory activity was also predicted by low baseline inflammatory parameters and anti-CCP negativity. In mixed model analysis, high baseline PGA VAS, HAQ and DAS28 were particularly associated with increased pain at baseline, but also with greater improvement in pain over time.
Several recent studies have reported that although pain outcome has improved in more recent years since the introduction of more extensive anti-rheumatic treatment and biologic therapy, there is still a group of patients with persistent pain (3, 4, 7). This points to substantial unmet needs regarding pain management beyond anti-rheumatic treatment, and highlights the importance of improved management of these patients. Such pain management might include encouraging patients to engage in physical activity (28), pain-coping strategies (29), and pain-modulating treatment, with for example serotonin and noradrenaline reuptake inhibitors (30). Treating relevant comorbidities, such as sleep disorders (31) and mental illness (32), when applicable, might also improve pain outcome.
Worse baseline PROs have previously been associated with increased pain levels later during the disease course (7). For example, in a recent study, RA patients with features of neuropathic pain also had higher self-reported global disease activity, disability, and TJC (5). In the present study, PROs were the strongest predictors of increased pain over time. Other predictors of unacceptable pain in this study were low age and female sex, both previously reported as risk factors for pain (3, 5, 7, 33). For instance, female sex has been reported to be a risk factor for chronic widespread pain in patients with RA (5, 33). In another study, women had significantly higher pain scores than men only at lower disease activity levels (34), suggesting that pain differences by sex might not be present in patients with more active disease. Moreover, in a study investigating pain predictors one year after treatment initiation, pain (derived from the Short-Form 36 questionnaire) was actually associated with male sex and higher age in patients receiving biologic DMARDs (35), suggesting that female sex is a risk factor for increased pain mainly in certain groups, e.g. in patients with low disease activity.
Furthermore, chronic widespread pain has been reported to be present in approximately 12% of the general population, with a higher prevalence in women compared to men (36), suggesting that susceptibility to sensitisation and nociplastic pain might be present early in the disease course in a subset of patients. This group might be identified at an early disease stage, as patients with severe PROs despite inflammation control.
Discrepancies between PROs and objective disease activity measures regarding their importance for pain have been reported earlier (7, 37). Furthermore, in a cross sectional study investigating factors associated with non-nociceptive pain, a negative association with anti-CCP positivity was found (5). In another study, predictors of satisfactory improvement in pain were reported to be anti-CCP positivity and symmetric arthritis (38). These studies are in line with our results, where approximately 20 percent of patients had unacceptable pain despite low inflammatory activity during the follow-up. These patients also had lower inflammatory parameters at baseline, and were more likely to be anti-CCP negative, strengthening the concept that there is a group of patients with RA with low disease activity, e.g. low SJC and low laboratory markers of inflammation, as well as anti-CCP negativity, that is more likely to experience a disease course characterized by high pain levels and disability. The current results suggest that patients with anti-CCP antibodies, and more severe clinical disease, respond better to treatment, resulting in reduced long term pain. Although anti-CCP positivity seems to be protective of long term non-inflammatory pain, the relation appears to be the opposite when it comes to radiographic progression, where anti-CCP positivity has been reported to be a predictor (39, 40).
The uncoupling between pain and inflammation supports that non-inflammatory mechanisms contribute to pain in RA. Furthermore, the prevalence of concomitant fibromyalgia in RA has been reported to be up to 25% (41), as compared to 2% in the general population (42), and in one report, investigating neuropathic pain in patients with RA initiating or escalating anti-rehumatic therapy, 23% and 12% of the patients were diagnosed with possible neuropathic pain, and probable neuropathic pain, respectively (43). In the latter group, significantly more patients fulfilled the classification criteria for fibromyalgia. These results all indicate that pain in RA is multifactorial and might include central pain mechanisms, encompassing an increased risk of secondary fibromyalgia and persistent nociplastic pain, presumably sharing common mechanisms of disease development, such as central sensitisation.
Some limitations of the present study should also be noted. Data on comorbidities before RA diagnosis were not available, and therefore the effects of other prevalent diagnoses affecting pain, e.g. concomitant fibromyalgia, depression and osteoarthritis, on the results cannot be excluded. Misclassification of individuals with primary fibromyalgia as RA in this study is however unlikely, as even patients with unacceptable pain and low inflammation at follow-up had active disease at inclusion, with median SJCs above 6, and all patients fulfilled the 1987 RA classification criteria at inclusion. The small number of patients was another limitation of the study, as it affects the statistical power, especially for the analyses of unacceptable pain with low inflammation. It would also have been interesting to investigate predictors of unacceptable pain with the strict definition of low inflammation, but the number of patients in this subset was too small. This would be of interest to study in a larger cohort. Moreover, the majority of patients were included before the practice of treat to target was implemented and before early treatment with biologic DMARDs was standard of care in severe cases, and the results of this study might therefore not be fully applicable to patients diagnosed in the more recent period. However, studies of more recent cohorts demonstrate that pain remains a major problem in patients with RA (4–6, 8). Finally, pain in RA may also be related to joint damage, which was not analysed in detail in this study, but would be interesting to examine in the future. Yet, albeit not significant, our result points mainly towards a negative association between pain and baseline presence of erosions.
The strengths of this study include a systematic longitudinal follow-up of patients from a defined period of time and a defined catchment area. Therefore, there should not be a major risk of selection bias, and the results could be generalized to patients with RA seen in clinical practice. Furthermore, all joint assessments were performed by the same rheumatologist for all patients, using a structured protocol. Finally, the definition of unacceptable pain is a common and validated measure, which makes it possible to compare the results of this study to other reports.