Abstract
Objective
Studies of dermatomyositis (DM) are frequently limited to single-centre cohorts. We used two large nationally representative US cohorts to conduct a descriptive epidemiological study of the characteristics, treatments and outcomes of patients with incident DM.
Methods
This retrospective study identified two DM inception cohorts using (1) commercial claims and (2) electronic health record (EHR) data from the Excellence Network in Rheumatology to Innovate Care and High-impact research (ENRICH), a community rheumatology practice-based research network. Patient characteristics, treatments and healthcare utilisation were assessed using the 18 months before and 12 months after diagnosis in claims and the 12 months before and after diagnosis in EHR data.
Results
We identified 2475 patients (claims) and 1196 patients (EHR) with incident DM. Among 998 patients in the EHR cohort with available laboratory data, 472 had available myositis panel results, with 165 (35.0%) having a positive myositis-specific antibody. Glucocorticoid use was common, 68.7% and 73.8% in the two cohorts, respectively, with initial doses most often >20 mg/day; among glucocorticoid users, mean cumulative dose was 1407 mg in the claims cohort. Hydroxychloroquine, methotrexate and mycophenolate were the most commonly used immunomodulatory therapies. During follow-up in the claims data cohort, incidence per 1000 person-years was 92.2, 15.3, 6.4, 2.9 and 2.1 for all-cause hospitalisation, malignancy, interstitial lung disease, gastrostomy tube placement and myocarditis, respectively.
Conclusion
Administrative claims and EHR data can be leveraged to assess treatment patterns and longitudinal outcomes/disease manifestations in incident dermatomyositis cohorts. This study highlights a high burden of glucocorticoid exposure, significant heterogeneity in treatment and high healthcare utilisation in this population.
Keywords: Dermatomyositis; Glucocorticoids; Treatment; Lung Diseases, Interstitial; Epidemiology
WHAT IS ALREADY KNOWN ON THIS TOPIC
Dermatomyositis carries substantial morbidity and can affect multiple organ systems, but many existing studies characterising the disease and its treatment are from single centres.
WHAT THIS STUDY ADDS
Using two complementary data sources—administrative claims data from a commercial insurer and electronic health record data from a rheumatology clinical practice network—we evaluated patient characteristics in large, nationally representative dermatomyositis cohorts. Assessment of treatment patterns revealed substantial treatment variability and a high burden of glucocorticoid use, with frequent high-dose use and large cumulative exposure. Claims data allowed assessment of incident disease manifestations and outcomes, including hospitalisations and mortality.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
This study demonstrates the need for more effective therapies to reduce glucocorticoid dependence and shows the potential for real-world data sources to assess long-term patient outcomes in patients with dermatomyositis.
Introduction
Dermatomyositis (DM) is a chronic, systemic, autoimmune inflammatory myopathy that is characterised by muscle inflammation and rash. Disease manifestations, such as interstitial lung disease (ILD) and myocarditis, can be severe and life-threatening.1 2 Patients with DM have been shown to have adverse outcomes including an increased risk for malignancy and increased healthcare utilisation and medical costs.3 4
Despite the individual and societal burden of DM, descriptions of the treatment patterns and outcomes of patients with DM in the USA have not been well characterised. Most reports describe single-centre experiences, often from academic centres.5,10 One of the few US studies using administrative data was conducted well over a decade ago,11 while a second was confined to the US veterans population.12
In this study, we used two independent but complementary data sources—(a) administrative claims data and (b) electronic health records (EHR) from a large US practice-based research network (PBRN) of community rheumatology providers. Administrative claims data is particularly well suited to evaluate treatment patterns and long-term health outcomes for patients seen by a range of different providers. In contrast, use of a rheumatology EHR provides more granular information for DM patients seeing a rheumatologist, including laboratory testing results. Our goal was to describe patient characteristics, treatments, testing and outcomes in newly diagnosed dermatomyositis.
Methods
This descriptive retrospective cohort study of patients with newly diagnosed DM included analyses in two independent data sources: enhanced health plan claims data (Komodo Health’s Healthcare Map research database) and EHR data from the Excellence Network in Rheumatology to Innovate Care and High-impact research (ENRICH). Komodo Health includes commercial claims data from more than 150 payers and provides claims data for more than 330 million patients. Available data includes diagnosis and procedure codes from inpatient and outpatient encounters, infusion procedures, as well as prescription fill data from pharmacy data. ENRICH EHR data includes data from a large US community rheumatology practice-based research network consisting of >700 community rheumatology providers that all use 1 of 2 selected EHRs.13 Available data includes diagnosis and procedure codes, laboratory results, and medication and infusion orders.
DM cohort construction in health plan claims data
The health plan claims data DM cohort included patients with index dates from 1/7/2017 to 31/12/2022 and used data from 1/1/2016 to 31/12/2023. Patients with DM were identified based on a single ICD-10-CM inpatient diagnosis code for DM or two outpatient diagnosis codes for DM separated by 7–365 days, following conventions from prior DM studies.1112 14,17 Diagnosis codes for DM included ICD-10 M33.1* (other dermatomyositis) or M33.9* (dermatopolymyositis, unspecified), but not M33.0 (juvenile dermatomyositis) nor M33.2* (polymyositis). Notably, the codes included have been previously found to have greater specificity than other myositis diagnosis codes and have shown high positive predictive value (PPV 95%) in the inpatient setting.14 The presence of multiple outpatient codes has been shown to have improved specificity, with ≥2 outpatient codes having particularly good PPV (85–96%) when from a dermatologist or rheumatologist.18 19
The index date was the date of the single inpatient diagnosis code or the date of the second outpatient diagnosis code (online supplemental Figure 1). Patients were required to be aged ≥18 years old at the index date. To ensure that identified DM cases were incident rather than prevalent cases, we required ≥18 months of preceding continuous medical and pharmacy coverage (baseline period) prior to the index date, excluded patients with DM or any other idiopathic inflammatory myositis diagnosis on or before the first diagnosis date (dermatomyositis or polymyositis ICD-10 M33.*; inclusion body myositis ICD-10 G72.41; other inflammatory and immune myopathies, not otherwise classified ICD-10 G72.49) and excluded patients with any immunomodulatory drug use other than hydroxychloroquine or glucocorticoids (often used before a definitive diagnosis) on or before the first DM diagnosis using all available data.
