Clinical Characteristics of Children With Juvenile Dermatomyositis: The Childhood Arthritis and Rheumatology Research Alliance Registry

Authors


Abstract

Objective

To investigate aspects of juvenile dermatomyositis (DM), including disease characteristics and treatment, through a national multicenter registry.

Methods

Subjects meeting the modified Bohan and Peter criteria for definite juvenile DM were analyzed from the cross-sectional Childhood Arthritis and Rheumatology Research Alliance (CARRA) Registry between 2010 and 2012 from 55 US pediatric rheumatology centers. Demographics, disease characteristics, diagnostic assessments, and medication exposure data were collected at enrollment.

Results

A total of 384 subjects met the criteria for analysis. At enrollment, the median Childhood Myositis Assessment Scale score was 51 (interquartile range [IQR] 46–52), the median Childhood Health Assessment Questionnaire score was 0 (IQR 0–0.5), and the median physician and subject global assessment scores were 1 (IQR 0–2) and 1 (IQR 0–3), respectively, out of a maximum of 10. Of the diagnostic assessments, magnetic resonance imaging was more likely than electromyography or muscle biopsy to show abnormalities. A total of 329 subjects had ≥2 diagnostic studies performed, and >34% of these subjects reported ≥1 negative study. Ninety-five percent had been treated with corticosteroids and 92% with methotrexate, suggesting that these medications were almost universally prescribed for juvenile DM in the US.

Conclusion

In 2 years, the ongoing CARRA Registry has collected clinical data on 384 children with juvenile DM and has the potential to become one of the largest juvenile DM cohorts in the world. More research is needed about prognostic factors in juvenile DM, and differences in therapy based on manifestations of disease need to be explored by practitioners. This registry provides the infrastructure needed to advance clinical and translational research and represents a major step toward improving outcomes of children with juvenile DM.

INTRODUCTION

Juvenile dermatomyositis (DM), the most common inflammatory myopathy of childhood, is a rare multisystemic autoimmune vasculopathy, primarily characterized by proximal muscle weakness and pathognomonic rashes. It affects an estimated 3.2 per 1 million children in the US per year, with the mean onset at age 7 years ([1]). Other organ systems, particularly gastrointestinal, pulmonary, and cardiac, may also be involved. Historically, outcomes in juvenile DM have varied dramatically, with one-third of patients spontaneously recovering, one-third developing moderate to severe disability, and the remaining one-third eventually succumbing to the illness ([2-4]). Since the 1970s, with the introduction of corticosteroids as the mainstay of treatment in juvenile DM, functional outcomes and mortality have dramatically improved ([2]). Despite these advances, manifestations such as calcinosis and lipodystrophy remain significant sources of long-term morbidity and a continued challenge for treating physicians.

Gaps in knowledge regarding multiple facets of juvenile DM exist, particularly concerning epidemiologic features, pathogenesis, treatment, and functional outcomes. Furthermore, because of the rarity of the disease, performing high-quality clinical and translational research in juvenile DM has been difficult. In 2010, the Childhood Arthritis and Rheumatology Research Alliance (CARRA) initiated a multicenter observational registry in the US to create a foundational clinical database for the major rheumatic diseases of childhood, including juvenile DM.

The objective of this study was to enhance knowledge regarding clinical features, methods of disease assessment, treatment regimens, and functional outcomes in children with juvenile DM. This study is a cross-sectional analysis of subjects at the time of enrollment into the registry; longitudinal cohort analysis of followup visits will be the subject of future study. In addition, we aimed to describe the basic infrastructure for one of the world's largest ongoing multicenter juvenile DM registries. The long-term goals of this research include the establishment of a repository of biologic specimens, a data-intensive resource for the development of standardized treatment protocols ([5]), and a foundation for future observational studies, interventional trials, and comparative effectiveness research. Initial data from the juvenile DM cohort (prevalent and incident cases) enrolled in the first 2 years of this ongoing study are described here.

