Macrophage activation syndrome in children with systemic lupus erythematosus and children with juvenile idiopathic arthritis

Authors


Abstract

Objective

To describe patient demographics, interventions, and outcomes in hospitalized children with macrophage activation syndrome (MAS) complicating systemic lupus erythematosus (SLE) or juvenile idiopathic arthritis (JIA).

Methods

We performed a retrospective cohort study using data recorded in the Pediatric Health Information System (PHIS) database from October 1, 2006 to September 30, 2010. Participants had International Classification of Diseases, Ninth Revision, Clinical Modification diagnosis codes for MAS and either SLE or JIA. The primary outcome was hospital mortality (for the index admission). Secondary outcomes included intensive care unit (ICU) admission, critical care interventions, and medication use.

Results

A total of 121 children at 28 children's hospitals met the inclusion criteria, including 19 children with SLE and 102 children with JIA. The index admission mortality rate was 7% (8 of 121 patients). ICU admission (33%), mechanical ventilation (26%), and inotrope/vasopressor therapy (26%) were common. Compared to children with JIA, those with SLE had a similar mortality rate (6% versus 11%, respectively; exact P = 0.6). More patients with SLE than those with JIA received ICU care (63% versus 27%; P = 0.002), received mechanical ventilation (53% versus 21%; P = 0.003), and had cardiovascular dysfunction (47% versus 23% received inotrope/vasopressor therapy; P = 0.02). Children with SLE and those with JIA received cyclosporine at similar rates, but more children with SLE received cyclophosphamide and mycophenolate mofetil, and more children with JIA received interleukin-1 antagonists.

Conclusion

Organ system dysfunction is common in children with rheumatic diseases complicated by MAS, and more organ system support is required in children with underlying SLE than in children with JIA. Current treatment of pediatric MAS varies based on the underlying rheumatic disease.

Macrophage activation syndrome (MAS) and hemophagocytic lymphohistiocytosis (HLH) are incompletely understood conditions associated with uncontrolled and ineffective immune activation (1). Both conditions have high mortality rates, particularly when untreated (1–4). HLH encompasses two related conditions, a primary form in genetically predisposed patients typically diagnosed in infancy or early childhood, and a secondary form that may develop in response to an inflammatory stimulus, such as infection (often viral) (1), malignancy (5), or rheumatic disease (2). MAS is a known complication of pediatric rheumatic disorders, particularly juvenile idiopathic arthritis (JIA) (3) and systemic lupus erythematosus (SLE) (6, 7). It has been suggested that MAS is the same disorder as the secondary or “reactive” form of HLH (8).

Diagnostic and treatment approaches for children with HLH are generally based on HLH-04, a recently updated study protocol that includes initial treatment with dexamethasone, cyclosporine, and etoposide for both the primary and secondary forms (2). Alternatively, children with MAS and a known rheumatic disease may receive varying treatments, including high-dose methylprednisolone, cyclosporine, cyclophosphamide, mycophenolate mofetil (MMF), azathioprine, rituximab, or interleukin-1 (IL-1) antagonists (7, 9). These immunosuppressive medications are typically chosen because they have activity against both the underlying rheumatic disease and MAS. Hospital or intensive care unit (ICU) management is often needed for organ system dysfunction or for secondary infections related to immunosuppression in children with HLH or MAS (2, 3). Recent reports of MAS complicating SLE have prompted concern that this association may be underrecognized (6, 7).

Pediatric rheumatic diseases are rare, and MAS is not a frequent complication. Therefore, large collaborations have been necessary for the collection of data on cohorts of children with MAS and rheumatic diseases (3, 7). The Pediatric Health Information System (PHIS) database has unique advantages for researchers studying hospitalized children with rare diseases and was recently used for studies of Kawasaki disease (KD) (10) and Henoch-Schönlein purpura (11, 12).

The purpose of this study was to describe subject demographics, therapeutic interventions, and outcomes in a cohort of children with SLE or JIA complicated by MAS, and to evaluate variation in treatment, costs, and outcomes across these disease groups.

PATIENTS AND METHODS

Study design.

We conducted a retrospective cohort study of patients with data recorded in the PHIS database, which was developed by the Children's Hospital Association (formerly known as the Child Health Corporation of America) (13). The subjects were children who received inpatient care for MAS and either SLE or JIA at a PHIS hospital.

Setting.