We excluded patients missing age or sex or with diagnoses of hepatitis C or HIV prior to the index date as these conditions may affect treatment and outcomes. We also excluded patients with malignancy other than non-melanoma skin cancer (NMSC) prior to the index date given effects on outcomes and treatment and the challenges related to separating malignancies associated with versus not associated with DM. This exclusion led to the exclusion of some patients with malignancy-associated DM. Malignancy-associated DM was intentionally excluded from our study objectives as this was not our population of interest. For healthcare utilisation analyses, we also required at least 12 months of continuous medical and pharmacy coverage after the index date in a subgroup analysis described below. The continuous enrolment period reflects the availability of various insurance sources for each individual, ensuring comprehensive tracking across all applicable plans.
ENRICH EHR cohort identification
The ENRICH cohort included patients with index dates 1/10/2015 to 30/9/2022, with available data 1/10/2014 to 30/9/2023 used to ensure sufficient preceding and follow-up data. Patients were identified based on two DM diagnosis codes from rheumatology providers separated by 7–365 days, using the same diagnosis code criteria as in the claims cohort.
Similar to the claims data approach, the index date was the date of the second diagnosis code (online supplemental Figure 2), and DM patients were required to be ≥18 years old at the index date. We excluded patients with DM diagnoses prior to the first diagnosis date, patients with any other idiopathic inflammatory myositis diagnosis on or before the index date (see codes above), and patients with immunomodulatory drug use other than hydroxychloroquine or glucocorticoids on or before the first DM diagnosis using all available data to capture incident rather than prevalent DM. Patients were also required to have a new consultation procedure code (ie, a new visit, CPT 99 202–99205 or CPT 99 242–99245) on or in the 12 months before the first DM diagnosis code to capture newly diagnosed patients, not patients with longstanding DM treated by rheumatologists.
As in the claims cohort, we excluded patients with missing age or sex and with diagnoses of hepatitis C, HIV or malignancy other than NMSC prior to the index date. We explored the possibility of a subgroup analysis that required the last follow-up visit to be ≥18 months from the first diagnosis date to ensure adequate follow-up time to capture testing results and initial treatments. However, given that all results were very similar to the main cohort, and this restriction reduced the sample size to 650 patients (54% of the total), this subgroup analysis was not included in the final results.
Assessing patient characteristics at baseline in both cohorts
In both cohorts, patient age, sex, race, ethnicity and geographic region were assessed on the index date. As shown in online supplemental Figures 1 and 2, different capture windows were used in the two datasets for other variables given the different structure of the data. In the claims-based cohort, patient comorbidities were captured based on diagnosis codes and comorbidity-specific treatments during the 18 month baseline period to ensure adequate data availability prior to the first DM diagnosis (given that comorbidities were captured from encounters with both rheumatologists and non-rheumatologists). In the ENRICH EHR cohort, since only comorbidities recorded within a rheumatology encounter were available, the 12 months before and after the index date was used with the expectation that some comorbidities or health conditions pre-existing or co-occurring with myositis might not be clinically assessed by the rheumatologist prior to the index date or at the initial visits.
Assessing DM testing, ILD, healthcare utilisation, treatment and incident outcomes in the health plan claims cohort
In the health plan cohort, diagnostic testing, prevalent ILD, healthcare utilisation and medication use in the 18 month baseline period and 12 month follow-up period were assessed, restricted to patients with at least 12 months of follow-up (online supplemental Figure 1). Testing and healthcare utilisation measures based on procedure codes included cancer-related screening (colonoscopy, mammography among females and prostate-specific antigen testing among males), CT chest, pulmonary function tests (PFT), echocardiogram, electromyography (EMG), muscle and skin biopsy, feeding tube placement, and outpatient visits with different provider specialties which included a diagnosis of DM. Prevalent ILD was assessed during the baseline period and 12-month follow-up period based on a set of validated diagnoses, allowing time post-index for diagnostic testing to reveal a diagnosis of ILD.20 Medication use was assessed based on prescription fills or infusion procedure codes during this time-period. Initial glucocorticoid dose was estimated using the first prescription in the capture window, and cumulative glucocorticoid dose in the 30 month time frame was calculated using all filled prescriptions.
Using all available data after the index date, we assessed all-cause hospitalisation (not including hospitalisations with a new diagnosis of DM on the index date) as well as incident malignancy, ILD (beginning >12 months after index, using an algorithm with PPV 96% for incident ILD when studied in people with RA),20 all myocarditis, and hospitalised myocarditis (PPV 74%)21 based on diagnosis codes, and feeding tube placement based on procedure codes, excluding patients with diagnoses of these conditions at baseline.
