A randomized controlled trial of early intervention with intraarticular corticosteroids followed by sulfasalazine versus conservative treatment in early oligoarthritis

Authors


Abstract

Objective

To determine the outcome after 52 weeks of early intervention with intraarticular corticosteroid injections followed by sulfasalazine versus conservative therapy in patients with recent-onset oligoarthritis in a randomized controlled trial.

Methods

Patients with ≤4 joints with clinical synovitis (disease duration ≤12 months) were randomized to early intervention (EI) with intraarticular methylprednisolone into all synovitic joints or to conservative treatment (CT) with nonsteroidal antiinflammatory drugs alone. Sulfasalazine was administered in both groups for persistent disease or disease that evolved into a polyarthritis. Primary outcome was complete response (CR) defined as the absence of synovitis at 52 weeks. Secondary outcomes included CR at weeks 4 and 12, function (Health Assessment Questionnaire), pain (0–100-mm visual analog scale), and work status.

Results

Fifty-nine patients (34 men, 25 women; mean age 32.9 years; median early morning stiffness 30 minutes) were randomized. At baseline, two-thirds reported that they were work impaired. At 52 weeks, 81% of patients in the EI group achieved CR compared with 57% in the CT group (χ2 = 3.833, 1 df, P = 0.05). In addition, 45% of patients in the EI group received sulfasalazine as opposed to 14% in the CT group (χ2 = 5.156, 1 df, P = 0.019). There were no differences in physical disability or work impairment between the treatment groups.

Conclusion

Oligoarthritis has a significant impact on function and work ability. Patients treated with EI using intraarticular corticosteroids followed by sulfasalazine therapy if resistant demonstrated reduced synovitis 12 months after treatment compared with those initially treated with more conservative therapy.

INTRODUCTION

The term oligoarthritis refers to an inflammatory arthritis affecting ≤4 joints. It is the most common pattern of disease seen in the seronegative arthritides such as reactive arthritis (ReA), psoriatic arthritis, and inflammatory bowel disease–associated arthritis, as well as in so-called undifferentiated arthritis. It often affects young patients and causes a high level of morbidity (1, 2). The outcome in patients with oligoarthritis is variable, with some studies describing as many as 75% of patients with ReA having persistent disease at 6 months (2), whereas other cohort studies have suggested rates of persistence as low as 15% (3). Furthermore, it is known that in a proportion of patients with an undifferentiated oligoarthritis, the disease evolves into diseases such as rheumatoid arthritis (RA) (4). Therefore, it is hypothesized that early intervention may alter the disease course and avoid persistence.

Conventional treatment includes the use of nonsteroidal antiinflammatory drugs (NSAIDs), intraarticular corticosteroids, and systemic therapy with disease-modifying antirheumatic drugs (DMARDs) such as sulfasalazine (SSZ) in patients with resistant or progressive disease. Previous studies have shown that patients with oligoarthritis present as frequently to early arthritis clinics as those with RA (1, 5). However, there is a lack of evidence regarding treatment options in this group of patients and protocols have only recently begun to be evaluated. Corticosteroids effectively suppress synovitis with a fast mechanism of action and are commonly used for symptom control. In a previous open-label study, we reported the outcome of a protocol of intraarticular steroid injections into all clinically involved joints in patients presenting with a recent-onset oligoarthritis (6). Results demonstrated that despite early treatment and an overall initial improvement, the majority of patients were still symptomatic at 12 months. This finding indicates a significant level of disease persistence in a young population of patients and is consistent with other published studies (2, 3, 7). The purpose of the current study was to assess the efficacy of early intervention with intraarticular corticosteroids followed by therapy with SSZ in patients with persistent disease compared with initial NSAID therapy alone followed by later intervention with intraarticular corticosteroid and SSZ in cases of disease persistence. The study assessed remission rates at 52 weeks in the 2 groups and evaluated the impact of treatment on function and work status.

PATIENTS AND METHODS

Study design and randomization protocol.

The study was a 52-week, single-center, randomized controlled trial and was approved by the local ethics committee. All patients were randomized into 1 of the 2 treatment groups via a random number sequence in blocks of 4 patients to ensure similar numbers in each treatment group. Once informed, written consent was obtained, and the physician contacted an independent person who then gave the physician the group allocation. The randomization sequence was conducted at a separate site by a statistician unconnected to the final analysis of the results.

