Article Text

Extended report
SF-36 summary and subscale scores are reliable outcomes of neuropsychiatric events in systemic lupus erythematosus
  1. J G Hanly1,
  2. M B Urowitz2,
  3. D Jackson3,
  4. S C Bae4,
  5. C Gordon5,
  6. D J Wallace6,
  7. A Clarke7,
  8. S Bernatsky8,
  9. A Vasudevan9,
  10. D Isenberg10,
  11. A Rahman10,
  12. J Sanchez-Guerrero11,
  13. J Romero-Diaz11,
  14. J T Merrill12,
  15. P R Fortin2,
  16. D D Gladman2,
  17. I N Bruce13,
  18. K Steinsson14,
  19. M Khamashta15,
  20. G S Alarcón16,
  21. B Fessler16,
  22. M Petri17,
  23. S Manzi18,
  24. O Nived19,
  25. G Sturfelt19,
  26. R Ramsey-Goldman20,
  27. M A Dooley21,
  28. C Aranow22,
  29. R Van Vollenhoven23,
  30. M Ramos-Casals24,25,
  31. A Zoma26,
  32. K Kalunian24,
  33. V Farewell3,
  34. for the Systemic Lupus International Collaborating Clinics (SLICC)
  1. 1Department of Medicine and Department of Pathology, Division of Rheumatology, Queen Elizabeth II Health Sciences Centre and Dalhousie University, Halifax, Nova Scotia, Canada
  2. 2Centre for Prognosis Studies in the Rheumatic Diseases, Toronto Western Hospital and University of Toronto, Ontario, Canada
  3. 3MRC Biostatistics Unit, Institute of Public Health, University Forvie Site, Cambridge, UK
  4. 4Department of Rheumatology, Hanyang University Hospital for Rheumatic Diseases, Seoul, Korea
  5. 5Rheumatology Research Group, School of Immunity and Infection, College of Medical and Dental Sciences, University of Birmingham, Birmingham, UK
  6. 6Cedars–Sinai/David Geffen School of Medicine at UCLA, Los Angeles, California, USA
  7. 7Divisions of Clinical Immunology/Allergy and Clinical Epidemiology, Montreal General Hospital, McGill University Health Centre, Montreal, Quebec, Canada
  8. 8Divisions of Rheumatology and Clinical Epidemiology, Montreal General Hospital, McGill University Health Centre, Montreal, Quebec, Canada
  9. 9Department of Medicine, SUNY Downstate Medical Center, Brooklyn, New York, USA
  10. 10Centre for Rheumatology Research, University College London, London, UK
  11. 11Instituto Nacional de Ciencias Medicas y Nutrición, Mexico City, Mexico
  12. 12Department of Clinical Pharmacology, Oklahoma Medical Research Foundation, Oklahoma City, Oklahoma, USA
  13. 13Arthritis Research UK Epidemiology Unit, School of Translational Medicine, Manchester Academic Health Sciences Centre, The University of Manchester, Manchester, UK
  14. 14Center for Rheumatology Research, Landspitali University Hospital, Reykjavik, Iceland
  15. 15Lupus Research Unit, The Rayne Institute, St Thomas' Hospital, King's College London School of Medicine, London, UK
  16. 16Department of Medicine, University of Alabama at Birmingham, Birmingham, Alabama, USA
  17. 17Department of Rheumatology, Johns Hopkins University, Baltimore, Maryland, USA
  18. 18Division of Rheumatology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA
  19. 19Department of Rheumatology, University Hospital Lund, Lund, Sweden
  20. 20Northwestern University and Feinberg School of Medicine, Chicago, Illinois, USA
  21. 21University of North Carolina, Chapel Hill, North Carolina, USA
  22. 22Columbia University Medical Center, New York, USA
  23. 23Department of Rheumatology, Karolinska Institute, Stockholm, Sweden
  24. 24UCSD School of Medicine, La Jolla, California, USA
  25. 25Servicio Enfermedades Autoinmunes Hospital Clínico y Provincial, Barcelona, Spain
  26. 26Lanarkshire Centre for Rheumatology, Hairmyres Hospital, East Kilbride, Scotland, UK
  1. Correspondence to Dr J G Hanly, Division of Rheumatology, Nova Scotia Rehabilitation Centre (2nd Floor), 1341 Summer Street, Halifax, NS B3H 4K4, Canada; john.hanly{at}