Assessing treatments, testing results, ILD and patient-reported outcomes in the EHR cohort
In the EHR cohort, treatments, testing results and prevalent ILD (using an established algorithm with PPV 77% for prevalent ILD among people with RA)20 were assessed using data from visits in the 12 months before and 12 months after the index date. Again, a longer 18 month look-back was not used in this cohort because in this rheumatology-specific EHR all relevant testing was expected to occur within 1 year prior to the index date. Treatments were identified based on provider prescriptions (rather than fill data) and administered infusions. Initial glucocorticoid dose was derived from the first written prescription in the capture window. Cumulative glucocorticoid dose was not assessed given the absence of fill data. The results of laboratory testing were assessed using all available structured data in the specified time frame, using results closest to the index date if multiple laboratory results were available. Maximum levels of muscle enzymes were also assessed during the 24 month period surrounding the index date. Results were reported only among patients with any of the above laboratory results available as structured data during the study period.
Statistical analyses
In claims data, incidence rates of all-cause hospitalisation and incident malignancy, ILD, myocarditis and feeding tube placement were calculated based on the number of events divided by the amount of person-years of follow-up. No minimum follow-up was required in these analyses to avoid immortal time bias, and follow-up ended at the earliest of end of continuous enrolment in medical and pharmacy coverage, death, or at the occurrence of an outcome. Sankey plots showing medication changes over time were created with R package ‘highcharter’. All other analyses were conducted using SAS 9.4.
All human and animal studies have been approved by the appropriate ethics committee and have therefore been performed in accordance with the ethical standards laid down in the 1975 Declaration of Helsinki and its later amendments. The study was approved and deemed exempt by Advarra Institutional Review Board (Pro00062070 and Pro00070131) and individual patient consent was waived.
Results
In the health plan claims data, we identified 11 482 patients who met DM inclusion within the identification period. After applying exclusions, the final cohort included 2475 patients with incident DM, with a requirement for an 18 month baseline period leading to the most attrition (online supplemental Figure 3). Notably, there were 361 patients (12.7%) excluded because of diagnoses of malignancy other than NMSC prior to the index date. Of the 2475 patients with incident DM, 352 (14.2%) were defined based on an inpatient diagnosis; 1938 (78.3%) had at least 12 months of consecutive medical and pharmacy coverage after the index date for healthcare utilisation analyses.
In the ENRICH EHR, we identified 2605 patients with DM meeting inclusion criteria within the identification period. After applying exclusions, the final cohort included 1196 patients with incident DM (online supplemental Figure 4). There were 54 (4.3%) patients excluded because of diagnoses of malignancy other than NMSC prior to the index date.
Patient characteristics of the two cohorts are shown in table 1. Mean age was 53.4 (SD 15.5) in the health plan claims data and 56.2 (SD 15.7) in ENRICH with 74.8% and 77.9% female patients. Comorbidities were common with hyperlipidaemia in 34.1% and 28.0% and diabetes in 17.0% and 16.6%, respectively.
Table 1. Select baseline characteristics of two incident dermatomyositis cohorts.
| Commercial health plan cohort n=2475 |
Rheumatology EHR cohort n=1196 |
|
|---|---|---|
| Age, years, mean (SD) | 53.4 (15.5) | 56.2 (15.7) |
| Sex, female | 1852 (74.8) | 932 (77.9) |
| Race/ethnicity | ||
| White, non-Hispanic | 1292 (52.2) | 703 (58.8) |
| Black or African American, non-Hispanic | 234 (9.5) | 96 (8.0) |
| Hispanic or Latino | 244 (9.9) | 112 (9.4) |
| Asian or Pacific Islander | 96 (3.9) | 26 (2.2) |
| Other/unknown | 609 (24.6) | 259 (21.7) |
| Geographic region | ||
| Midwest | 566 (22.9) | 141 (11.8) |
| Northeast | 617 (24.9) | 135 (11.3) |
| West | 345 (13.9) | 126 (10.5) |
| South | 946 (38.2) | 794 (66.4) |
| Rheumatology provider visit* | 1009 (52.1) | 1196 (100.0) |
| Select comorbidities | ||
| Diabetes mellitus | 421 (17.0) | 198 (16.6) |
| Hyperlipidaemia | 844 (34.1) | 335 (28.0) |
| Ischaemic heart disease | 228 (9.2) | 12 (1.0) |
| Osteoporosis | 121 (4.9) | 150 (12.5) |
| ILD diagnosis codes | 212 (10.9) | 60 (5.0) |
n (%) except as indicated. Comorbidities were assessed in the 18 months prior to the index date in the claims cohort. In the EHR cohort, comorbidities were assessed in the 12 months before and after the index date, given that all comorbidities came from rheumatology encounters, with the potential for some comorbidities not to be identified until after the index date. ILD (set of validated diagnoses), medication use and laboratory results were assessed in the 18 months before and 12 months after index in claims and the 12 months before and after index in the EHR cohort.
686 (35.4%) of patients in the claims cohort saw providers whose specialty could not be determined.
EHR, electronic health records; ILD, interstitial lung disease.
Diagnostic testing, treating providers and laboratory results
Frequency of various diagnostic tests in the claims data in the 18 month baseline period and 12 month follow-up period was assessed (online supplemental Table 1). Cancer screenings were frequently performed with colonoscopy performed in 33.2% of patients, mammography in 66.5% of females and prostate-specific antigen testing in 53.7% of males. Dermatomyositis diagnostic testing was highly variable. CT chest was performed in 51.5% of patients, PFTs in 27.0%, and EMG in 25.1%. Skin biopsy (51.5%) was more frequent than muscle biopsy (11.7%). Prevalent ILD (diagnosis during the baseline period or 12 months after index) was observed in 10.9% of patients in the health plan claims data.