Participants and baseline assessment.

Patient recruitment was drawn from the Leeds Early Arthritis Clinic (EAC), which is part of the Leeds Early Arthritis Project (LEAP) network of clinics. The majority of patients presenting to the EAC come from a primary care setting where physicians are encouraged to refer any patients with a history of inflammatory joint symptoms (i.e., swollen or tender joints associated with early morning stiffness responsive to NSAIDs) as early as possible. Patients can be fast tracked after discussion, thereby allowing for a prompt specialist review that may occur within hours or days of symptom onset. Consecutive new patients presenting to the EAC with objective evidence of synovitis in ≤4 joints (excluding a distal interphalangeal joint alone) of <12 months' duration were invited to participate in the study. Synovitis was defined as the presence of at least 2 of the following 3 clinical criteria: swelling, tenderness, or decreased range of movement. Patients were excluded from the study if they had had any serious infectious disease within the previous 3 months; had unstable systemic disease; were receiving treatment with immunosuppressive drugs, cytotoxic drugs, or DMARDs as well as corticosteroids; or had a diagnosis of gout.

Patients were assessed according to the standard LEAP protocol (6) including swollen and tender joint counts, which were evaluated by the same trained metrologist in all cases. Blood tests were performed at every visit and included C-reactive protein level (CRP) measurements. In addition, blood was tested at baseline for rheumatoid factor (RF), HLA–DRβ1*04 or *01 alleles, and HLA–B27. A comprehensive microbiologic assessment was performed where clinically indicated, including rubella and parvovirus IgM titers and anti-streptolysin oligonucleotide (ASO) titers, and serum testing was performed for Salmonella, Campylobacter, and Yersinia if there was a recent history of diarrhea. First-voided early morning urine was examined for Chlamydia using a polymerase chain reaction (PCR) method (8) according to protocol in men age <60 years and premenopausal women.

Clinical protocol.

Assessment for clinical synovitis was carried out by 2 rheumatologists and consensus was reached where there was disagreement. Patients with a history suggestive of a genitourinary infection or those found to have urine PCR tests positive for Chlamydia were referred directly to genitourinary medicine for assessment, and all such patients commenced appropriate antibiotic treatment in addition to the study therapy. Once patients had been assessed, they were randomized to 1 of the 2 treatment arms. Patients were examined at baseline and weeks 4, 12, 26, and 52. In the early intervention (EI) group, patients received an intraarticular injection of up to 160 mg of methylprednisolone into all joints with clinical synovitis (Table 1). All joints deemed to have clinical synovitis at the next 2 visits were reinjected with corticosteroids. In addition, NSAIDs were permitted for the duration of the study. In the conservative treatment (CT) group, NSAIDs were initiated at baseline. If patients were already taking an NSAID, the treatment was changed to a different preparation to see whether efficacy could be improved. Where present, effusions were aspirated on 2 occasions, but no intraarticular corticosteroids were administered. The third time a joint needed aspiration, the patient was considered to have failed treatment. At this stage, intraarticular corticosteroids were administered as needed. In both groups, SSZ was initiated in patients whose disease evolved into a polyarthritis (i.e., ≥5 synovitic joints) and in those with persistent disease requiring intraarticular corticosteroids to a single joint on at least 2 occasions. Clinical assessments and joint injections were performed at all end points by the same clinician (HM-O) who was blinded to the randomization code but not to the treatment.

Table 1. Dose of intraarticular methylprednisolone used per joint
JointDose, mg
Proximal or distal interphalangeal10
Metacarpophalangeal10
Wrist40
Elbow40
Shoulder40
Knee80
Ankle40
Metatarsophalangeal10
Interphalangeal10

Outcome measures.

The primary outcome measure was the absence of synovitis, defined as complete response (CR) at week 52, and was determined by clinician assessment. Secondary measures included CR at weeks 4 and 12, and for weeks 4, 12, and 52 the following were measured: Health Assessment Questionnaire score (HAQ; scored on a 0–24 scale) (9); physician's and patient's disease assessment scores (0–100-mm visual analog scale [VAS]); patient's perception of pain (VAS); and use of concomitant medication (i.e., DMARDs) at weeks 4, 12, and 52. In addition, work ability was assessed at every end point using an in-house developed questionnaire. Patients were asked to give a yes/no answer to the following questions: “Are you able to work at the moment? If not, please indicate whether you are retired”; “If able to work, are you in work?”; and “Do you feel your job is impaired by your current condition?” In addition, patients were asked to indicate whether they were on sick leave at the time of the interview and to specify the number of days of absence from the work place since symptoms started. In the event of job loss, patients were asked to specify whether this had been due to their musculoskeletal condition.