Objective To examine change in health-related quality of life in association with clinical outcomes of neuropsychiatric events in systemic lupus erythematosus (SLE).

Methods An international study evaluated newly diagnosed SLE patients for neuropsychiatric events attributed to SLE and non-SLE causes. The outcome of events was determined by a physician-completed seven-point scale and compared with patient-completed Short Form 36 (SF-36) health survey questionnaires. Statistical analysis used linear mixed-effects regression models with patient-specific random effects.

Results 274 patients (92% female; 68% Caucasian), from a cohort of 1400, had one or more neuropsychiatric event in which the interval between assessments was 12.3±2 months. The overall difference in change between visits in mental component summary (MCS) scores of the SF-36 was significant (p<0.0001) following adjustments for gender, ethnicity, centre and previous score. A consistent improvement in neuropsychiatric status (N=295) was associated with an increase in the mean (SD) adjusted MCS score of 3.66 (0.89) in SF-36 scores. Between paired visits when the neuropsychiatric status consistently deteriorated (N=30), the adjusted MCS score decreased by 4.00 (1.96). For the physical component summary scores the corresponding changes were +1.73 (0.71) and −0.62 (1.58) (p<0.05), respectively. Changes in SF-36 subscales were in the same direction (p<0.05; with the exception of role physical). Sensitivity analyses confirmed these findings. Adjustment for age, education, medications, SLE disease activity, organ damage, disease duration, attribution and characteristics of neuropsychiatric events did not substantially alter the results.

Conclusion Changes in SF-36 summary and subscale scores, in particular those related to mental health, are strongly associated with the clinical outcome of neuropsychiatric events in SLE patients.

View Full Text

Statistics from

The prevalence of neuropsychiatric disease in patients with systemic lupus erythematosus (SLE) varies from 21% to 95% and recent studies indicate a frequency at the lower end of this range.1,,5 There is a spectrum of potential neuropsychiatric events, as reflected by the American College of Rheumatology (ACR) case definitions for 19 neuropsychiatric syndromes.6 Only 13–38% of all neuropsychiatric events are attributable to SLE depending upon the stringency of the decision rules for determining attribution.7,,11 Regardless of attribution, both cross-sectional3 8 and longitudinal7 11 studies have demonstrated that neuropsychiatric events are independently associated with lower self-reported health-related quality of life (HRQoL).

Clinical trials are necessary to address the optimal management of neuropsychiatric events in SLE patients. Challenges to planning and executing such studies include the lack of validated outcome measures for diverse neuropsychiatric events. Specific outcomes for individual neuropsychiatric events in addition to generic outcomes to allow a comparison with change across all neuropsychiatric events and other manifestations of SLE are required. The objective of the present study was to determine if the Short Form 36 (SF-36) health survey is an appropriate outcome measure for the study of neuropsychiatric events in SLE. In particular, we examined the association between changes in patient-derived SF-36 summary and subscale scores and physician-determined outcomes of neuropsychiatric events in an international, longitudinal cohort study of neuropsychiatric events in SLE patients.

Patients and methods

Research network

This prospective study was performed by the Systemic Lupus International Collaborating Clinics (SLICC)12 between October 1999 and May 2009. The study was approved by the Capital Health Research Ethics Board, Halifax, Nova Scotia, Canada and by the institutional research ethics boards of participating centres in accordance with the Declaration of Helsinki's guidelines for research in humans.