Box 1. Significance & Innovations

  • This ongoing registry provides the infrastructure needed to significantly advance the clinical and translational research of juvenile dermatomyositis (DM), including prospective observational studies, interventional trials, and comparative effectiveness research, with the ultimate goal of improving the outcomes of juvenile DM worldwide.
  • The development of multicenter collaborative networks such as the Childhood Arthritis and Rheumatology Research Alliance Registry will be integral in advancing research, understanding prognosis, and enhancing therapeutic development for rare rheumatic diseases in childhood.

PATIENTS AND METHODS

Study population

The CARRA Registry is a multicenter registry developed within the US to capture information about multiple rheumatic diseases, including juvenile idiopathic arthritis, mixed connective tissue disease, systemic lupus erythematosus, juvenile DM, vasculitis, scleroderma, sarcoidosis, and juvenile primary fibromyalgia. First initiated on May 28, 2010, subjects with diagnosed rheumatic diseases were included from 55 pediatric rheumatology centers within the US. Institutional review board approval was obtained at each enrolling site. Subjects and/or a parent/legal guardian were required to provide informed consent (in English or Spanish).

Subjects were eligible for inclusion into the registry if they were age <21 years, with onset of juvenile DM prior to age 18 years, and met definite or probable criteria for juvenile DM as defined by the modified Bohan and Peter criteria ([6]). The Bohan and Peter criteria were modified by the CARRA Registry project investigators to include magnetic resonance imaging (MRI) as an acceptable diagnostic modality for juvenile DM in order to reflect common practice and widen subject inclusion. The CARRA Registry was also modified after inception to include subjects with a diagnosis of probable juvenile DM. Subjects without diagnostic evidence (on MRI, electromyography [EMG], or biopsy) were excluded from this analysis but are included in the CARRA Registry for possible future comparison analysis. Subjects were not required to be recently diagnosed. For the purposes of our analyses, subjects in the CARRA Registry with juvenile DM who were recruited prior to May 3, 2012 were included in this study. Subjects were excluded from analysis if age at onset could not be clearly identified as prior to 18 years or if they did not meet the criteria for a diagnosis of definite juvenile DM by these modified standards.

Diagnosis of definite juvenile DM (modified Bohan and Peter criteria).

A diagnosis of definite juvenile DM consisted of classic skin involvement for juvenile DM and at least 3 of the following: 1) muscle weakness, 2) elevation of muscle enzyme(s), 3) abnormal EMG suggestive of inflammatory myopathy, 4) abnormal muscle biopsy sample suggestive of inflammatory myopathy, and 5) MRI evidence of myositis (modification introduced by the CARRA Registry investigators) ([6, 7]).

Data collection

Clinical data were collected from the subjects/guardians and medical providers using both general and juvenile DM–specific case report forms at the time of enrollment. In addition, the subjects' medical records were reviewed for previously obtained disease-specific clinical information. Subjects diagnosed prior to May 28, 2010 were recruited retrospectively. Data regarding demographics, disease characteristics, diagnostic assessment, and medication exposure were collected and recorded by the recruiting physician using parental information, chart review, and physician recall. Baseline measures of muscle strength, physical functioning, and quality of life were performed, including the Childhood Myositis Assessment Scale (CMAS; normal score 52) ([8]), the Childhood Health Assessment Questionnaire (C-HAQ; where 0 = normal and 3 = worst) ([9, 10]), health-related quality of life measure ([11]), the American College of Rheumatology (ACR) functional class rating ([12]), global disease assessments, and pain scores (range 0–10, where 10 = worst pain). Followup data were additionally obtained through the registry but not included in this analysis. Data were pooled and stored in a secure centralized electronic database and deidentified prior to analysis.

