National Trends in Pediatric Lupus Hospitalizations in the United States: 2000-2009

Abstract

Objective

In the setting of recent healthcare advances and emphasis on reduced spending, we aimed to characterize US trends in inpatient healthcare utilization and mortality for pediatric SLE.

Methods

We performed a retrospective, serial, cross-sectional analysis of the national Kids’ Inpatient Database (years 2000, 2003, 2006 & 2009). We identified patients aged 2 up to 21 years with SLE using an International Classification of Diseases, Ninth Revision [ICD-9] code of 710.0 listed as a discharge diagnosis. Using sampling weights, we estimated trends in hospitalization, inpatient mortality, procedure rates and length of stay (LOS). We analyzed patient and hospital-specific risk factors for mortality and LOS, and compared these outcomes to those without SLE.

Results

We identified 26,903 estimated pediatric SLE hospitalizations. The hospitalization rate of 8.6 (95% confidence interval [CI] 7.6-9.6) per 100,000 population and mean LOS of 5.9 days (95%CI 5.6-6.2) were stable over time. We found a significant downward trend in mortality, decreasing from 1% to 0.6% (p=0.04), which paralleled a less pronounced trend for those without SLE. The rate of dialysis, blood transfusions, and vascular catheterization procedures increased. Patients with SLE nephritis and non-White race were at risk for increased healthcare utilization and death.

Conclusion

Pediatric SLE hospitalization rate and LOS remained stable, but inpatient mortality decreased as the rate of common therapeutic procedures increased. More research is needed to understand the drivers of these relationships.

Introduction

Systemic lupus erythematosus (SLE) is a chronic autoimmune condition with potential for significant morbidity and mortality, thereby requiring frequent healthcare contact for optimal management and outcomes. Pediatric-onset SLE represents 15-20% of overall SLE cases, and is associated with an increased risk for aggressive clinical course and major organ damage in comparison to adults [1]. Pediatric SLE hospitalizations are significant events, indicating more severe disease, disrupting psychosocial and school functioning, and contributing to high healthcare utilization. Further understanding of the factors contributing to hospitalization in pediatric SLE may therefore improve clinical outcomes, quality of life and healthcare utilization for these patients.

Estimates of healthcare utilization and mortality for adults with SLE are established in the literature. Several studies of adults with SLE in the United States (US) have found high resource utilization in these patients [2, 3], and a study of US hospitalizations for adult SLE found that 1 in 30 hospitalizations resulted in death [4]. Few studies also describe increased healthcare utilization for children with SLE [5-7]; however, the relationship of resource utilization and outcomes such as mortality remains unclear. Importantly, these parameters may also be changing over time, given recent advances in SLE treatment, such as the recommendations for use of mycophenolate mofetil therapy [8, 9], other regimens for reduced exposure to cyclophosphamide toxicity [8-11], and the introduction of B-cell targeted Rituximab therapy. Additionally, there has been a general emphasis on reducing healthcare utilization while improving the quality of care.

We used a retrospective, serial cross-sectional design to analyze data from the nationally representative Kids’ Inpatient Database (KID) to examine US trends in inpatient healthcare utilization and mortality associated with pediatric SLE. Specifically, we aimed to: 1) characterize the national pediatric SLE inpatient cohort 2) describe trends in the hospitalization rate, inpatient mortality rate, procedure rate and length of stay (LOS) for pediatric SLE; 3) identify patient and hospital-specific risk factors for the outcomes of mortality and LOS 4) compare these outcomes for patients with and without pediatric SLE.

Patients and Methods

Study Design & Setting

Children and adolescents with SLE were identified from a serial cross-sectional analysis of pediatric discharges in years 2000, 2003, 2006 and 2009, using data from the Kids’ Inpatient Database (KID), Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality. HCUP is the largest collection of multi-year, all-payer, encounter-level, health care data available in the US. The KID contains discharge data from pediatric inpatient admissions, including patient and hospital-specific variables as well as healthcare utilization variables such as discharge diagnoses (up to 25 per admission), procedures performed (up to 25 per admission) and LOS. The KID, available every 3 years since 1997, is a stratified sample of all pediatric discharges for children and adolescents ages less than 21 years from states that participate in HCUP. Participating states provide discharge-level data on all inpatient discharges from community non-rehabilitation hospitals (nonfederal, short-term, general, and specialty hospitals) in that state. Patient-level data is not reported and therefore individual patients cannot be tracked. The KID 2000 includes 2784 hospitals in 27 states; KID 2003 includes 3438 hospitals in 36 states; KID 2006 includes 3739 hospitals in 38 states; KID 2009 includes 4121 hospitals in 44 states. The KID was specifically designed to report pediatric hospital use and outcomes for rare pediatric conditions, and samples 10% of uncomplicated births and 80% of all other pediatric discharges from all participating hospitals to allow adequate representation of these conditions. National estimates with 95% confidence intervals (95% CI) can be calculated using sampling weights, which are based on hospital characteristics. More detailed information on the KID database can be found at http://www.hcup-us.ahrq.gov/kidoverview.jsp.

