Disability Related to Function of the Upper Extremities in Early Rheumatoid Arthritis – Course and Relation to Other Disease Parameters Over 10 Years: A Cohort Study

Background: The objective of this study was to investigate the course of disability related to the upper extremities (UE) in early rheumatoid arthritis (RA), and to assess correlations between such disability and clinical parameters, including grip force. Methods: In an inception cohort of patients with early RA (N=222), disability of the UE was assessed using a subscore of the Health assessment questionnaire disability index (HAQ-DI), and average grip force of the dominant hand was measured. Changes between consecutive follow-up visits in the HAQ-DI-UE subscore were assessed using the paired samples t-test, and correlations with key disease parameters using Spearman’s rank test. The relation between joint involvement and HAQ-DI-UE was examined using multivariate linear regression analysis. Results: The HAQ-DI-UE decreased signicantly from inclusion to the 6-month follow-up (mean change -0.26; 95% CI -0.18 to -0.34), and increased signicantly after 2 years. There were fairly strong correlations for HAQ-DI-UE with grip force (r: -0.50 to -0.62), patient’s global assessment (r:0.58 to 0.64) and patient’s assessment of pain (r:0.54 to 0.60) at all time points up to 5 years, but only moderate to weak correlations with swollen joints, CRP and ESR. At inclusion wrist synovitis and tender PIP joints had both an independent impact on HAQ-DI-UE, whereas tenderness of the shoulder and the wrist had a greater importance at 6 months. Conclusions: Disability and pain

the rst year after diagnosis (6). However, grip force remained signi cantly reduced compared to the general population, even in patients in clinical remission (6).
PROMs, e.g. HAQ, and RA disease activity measures (e.g. DAS28), have been found to be moderately correlated (17). A moderate association has also been found between HAQ-DI and grip force in patients with RA (18). Therefore, it is important to study the course of disability as a separate outcome in RA (11).
HAQ-DI is a general measure of disability (13,14), and only a subset of the questions are related to ADL that are affected by hand function. Pain and limited range of movements of joints have the greatest impact on individual sub dimensions of the HAQ-DI (19). Studies have shown that higher HAQ-DI scores at baseline are associated with long-term disability (20,21), especially for women, older patients and those with high pain scores (21). Decreased grip force, a great number of swollen and tender joints in the upper extremities (UE) and limitations in wrist and shoulder motion were related to many of the sub dimensions and explained increased disability with higher HAQ scores (19,22). Previous studies have addressed HAQ overall, or individual subdimensions, and not dimensions re ecting disability related to the UE.
The objective of this study was to speci cally investigate the course of disability related to the UE in early RA, using a subset of the HAQ-DI (HAQ-DI-UE), and to assess relations between HAQ-DI-UE and disease parameters. Furthermore, correlations between HAQ-DI-UE and grip force were studied.