The Children's Hospital Association is a business alliance of 43 children's hospitals, and PHIS contains administrative data, including patient demographics, diagnoses, procedures, and charges. In addition, a subset of PHIS hospitals submits “Level II” data, which are detailed data on billed services, including pharmacy, clinical services, imaging, laboratory, supply, and room charges (11). Clinical Transaction Classification (CTC) codes are used to identify the detailed billing services received by patients (11, 14, 15). All PHIS data are de-identified and checked for reliability and validity prior to release (14). We obtained data regarding patients admitted to PHIS member hospitals who met our inclusion criteria. Our data use agreement limits reporting of hospital-level data, but general information on 36 PHIS hospitals has been published previously (11), and several studies of uncommon diseases in children admitted to PHIS hospitals have been published (10, 11, 16, 17).

Selection of participants.

We identified children who were treated at a PHIS hospital and had International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) discharge diagnosis codes for MAS (288.4 [hemophagocytic syndromes]) and either SLE (710.0) or JIA (714.30–714.33) from October 1, 2006 to September 30, 2010. We grouped all categories of JIA together because misclassification of the type of JIA in administrative data was more likely than MAS in a child with a category of JIA other than systemic arthritis. The 288.4 diagnosis code for hemophagocytic syndromes was adopted on October 1, 2006 (18), replacing the more generic code 288.0 (leukopenia), which had previously been used for patients with MAS. We defined the index admission as the first hospital admission at which each individual patient met the inclusion criteria.

Covariates and outcome measures.

Unless otherwise noted, all analyses were of each patient's index admission. The ICD-9-CM procedure and diagnosis codes used to identify diagnostic and therapeutic interventions and complications are available from the author upon request. The primary outcome was hospital mortality (for the index admission). Secondary outcomes and covariates included ICU admission, length of hospital stay, hospitalization costs, critical care interventions, central nervous system (CNS) evaluations and therapies, gastrointestinal hemorrhage and blood product transfusions, and immunosuppressive medication use. All medication use was ascertained using pharmacy CTC codes, and each drug had an associated “day of service” code indicating the day of the hospital stay on which that drug was ordered. The inotrope/vasopressor therapy variable included any use of dopamine, dobutamine, epinephrine, norepinephrine, or phenylephrine. The antiepileptic therapy variable included any use of phenobarbital, phenytoin, fosphenytoin, levetiracetam, or valproic acid. Intravenous immunoglobulin (IVIG) was coded using both CTC and ICD-9-CM codes. Other blood products were coded using ICD-9-CM procedure codes. Adjusted total inpatient costs were calculated by PHIS from charges using hospital-specific costs-to-charges ratios, and are adjusted to the Center for Medicare wage index (14).

Evaluation of patient selection criteria.

We validated our patient selection criteria by reviewing the electronic medical records of patients who met our inclusion criteria at a single PHIS hospital to confirm the provider diagnosis of SLE or JIA and MAS.

Statistical analysis.

We compared categorical covariates and outcome measures using the chi-square test or Pearson's exact test where appropriate, and length of hospital stay and costs using Wilcoxon's rank sum test. We used Kaplan-Meier analysis to estimate the distribution of time (in days after admission) to the first use of cyclosporine and the first ferritin level test, and the log rank test to test for differences by diagnosis type. A ferritin level is a component of the diagnostic guidelines for HLH (2), which may be applied to children with rheumatic diseases being evaluated for MAS (7).

P values less than 0.05 were considered significant. All analyses were performed using Stata software, version 11 (StataCorp). This study was reviewed by the University of Utah School of Medicine Institutional Review Board, and informed consent was not required.

RESULTS

Characteristics of the study subjects.

We identified 7,436 inpatient admissions at PHIS hospitals of 3,567 unique patients with diagnoses of SLE or JIA, and 1,379 admissions of 628 patients with a diagnosis code for MAS. Of those, 198 admissions of 123 unique patients had both a diagnosis of SLE or JIA and a diagnosis code for MAS. Subsequent admissions were then excluded, and 2 patients were excluded for missing disposition at the index admission, leaving a total of 121 patients. A median of 4 patients (range 1–18, interquartile range [IQR] 2–5) per hospital were treated at 28 hospitals over 48 calendar months. Each of the 28 hospitals listed at least one staff pediatric rheumatologist on their publicly available web site on June 21, 2012.