The Institutional Review Board at the Children's Hospital of Philadelphia deemed this study not human subjects research due to the de-identified nature of the data.

Identification of Sample

We identified hospitalizations for patients with SLE ages 2 up to 21 years in the KID using an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code of 710.0 listed as a discharge diagnosis. Given the systemic nature of SLE, we included those with a diagnosis of SLE in any position to optimally capture the broad range of reasons for hospitalization. We also identified a subgroup of patients with SLE nephritis using a combination of SLE ICD-9-CM code 710.0 and one of several nephritis ICD-9-CM codes as previously described in the literature [12-16] (Supplementary Table 1). We further characterized the SLE cohort by identifying the most common principal discharge diagnoses, using Clinical Classifications Software (CCS) codes to first identify broad categories [17], then individual ICD-9-CM codes to tabulate the principal discharge diagnoses. We sought to exclude neonatal SLE patients by restricting our analyses to hospitalizations for patients aged 2 years and above. We included patients aged 18 up to 21 years because this age group frequently receives care in the pediatric setting.

Outcome Variables

We calculated national estimates with 95% CI and evaluated for temporal trend for the outcomes of hospitalization rate, inpatient mortality rate, procedure rate and average LOS for years 2000, 2003, 2006 and 2009. We examined patient and hospital-specific risk factors for the outcomes of mortality rate and LOS, and conducted secondary analyses in the SLE nephritis subgroup. We also compared these outcomes to those without a SLE diagnosis. The annual hospitalization rate was calculated by dividing the estimated number of pediatric SLE hospitalizations by the age-matched US population estimate for that year (obtained from US Census data), and multiplying by 100,000. Annual inpatient mortality rate was calculated by dividing the number of pediatric SLE inpatient deaths by the total number of pediatric SLE hospitalizations for that year. We tabulated the most common procedures (including primary and secondary) by using Clinical Classifications Software (CCS) codes to first identify broad categories [17], then individual ICD-9-CM codes to form procedure groupings (Supplementary Table 2). The procedure rate was defined as the number of procedures performed per 1000 hospitalizations. Data for outcome variables was missing as follows: mortality <0.1%, LOS <0.1%.

Demographic Variables

We included the following patient-specific characteristics in the analysis: presence of SLE nephritis (yes/no), age, race/ethnicity, sex and health insurance coverage. We identified patients with SLE nephritis based on ICD-9-CM codes as described above. We defined 3 age groups based on the American Academy of Pediatrics developmental stages[18]: children 2-11 years, adolescents 12-17 years, and older adolescents 18-20 years. Self-reported race/ethnicity was categorized into the following mutually exclusive groups: White, Black, Hispanic and other (includes Asian/Pacific Islander and Native American). Health insurance coverage was categorized into 3 groups based on primary payer: Medicaid, private and other (includes Medicare, self-pay and no charge). We included the following hospital-specific characteristics in the analysis: hospital type (general or children's), teaching status, hospital location (rural or urban) and hospital region (Northeast, Midwest, South or West). The classification schema of the National Association of Children's Hospitals and Related Institutions (NACHRI) was used to categorize hospital type into 2 groups: children's hospital (includes children's general hospital, children's specialty hospital and children's unit in a general hospital) and general hospital (does not have a children's unit). Hospitals were considered to be teaching hospitals if they met any one of the following criteria: residency training approval by the Accreditation Council for Graduate Medical Education (ACGME); membership in the Council of Teaching Hospitals (COTH); a ratio of full-time equivalent interns and residents to beds of 0.25 or higher. Demographic data were missing for less than 10% of hospitalizations, with the exception of race/ethnicity data (16%), which was retained in the analysis due to the known differential prevalence and severity of SLE according to race/ethnicity [1, 19, 20].