Patients fulfilled the ACR classification criteria for SLE13 and provided written informed consent. The date of diagnosis was the time when four or more of the ACR criteria were first recognised and patients were enrolled up to 15 months thereafter. Data collection included age, gender, ethnicity, education, medication use, SLE disease activity index (SLEDAI)14 and SLICC/ACR damage index (SDI).15 HRQoL was measured by the SF-36.16 Laboratory data included a complete blood count, serum creatinine, urinalysis, anti-double-stranded DNA, C3 and C4.

Neuropsychiatric events

An enrolment window within which all neuropsychiatric events were captured extended from 6 months before the diagnosis of SLE up to the enrolment date. Neuropsychiatric events were characterised using the ACR nomenclature and case definitions for 19 neuropsychiatric syndromes.6 Screening for neuropsychiatric syndromes was done by clinical evaluation and investigations were performed if clinically warranted.

Patients were reviewed annually following enrolment with a 6-month window around the anticipated assessment date. New neuropsychiatric events since the previous study visit and their attribution were determined. For the purpose of this study, patients were required to have a neuropsychiatric event(s) at one assessment, a physician-defined outcome of the event at a subsequent assessment and a completed SF-36 questionnaire at both assessments. Those patients who developed a new or recurring neuropsychiatric event(s) between these assessments were excluded as such events would not have been reflected in the previous SF-36 scores.

Supplementary information was recorded as per the ACR glossary for neuropsychiatric syndromes6 to identify other potential causes (‘exclusions’) or contributing factors (‘associations’) for each of the neuropsychiatric events.6 These ‘non-SLE factors’ were used partly to determine the attribution of neuropsychiatric events. Patients could have more than one type of neuropsychiatric event, and repeated episodes of the same event within the enrolment window or at follow-up annual assessments were recorded once. The date of onset of the neuropsychiatric event was the time of the first episode within the assessment period.

Attribution of neuropsychiatric events

All neuropsychiatric events were recorded and attribution to SLE was determined by decision rules of different stringency (models A and B) as described in detail elsewhere.8 10 Neuropsychiatric events that fulfilled the criteria for model A (the most stringent) or for model B (the least stringent) were attributed to SLE. By definition, all neuropsychiatric events attributed to SLE using model A were included in the group of neuropsychiatric events attributed to SLE using model B. Those events that did not fulfil these criteria were attributed to non-SLE causes.

Outcome of neuropsychiatric events

A physician-generated seven-point Likert scale for neuropsychiatric events comparing the change in neuropsychiatric status between the onset of the event and the time of study assessment was available for each neuropsychiatric event (1, patient death; 2, much worse; 3, worse; 4, no change; 5, improved; 6, much improved; 7, resolved). Due to the occurrence of multiple concurrent neuropsychiatric events in the same patient, there were five possible patient-level outcomes for each pair of assessments: 0, no difference (single event) or no consistent change (multiple events) in neuropsychiatric status; 1, all events improve; 2, all events worsen; 3, some but not all events improve and none worsen; 4, some but not all events worsen and none improve. For the purpose of the primary analyses the patients' paired visits with outcomes 3 and 4, of which there were only 71, were combined with those patients paired visits with neuropsychiatric events without a consistent change, outcome 0.

A second outcome measure was based on a patient-generated SF-36 questionnaire that provided mental component summary (MCS) and physical component summary (PCS) scores, subscale scores16 and SF-6D scores.17 The difference between the scores at the paired visits was used as the response variable. The physician-determined outcome of neuropsychiatric status defined explanatory variables for the regression analyses, which were also adjusted for other relevant variables. The SF-36 scores were not available to the physicians at the time of their evaluation of neuropsychiatric status.

Statistical analyses

Linear mixed-effects models with random intercepts were used so that patient/visits involving the same patient are not assumed to be independent. In particular, conditional on this random intercept, we assume the variance of the change in health scores is constant and that the patients' scores are uncorrelated.