Statistical analysis.

This study was designed as part of a large imaging project in early oligoarthritis. The sample size calculations were based on a previous study (6) where ultrasound examination found evidence of subclinical inflammation that determined disease persistence (10). Based on this finding and expecting 35% of suboptimally treated patients to develop persistent disease as opposed to 15% of the optimally treated group, it was estimated that 150 patients would be required to show a statistical difference with α level of 5% and power of 80%. Ultrasound results are discussed elsewhere (10). Our protocol included the clinical data as coprimary analyses. Recruitment was, however, much slower than anticipated and the study was closed before the numbers required for analysis of the ultrasound data could be achieved. As such, the analysis was undertaken focusing on the clinical outcome. Parametric and nonparametric tests were applied as necessary using the chi-square and Mann-Whitney U tests where appropriate. The intent-to-treat principle with last observation carried forward (LOCF) was used in the event of missing data.

RESULTS

A total of 72 patients were screened between December 1998 through May 2002, of which 66 were eligible to participate in the trial. Of these 66, 60 patients were randomized, although 1 patient refused consent after randomization. A total of 59 patients (34 men, 25 women; mean age 32.9 years; median early morning stiffness 30 minutes) were randomized and started treatment in either the EI group (n = 31) or CT group (n = 28). The participant flow chart is presented in Figure 1.

Figure 1.

Participant flow diagram. DMARD = disease-modifying antirheumatic drug.

Baseline characteristics.

A total of 27 (46%) patients had CRP levels >10 mg/liter (median CRP level 9 mg/liter), with a median pain score of 38 mm and a median disease activity of 44 mm. Patients reported a median HAQ score of 4 (range 0–13). Although there was individual variation between the groups (54% of the patients in the EI group had elevated CRP levels compared with 36% in the CT group and the median pain score was higher in the CT group than in the EI group [43 mm versus 38 mm]), there were no statistically significant differences between them (Table 2). At study entry, the median symptom duration was 10 weeks (range 1–68), with no difference between groups. Sixteen (52%) of the 31 patients in the EI group and 16 (57%) of the 28 patients in the CT group presented with a monarthritis. Four patients were RF positive at baseline, 2 in each group. HLA–B27 antigen results were available for 87% of the EI group and 93% of the CT group; of these, 37% (10 of 27) and 23% (6 of 26), respectively, were B27 positive. In addition, 64% (14 of 22) of the EI group and 63% (15 of 24) of the CT group were either DR1 and/or DR4 positive (Table 2).

Table 2. Patient characteristics at baseline*
CharacteristicEI groupCT group
  • *

    EI = early intervention; CT = conservative treatment; HAQ = Health Assessment Questionnaire; CRP = C-reactive protein; RF = rheumatoid factor.

Number of patients3128
Sex, male:female18:1316:12
Age, mean ± SD years34.4 ± 11.531.4 ± 10.3
Early morning stiffness, median minutes3030
Disease duration, mean weeks1210
Baseline HAQ score, median (range)4 (0–13)3.5 (0–13)
Swollen joint count, median11
Tender joint count, median23
CRP >10 mg/liter, no. (%)17 (54.8)10 (35.7)
RF >40 IU/liter, no. (%)2 (6.4)1 (3.5)
HLA–B27 positive, no.3723
HLA–DR4/1 positive, no.6463

Microbiologic assessment yielded positive results in 14 patients, 7 in each group. Chlamydia trachomatis was identified in 9 patients (5 in the EI group, 4 in the CT group) who had positive PCR tests in first-voided urine samples and positive serology. In addition, positive serology for Chlamydia pneumoniae was identified in 1 patient (EI group). Three patients had positive serology for Campylobacter (1 in the EI group and 2 in the CT group). ASO titers were found to be elevated in 1 patient in the EI group who had a recent history suggestive of streptococcal infection. Details of clinical diagnosis at week 52 are given in Table 3.