Statistical Analysis

We performed all analyses and statistical comparisons using Stata 12 (Stata Corp., College Station, TX). National estimates with 95% CI were calculated using the discharge-level statistical weights provided by KID. We used a marginal model utilizing the Stata survey command to analyze the data at the population level, accounting for multi-level clustering (eg. hospital, region) and to incorporate the sampling weights. Patient and hospital-specific demographic variables were tabulated for pediatric SLE discharges for 2000, 2003, 2006 and 2009. We tabulated the frequencies of the principal discharge diagnoses. Hospitalization rate was evaluated for temporal trend using univariable Poisson regression. Mortality was evaluated for temporal trend, adjusting for demographic variables, using multivariable logistic regression. To evaluate for temporal trend in procedure rate and average LOS, we used a multivariable generalized linear regression model that adjusted for demographic variables, utilizing the gamma family of distributions with a log link. This model provides robust estimates for highly skewed outcomes such as inpatient LOS [21]. We tested for differences in temporal trends of mortality and LOS for those with and without a SLE diagnosis using an interaction term between diagnosis group (SLE vs non-SLE) and year. We performed secondary analyses of the outcomes for patients with SLE nephritis. Throughout the analysis, all testing was 2-sided, with a threshold for statistical significance of p<0.05.

Results

Demographics & Principal Discharge Diagnoses

Of the approximately 29 million total pediatric hospitalizations in the KID for the years of study, an estimated 26,903 (95%CI 23,802- 30,005) SLE hospitalizations were identified. Of these hospitalizations, 85% were female, 45% were ages 18 – 20 years, and 81% were of non-White race/ethnicity. Medicaid was identified as the primary payer in 46%. Over half of the hospitalizations were for patients with SLE nephritis (57%), and this proportion was stable over time (range 55-57%, p=0.32). Additional demographic characteristics are listed in Table 1. The most commonly listed principal discharge diagnoses were SLE (45% with ICD-9-CM 710.0), infection (11%) and nephritis (4%).

Table 1.

Demographics for Hospitalizations with a Diagnosis of SLE

N=26903 for All Years	N (%)
Patient-Specific Factors
Age (y)
2-12	2337 (8.7)
12-17	12379 (46.0)
18-20	12187 (45.3)
Femalea	22814 (84.8)
Race/Ethnicitya
White	5140 (19.1)
Black	8853 (32.9)
Hispanic	5902 (21.9)
Other	2733 (10.2)
Primary Payera
Medicaid	12268 (45.6)
Private	10411 (38.7)
Other	4171 (15.5)
SLE Nephritis
No	11705 (43.5)
Yes	15198 (56.5)
Hospital-Specific Factors
Locationa
Urban	24858 (92.4)
Rural	1025 (3.8)
Region
Northeast	5165 (19.2)
Midwest	4246 (15.8)
South	10478 (38.9)
West	7014 (26.1)
Teaching Statusa
Teaching	20608 (76.6)
Non-Teaching	5281 (19.6)
Typea
Children's Hospital	14958 (55.6)
General Hospital	9990 (37.1)

^aThe percentages for these categories do not add up to 100% due to missing demographic data as follows: sex 0.3%, race/ethnicity 15.9%, primary payer 0.2%, location=3.8%, teaching status=3.8%, hospital type=7.3%.

Hospitalization and Inpatient Mortality Rates

The annual hospitalization rate for pediatric SLE was stable over the years of study at an average of 8.6 per 100,000 population (95%CI 7.6-9.6) (Table 2). The average inpatient mortality rate across all years was 1% (95%CI 0.8-1.1), and there was a statistically significant decrease to a low of 0.6% in 2009 (odds ratio (OR)=0.95, 95% CI 0.900-0.998, p=0.04) (Table 2). Factors associated with death were: SLE nephritis, age 18-20 years, Black race and hospital location in South region (Table 3). For the subgroup of patients with SLE nephritis, the average inpatient mortality rate was 1.4% (95%CI 1.1-1.6), and there was a statistically significant decrease from 1.5% in 2000 to 0.7% in 2009 (OR=0.93, 95%CI 0.88-0.99, p=0.01) (Figure 1). Mortality for all other KID hospitalizations also showed a statistically significant, but very minor decrease over time, from 0.34% in 2000 to 0.27% in 2009 (OR=0.97, 95%CI 0.96-0.97, p<0.001) (Figure 1).

Table 2.

Temporal Trends in SLE Hospitalization Rate, Mortality and LOS

Year	Discharges, N (95% CI)	Hospitalization Ratea (95%CI)	Mortality, % (95% CI)	Mean LOS, days (95% CI)
All Years	26903 (23802- 30005)	8.6 (7.6- 9.6)	1.0 (0.8-1.1)	5.9 (5.6- 6.2)
2000	5962 (5096-6827)	7.7 (6.6- 8.8)	1.0 (0.7- 1.4)	5.6 (5.2- 6.1)
2003	6701 (5844-7559)	8.6 (7.6- 9.7)	1.2 (0.8- 1.5)	6.1 (5.7- 6.5)
2006	7344 (6416- 8273)	9.3 (8.2- 10.5)	1.2 (0.9- 1.5)	6.3 (5.8- 6.8)
2009	6896 (6021- 7770)	8.7 (7.6- 9.7)	0.6 (0.4- 0.8)	5.7 (5.3- 6.0)
p-value for trendb	-	0.14	0.04	0.45

^aSLE hospitalizations per 100,000 children in population aged 2 up to 21yrs of age, based on population estimates from US Census data.