A total of 274 patients (20% of all patients) who had one or more neuropsychiatric event with a documented clinical outcome and two completed SF-36 questionnaires at the appropriate assessments was identified from a cohort of 1400 newly diagnosed SLE patients. Patients were predominantly women (92.3%) with a mean±SD age of 37.1±13.2 years and a wide ethnic distribution although they were predominantly Caucasian (table 1).

Table 1

Demographic and clinical manifestations of 274 SLE patients

At enrolment the mean disease duration was 5.9±4.0 months. The prevalence of individual ACR classification criteria at baseline reflected an unselected patient population and ‘neurological disorder’, which includes seizures and psychosis only, was present in 25 (9%) of the 274 patients. The mean SLEDAI and SDI scores at enrolment were 5.1±5.4 and 0.31±0.70, respectively, indicating moderate global disease activity and minimal cumulative organ damage, respectively. Therapy at enrolment reflected the typical range of lupus medications. The mean interval between assessments was 12.3±1.94 months.

The characteristics of the patients who did not contribute to the analysis were generally similar to those who did. For example, 88% were women and their age at enrolment was 33.6±13.3 years. Their mean SLEDAI and SDI scores at enrolment were 5.5±5.5 and 0.29±0.75. However some regions, were underrepresented in the sample that did not contribute to the analysis (Canada 19%; USA 30%; Mexico 16%; Europe 23%; Asian 13%) and this has some implications for their ethnicity (Caucasian 42%; Hispanic 20%; Asian 19%; black 16%; other 4%). All analyses adjust for region and ethnicity effects and thus the main impact of any difference would be to limit the power for testing interaction effects.

Frequency, attribution of neuropsychiatric events

In 274 patients there were 587 pairs of patient visits that met the criteria in order to contribute to the analysis. One hundred and twenty-four of these patients contributed a single pair of visits and 150 contributed two or more pairs. Multiple events at the previous study visit were common and, in total, 912 events were included in the analysis, which encompassed 17 of the 19 ACR case definitions (table 2); Guillain–Barré syndrome and aseptic meningitis did not occur in any patient.

Table 2

The number of neuropsychiatric events by attribution over the period of study (N=912)

The most frequent events were headache (437 (48%) of 912 events), mood disorders (182 (20%)), cognitive dysfunction (63 (6.9%)), anxiety disorder (58 (6.4%)), cerebrovascular disease (31 (3.4%)), polyneuropathy (31 (3.4%)), mononeuropathy (31 (3.4%)) and seizures (26 (2.9%)). The remaining 11 neuropsychiatric syndromes had a prevalence of less than 6% of all neuropsychiatric events.

Neuropsychiatric events attributed to SLE using alternative attribution models varied from 15% (model A) to 28% (model B) of the 912 neuropsychiatric events (table 2). Of these, 91% affected the central nervous system and 9% involved the peripheral nervous system; 83% were diffuse and 17% were focal events. Using attribution models A and B, respectively, the most frequent neuropsychiatric events attributed to SLE were mood disorders (24–30%), mononeuropathy (15–12%), cerebrovascular disease (14–12%) and cognitive dysfunction (10–15%).

Physician-determined outcome scores for neuropsychiatric events

A summary of neuropsychiatric outcomes at individual patient visits is provided in table 3.

Table 3

Outcome of neuropsychiatric events in SLE patients at individual physician-generated assessments

Change in SF-36 summary and subscale scores and outcome of neuropsychiatric events

The overall difference in the change in MCS scores in three patient groups (all neuropsychiatric events improved, all neuropsychiatric events deteriorated and neuropsychiatric events without a consistent change) reached statistical significance (p<0.0001) following adjustments for gender, ethnicity, research centre and previous SF-36 MCS score (table 4).