Patients And Methods
Patients. An inception cohort of patients with early RA (symptom duration ≤ 12 months), recruited in 1995-2005, was investigated. The patients were diagnosed with RA by a rheumatologist and ful lled the 1987 American College of Rheumatology (ACR) classi cation criteria for RA (23). The study included individuals from a de ned area, the city of Malmö, Sweden (population 260 000 in 2000). Patients were recruited from the rheumatology outpatient clinic of Malmö University Hospital, which was the only hospital serving the city, and from the four rheumatologists in private practice in Malmö. All patients gave their written informed consent to participate, and the study was approved by the Regional Ethical Review Board for southern Sweden (Lund, Sweden).
Clinical assessment. Patients were managed according to usual care, with no pre-speci ed protocol for pharmacotherapy or rehabilitation. The patients were included before the current practice of treat-totarget (24) was implemented, and before early treatment with biological disease-modifying antirheumatics drugs (bDMARDs) came into widespread use.
In a structured follow-up program, all patients were examined by the same rheumatologist. Visits were scheduled at 6, 12 and 24 months as well as 5 and 10 years after inclusion. Using a standardized protocol, individual joints were assessed as swollen/not swollen and tender/not tender, and standard 28joint swollen joint counts (SJC) and tender joint counts (TJC) were obtained. Disability was assessed using the HAQ-DI (12). The Swedish validated translated version of the HAQ-DI (25) was used. Patient reported pain and patients' global assessment of disease activity were assessed using Visual Analogue Scales (VAS; scale 0-100). Information on treatment was obtained as previously described (6). In brief, information on current treatment with disease-modifying anti-rheumatic drugs (DMARDs) and corticosteroids was obtained through a structured interview at each visit. Data on treatment with bDMARDs during the study period was obtained through linkage to a regional biological register (26).
Most patients were started on a DMARD at the time of diagnosis. The use of Methotrexate as a rst choice increased gradually during the study period. A limited number of patients with severe, refractory disease were treated with bDMARDs after their introduction in 1999. Blood samples were obtained at the visit when the joint assessment was performed (within 1 hour). C-reactive protein (CRP) and the Erythrocyte Sedimentation Rate (ESR) were analyzed using standard methods at the Department of Clinical Chemistry, Malmö University Hospital.
Assessment of grip force. Grip force (Newton, N) was measured by using the electronic instrument Grippit (AB Detektor, Gothenburg, Sweden). This was performed at the same visit as the joint assessment (within 1 hour). The patient was seated comfortably in a chair without armrests, with the shoulder, arm and hand in standard positions as previously described (27). The other arm was resting on the table. Standardized instructions were given. When using this procedure, the test-retest scores for Grippit measures have been demonstrated to be high (27). The grip force was measured alternately in the dominant hand and the non-dominant hand three times, and the mean of the three measurement values from each hand was used. Average values of the ten second uninterrupted grip were obtained, as previously described (6). Average grip force values of the dominant hand at inclusion and at the 6 month, 1, 2, 5, and 10 year follow-up visits were compared to the expected, based on age-and sex-speci c reference values from a convenience sample from a cross-sectional study of volunteers in the region of Oslo, Norway (28). Grip force values for each patient were expressed as % of the expected, based on the reference values.
Disability related to the upper extremities. To estimate disability based on self-reported activity limitations during the study period we used the validated Swedish version (12) of the HAQ (25). This assessment instrument included 20 questions divided into eight domains: Dressing and grooming, Arising, Eating, Walking, Hygiene, Reach, Grip, and Other usual activities, and is used in practice to calculate the HAQ disability index (HAQ-DI); range 0-3). In previous studies, subscore for lower extremities (HAQ-DI-LE), which included 10 questions that cover activities that are mainly dependent on function of the lower extremities (LE), has been calculated (29)(30)(31). To more speci cally address disability of the upper extremities, we computed a subscore, HAQ-DI upper extremities (HAQ-DI-UE), which includes the remaining 10 questions on activities mainly performed using the UE (Additional File 1, Supplementary Table 1).