Statistical analysis

Statistical analysis was conducted using Stata software, version 10.0 (StataCorp). All data analyses were preceded by extensive data checking and verification to identify and resolve the reasons for missing values, inconsistencies, and out of range values. Descriptive statistics were computed to summarize each variable, including the mean, median, SD, and interquartile range (IQR) for continuous variables, and frequencies (percentages) for categorical variables. Chi-square and Wilcoxon's rank sum tests were used to evaluate the associations of disease activity with duration of illness. All tests were 2-sided and P values less than 0.05 were considered statistically significant.

RESULTS

Between May 28, 2010 and May 11, 2012, 384 subjects meeting the modified Bohan and Peter criteria for definite juvenile DM were enrolled from 55 sites in the US. A total of 483 subjects were enrolled in the CARRA Registry at this point with a diagnosis of probable or definite juvenile DM. Eighteen subjects (5%) were enrolled within the first 6 months of onset of symptoms, 15% were enrolled in the first year of symptom onset, and 30% were enrolled within 2 years of symptom onset. Most children were evaluated by a pediatric rheumatologist within 1 year of symptom onset (median duration 5 months, IQR 2–11 months). Demographics and disease characteristics are shown in Table 1.

Table 1. Demographic features and disease characteristics of 384 subjects with juvenile dermatomyositis*
 At enrollmentPresent ever
  1. Values are the number (percentage) unless indicated otherwise (categorical variables). Median (interquartile range [IQR]) values are continuous variables. Data shown were available on 384 subjects unless otherwise specified. ANA = antinuclear antibody; ILD = interstitial lung disease; GI = gastrointestinal.
  2. aRange 1.8–21.4 years.
  3. bRange 0.2–16.6 years.
  4. cRange 0.1–17.4 years.
  5. dRange 0–7.3 years.
Age, median (IQR) years10.8 (7.8–14.8)a 
At disease onset5.7 (3.5–9.3)b 
Disease duration, median (IQR) years3.8 (1.6–6.7)c 
Time to rheumatologist from disease onset, median (IQR) years0.4 (0.2–0.9)d 
Sex  
Female275 (72) 
Male109 (28) 
Race  
White300 (78) 
African American45 (12) 
American Indian2 (<1) 
Asian5 (1) 
Pacific Islander1 (<1) 
Multiracial12 (3) 
Unknown19 (5) 
Ethnicity (n = 379)  
Non-Hispanic329 (87) 
Hispanic50 (13) 
Family history of autoimmunity (n = 360)83 (23) 
Positive ANA (n = 299)186 (62) 
Elevated muscle enzymes (n = 344)44 (13)324 (94)
Arthritis (n = 335)20 (6)126 (38)
Calcinosis (n = 337)38 (11)50 (15)
Dysphagia/dysphonia (n = 330)5 (2)82 (25)
ILD (n = 328)2 (<1)11 (3)
GI ulceration (n = 332)1 (<1)10 (3)
Cardiac involvement (n = 326)4 (1)8 (2)
Weakness (n = 350)  
None250 (71) 
Mild75 (21) 
Moderate20 (6) 
Severe5 (1) 
Periungual telangiectasia (n = 343)141 (41) 
Malar or facial erythema (n = 350)103 (29) 
Contractures (n = 347)37 (11) 
V or shawl sign (n = 348)18 (5) 
Lipodystrophy (n = 348)17 (5) 
Skin ulcer (n = 348)18 (5) 
Gottron's sign, papules, or heliotrope (n = 350)153 (44) 

Notably, only 13% of patients had current muscle enzyme elevation, although 94% of the subjects had elevated enzymes at some point in the past. Arthritis was reported in 38% and a history of calcinosis was reported in 15%. Dysphagia or dysphonia was reported in one-quarter of the subjects, but had resolved prior to enrollment in 23%. Less than 4% of subjects were reported by their physician to have a history of interstitial lung disease, cardiac involvement, or gastrointestinal ulceration.