Table 4

The mean (±SEM) difference* in change in SF-36 MCS, PCS and SF-6D scores between patients visits whose neuropsychiatric events all improve (1) or all worsen (2) compared with patients lacking a consistent change in neuropsychiatric events (0, 3, 4)

Patients whose neuropsychiatric status consistently improved (N=295) had their estimated adjusted mean (SD) MCS score increase by 3.66 (0.89) more than in patients without a consistent change (N=262) over the same interval. In contrast, for patients whose neuropsychiatric status consistently deteriorated (N=30), the MCS score decreased by an estimated adjusted mean of 4.00 (1.96). For the SF-36 PCS scores the corresponding changes associated with improvement and deterioration in neuropsychiatric status were +1.73 (0.71) and −0.62 (1.58) (p<0.05), respectively. The results are also shown for SF-6D19 in table 4, and qualitatively similar results are obtained using this measure of the SF-36 MCS. Sensitivity analyses were conducted by modifying the definition of clinical change in neuropsychiatric status. Expanding the definition of improvement in neuropsychiatric status, by using all five categories of change as in table 3 and refitting the models, did not substantially change the estimated effects associated with consistent improvement or worsening (see supplementary table S1, available online only).

Changes in the eight SF-36 subscales were in the same direction as the summary scores (p<0.05; with the exception of role physical) (table 5). Unadjusted mean SF-36 subscale scores for the patient/visits for each of the three groups in table 4 are shown in figure 1 using a ‘spydergram’.20

Figure 1

Unadjusted mean Short Form 36 (SF-36) health survey subscale scores in patient/visits in three groups: patients lacking a consistent change in neuropsychiatric events (top panel; outcomes 0, 3 and 4; N=262); patients whose neuropsychiatric events all improve (middle panel; outcome 1; N=295); patients whose neuropsychiatric events all worsen (bottom panel; outcome 2; N=30). Dashed lines denote subscale scores from the first visit of the pair and solid lines denote the second set of scores. The SF-36 subscales are: BP, bodily pain; GH, general health; MH, mental health; PF, physical function; RE, role emotion; RP, role physical; SF, social function; VT, vitality.

Table 5

The mean (±SEM) difference* in change in SF-36 subscale scores between patient visits whose neuropsychiatric events all improve (1) or all worsen (2) compared with patients lacking a consistent change in neuropsychiatric events (0, 3, 4)

Additional variables at the first of the two visits were considered for adjustment: age, education, medications, global SLE disease activity, cumulative organ damage, attribution of neuropsychiatric events, diffuse/focal classification of neuropsychiatric events and disease duration. As headache was such a common event, whether or not it was present was also investigated as a potential variable for adjustment. In each instance the additional variables were entered into the statistical model both with and without an interaction with change in neuropsychiatric status. Again, no substantial changes were observed.

For the MCS, the only notable predictor was SLEDAI (non-neuropsychiatric score estimated without the neuropsychiatric variables) in which there was evidence of an interaction with change in neuropsychiatric status (p<0.01, 2 df test). The dominant interaction with the change in SF-36 score occurred in patients whose neuropsychiatric status deteriorated compared with patients without a consistent change. The mean change in SF-36 MCS was estimated to decrease by approximately 2 points for each 1 unit increase in SLEDAI. As the SD of SLEDAI (non-neuropsychiatric) across the entire sample of patient visits is 4.4, this could translate into an important quantitative effect. However, there were only 30 patients who had a consistent deterioration in neuropsychiatric status between their two assessments (table 2) and this estimate is thus uncertain, with an approximate 95% CI of 0.7 to 3.7, and should be interpreted with some caution. The comparable (non-significant) estimated effect of SLEDAI (non-neuropsychiatric) on the mean difference in SF-36 MCS score between patients with improved neuropsychiatric status and those with no consistent change was estimated to be a decrease of 0.3 per unit change in SLEDAI (non-neuropsychiatric) with a CI of −0.1 to 0.7.