In this cohort of children with their first hospital admission for MAS, JIA was much more common than SLE (Table 1). Of the 102 children with JIA, 90 (88%) had an ICD-9-CM diagnosis code for systemic JIA (714.30), 8 (8%) for polyarticular JIA (714.31), 3 (3%) for pauciarticular JIA (714.32), and 1 (1%) for monarticular JIA (714.33). Two children had diagnoses of both JIA and KD at the index admission. No patients had codes for both SLE and JIA.

Table 1. Characteristics of the patients in the pediatric MAS cohort at hospital admission, by diagnosis*
 SLE (n = 19)JIA (n = 102)
  • *

    Except where indicated otherwise, values are the number (%) of patients. MAS = macrophage activation syndrome; SLE = systemic lupus erythematosus; JIA = juvenile idiopathic arthritis; ICU = intensive care unit; IQR = interquartile range; HSCT = hematopoietic stem cell transplant.

Age  
 0–364 days1 (5)1 (1)
 1–<5 years0 (0)31 (30)
 5–<13 years2 (11)37 (36)
 13–<18 years16 (84)33 (32)
Sex  
 Female12 (63)65 (64)
 Male7 (37)37 (36)
Race/ethnicity  
 White/non-Hispanic4 (21)51 (50)
 Hispanic/Latino6 (32)16 (16)
 Black1 (5)17 (17)
 Other8 (42)16 (16)
 Missing0 (0)2 (2)
Insurance status  
 Government8 (42)41 (40)
 Private6 (32)43 (42)
 Other5 (26)15 (15)
 Missing0 (0)3 (3)
Admission year  
 2006–20073 (16)36 (35)
 20084 (21)27 (26)
 200910 (53)16 (16)
 20102 (11)23 (23)
Mortality2 (11)6 (6)
ICU stay, days  
 Any12 (63)28 (27)
 Range0–1550–77
 Median (IQR)8 (0–13)0 (0–1)
Length of hospital stay, days  
 Range6–3671–122
 Median (IQR)25 (12–80)6 (4–15)
Adjusted total costs  
 Range, × $1,0009.7–2,230.31.8–674.2
 Median (IQR), × $1,000126.7 (28.2–359.7)16.9 (8.4–41.9)
 Missing0 (0)9 (9)
HSCT0 (0)2 (2)

Children with JIA were more likely to be younger than 5 years old than were children with SLE (Table 1). In the entire cohort, nearly two-thirds (64%) of the patients were female. Nonwhite race/ethnicity (54%) was more common than white race/ethnicity (46%), and ∼18% of the patients were of Hispanic/Latino ethnicity.

Primary and secondary outcomes.

The hospital mortality rate for the entire cohort was 7% (Table 1). Children with JIA and those with SLE had similar mortality rates (6% versus 11%; exact P = 0.6).

ICU admission was common (33% of the entire cohort). The median length of hospital stay was 8 days (IQR 5–19 days), and the median hospital cost was $21,574 (IQR $9,501–59,781). Children with SLE had a higher ICU admission rate (63% versus 27%; P = 0.002), longer length of hospital stay (median 25 days versus 6 days; P < 0.001), and higher costs (median $126,700 versus $16,886; P < 0.001) than children with JIA.

Mechanical ventilation (26%) and inotrope/vasopressor therapy (26%) were common (Table 2). Higher percentages of children with SLE received mechanical ventilation (53% versus 21%; P = 0.003) and cardiovascular support (inotrope/vasopressor therapy) (47% versus 23%; P = 0.02) than children with JIA. Three patients with JIA received extracorporeal membrane oxygenation for vital organ failure, and only 1 survived.

Table 2. Selected diagnoses and interventions in the pediatric MAS cohort, by diagnosis*
 SLE (n = 19)JIA (n = 102)
  • *

    Values are the number (%) of patients. CRRT = continuous renal replacement therapy; ECMO = extracorporeal membrane oxygenation; CNS = central nervous system; ICP = intracranial pressure; CT = computed tomography; MRI = magnetic resonance imaging; MRA = magnetic resonance angiography; EEG = electroencephalography; PRBCs = packed red blood cells; GI = gastrointestinal (see Table 1 for other definitions).