Disease characteristics reported by time elapsed from symptom onset are provided in Table 2. For subjects enrolled within 2 years of symptom onset, the median C-HAQ score was 0.25 points higher than for subjects enrolled later in their disease course (P < 0.001). Fifty-four percent of subjects assessed reported no impairment of the C-HAQ score at enrollment, and 86% had scores <1. Median physician and subject global assessment scores at enrollment were 1 (IQR 0–2, range 0–10) and 1 (IQR 0–3, range 0–10), respectively. Sixty-four percent of subjects reported excellent or very good quality of life compared with only 4% reporting poor or very poor quality of life. Thirty-two percent of subjects reported fair quality of life. At the time of enrollment, 81% of the subjects were classified as ACR functional class I, with only 7% in class III (able to perform usual self-care activities, but limited in vocational and avocational activities) or IV (limited ability to perform usual self-care activities). However, for the duration of their illness prior to enrollment, 29% of subjects were classified by their providers at some point as ACR functional class IV and 19% as class III. Eighteen percent of the subjects reported no limitations in self-care or avocational activities at any point in their disease.

Table 2. Disease characteristics by duration of disease (time calculated from reported onset of symptoms to subject enrollment)*
 Disease duration <2 years (n = 117)Disease duration ≥2 years (n = 259)P
  1. Values are the number (percentage) unless indicated otherwise (categorical variables). Median (interquartile range [IQR]) values are continuous variables. CMAS = Childhood Myositis Assessment Scale; C-HAQ = Childhood Health Assessment Questionnaire; ACR = American College of Rheumatology.
CMAS score (n = 218), median (IQR)47 (43–51)52 (48–52)< 0.001
C-HAQ score (n = 368), median (IQR)0.25 (0–0.87)0 (0–0.25)< 0.001
Physician's global assessment score (n = 329), median (IQR)1 (0–3)1 (0–2)0.002
Subject's global assessment score (n = 371), median (IQR)2 (1–4)1 (0–3)< 0.001
ACR functional class (n = 350)  < 0.001
Class I74 (67)208 (87) 
Class II24 (22)19 (8) 
Class III11 (10)7 (3) 
Class IV2 (2)5 (2) 
Weakness (n = 345)  < 0.001
None55 (50)191 (81) 
Mild38 (35)37 (16) 
Moderate14 (13)5 (2) 
Severe2 (2)3 (1) 
Periungual telangiectasia (n = 338)53 (50)85 (37)0.020
Malar or facial erythema (n = 345)42 (39)60 (25)0.008
Gottron's sign, papules, or heliotrope (n = 345)53 (50)97 (41)0.128

Longer time since symptom onset (≥2 years) was associated with higher CMAS scores, lower C-HAQ scores, and lower physician and subject global disease activity scores (Table 2). Lower current ACR functional class scores were associated with longer duration of illness (≥2 years of symptoms) at baseline by chi-square analysis. Active periungual telangiectasias were associated with a disease duration <2 years (Table 2). Malar rash was also associated with a shorter disease duration (P = 0.008), with 39% of subjects with duration of disease <2 years having active rash compared to 25% of those beyond the first 2 years of onset. The presence of Gottron's papules was not associated with disease duration (50% versus 41%; P = 0.128).

Diagnostic studies commonly ordered to diagnose juvenile DM include EMG, muscle biopsy, and MRI. Overall, MRI was the most commonly performed study, with 90% of subjects receiving an MRI at some point prior to enrollment. MRI (91%) was also more likely than EMG (50%) or muscle biopsy (76%) to reveal abnormalities consistent with juvenile DM (Table 3). EMG was the most likely to show nondiagnostic results, with half of the children requiring an additional diagnostic study to confirm juvenile DM. Half of the registry subjects had ≥2 studies performed and more than one-third of these subjects reported ≥1 negative study.