For PCS, the only strong additional potential predictor was the patients' age (p<0.0001). A strong and negative association between age and change in the PCS score was observed as one might expect but this did not affect estimates linked to the change in neuropsychiatric status.


Different outcome measures have been used in the assessment of patients treated for neuropsychiatric SLE. These include clinical assessment,21 22 neuropsychological evaluation of cognition23 24 and neuroimaging.22 25 Although neuropsychiatric events in SLE patients are known to impact negatively on HRQoL,3 7 8 11 the use of standardised measures of HRQoL have not been validated in the assessment of change in neuropsychiatric status in SLE. We have examined the change in SF-36 summary and subscale scores in SLE patients who have had a clinically significant change in neuropsychiatric status over 1 year. In essence, we examined the face validity of SF-36 scores for measuring the clinical outcome of neuropsychiatric events. The results indicate that SF-36 scores, in particular those that reflect self-reported mental health, change in the appropriate direction in association with both clinical improvement and deterioration in a variety of neuropsychiatric events.

The study was conducted within a large, international, inception cohort of SLE patients. The characteristics of the entire cohort have been published elsewhere7 8 and are similar to the subset of patients who were eligible for the current analysis. In contrast to many previous studies of neuropsychiatric SLE, ours was prospective with the specific objective of identifying the characteristics, attribution and outcome of neuropsychiatric events using a predefined annual data collection protocol. The multicentre, international study design provides a basis for assuming that our findings are applicable to the broader community of SLE patients.

Predefined outcomes in studies of therapeutic interventions for neuropsychiatric SLE have been variable and largely determined by the specific neuropsychiatric event under study. Therefore, in the evaluation of interventions to improve cognitive performance23 24 sequential neuropsychological assessment was the primary outcome of brain health. The efficacy of treating neurological manifestations of neuropsychiatric SLE have been determined by sequential neuroimaging alone25 26 or used in combination with clinical variables to provide a composite outcome.27 28 To our knowledge none of the previous approaches has been validated. Given the diversity of neuropsychiatric events in SLE and the lack of clinical trial evidence to support specific interventions, there is a need to determine optimal outcome measures.

Rather than developing a measure for change in a specific neuropsychiatric manifestation, we wished to validate a change in patient self-reported HRQoL as a generic measure for improvement and deterioration in neuropsychiatric status over time. This was a logical step following upon the previous work by ourselves and others showing that neuropsychiatric events have been associated with lower SF-36 scores in both cross-sectional and longitudinal studies of SLE patients.3 7 8 11 29 Physician-determined change in neuropsychiatric status was significantly correlated with change in patient self-reported SF-36 scores. Not surprisingly, the magnitude of change in SF-36 scores was greater for the MCS score than the physical score. Although the SF-36 may be affected by other variables such as global disease activity, cumulative organ damage and concurrent medications, adjustment for these and other variables in the analysis did not negate the association with neuropsychiatric events, with the possible exception of SLE disease activity outside of the nervous system in patients who had deterioration in neuropsychiatric status. It was also apparent that attribution of neuropsychiatric events to SLE and non-SLE causes did not change the results, indicating that this is a valid outcome for all neuropsychiatric events in SLE patients, even though the majority of events are not attributed to SLE.

Previous studies have indicated that in SLE patients the minimum clinically important difference for SF-36 MCS and PCS summary scores is 2.5–5.0 and for subscale scores is 5.0–10.0.19 30 The adjusted differences in SF-36 scores in our analysis, which would be expected to be less than unadjusted changes, fall within the minimum clinically important difference for the MCS score and most of the subscale scores of the SF-36. Furthermore, the estimated SE associated with a single observation of a change in SF-36 summary scores between visits, with adjustment for gender, ethnicity, institution and previous score, is approximately 10. Therefore, if a clinical trial was designed to detect a difference in the mean change of 3.66 (the mean difference in MCS scores between patients whose neuropsychiatric events all improve compared with patients without a consistent change in neuropsychiatric events) between two equal sized treatment groups, the sample size to achieve 80% power when testing at the 5% level would be approximately 120 patients per group. If no adjustment for previous score was planned, then approximately 155 patients per group would be required.