Critical care  
 Mechanical ventilation10 (53)21 (21)
 Nitric oxide0 (0)6 (6)
 Dialysis or CRRT5 (26)6 (6)
 Inotrope/vasopressor therapy9 (47)23 (23)
 ECMO0 (0)3 (3)
CNS involvement/treatment  
 Ischemic stroke0 (0)0 (0)
 Intracranial hemorrhage2 (11)1 (1)
 Hypertonic saline7 (37)7 (7)
 Mannitol2 (11)0 (0)
 Antiepileptic agent5 (26)7 (7)
CNS diagnostics  
 ICP monitor1 (5)2 (2)
 Brain CT9 (47)7 (7)
 Brain MRI and/or MRA9 (47)7 (7)
 Any CNS imaging12 (63)13 (13)
 EEG5 (26)6 (6)
 CNS imaging or EEG12 (63)14 (14)
Hyponatremia (electrolyte disorder)5 (26)4 (4)
Bleeding/transfusion  
 PRBCs7 (37)14 (14)
 Platelets3 (16)5 (5)
 Plasma3 (16)7 (7)
 Activated factor VIIa1 (5)1 (1)
 GI hemorrhage3 (16)0 (0)
Bone marrow biopsy6 (32)39 (38)

We defined concern for CNS involvement as receipt of brain magnetic resonance imaging or computed tomography scan or an electroencephalogram (EEG). Overall, 21% of the subjects underwent CNS imaging or an EEG, with higher rates among patients with SLE (63%) than those with JIA (14%) (P < 0.001) (Table 2). Concern for CNS involvement, as evidenced by CNS imaging or EEG, was associated with hospital mortality: 27% (7 of 26) of those who underwent CNS imaging or an EEG died versus 1% (1 of 95) of those who did not (exact P < 0.001).

Hyponatremia occurred in 7% of the patients, with higher rates among children with SLE than children with JIA (26% versus 4%, exact P = 0.005) (Table 2). Administration of hypertonic saline (in any concentration greater than 0.9%) was fairly common (12%), again with greater use in children with SLE than in those with JIA (37% versus 7%; P < 0.001) (Table 2). However, most children who received hypertonic saline (10 of 14 [71%]) did not have a diagnosis code for hyponatremia. Only 2 children (both with SLE) received mannitol, and only 3 children (1 with SLE and 2 with JIA) were monitored for intracranial pressure.

Blood product transfusion (packed red blood cells, platelets, or plasma) was common (20% of all patients), with higher rates among children with SLE than children with JIA (42% versus 16%; P = 0.008) (Table 2).

Immunosuppressive treatments.

The immunosuppressive medications given to patients with each diagnosis are shown in Table 3. Most patients (93%) received corticosteroids, most often methylprednisolone (83%). More children with SLE than those with JIA received dexamethasone (32% versus 14%; P = 0.05), the corticosteroid recommended in the HLH-04 protocol. Children with SLE were more likely to receive cyclophosphamide (21% versus 3%; exact P = 0.01) and MMF (32% versus 2%, exact P < 0.001) than children with JIA.

Table 3. Immunosuppressive medications received by the patients in the pediatric MAS cohort, by diagnosis*
 SLE (n = 19)JIA (n = 102)
  • *

    Except where indicated otherwise, values are the number (%) of patients. IL-1 = interleukin-1; IVIG = intravenous immunoglobulin; 95% CI = 95% confidence interval (see Table 1 for other definitions).

Any steroid19 (100)93 (91)
Dexamethasone6 (32)14 (14)
Methylprednisolone17 (89)83 (81)
Cyclophosphamide4 (21)3 (3)
Mycophenolate mofetil6 (32)2 (2)
Azathioprine1 (5)0 (0)
IL-1 blockade0 (0)15 (15)
Rituximab2 (1)1 (1)
Intrathecal hydrocortisone0 (0)0 (0)
Intrathecal methotrexate0 (0)0 (0)
IVIG7 (37)18 (18)
Plasmapheresis2 (11)0 (0)
Cyclosporine and/or etoposide  
 Neither cyclosporine nor etoposide9 (47)55 (54)
 Etoposide alone1 (5)2 (2)
 Cyclosporine alone8 (42)43 (42)
 Cyclosporine and etoposide1 (5)2 (2)
Time to first dose of cyclosporine, estimated % given drug (95% CI)  
 3 days after admission21 (8–47)33 (25–43)
 7 days after admission26 (12–52)43 (33–54)
 14 days after admission40 (21–66)51 (39–64)

Fifteen patients at 8 hospitals, all with underlying JIA, received an IL-1 antagonist. Of these 15 patients, 14 also received corticosteroids, 5 received cyclosporine, and 1 received etoposide. IL-1 antagonists tended to be given early in the hospital stay: 11 of 15 patients received the IL-1 antagonist by hospital day 3, and 13 of 15 patients received it by hospital day 6.