Statistics. Changes in HAQ-DI-UE between two consecutive follow-up visits were assessed using the paired samples t-test. Correlations between HAQ-DI-UE and key disease parameters, i.e. grip force, ESR, CRP, visual analogue scales for patient global assessment (VAS global) and VAS pain, swollen and tender joint counts (SJC and TJC), at each visit were assessed using Spearman's rank test. Furthermore, the distributions of HAQ-DI-UE at different time points in patients with and without current synovitis, and with and without tenderness, of ≥ 1 joint in each joint group (i.e. shoulders, elbows, wrists, metacarpophalangeal (MCP) and proximal interphalangeal (PIP) joints were compared using the Mann-Whitney test. To further assess the relation between joint involvement (as above) and HAQ-DI-UE at inclusion and 6 months of follow up were examined using linear regressions analysis. Normality of distribution for the residuals was examined using the Shapiro-Wilk test. Covariates with p-values of < 0.10 in the bivariate models were included in multivariate models. Collinearity between covariates was examined using Spearman's test. In cases of major collinearity, the covariate with the strongest association with the dependent variable was selected for the multivariate analysis.
Relation between joint involvement and HAQ-DI-UE. Among patients with data on HAQ-DI at inclusion, 192 patients had synovitis of ≥ 1 MCP joint in the UE, whereas presence of synovitis of wrists, PIP, elbows and shoulders was observed in 153, 139, 20 and 19 patients, respectively. Whereas tenderness at inclusion was more common compared to clinical synovitis in the shoulders (n = 77) and elbows (n = 34), respectively, the reverse was the case for MCP joints, wrists and PIP joints, where tenderness was observed in 136, 118 and 119 patients, respectively.
Proportions with current joint involvement decreased somewhat over time, in particular PIP joint synovitis and tenderness (Supplementary Tables 3 and 4). Overall, there was a greater reduction in tenderness of different joints in the UE from inclusion until 10 years of follow up, compared to the reduction in synovitis (Additional le 1 ( Supplementary Tables 3 and 4)).
At inclusion, higher HAQ-DI-UE scores were found in those with current wrist synovitis MCP synovitis and shoulder synovitis, compared to those without such joint involvement (Fig. 4A). There were also signi cant associations between tenderness of elbows, wrists, MCP joints and PIP joints with higher HAQ-DI-UE (Fig. 3B).
Patients with wrist synovitis had signi cantly higher HAQ-DI-UE scores at 6 months, 1 year and 5 years, but not at 2 and 10 years (Additional le 1 (Supplementary table 3)). Tenderness was associated with higher HAQ-DI-UE, in wrists and elbows at all time points from inclusion to 10 years of follow up and in MCP and PIP joints through 5 years and in shoulders from 6 months to 10 years (Additional le 1 (Supplementary Table 4)).
Joint involvement and disability related to the upper extremities -multivariate analyses. In bivariate and multivariate linear regression models on the relation between joint involvement and HAQ-DI-UE at inclusion and after 6 months, the residuals were normally distributed (Shapiro-Wilk statistic > 0.90). Models including data at other time points did not ful ll normality of distribution for the residuals.
Results of the linear regression analyses at inclusion and at 6 months are shown in Table 2. In multivariate analysis, there was a signi cant association between wrist synovitis and higher HAQ-DI-UE at inclusion (B = 0.37; 95% CI 0.20 to 0.54) but not at 6 months ( Table 2). The presence of PIP tenderness was associated with higher HAQ-DI-UE at inclusion, independent of other joint involvement (  HAQ-DI-UE = Health Assessment Questionnaire Upper Extremities; MCP joints = metacarpophalangeal; PIP joints = proximal interphalangeal joints; CI = Con dence intervals.