Assessing providers caring for patients with DM based on diagnosis codes within the 12 months before or after the index date, we identified a rheumatology outpatient visit for 50.4% of patients, primary care for 46.3%, dermatology visit for 28.4%, pulmonary visit for 7.6%, and neurology visit for 6.7%.
Among DM patients seeing rheumatologists in the ENRICH EHR data, prevalent ILD was present in 60 (5.0%) patients based on diagnosis codes in the 12 months before or after the index date. Laboratory testing results are shown in table 2 among the 988 patients with some structured laboratory data available. Mean maximum CK was 582 (SD 1477), with a median maximum of 135 U/L (IQR 82–341) using all available data. Among those with any structured laboratory values available, ANA results were available in 533 (53.9%) and positive in 255/533 (53.9%). Availability of myositis-specific antibody lab test results increased in more recent years, and different myositis-specific antibodies were available at varying rates depending on the composition of antibody panels. The most frequent myositis-specific antibodies were anti-TIF1 gamma (41/240, 17.1%), anti-NXP2 (39/336, 11.6%), anti-Jo-1 (31/620, 5.0%), anti-MDA5 (23/337, 6.8%) and anti-Mi-2 (22/462, 4.8%). Among those with any structured laboratory results available, 745/988 (74.6%) had available results for a myositis-specific or myositis-associated antibody and 472/988 (47.8%) had available results for non-Jo-1 myositis-specific antibody testing (suggestive of panel testing for myositis antibodies). Of these patients, 35.0% (165/472) had a positive myositis-specific antibody.
Table 2. Muscle biomarker and serologic testing in the ±12 months from index, using EHR data.
| Laboratory testing | Patients with any structured laboratory data available n=988 |
|---|---|
| Muscle enzymes | |
| CK (U/L), mean (SD), n=661 | 268.2 (943.2) |
| CK (U/L), median (IQR), n=661 | 102.0 (51.0, 137.0) |
| CK (U/L), maximum (all available data), mean (SD), n=944 | 582.5 (1477.3) |
| CK (U/L), maximum (all available data), median (IQR), n=944 | 135.0 (82.0, 341.0) |
| Aldolase (U/L), mean (SD), n=520 | 7.5 (10.0) |
| LDH (U/L), mean (SD), n=92 | 247.2 (159.9) |
| None of the above muscle enzymes in the 24-month capture window | 208 (21.1%) |
| Autoantibody positivity among those with test results available | |
| ANA | 255/533 (47.8%) |
| Myositis-associated antibodies (positive/patients ever tested) | |
| SSA | 65/580 (11.2%) |
| U1 RNP | 12/198 (6.1%) |
| Smith | 11/496 (2.2%) |
| PM/SCL | 10/198 (5.1%) |
| KU | 9/266 (3.4%) |
| U2 RNP | 1/148 (0.7%) |
| Fibrillarin | 1/158 (0.6%) |
| Myositis-specific antibodies (positive/patients ever tested) | |
| TIF1-gamma | 41/240 (17.1%) |
| NXP-2 | 39/336 (11.6%) |
| Jo-1 | 31/620 (5.0%) |
| MDA5 | 23/337 (6.8%) |
| Mi2 | 22/462 (4.8%) |
| SRP | 13/441 (2.9%) |
| Pl7 | 9/444 (2.0%) |
| SAE | 8/99 (8.1%) |
| Pl12 | 7/437 (1.6%) |
| OJ | 6/445 (1.3%) |
| EJ | 2/445 (0.4%) |
| cN-1A | 2/7 (28.6%) |
| HMGCR | 0/30 (0.0%) |
If a patient had multiple values, the closest value to the index date was used.
ANA, antinuclear antibody; CK, creatine phosphokinase or creatine kinase; EHR, electronic health record; LDH, lactate dehydrogenase.
Treatment of dermatomyositis
The frequency of different treatments for dermatomyositis is shown in figure 1 and online supplemental Table 2). The most common treatment in both cohorts was hydroxychloroquine (35.6% and 41.5%, respectively), followed by methotrexate, mycophenolate and azathioprine. IVIG was used by 10.0% of patients in the health plan data and 2.3% in ENRICH, and rituximab in 4.0% and 6.3%, respectively. Other medications were used infrequently, including JAK inhibitors (0.1% and 1.5%). Lack of any non-glucocorticoid immunomodulatory therapy was more common in the claims data (38.6%) than in the ENRICH rheumatologist EHR data (17.6%). Additionally, among those with ≥18 months of follow-up data available after the first DM diagnosis in ENRICH, lack of any non-glucocorticoid immunomodulatory therapy was found in only 9.8%. Changes in myositis treatment over time in each cohort are shown in figure 2 and online supplemental Tables 3 and 4.
Figure 1. Immunomodulatory therapy use in the claims and EHR dermatomyositis cohorts Immunomodulatory therapy use based on prescription fills and infusions in the 18 months before and 12 months after the index date in the health plan claims data dermatomyositis cohort and based on electronic health record (EHR) data in the 12 months before and after the index date in the ENRICH rheumatology EHR dermatomyositis cohort. As patients with immunomodulatory therapies other than hydroxychloroquine prior to the first dermatomyositis diagnosis date were excluded, these therapies reflect use starting at the first DM diagnosis through 12 months after the index date. EHR, electronic health record; IVIG, intravenous immunoglobulin; JAKi, janus kinase inhibitor; TNFi, tumour necrosis factor inhibitor.