The proportions of children with SLE and children with JIA receiving IVIG did not differ significantly (37% versus 18%; exact P = 0.06), and few patients received plasmapheresis (2 patients, both with SLE) (Table 3).

Approximately half (47%) of all patients received either cyclosporine or etoposide, with most of those (42%) receiving cyclosporine alone, only 2% receiving etoposide alone, and 2% receiving both cyclosporine and etoposide. Equivalent proportions of children with SLE and children with JIA received cyclosporine (47% versus 44%; P = 0.8). We did not find a difference between children with SLE and those with JIA with regard to the number of days between admission and the first dose of cyclosporine (P = 0.196 by log rank test). Two patients with JIA received a hematopoietic stem cell transplant at the index hospital during the index admission.

Figure 1 shows the proportion of patients who received corticosteroids, IVIG, cyclosporine, etoposide, and IL-1 antagonists in each year of the study. Other than a modest decrease in corticosteroid use over time, there did not appear to be consistent trends in immunosuppressant use over time.

Figure 1.

Immunosuppressive therapies administered to patients in the pediatric macrophage activation syndrome cohort, by hospital admission year. IVIG = intravenous immunoglobulin; IL-1 = interleukin-1.

Laboratory testing.

Overall, 92% of the patients (111 of 121) had their ferritin level checked during hospitalization, with no difference between children with SLE (18 of 19 [95%]) and those with JIA (93 of 102 [91%]; exact P = 1.00). Figure 2 shows the time, in days after hospital admission, to the first evaluation of ferritin level. Children with JIA had their ferritin level checked earlier in the hospital stay than did children with SLE (P = 0.017 by log rank test). We estimate that after one week of hospitalization, 37% of children with SLE (95% confidence interval [95% CI] 20–62%) and 73% of children with JIA (95% CI 64–82%) had had their ferritin level evaluated.

Figure 2.

Time to first evaluation of ferritin level in the patients in the pediatric macrophage activation syndrome cohort with systemic lupus erythematosus (SLE) or juvenile idiopathic arthritis (JIA).

Patient selection criteria validation.

We identified all patients (n = 5) at one PHIS hospital using our selection criteria, and confirmed that the medical record reported diagnoses of SLE or JIA and MAS in all 5 of these patients.

DISCUSSION

To our knowledge, this multicenter study is the largest reported cohort of hospitalized children with MAS complicating a chronic rheumatic disease. ICU admission and organ system dysfunction were common in children with SLE and those with JIA, and more organ system support was required in children with underlying SLE than in those with JIA. As expected given typical treatments for the two underlying rheumatic diseases, children with SLE or JIA complicated by MAS received cyclosporine at similar rates, but more children with SLE received cyclophosphamide and MMF, and more children with JIA received IL-1 antagonists.

More hospitalized children with MAS in our study had JIA than SLE. This is consistent with previous reports that MAS may be more common in children with JIA than in children with SLE (3). Other investigators have suggested that MAS is not recognized as readily in children with SLE as in children with other rheumatic disorders (6). Our finding that ferritin levels were evaluated earlier in the hospital stay in children with JIA than in children with SLE lends support to the concern about delayed diagnosis of MAS in children with SLE and to recent efforts to develop diagnostic criteria for MAS in children with SLE (7).

The index admission mortality rate of 7% in our cohort is similar to reported mortality rates of 7–22% from case series that included readmission data (3, 4, 7, 19). The age and sex distributions in our data set fit with what is known about the epidemiology of SLE (7) and JIA (3). The ICU admission rate in our study was 33% overall (compared to 42% in a previous study [3]) with a higher rate (63%) in children with SLE (consistent with 57% in a previous study [7]).

We found that children with SLE had more organ system dysfunction, longer hospital stays, and higher costs at their index hospitalization for MAS than children with JIA. Other investigators have presented the hypothesis that children with SLE and MAS have more organ system dysfunction (6), which is also supported by the findings of a recent registry study (7). This is the first study to compare children with SLE and those with JIA in a large cohort, and the first analysis to report medication, imaging, laboratory, and cost data.