There are limitations to the current analysis. First, some neuropsychiatric events were either absent or underrepresented, which emphasises the relatively low frequency of some neuropsychiatric manifestations. In previous studies2,,5 18 approximately half of the 19 neuropsychiatric syndromes in the ACR case definitions occurred in only 1–2% of SLE patients. Second, patients with severe or very active neuropsychiatric disease may have been unable to complete the SF-36 at the time when the neuropsychiatric disease was at it worst. In such cases, the magnitude of change in scores would probably have been even higher that what was found. Third, some psychosocial variables such as fibromyalgia, helplessness and low socioeconomic status were not available in our dataset. Fourth, the physician assessment of some neuropsychiatric events (eg, headache and mood disorders) is dependent in part on patients' subjective complaints, which are potential confounders in the interpretation of parallel changes in SF-36 scores. However, for those neuropsychiatric events (eg, cerebrovascular disease, myelopathy and neuropathies) with more objective outcomes, there were comparable changes in SF-36 scores. Finally, the small number of patients30 with outcome 2 (all worsen) for neuropsychiatric events, which may have been due to the lengthy interval between assessments that provided ample time for treatment, limits the generalisability of the findings for this group.

Despite these limitations the results of our study indicate that changes in SF-36 summary and subscale scores, in particular those related to mental health, are strongly associated with the clinical outcome of neuropsychiatric events in SLE patients. Therefore, changes in SF-36 should be an outcome measure in any clinical trial or study that examines the efficacy of therapeutic interventions for neuropsychiatric disease in SLE patients.


View Abstract

Supplementary materials

  • Web Only Data

    Files in this Data Supplement:


  • Funding JGH (Canadian Institutes of Health Research grant MOP-57752, Capital Health Research Fund); MBU (Canadian Institutes of Health Research grant MOP-49529, Lupus Foundation of Ontario, Ontario Lupus Association, Lupus UK, Lupus Foundation of America, Lupus Alliance Western New York, Conn Smythe Foundation, Tolfo Family (Toronto)); DJ (MRC (UK) grant U.1052.00.006.00001.01); SCB (Korea Healthcare technology R&D project, Ministry for Health, Welfare and Family Affairs, Republic of Korea (A080588)); CG (Lupus UK, arthritis research campaign, Wellcome Trust Clinical Research Facility in Birmingham, UK); AC (Fonds de la recherche en sante de Quebec National Scholar, Singer Family Fund for Lupus Research); SB (Canadian Institutes of Health Research Junior Investigator Award; Fonds de la recherche en santé du Québéc Jeune Chercheure; Canadian Arthritis Network Scholar Award; McGill University Health Centre Research Institute); GSA (University of Alabama at Birmingham, grant P60AR48095); DDG (Canadian Institutes of Health Research); PRF (Distinguished Senior Research Investigator of the Arthritis Society and Arthritis Centre of Excellence); INB (supported by the Manchester Academic Health Sciences Centre and the Manchester NIHR Biomedical Research Centre); MP (Hopkins Lupus Cohort grant AR 43727, Johns Hopkins University General Clinical Research Center grant MO1 RR00052); SM (National Institutes of Health research grants R01 AR46588, K24 AR002213 and M01 RR000056; ON (Swedish Medical Research council grant 13489)); GS (Swedish Medical Research council grant 13489); RR-G (National Institutes of Health research grants UL1RR025741; K24 AR02318; P60 AR 48098); VF (MRC (UK) grant U.1052.00.009.00001.01).

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the Capital Health Research Ethics Board and the research ethics boards at each of the participating centres.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.