Figure 2. Sankey Plot showing changes in treatment over time. (A) Treatment changes over time in health claims cohort through 3 years of follow-up. (B) Treatment changes over time in ENRICH EHR through 3 years of follow-up. CDMARD, conventional disease modifying anti-rheumatic drug; IVIG, intravenous immunoglobulin. Panel A column Ns: 1=1217, 2=677, 3=291, 4=120. Panel B column N’s:1=999; 2=459; 3=189; 4=59. Note: If multiple cDMARDs were administered on the same day, they were categorised as cDMARDs. If bDMARD, IVIG or tsDMARD was given alongside cDMARD on the same day, it was classified as the bDMARD, IVIG or tsDMARD generic plus combo.
Glucocorticoid use was common, present in 68.7% in the health plan data cohort and 73.8% in ENRICH (table 3). Initial glucocorticoid doses were >20 mg in 36.0% of glucocorticoid users in the health plan data and 58.9% in ENRICH. Among glucocorticoid users, cumulative glucocorticoid use in the 18 months prior to the index date and 12 months after the index date in the health plan claims data was 1407 mg (SD 1981) with median 820 mg (IQR 284–1900), and approximately 24% of patients in both cohorts remained on glucocorticoids 1 year after index. Opioids were used by 39.9% in the claims cohort data and 21.7% in ENRICH, although with potential under-capture in ENRICH if prescribed by a non-rheumatology provider (also evident in the low frequency of patients with ≥2 unique opioid prescriptions in ENRICH).
Table 3. Index glucocorticoid and non-immunomodulatory therapy use in claims (18-month baseline and 12 months after index) and EHR (12 months before and after index) data.
| Administrative claims data, n=1938 | EHR (ENRICH), n=1196 |
|
|---|---|---|
| Glucocorticoids, % | 1332 (68.7) | 883 (73.8) |
| Glucocorticoid initial dose (among users) | ||
| <10 mg/day | 441 (33.1) | 50 (5.8) |
| 10 to <20 mg/day | 411 (30.9) | 304 (35.3) |
| ≥20 mg/day | 480 (36.0) | 508 (58.9) |
| Unknown | 0 | 21 |
| Cumulative GC dose in mg (among users, entire study period), mean (SD) | 1407 (1981) | N/A |
| Cumulative GC dose in mg (among users, 12 months after index), mean (SD) | 1313 (1912) | N/A |
| Glucocorticoid use at 1 year after index | 469 (24.2) | 282 (23.6) |
| Opioids† | 773 (39.9) | 259 (21.7) |
| Opioids≥2 prescription fills/written prescriptions† | 618 (31.9) | 87 (7.3) |
| NSAIDs | 685 (35.3) | 492 (41.2) |
n (%) unless otherwise indicated.
Based on prescription fill (claims data) or active prescription (EHR) in the 9–12 months after index.
Opioids were captured based on filled prescriptions for opioids (claims) or unique written prescriptions in EHR records during the 12-month follow-up period, noting that for EHR data prescriptions may not be captured if written by other providers.
EHR, electronic health record; GC, glucocorticoid; NSAID, nonsteroidal anti-inflammatory drugs.
Patient outcomes and incident disease manifestations/comorbidities
In the health plan claims data, the mean follow-up duration was 862 days (SD 589). The incidence of the first all-cause hospitalisation after the index date was 92.2/1000 person-years (table 4). The incidence of malignancy, ILD, myocarditis and gastrostomy tube placement was 15.3, 6.4, 2.1 and 2.9 per 1000 person-years, respectively. Cumulative incidence of these outcomes is shown in online supplemental Figure 4.
Table 4. Rates of events of interest and diagnoses in patients with dermatomyositis using administrative claims data.
| Event type (all events) | Number events | Person-years | Incidence rate per 1000 patient-years (95% CI) |
|---|---|---|---|
| Incident all-cause hospitalisation | 467 | 5065 | 92.2 (84.1, 100.9) |
| Mortality | 106 | 5852 | 18.1 (14.9, 21.8) |
| Incident malignancy | 87 | 5693 | 15.3 (12.3, 18.8) |
| Incident ILD* | 33 | 5189 | 6.4 (4.5, 8.8) |
| Incident myocarditis† | 12 | 5802 | 2.1 (1.1, 3.5) |
| Hospitalised myocarditis | 5 | 5822 | 0.9 (0.3, 1.9) |
| Gastrostomy tube placement‡ | 17 | 5828 | 2.9 (1.8, 4.6) |
n=2475 in DM cohort, unless otherwise specified. In patients lost to follow-up before 6 months, 46, 6, 0, 0, 0, 0 and three contributed to the event types of interest, respectively.
Based on approach used by Meehan et al, Arthritis Res Ther. 2022 Jan 3;24(1); analysis excludes patients with baseline (prevalent) ILD (n=2182 in analysis cohort).
Analysis excludes patients with any prior diagnosis code for myocarditis (n=2466 in analysis cohort).
Analysis excludes patients with any prior procedure code for gastrostomy tube placement (n=2474 in analysis cohort).
CI, confidence interval; DM, dermatomyositis; ILD, interstitial lung disease.