Table 3. Clinical diagnoses of patients at week 52*
DiagnosesEI group (n = 31)CT group (n = 28)
  • *

    IBD = inflammatory bowel disease; see Table 2 for additional definitions.

Undifferentiated oligoarthritis67
 Monarthritis36
Reactive arthritis  
 Genitourinary65
 Enteropathic11
 Other (poststreptococcal)11
IBD-associated arthritis20
Psoriatic arthritis31
Undifferentiated spondylarthritis33
Rheumatoid arthritis23
Other  
 Osteoarthritis10
 Sarcoidosis10
 Diabetic Charcot's joint10
 Connective tissue disease10
 Osteogenesis imperfecta01

Primary and secondary analyses.

At week 52, 25 (81%) of 31 patients in the EI group had no clinical evidence of synovitis (CR) compared with 16 (57%) of 28 in the CT group (χ2 = 3.833, 1 df, P = 0.05). Although both groups demonstrated a reduction in synovitis, these differences were not statistically significant at weeks 4 and 12 (Figure 2). Of the 4 patients that were positive for RF at baseline, 3 patients' disease evolved into a polyarthritis and fulfilled American College of Rheumatology (ACR; formerly the American Rheumatism Association) criteria for RA (11) at 52 weeks. The fourth patient's disease remained an undifferentiated oligoarthritis. Function as measured by HAQ scores improved over the study period in both groups (median score of 1 at week 4 for EI group compared with 3 for the CT group), although this was not statistically significant as either a raw score (Figure 3A) or as a percentage change in the baseline data. On assessment of pain, there was a short-term improvement in the patients receiving steroids who reported less pain at 4 weeks (Mann-Whitney U test score 237, P = 0.012), although this response was not maintained at weeks 12 and 52 (Figure 3B). There were no significant differences between the EI and the CT groups in physician disease assessment score, patient disease assessment score, or the swollen joint count at either 12 or 52 weeks (Table 4). Of note, 14 (45%) patients in the EI group were prescribed a DMARD as opposed to 4 (14%) patients in the CT group (χ2 = 5.156, 1 df, P = 0.019). As previously described, SSZ was the DMARD of choice and was tolerated in all patients with the exception of 3 who received azathioprine (1 patient with inflammatory bowel disease) and methotrexate and hydroxychloroquine (2 cases of intolerance to SSZ). In addition, 22 (70%) patients in the EI group received an NSAID at some point during the study period.

Figure 2.

Primary analysis: the percentage of patients with clinician-assessed synovitis. * Statistically significant difference (P = 0.05). EI = early intervention; CT = conservative treatment.

Figure 3.

Secondary analysis. A, The median Health Assessment Questionnaire (HAQ) scores (0–24 scale). Data are presented as means and 95% confidence intervals. There were no significant differences between the groups. B, Patients' self assessment of pain during the study period using a 0 (no pain) to 100 (worst pain imaginable) visual analog scale. The graph is presented as medians ± 95% confidence intervals and * indicates a statistical difference. EI = early intervention; CT = conservative treatment.

Table 4. Secondary outcome measures*
 Baseline12 weeks52 weeks
EI groupCT groupEI groupCT groupEI groupCT group
  • *

    Values are the median (range) unless otherwise indicated. DMARDs = disease-modifying antirheumatic drugs; DAS = disease assessment score; see Table 2 for additional definitions.

  • CRP values <5 mg/liter were converted to zero. None of the secondary analyses were significant.

Patient pain38 (5–95)43 (10–100)23.5 (0–73)18 (0–95)5 (0–73)19 (0–83)
Patient disease activity51 (7–80)47.5 (14–95)32 (0–75)17 (0–64)1.5 (0–58)11.5 (0–65)
Use of DMARDs, no.0060144
Physician DAS45 (10–80)41 (18–81)23 (0–86)15.5 (0–69)4 (0–74)9 (0–75)
 CRP >10 mg/liter, no. (range)17 (0–132)10 (0–65)16 (0–32)15 (0–42)18 (0–49)13 (0–13)
Swollen joint count1 (1–4)1 (1–4)1 (1–6)1 (1–5)0 (0–1)0 (0–7)