Discussion
In this study of patients with early RA, the HAQ-DI-UE decreased signi cantly from inclusion to the 6month follow-up. HAQ-DI-UE levels were stable between 6 month and 2 years, whereas between 2 years and 5 years, there was a signi cant increase. Furthermore, there were strong correlations for HAQ-DI-UE with grip force and PROMs at all time points, and moderate to weak correlations with joint counts and laboratory parameters of in ammation. The strong and consistent negative association between grip force and disability related to the UE underlines the importance of the grip in this context.
Involvement of several joint groups contributed to disability related to the upper extremities. At inclusion wrist synovitis and tender PIP joints had both an independent impact on HAQ-DI-UE, whereas tenderness of the shoulder and the wrist had a greater importance at 6 months.
Our results suggest that patients with poor PROMs in early RA are at increased risk of persistent limitations in daily life activities that are performed using the UE. Previous studies have investigated disability overall, and not the HAQ-DI-UE. Disease activity, measured as swollen and tender joint counts, as well as pain, have both been shown to be associated with the total HAQ score (32). Several other studies demonstrated similar results, with associations between high disease activity and impaired PROs like HAQ (17,(33)(34)(35).
One of these compared two large inception cohorts of patients with early RA, the Early Rheumatoid Arthritis Study (ERAS, inclusion period 1986-1997), and the Norfolk Arthritis Register (NOAR, 1990(NOAR, -1994, both from the United Kingdom (33). Both these studies analyzed trajectories, i.e. patterns of progression of disability (33). The overall pattern for HAQ-DI was similar to our observations on HAQ-DI-UE. They also reported an association with DAS28 (33) but did not investigate the importance of the subcomponents (swollen, tender, ESR, VAS pain and VAS Global), in contrast with our study.
In a study based on the Canadian Early Arthritis Cohort (CATCH), the HAQ-DI and DAS28 decreased signi cantly over time from baseline to 24 month (34), but no further follow-up was reported. The Swedish TIRA study (Early Interventions in Rheumatoid Arthritis) group reported that disability decreased signi cantly during the rst year after inclusion in their inception cohort (1). This was followed by gradually worsening of disability up to the 8-year follow-up (1). In that study, grip force and pain intensity were major contributors to disability measured using HAQ-DI (1), similar to our ndings on HAQ-DI-UE. Taken together, these studies suggest that although disability may decrease in early RA, subsequent worsening of disability is common, in particular in those with active disease and severe pain (1,33,34).
In a study of early RA, performed by the Swedish BARFOT (Better Anti-Rheumatic FarmacOTherapy) group, outcomes including the disease activity score, VAS pain, Patient Global Assessment (PatGA) and HAQ were followed during ve years (35). Interestingly, patterns were similar in patients diagnosed 1992-the latter group was more actively treated (35). The authors suggest that other mechanisms than in ammation might be of importance for persistent disability and pain (35). This is in accordance with the results of the present study, with limited correlation between HAQ-DI-UE and markers of in ammation, in particular with long term follow-up.
A major impact of reduced grip force on disability has been demonstrated in several previous studies (1,3,18,19,32), including a survey of the cohort investigated in the present study (6). This suggests that hand training should be bene cial in early RA. Indeed, structured rehabilitation, including tailored exercise program for the distal UE have been shown to improve the grip strength and reduce the impact of the disease on the individual (36)(37)(38). Furthermore, multiprofessional interventions may prevent progressive disability in RA (1,36,37,39).
Our results also showed a varying contribution of joint involvement to disability related to the UE. Like in the present cohort, a Japanese study of 3457 patients with established RA that analyzed the importance of change in joint involvement over time found that involvement of wrist and shoulder joints contributed signi cantly to worse HAQ scores in patients with RA, whereas for the small joints in the hand (MCP and PIP), the effect was modest (22). In our study, there was a greater importance for PIP joint involvement, which may characterize early RA. Furthermore, several studies found no (19,40,41), or limited (22), impact of involvement of the small joints in the hands on HAQ-DI. This, contrasts with our result that PIP joint tenderness was independently associated with HAQ-DI-UE at baseline. Impairment of small joints could in uence particular subdimensions of the HAQ score (22), which might explain our nding.
A German study of RA patients describes that wrist involvement is strongly associated with disability, assessed by HAQ-DI (42). In a large cohort of 4 530 patients with established RA, the most frequently affected joint was shown to be the wrist (45.5%). Together with shoulders and elbows, wrist joints accounted for the greatest contributions to disease activity, HAQ-DI and long term functional prognosis (41). Several other studies demonstrated a major impact of wrist involvement on grip force in RA (19,22,(43)(44)(45).
A limitation of the study is that the HAQ-DI-UE subscore of the HAQ-DI has not been studied before, and is not a validated outcome measure. A corresponding subscore for the lower extremities (LE), (HAQ-DI-LE) which has been used in previous studies (29)(30)(31), included 10 questions that cover activities that are mainly dependent on function of the LE. Due to the heterogeneity of RA and the differential impact on various joints in individual patients, there is a rational for constructing separate questionnaires, sorting the 20 sub dimensions in two groups, one for the LE and one for the UE. Further studies should evaluate the utility of this approach.
Another limitation in the present study is related to the loss of patients that for various reasons have not been assessed at all follow-up visits. This may lead to underestimation of long term disability in RA. However, changes over time were assessed using paired analyses of patients with data at two consecutive visits. Furthermore, due to the limited sample size, some factors that contribute to higher HAQ-DI-UE may not be identi ed in this study.
Strengths of this study include the structured longitudinal follow-up of an inception cohort from a de ned catchment area. Therefore, selection bias is not a major issue in this study, and the results could be generalized to patients with RA seen in clinical practice. On the other hand, they may not apply to other ethnic or geographic settings, or patients managed using completely different strategies for pharmacologic treatment and rehabilitation. The protocol includes detailed documented joint assessments performed by the same rheumatologist at all follow-up visits, and structured assessment of PRO's including the widely used HAQ-DI (46)(47)(48)(49) at all visits.

Conclusion
In conclusion, in this study of patients with early RA, disability related to the upper extremities decreased signi cant from inclusion to the 6 month follow-up, and increased signi cantly again after 2 years. HAQ-DI-UE scores correlated strongly with grip force and patient reported outcomes at all time points and to a lesser extent with joint counts and laboratory markers of in ammation. The results underline the major impact of pain and reduced grip strength in early RA.

Declarations
Ethics approval and consent to participate: All procedures performed in studies involving human participants were in accordance with the ethical standards of the regional research committee (Regional Ethical Review Board for southern Sweden, Lund, Sweden-LU 410-94, January 30. 1995 and LU 311-02, June 10, 2002) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. All patients gave their written informed consent to participate, and the study was approved by the Regional Ethical Review Board for southern Sweden (Lund, Sweden).

Consent for publication:
Not applicable