Discussion
This retrospective study of two independent real-world data sources assessed the characteristics, treatment patterns, and long-term outcomes of patients with incident dermatomyositis. Through the inclusion of a cohort of patients from commercial health plan data and a separate cohort from community rheumatology practices, this study leveraged the advantages of each data source to produce a comprehensive account of real-world practice patterns. We identified a high burden of glucocorticoid exposure and found significant heterogeneity in the initial treatment of dermatomyositis, including conventional immunomodulatory therapies as well as advanced therapies such as biologic medications and IVIG, highlighting the need for improved treatment approaches. Additionally, use of administrative claims data allowed an assessment of long-term outcomes in this population.
While patients were identified based on DM diagnosis codes either from any healthcare provider or from rheumatologists (for the EHR cohort), reassuringly, the cohorts identified are comparable to other contemporary cohorts. We found similar age, sex, ILD prevalence and rates of immunosuppressive medication use compared with a contemporary Canadian cohort.22 Average values for maximum creatine kinase were modestly elevated, which may reflect the fact that a substantial proportion of patients with DM are amyopathic or have normal creatine kinase even in the presence of muscle weakness.23 In addition, creatine kinase values obtained during hospitalisations or prior to establishing care in the rheumatology practice may have been missed.
We found positive myositis-specific or myositis-associated antibodies in around 75% of the EHR cohort, with myositis-specific antibodies present among 35% of those who had myositis antibody panel results. In comparison, other studies report positive myositis-specific antibodies in 60–80% although lower rates are also reported depending on the cohort.24 25 Testing was likely mostly conducted in commercial laboratories which may explain lower rates of positivity. Finding the highest frequencies of TIF-1 gamma, NXP-2, Mi-2 and Jo-1 antibodies is expected for a DM cohort.24 26 27 Notably, we excluded those with malignancy at baseline, which may lead to lower rates of TIF-1 gamma positivity or NXP-2 antibodies. As expected, overall rates of Jo-1 and other synthetase antibodies were lower in our DM cohorts than in other general myositis cohorts27 because physicians may use polymyositis or other myositis diagnosis codes for many patients with antisynthetase syndrome, particularly those without substantial skin involvement. Many studies reporting antibody positivity are derived from single-centre tertiary practices or registry studies. Our results reflect the capture of myositis of autoantibody positivity on a large scale at the community-practice level, providing valuable context for the prevalence of antibody subtypes in the US dermatomyositis population using commercially available tests.
We found considerable heterogeneity in medication use among both cohorts. Among the conventional immunosuppressive medications, hydroxychloroquine, methotrexate, mycophenolate and azathioprine were each common as initial treatment. Advanced therapies such as rituximab, other biologic medications and IVIG were much less frequently used and tended to be added later. In comparison to an older claims study using data from 2004 to 2008, however, the frequency of IVIG (approximately 10% vs 1.5%) and rituximab use (approximately 5% vs 0.5%) was much greater in our modern cohort,11 in line with results from another recently published study.28 The overall variability in treatment approach may reflect the lack of guidelines for treatment in dermatomyositis and the need for evidence-based treatment algorithms. Similar to previous studies, we found high rates of glucocorticoid use, reflecting the important role of glucocorticoids in treatment of early disease.22 29 Using claims data, we were able to provide more detailed measures of glucocorticoid use than prior studies; the high cumulative glucocorticoid exposure (mean of 1400 mg) highlights the need for more effective glucocorticoid-sparing approaches or earlier use of effective therapies. Opioid use was also surprisingly common, although the lower frequency in the EHR cohort (particularly for multiple unique records) suggests that some of this prescribing may not be for dermatomyositis, or at least is not written by rheumatologists.
The claims-based cohort allows the assessment of long-term outcomes which cannot be fully captured in single-centre cohorts or most registries. We identified an incidence rate of 92 hospitalisations/1000 person-years compared with approximately 60 per 1000 person-years in previous studies of the general population,30 reflecting the substantial healthcare utilisation for this population. Mortality rates of 18/1000 person-years are also substantially higher than rates of 4–9/1000 person-years in the similarly aged general population,31 similar to what was found in a previous study of commercially insured patients with dermatomyositis.28 Future studies evaluating risk factors for these outcomes may help identify high-risk patients. We excluded 12.4% of patients with a diagnosis of malignancy (other than NMSC) in the enrolment period, which is consistent with prevalent malignancy rates in a Canadian cohort,22 and found a cumulative incidence of malignancy at 3 years after index of 4%, similar to rates in other studies and demonstrating the smaller but still substantial risk for malignancy during follow-up, although some misclassification of malignancy is possible given the use of a single diagnosis code.32 Prevalent ILD was identified in 10.9% of patients, supporting a significant comorbidity burden that warrants screening, although noting that ILD risk may differ substantially in those with different myositis-specific antibodies. Myocarditis was infrequent and may be even less frequent than our estimates given the possibility that elevations of troponin due to skeletal muscle inflammation could lead to a misdiagnosis of myocarditis. Overall, these results demonstrate the potential for assessing a variety of long-term outcomes in this population.
This study highlights the complementary nature of claims and EHR-based cohorts, allowing the identification of trends that may not be possible in either cohort alone. For example, the claims-based cohort allows the capture of prescriptions that may not be prescribed by the treating physicians within the EHR and may allow for more complete capture of comorbidities. The ability to assess long-term outcomes is an additional advantage of the claims-based cohort. In contrast, the EHR cohort captures laboratory values and antibody positivity rates, allowing for investigations by antibody subtypes. In the future, building linked claims and EHR cohorts in immune-mediated myopathies would allow for research into comorbidity patterns and pharmacoepidemiology unique to antibody subtypes currently not possible in the separate EHR and claims cohorts.