Table 3. Evaluation of diagnostic modalities in juvenile DM*
 No. (%)
  1. DM = dermatomyositis; EMG = electromyography; MRI = magnetic resonance imaging.
Muscle biopsy (n = 330) 
Performed165 (50.0)
Consistent with juvenile DM125 (75.8)
Nondiagnostic40 (24.2)
EMG (n = 323) 
Performed102 (31.6)
Consistent with juvenile DM51 (50.0)
Nondiagnostic51 (50.0)
MRI (n = 333) 
Performed299 (89.8)
Consistent with juvenile DM272 (91.0)
Nondiagnostic27 (9.0)
Combination of ≥2 studies (n = 329) 
Performed166 (50.5)
3 studies consistent with juvenile DM3 (1.8)
2 studies consistent with juvenile DM106 (63.9)
1 study consistent with juvenile DM57 (34.3)

In terms of medications, 359 subjects (96%) were treated with corticosteroids during their course of treatment and 347 subjects (93%) were treated with methotrexate. Sixteen subjects had never been exposed to corticosteroids; 2 of them were newly diagnosed subjects, with the rest receiving rheumatologic care from 1.6–10.2 years at the time of enrollment. Current use of long-term daily steroids was associated with symptom onset <2 years prior (P < 0.001), with 75% of subjects with onset of symptoms <2 years prior currently taking daily steroids compared to 35% of those who had onset of symptoms ≥2 years prior. Current use of methotrexate was also associated with onset of symptoms <2 years prior (P < 0.001 by chi-square test). Fifty-four percent of subjects with onset of symptoms ≥2 years prior remained on subcutaneous or oral methotrexate, and 40% were previously treated with methotrexate. Ninety-four percent of those with symptom onset ≥2 years prior had been treated with methotrexate. Of those who were in the first 2 years of disease, 84% were still currently treated with methotrexate. Approximately 40–60% of all subjects had been exposed to pulse corticosteroids, hydroxychloroquine, or intravenous gamma globulin during the course of their disease. The rate of pulse steroid usage did not change with duration of illness, but the percentage of subjects currently reporting pulse steroid usage was higher in those in the first 2 years of illness (21% compared to 8%). Subjects with ≥2 years of illness had higher rates of exposure to hydroxychloroquine (60% versus 37%) and intravenous gamma globulin (46% versus 28%). Anti–tumor necrosis factor therapy usage was infrequent; <5% of subjects were treated with etanercept, infliximab, or adalimumab at some point during their illness. Medication history in order of frequency of usage is shown in Table 4.

Table 4. Medication usage within the CARRAnet juvenile dermatomyositis cohort (n = 384)
 Current use, no. (%)Previous use, no. (%)aEver used, no. (%)Missing, no.
  1. aExcludes current use.
All corticosteroids185 (49.3)174 (46.4)359 (95.7)9
Daily corticosteroids173 (46.1)175 (46.7)348 (92.8)9
Methotrexate236 (62.9)111 (29.6)347 (92.5)9
Pulse corticosteroids45 (12.2)171 (46.5)216 (58.7)16
Hydroxychloroquine132 (35.5)65 (17.5)197 (53.0)12
Intravenous gamma globulin67 (18.0)83 (22.3)150 (40.2)11
Mycophenolate mofetil46 (12.4)16 (4.3)62 (16.7)13
Cyclosporine14 (3.7)25 (6.7)39 (10.5)13
Rituximab4 (1.1)19 (5.1)23 (6.2)12
Etanercept1 (0.3)15 (4.0)16 (4.3)11
Infliximab3 (0.8)7 (1.9)10 (2.7)11
Adalimumab5 (1.3)1 (0.3)6 (1.6)12
Cyclophosphamide0 (0.0)5 (1.4)5 (1.4)14

DISCUSSION

To date, this is the largest cohort of North American children with juvenile DM reported. Patient demographics, such as mean age at onset and female predominance, were similar to previously reported cohorts from both North America and around the world ([13-19]). As opposed to other autoimmune diseases, the distribution of race is comparable to the national demographics from the US Census Bureau from 2005–2009 (74.5% white, 12.4% black or African American, 0.8% American Indian/Alaskan, 4.4% Asian, and 0.1% Native Hawaiian/Pacific Islander), with no discernible ethnic/racial propensity ([20]). This distribution is similar to that seen in previously reported North American cohorts ([1, 17, 18]). Previous registries from other countries have reported different racial demographics that likely correspond to the demographics of the enrolling countries ([13, 14]).