Our estimate of CNS involvement during the index admission (in 21% of the patients), identified by the use of CNS imaging or EEG, was within the wide range reported in other studies (14–67%, with variable case definitions) (3, 7, 19, 20). Presumably, the imaging studies and EEGs were obtained in patients who had altered mental status, clinical seizures, or other manifestations of CNS disease. We found higher CNS involvement rates among children with SLE than those with JIA, as might be expected given the potential for CNS involvement in SLE without MAS. The index admission mortality rate was high among the patients with CNS involvement in our study (27%, compared to 1% in patients without CNS disease). CNS events, including intracranial hemorrhages and cerebral edema, may be important contributors to morbidity and mortality in children with MAS. The higher rate of dexamethasone use we found in children with SLE compared to those with JIA may reflect providers' decisions to administer a corticosteroid with better CNS penetration than methylprednisolone.

Hyponatremia is known to complicate active disease in children with rheumatic diseases and MAS (20, 21). We found higher rates of hyponatremia among children with SLE than those with JIA, consistent with the overall higher rates of organ dysfunction in those children. However, most of the children who received hypertonic saline did not have a diagnosis code for hyponatremia, suggesting that the hypertonic saline may have been given for cerebral edema, for seizure in the setting of suspected hyponatremia, or for the prevention of hyponatremia. Intracranial pressure monitoring and mannitol administration rates in our study were low, however, which suggests that few patients were known to have intracranial hypertension.

We found variation in the treatment regimen for MAS between children with SLE and those with JIA. These conditions are rare and incompletely understood, and variation in therapy is not unexpected in the absence of evidence to guide providers in their care of patients with varying disease characteristics. We found it interesting that only children with underlying JIA received IL-1 antagonists, perhaps due to recent reports of effect in those patients (9, 22) and little available data to support its use and safety profile in children with SLE. The variation in immunosuppressive medication use across diagnosis categories may highlight differences in the way providers approach children with MAS; children with SLE, who tended to have more organ system dysfunction, more often received therapy that resembled the HLH-04 protocol (including dexamethasone and IVIG) than children with JIA.

There are several study limitations to consider. The PHIS database has advantages over traditional administrative databases, in that it includes information regarding medications and the use of imaging studies and laboratory tests, but it has the standard limitations of existing data. The database does not contain clinical information (e.g., laboratory or imaging results) beyond that discernible from CTC and ICD-9-CM codes. We identified patients based on diagnosis codes and not on patient symptoms, signs, laboratory test results, or measures of disease severity. Our analysis of medical records from one PHIS hospital (n = 5 patients) suggests that our inclusion criteria were valid. Two patients in our study had diagnoses of both JIA and KD, and we classified those patients as having JIA based on reports of coronary dilation in JIA and a concern that some children with JIA are misdiagnosed as having KD (23).

Our analyses of index admission outcomes were limited because the database does not contain physiologic data, and our ability to adjust for severity of illness at hospital or ICU admission was limited. We were concerned about induced bias if we included treatment variables (medications, critical care interventions, etc.) in an outcome model, and chose to present the data descriptively and in bivariate analyses to facilitate hypothesis generation for other studies.

It is possible that selection bias is present in comparisons of MAS severity between children with JIA and children with SLE. If providers tended to diagnose MAS in a child with JIA sooner than in a child with SLE, delaying diagnosis in the child with SLE until the MAS was more severe, then MAS in SLE would appear to be more severe overall. In children with either JIA or SLE, less severe or “subclinical” (24–26) cases of MAS may not be diagnosed if they respond to immunosuppressive treatments given for a possible flare of the underlying disease. It is not known if this happens differentially between children with JIA and those with SLE.

Genetic abnormalities associated with HLH, such as perforin and syntaxin mutations and MUNC13-4 anomalies, are not coded in the PHIS database (2). ICD-9-CM diagnosis codes may have limited sensitivity for viral infections and other conditions that may trigger episodes of HLH or MAS (2, 3, 27). Because our study only includes hospitalized patients diagnosed with MAS, the incidence of MAS complicating SLE or JIA cannot be estimated.

Among children with MAS complicating rheumatic diseases, organ system dysfunction is common during the index admission. Children with underlying SLE appear to have more organ system dysfunction at their index MAS episode than children with underlying JIA. Children with underlying SLE and those with underlying JIA tend to receive different treatments for their MAS.

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. Bennett 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. Bennett, Fluchel, A. O. Hersh, Hayward, A. L. Hersh, Brogan, Srivastava, Bratton.

Acquisition of data. Korgenski.

Analysis and interpretation of data. Bennett, Fluchel, A. O. Hersh, A. L. Hersh, Brogan, Srivastava, Stone, Mundorff, Casper, Bratton.

Ancillary