There are also limitations to consider. First, this study applied diagnosis code-based algorithms for the identification of incident cases of DM. While this approach has been applied previously and inpatient codes for DM have been validated, there is the potential for misclassification, particularly in the outpatient setting. This misclassification is expected to be lower in the EHR cohort given diagnoses in this cohort were from a rheumatologist. The study purposefully excluded those with malignancy or forms of myositis other than DM; the findings may not be generalisable to those with non-DM idiopathic inflammatory myopathies or those with malignancy-associated DM (an important minority with DM). We also excluded people receiving immunosuppressive therapies before the first DM diagnosis to ensure incident disease was captured, but this approach excludes people with other autoimmune conditions prior to DM. There may be incomplete capture of test results or certain treatments in the rheumatology-specific EHR if not present in structured data, which may explain the lower rates of IVIG (potential for incomplete capture if given by home infusion) observed in the EHR cohort compared with reported rates in the claims cohort and other studies.33 This limitation demonstrates the potential benefits of future studies using linked EHR and claims data, and taking advantage of unstructured data, to harness the benefits of each individual type of data.
In conclusion, this retrospective study of two independent data sources allowed for the description of patient characteristics, real-world prescription patterns and disease outcomes of two incident cohorts of patients with dermatomyositis. The study highlights a high burden of glucocorticoid exposure, significant heterogeneity in the approach to treatment and high healthcare utilisation in this population.
Supplementary material
Footnotes
Funding: This study was funded by Pfizer. The ENRICH network receives infrastructure support from the National Institute of Arthritis and Musculoskeletal Disease (NIAMS) P30AR072583. TRR is supported by the National Institutes of Health Pharmacoepidemiology T32 (5T32GM075766-17). ER is supported by the National Institute of Arthritis and Musculoskeletal Disease (NIAMS) Rheumatology Research Training Grant T32-AR076951-04. BRE is supported by the US Department of Veterans Affairs CSR&D (IK2 CX002203) and the Rheumatology Research Foundation.
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent for publication: Not applicable.
Data availability free text: Data cannot be shared for privacy and ethical reasons.
Data availability statement
No data are available.
References
- 1.Chung MP, Lovell J, Kelly W, et al. Myocarditis in Patients With Idiopathic Inflammatory Myopathies: Clinical Presentation and Outcomes. J Rheumatol. 2023;50:1039–46. doi: 10.3899/jrheum.220989. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.George MD, Shah R, Kreider M, et al. Pulmonary function tests, interstitial lung disease and lung function decline in outpatients with classic and clinically amyopathic dermatomyositis. Br J Dermatol. 2017;176:262–4. doi: 10.1111/bjd.14771. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Bradford Rice J, White A, Lopez A, et al. Healthcare resource utilization and work loss in dermatomyositis and polymyositis patients in a privately-insured US population. J Med Econ. 2016;19:649–54. doi: 10.3111/13696998.2016.1151433. [DOI] [PubMed] [Google Scholar]
- 4.Digre KB, Bruce BB, McDermott MP, et al. Quality of life in idiopathic intracranial hypertension at diagnosis: IIH Treatment Trial results. Neurology (ECronicon) 2015;84:2449–56. doi: 10.1212/WNL.0000000000001687. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Pandya R, Kleitsch J, Lim D, et al. Clinical characteristics and symptom progression of dermatomyositis subtypes: A retrospective analysis of a prospective database. J Am Acad Dermatol. 2024;91:31–6. doi: 10.1016/j.jaad.2024.02.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Da Silva DM, Patel B, Werth VP. Dermatomyositis: A diagnostic dilemma. J Am Acad Dermatol. 2018;79:371–3. doi: 10.1016/j.jaad.2017.12.074. [DOI] [PubMed] [Google Scholar]
- 7.Mehta P, Rathore U, Naveen R, et al. Prevalent Drug Usage Practices in Adults and Children With Idiopathic Inflammatory Myopathies: Registry-Based Analysis From the MyoCite Cohort. J Clin Rheumatol. 2022;28:89–96. doi: 10.1097/RHU.0000000000001813. [DOI] [PubMed] [Google Scholar]
- 8.Kalluri M, Sahn SA, Oddis CV, et al. Clinical profile of anti-PL-12 autoantibody. Cohort study and review of the literature. Chest. 2009;135:1550–6. doi: 10.1378/chest.08-2233. [DOI] [PubMed] [Google Scholar]
- 9.Kronzer VL, Kimbrough BA, Crowson CS, et al. Incidence, Prevalence, and Mortality of Dermatomyositis: A Population-Based Cohort Study. Arthritis Care Res (Hoboken) 2023;75:348–55. doi: 10.1002/acr.24786. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Lin RR, Warp PV, Williams KN, et al. Sex-based differences in dermatomyositis: a retrospective study. Arch Dermatol Res. 2024;316:3–4. doi: 10.1007/s00403-024-03104-y. [DOI] [PubMed] [Google Scholar]