With regards to work status, all patients except 1 were of working age at the start of the study and 43 (73%) were employed. Two-thirds of these patients (63%) said that they were work impaired. Overall, there were 464 work days lost leading up to treatment, representing an average of 11.6 days lost per working participant from symptom onset. At baseline, men reported greater work impairment than women (20 [61%] of 33 versus 8 [32%] of 25; P = 0.027), although no difference was seen by week 52. There were 440 working days lost over the 52 weeks of treatment, representing an average of 10.7 days lost per working participant. No difference was seen between the 2 treatment groups in either perceived work impairment or working days lost during the study period. One person lost a job during the study as a result of symptoms, but gained employment within 8 weeks. One patient had to change duties. Of the patients who were not in work (37%) at the beginning of the study, 4 of 10 said that they were not able to work because of their symptoms.

DISCUSSION

Oligoarthritis is a common condition with a variable outcome that can lead to high morbidity. Although conventional treatments include the use of intraarticular corticosteroids, NSAIDs, and DMARDs, there is no evidence in favor of a particular approach at a given time. This study was designed to examine the outcome of patients randomized to either EI with intraarticular corticosteroids plus SSZ for polyarthritis or persistent disease, or a period of conservative treatment with NSAIDs alone in the first instance followed by intraarticular steroids and SSZ if disease evolved to a polyarthritis or there was clinical evidence of persistent disease. Although the number of patients with synovitis in both treatment groups was reduced over the study period, more patients in the EI group achieved CR with complete absence of synovitis at 1 year compared with the CT group. However, more patients in the EI group received SSZ, therefore it is difficult to discern from the results of this study whether the beneficial effect observed with early intervention was due to a direct effect of the early use of intraarticular corticosteroids, SSZ, or a combination of the 2.

Corticosteroid injections are widely used in clinical practice for short-term relief of symptoms in patients with recent-onset oligoarthritis. In our cohort there was a rapid improvement in pain in the group treated with intraarticular corticosteroids when compared with the group treated with oral NSAIDs, which provides support for the benefit of this approach. Furthermore, the maximum reduction in synovitis due to intraarticular corticosteroids would be expected to occur in the weeks soon after injections, but there was no significant difference between the 2 groups at the secondary time points of 4 and 12 weeks with respect to CR. In this study, SSZ was started when patients' disease evolved to a polyarthritis (i.e., ≥5 joints with clinical synovitis) or when there was evidence of persistent synovitis despite 2 previous injections with intraarticular corticosteroids. This necessarily led to more patients in the EI group starting SSZ and at an earlier stage in their disease than patients in the CT group. In the EI group, 45% of patients who completed the study received SSZ, which was significantly more than the 14% of patients who completed the study in the CT group. Taken together, these findings suggest that the long-term beneficial effect of the EI protocol is unlikely to be due solely to intraarticular corticosteroids. These findings also suggest that careful injection of all clinically synovitic joints with intraarticular corticosteroids alone does not prevent disease progression or persistence.

In a previous study, 63 patients with an early oligoarthritis (6) were treated with intraarticular corticosteroids into all joints with clinical synovitis and demonstrated improvement following these injections within 2 weeks, which was maintained at the 52-week followup. Although SSZ was also used, this was not part of the protocol. The rate of persistence of synovitis at 52 weeks in this cohort was 47%, which is similar to that of the CT group (43%) in the current study and again far in excess of the rate of persistence in the EI group (20%). It is of note that the rate of use of SSZ in this historical cohort was 25%, which is more in keeping with the CT group of the current study, suggesting that the improvement in outcome seen in the EI group may be due to the earlier use of DMARDs. Comparisons between the cohorts, however, need to be made with caution as there are some subtle differences between them. In the earlier study, the number of joints with clinical synovitis was higher (with a mean of 2.3 versus a median of 1 in the current study) and 22% of patients were seropositive for RF compared with 5% in the current study. Indeed, 21% of patients in the initial cohort fulfilled ACR criteria for RA at 52 weeks compared with 8% in the current study. Interestingly, persistence in these cohorts was found to be determined by the amount of subclinical synovitis detectable by ultrasound (10). This may explain why injection of only clinically involved joints with intraarticular corticosteroids did not prevent disease progression or persistence.