The relatively mild disease severity observed at enrollment likely reflects the disproportionate number of prevalent cases collected to date, and is similar to other international cross-sectional juvenile DM cohorts ([13, 14]). Given that the mean onset of symptoms was 4.5 years prior to enrollment, this cohort as a whole tended to skew toward minimal disease, with near-normal CMAS scores, low C-HAQ scores, and low median global assessment scores. Relatively low incidences of disease manifestations at the time of enrollment, such as elevated muscle enzymes (13%), weakness on physical examination (29%), or characteristic rash (44%), also reflect this phenomenon. These results will likely trend toward more severe disease as additional subjects are enrolled into this cohort closer to their initial disease presentation and prior to the initiation of treatment.

To examine this phenomenon more closely, we divided the cohort into subgroups based on time elapsed since symptom onset, with a cutoff of 2 years of disease. Those children with ≥2 years since symptom onset reported significantly less weakness, which corresponded with increased CMAS scores and decreased C-HAQ scores when compared to subjects with <2 years of disease. Although the presence of skin manifestations and periungual capillary changes showed a trend toward improvement, a significant difference between the subgroups was not found. It is interesting to note that skin manifestations seem to be more persistent than weakness in this cohort, although this observation is problematic because of possible enrollment bias and will need to be the subject of future analysis. Although this association of longer disease duration with lower disease activity is encouraging, the association is hampered by the possibility of bias in enrollment into the registry.

Antinuclear antibody positivity (62%) was consistent with other cohorts ([14, 16]). However, patients in this cohort experienced lower rates of complications when compared with other older mixed cohorts, including calcinosis (15%) and arthritis (38%) ([13, 14, 16, 17]). This may be partly an effect of early recognition and aggressive treatment when compared with older cohorts or may be a reflection of the wide range of subjects who were enrolled at any point during their disease course. When compared to subjects enrolled at diagnosis, rates of arthritis varied from 0–3% and rates of calcinosis varied from 9–12% after 2 years of treatment ([21]). Dates of development of calcinosis or arthritis are not included in the CARRA Registry, and we do not have information on when during the course of their disease these were found. Additionally, the inclusion of subjects with myositis syndromes that do not meet the modified Bohan and Peter criteria for the diagnosis of juvenile DM may skew the frequency of such complications in other groups ([14, 17, 19]). This large cohort consisting only of children with definite juvenile DM should provide a more accurate frequency of rare disease manifestations, such as gastrointestinal, pulmonary, and cardiac involvement.

In the Juvenile Dermatomyositis National Registry and Repository (UK and Ireland), EMG was used in only 8% of patients, muscle biopsy in 36%, and MRI in 68%. MRI was abnormal in 76% of patients and biopsy samples were abnormal in 89% of patients ([14]). In a European and Latin American registry, 60% of patients received EMG, which was abnormal in 90%, and muscle biopsy was performed in 55% of patients, of which 89% of samples were abnormal (consistent with the UK and Ireland) ([15]). In our cohort, MRI was the most common diagnostic modality used and was the most likely to show abnormalities consistent with juvenile DM. The false-negative rates for EMG (50%) and muscle biopsy (24%) alone were higher than expected if ascertainment is correct and higher than in the other 2 large international cohorts; however, there may have been more selectivity in who received studies in the other registries. Second, our registry did not include information on when these studies were obtained, and it is possible that some of these studies may have been obtained later in the course of disease, decreasing study sensitivity. Third, some differences may be due to increasing use of MRI in developed countries worldwide between the time periods represented in the comparison articles.