- 11.Smoyer-Tomic KE, Amato AA, Fernandes AW. Incidence and prevalence of idiopathic inflammatory myopathies among commercially insured, Medicare supplemental insured, and Medicaid enrolled populations: an administrative claims analysis. BMC Musculoskelet Disord. 2012;13:1–13. doi: 10.1186/1471-2474-13-103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bolender CM, Jimenez A, Clarke JT, et al. Incidence of Dermatomyositis in a Nationwide Cohort Study of US Veterans. JAMA Dermatol. 2022;158:1321–3. doi: 10.1001/jamadermatol.2022.3112. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Beukelman T, Mudano A, Stewart P, et al. Generating Real-World Evidence From the Excellence Network in Rheumatology. Pharmacoepidemiol Drug Saf. 2024;33:1–7. doi: 10.1002/pds.70067. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hannah JR, Gordon PA, Galloway J, et al. Validation of methods to identify people with idiopathic inflammatory myopathies using hospital episode statistics. Rheumatol Adv Pract. 2022;6:1–7. doi: 10.1093/rap/rkac102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Kwa MC, Ardalan K, Laumann AE, et al. Validation of International Classification of Diseases Codes for the Epidemiologic Study of Dermatomyositis. Arthritis Care Res (Hoboken) 2017;69:753–7. doi: 10.1002/acr.23010. [DOI] [PubMed] [Google Scholar]
- 16.Bernatsky S, Joseph L, Pineau CA, et al. Estimating the prevalence of polymyositis and dermatomyositis from administrative data: age, sex and regional differences. Ann Rheum Dis. 2009;68:1192–6. doi: 10.1136/ard.2008.093161. [DOI] [PubMed] [Google Scholar]
- 17.Silberzweig A, Cohn E, Strunk E, et al. Validity of dermatomyositis case identification using diagnostic codes. Medical Dermatology Society. 2024;1 [Google Scholar]
- 18.Silberzweig A, Strunk A, Flatley E, et al. Validity of Diagnosis Codes to Identify a Dermatomyositis Cohort from Electronic Health Records. Dermatology . 2025;241:216–22. doi: 10.1159/000543803. [DOI] [PubMed] [Google Scholar]
- 19.Allenzara A, Stachowski J, Jones EP, et al. Identification of idiopathic inflammatory myopathy research cohorts using international classification of disease (ICD) codes: A systematic review. Semin Arthritis Rheum. 2025;75:152830. doi: 10.1016/j.semarthrit.2025.152830. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Meehan M, Shah A, Lobo J, et al. Validation of an algorithm to identify incident interstitial lung disease in patients with rheumatoid arthritis. Arthritis Res Ther. 2022;24:1–11. doi: 10.1186/s13075-021-02655-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Wu L-Y, Shao S-C, Liao S-C. Positive Predictive Value of ICD-10-CM Codes for Myocarditis in Claims Data: A Multi-Institutional Study in Taiwan. Clin Epidemiol. 2023;15:459–68. doi: 10.2147/CLEP.S405660. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Osman M, Martins KJB, Wong KO, et al. Incidence and prevalence, and medication use among adults living with dermatomyositis: an Alberta, Canada population-based cohort study. Sci Rep. 2023;13:1–9. doi: 10.1038/s41598-023-43880-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Tang K, Zhang H, Jin H. Clinical Characteristics and Management of Patients With Clinical Amyopathic Dermatomyositis: A Retrospective Study of 64 Patients at a Tertiary Dermatology Department. Front Med (Lausanne) 2021;8 doi: 10.3389/fmed.2021.783416. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Hodgkinson LM, Wu TT, Fiorentino DF. Dermatomyositis autoantibodies: how can we maximize utility? Ann Transl Med. 2021;9:433. doi: 10.21037/atm-20-5175. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Gandiga PC, Zhang J, Sangani S, et al. Utilization patterns and performance of commercial myositis autoantibody panels in routine clinical practice. Br J Dermatol. 2019;181:1090–2. doi: 10.1111/bjd.18133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Gómez GN, Pérez N, Braillard Poccard A, et al. Myositis-specific antibodies and clinical characteristics in patients with autoimmune inflammatory myopathies: reported by the Argentine Registry of Inflammatory Myopathies of the Argentine Society of Rheumatology. Clin Rheumatol. 2021;40:4473–83. doi: 10.1007/s10067-021-05797-2. [DOI] [PubMed] [Google Scholar]
- 27.Chiang H-L, Tung C-H, Huang K-Y, et al. Association between clinical phenotypes of dermatomyositis and polymyositis with myositis-specific antibodies and overlap systemic autoimmune diseases. Medicine (Baltimore) 2021;100:e27230. doi: 10.1097/MD.0000000000027230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Barnes DM, Graham D, Borlenghi CE, et al. A retrospective natural history study in adult and juvenile patients with incident dermatomyositis and polymyositis using real world data. Clin Rheumatol. 2025;44:4237–47. doi: 10.1007/s10067-025-07614-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Saygin D, Oddis CV. Glucocorticoids in Myositis: Initiation, Tapering, and Discontinuation. Curr Rheumatol Rep. 2022;24:47–53. doi: 10.1007/s11926-022-01060-y. [DOI] [PubMed] [Google Scholar]
- 30.Tillmann BW, Fu L, Hill AD, et al. Acute healthcare resource utilization by age: A cohort study. PLoS One. 2021;16:e0251877. doi: 10.1371/journal.pone.0251877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Woolf SH, Schoomaker H. Life Expectancy and Mortality Rates in the United States, 1959-2017. JAMA. 2019;322:1996–2016. doi: 10.1001/jama.2019.16932. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Hsu J-L, Liao M-F, Chu C-C, et al. Reappraisal of the incidence, various types and risk factors of malignancies in patients with dermatomyositis and polymyositis in Taiwan. Sci Rep. 2021;11:1–11. doi: 10.1038/s41598-021-83729-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Miyazaki C, Ishii Y, Stelmaszuk NM. Disease burden and treatment sequence of polymyositis and dermatomyositis patients in Japan: a real-world evidence study. Clin Rheumatol. 2022;41:741–55. doi: 10.1007/s10067-021-05939-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
No data are available.