The present study has a number of limitations that need to be taken into account when interpreting these results. First, although recruitment was undertaken over a 2.5-year period, the expected sample size was not reached. The slow recruitment and associated resource implications forced an early termination with the consequent underpowering of the study, although a preliminary analysis revealed that the primary outcome had been achieved. In the absence of the expected ultrasound data, we used the clinical outcome as the primary outcome, and we acknowledge that this is technically a secondary analysis and may result in some difficulty in interpreting the results. Second, the decision to use LOCF analyses was based on a pragmatic, clinically based approach to the data based on previous experience. We do, however, acknowledge that this method makes the assumption that the dropout sample represents the group as a whole. However, we had anticipated that patients would self withdraw if they were healthy, and this was indeed the case for 4 patients, 2 on each arm. Of those patients lost to followup, the majority were not in remission on their last appointment and we therefore decided that the LOCF was a reflection of the clinical situation. Our choice for this data analysis was based on 2 factors: our analysis of the raw data indicated that there was approximately the same proportion of patients in remission for each group as there was with the LOCF, and our limited sample size made statistical modeling of the dropout group problematic. Third, our indicators of work disability were based on self reporting and have not been validated. Nonetheless, an important observation from this study was the as yet unrecognized high level of disability associated with oligoarthritis, with the majority of patients being work impaired at baseline as a result of their symptoms. This has important personal and socioeconomic implications for a predominantly young population of patients who represent, in their majority, an active workforce. The positive work outcome of this cohort after therapy underscores the importance of early assessment and treatment in these patients.

We believe that some important observations can be drawn from the examination of these cohorts. For example, an early assessment was performed in both groups (2 weeks in the cohort described by Green et al [6] and 4 weeks in the current study), which allowed for the prompt initiation of DMARD therapy. Indeed, the benefit observed in the EI group may have been multifactorial and due not only to the early use of corticosteroids but also of SSZ. Ideally, any effective therapy would cause a rapid and sustained suppression of the inflammation process, allowing for prompt remission and improved outcome. However, a considerable amount of patients (20% in the EI group and 43% in the CT group) in our study still had persistent disease after 52 weeks, which makes the case for more aggressive intervention at baseline in the hope of a better chance of remission. Future randomized research studies in these patients should look at the efficacy of an intensive therapeutic package at baseline, i.e., intraarticular steroids and DMARDs versus NSAIDs and DMARDs if there is persistent disease at 2–4 weeks, or tumor necrosis factor blockers versus intraarticular steroids and DMARDs.

It is proposed that clinical guidelines should be developed for the treatment of patients presenting with an oligoarthritis of recent onset. Based on the findings from the previous study (6) and the current study, and as a preliminary improvement over the current practice, while further studies are developed it is proposed that all synovitic joints should be injected with intraarticular corticosteroids at presentation and then reassessed at an early time point (e.g., 2–4 weeks). If persistent synovitis is found, DMARD therapy (e.g., SSZ) should be commenced without delay. If CR is achieved, DMARDs could be deferred because there is a high probability of long-term remission (6).

In conclusion, this is the first randomized study looking at the treatment of patients in the early stages of oligoarthritis. This study compares early intervention with intraarticular corticosteroids followed by SSZ for patients with persistent disease with a more conservative approach of only NSAIDs. Although this study was terminated without reaching the desired power, these results indicate a significant benefit from an early intervention protocol, and it is suggested that this is mainly a DMARD effect. We also report a high level of morbidity in patients with oligoarthritis in terms of work disability, which has been underrecognized. Long-term outcome of an EI approach is better when compared with historical cohorts (6) and these data have been used to recommend management guidelines for such patients.

AUTHOR CONTRIBUTIONS

Dr. Emery 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 design. Drs. Green, Wakefield, Proudman, and Emery.

Acquisition of data. Drs. Marzo-Ortega, Green, Wakefield, Proudman, and Emery.

Analysis and interpretation of data. Drs. Marzo-Ortega, Green, and Emery.

Manuscript preparation. Drs. Marzo-Ortega, Green, Proudman, and Emery.

Statistical analysis. Dr. Keenan.

Acknowledgements

We thank Sisters Claire Brown, Louise Cunningham, and Susan Burns for their help in metrology and data collection; Professor H. Zeidler (Medizinsche Hochschule Hannover Rheumatology, University of Hannover, Germany) for advice on microbiology assessments; Professors Dennis McGonagle and Philip Conaghan for helpful comments on the manuscript; and medical colleagues in the Yorkshire region for patient referral.

Ancillary