Corticosteroids and methotrexate appear to be standard first-line medications used by US pediatric rheumatologists for juvenile DM. This appears to be congruent with a previous survey of pediatric rheumatologists in CARRA ([22]). The European/Latin American cohort reported similar corticosteroid usage rates, but only 56% received methotrexate, perhaps reflecting the fact that this cohort was collected between 1980 and 2004, with higher percentages receiving methotrexate with time ([15]). In the study by Martin et al, more recent data from the UK and Ireland suggest use of steroids and methotrexate similar to our cohort ([21]). Cyclosporine was also used more frequently (25% versus 10%) in that cohort, particularly in Europe. Intravenous gamma globulin was used in 17% of patients compared to the CARRA cohort (39%) ([15]). Pulse corticosteroids, intravenous gamma globulin, and hydroxychloroquine have been used by approximately 40–60% of our subjects, and further investigation as to which subgroups receive these medications is warranted. In 2011, an oral presentation by Boulter et al at the 18th Pediatric Rheumatology European Society Congress reported an analysis of 43 exposures to etanercept, infliximab, and adalimumab in the UK/Ireland cohort; the size of the cohort at that time is not listed in the abstract. Anti–tumor necrosis factor therapy is infrequent in the CARRA Registry cohort and may represent an area of future investigation ([23]).

In its first 2 years, the ongoing CARRA Registry of children with rheumatic diseases has collected clinical data on 384 children with juvenile DM and has become one of the largest juvenile DM cohorts in the world. Clinical data have been collected from 55 pediatric rheumatology centers and entered into a single database. Data on followup visits are now being collected, while new patient enrollment continues and additional sites are being recruited for enrollment. Substudies of the CARRA Registry are being used to collect data on treatment protocols for juvenile DM. While many of the observations to date are consistent with previously described cohorts of patients with juvenile DM, this is the first US study to report the prevalence of disease manifestations and functional and quality of life outcomes in a cross-sectional manner. More importantly, when combined with a repository of linked biologic specimens, this registry will provide the infrastructure needed to significantly advance the clinical and translational research of juvenile DM, including prospective observational studies, interventional trials, and comparative effectiveness research. The ongoing development of the CARRA Registry represents a major step toward improving outcomes of children with juvenile DM worldwide.

AUTHOR CONTRIBUTIONS

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. Robinson 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. Robinson, Hoeltzel, Wahezi, Huber, Feldman, Reed.

Acquisition of data. Robinson, Hoeltzel, Wahezi, Reed.

Analysis and interpretation of data. Robinson, Hoeltzel, Wahezi, Becker, Kessler, Schmeling, Carrasco, Huber, Feldman, Reed.

Acknowledgments

The authors would like to thank all of the participants and hospital sites that recruited patients for CARRA. The authors thank the following CARRA Registry site principal investigators: L. Abramson, B. Adams, J. Birmingham, P. Blier, S. Bowyer, E. Chalom, F. Dedeoglu, P. Ferguson, M. Klein-Gitelman, D. Goldsmith, B. Gottlieb, T. Graham, G. Higgins, J. R. Hollister, J. Hsu, A. Huttenlocher, N. Ilowite, L. Imundo, C. Inman, T. Jerath, L. Jung, P. Kahn, D. Kingsbury, A. Lasky, T. Lehman, C. Lindsley, J. Lopez-Benitez, D. McCurdy, N. Moorthy, B. Myones, K. Nanda, J. Olson, K. O'Neil, K. Onel, K. Peterson, S. Prahalad, M. Punaro, A. Quintero, C. Rabinovich, A. Reed, S. Ringold, D. Rothman, N. Ruth, C. Sandborg, E. von Scheven, K. Schikler, D. Sherry, N. Singer, S. Spalding, R. Syed, K. Torok, R. Vehe, C. Wallace, J. Weiss, A. White, A. Yalcindag, and L. Zemel.

Ancillary