^bOutcomes evaluated for temporal trend using multivariable regression to adjust for patient and hospital-specific demographic variables (with the exception of hospitalization rate evaluated by univariable regression).

Table 3.

Risk Factors for Inpatient Mortality and LOS for SLE Hospitalizations

Risk Factor		Mortality OR, 95% CI	LOS RR, 95% CI
Nephritis	No	-	-
	Yes	2.71 (1.74-4.23)***	1.30 (1.21-1.40)***
Age (y)	2-12	-	-
	12-17	1.42 (0.58-3.48)	0.92 (0.81-1.04)
	18-20	2.85 (1.21-6.71)*	1.06 (0.93-1.20)
Sex	Female	-	-
	Male	1.51 (0.99-2.29)	1.03 (0.94-1.14)
Race	White	-	-
	Black	2.01 (1.12-3.62)*	1.09 (1.00-1.20)
	Hispanic	1.39 (0.73-2.64)	1.07 (0.97-1.18)
	Other	1.88 (0.94-3.75)	1.14 (1.001-1.30)*
Payer	Medicaid	-	-
	Private	0.89 (0.59-1.34)	0.88 (0.82-0.94)***
	Other	1.01 (0.65-1.58)	0.90 (0.82-0.999)*
Location	Rural	-	-
	Urban	0.98 (0.28-3.42)	1.33 (1.16-1.53)***
Region	Northeast	-	-
	Midwest	0.81 (0.34-1.95)	1.08 (0.92-1.28)
	South	2.03 (1.23-3.36)**	1.06 (0.92-1.21)
	West	1.77 (0.98-3.20)	1.11 (0.93-1.32)
Teaching	No	-	-
	Yes	1.06 (0.60-1.87)	1.10 (1.01-1.20)*
Type	General	-	-
	Children's	0.95 (0.59-1.52)	1.05 (0.95-1.17)

Odds ratios (OR) and relative risks (RR) for risk factors included in the multivariable regressions for the main outcomes are shown. The covariates included in the multivariable regressions are as follows: calendar year, presence of SLE nephritis, age group, sex, race/ethnicity, primary payer, hospital location, hospital region, hospital teaching status and hospital type. Reference groups are denoted by “-”. Significance for p-values is noted as follows:

^*p<0.05

^**p<0.01

^***p<0.001.

Figure 1.

Inpatient mortality for children and adolescents with a diagnosis of SLE showed a statistically significant decrease over the years of study (OR=0.95, 95% CI 0.900-0.998, p=0.04). Mortality for those with SLE nephritis also showed a statistically significant decrease (OR=0.93, 95%CI 0.88-0.99, p=0.01). Mortality for all other KID hospitalizations showed a statistically significant but very minor decrease (OR=0.97, 95%CI 0.96-0.97, p<0.001). The difference in rate of mortality decrease for hospitalizations with SLE diagnosis versus those without a SLE diagnosis was not statistically significant. Error bars indicate standard error; the error bars for the group of hospitalizations without a SLE diagnosis are so small that they are barely visible on the graph.

Procedure Rates

Approximately 60% of all SLE hospitalizations (N=16,056) had at least 1 billed procedure, and the median number of procedures per hospitalization was 2 (interquartile range 1-3). The most commonly billed procedures were medication infusion (12.8%), dialysis (8.4%), blood transfusions (7.5%), vascular catheterization (6.4%) and renal biopsy (6.3%) (Figure 2A). There was a statistically significant increase in the rates of dialysis, blood transfusion and vascular catheterization procedures performed over the years of study (Figure 2B).

Figure 2A.

Billed procedures comprising the most common 75% are shown (the remaining 25% are not shown).

Figure 2B.

Temporal trends for the 5 most commonly billed procedures (comprising over 40% of all procedures) are shown. The associated p-values for trend are as follows: medication infusion (p=0.20), dialysis (p=0.01), blood transfusion (p<0.001), vascular catheterization (p=0.002), renal biopsy (p=0.74).

Length of Stay

The mean LOS of 5.9 days (95%CI 5.6- 6.2) for SLE hospitalizations was stable over time (Table 2). Factors associated with increased LOS were: SLE nephritis, other race, urban location, and teaching hospital status (Table 3). Non-Medicaid payer was associated with decreased LOS. For patients with SLE nephritis, LOS was also stable over time at a mean of 6.7 days (95%CI 6.2-7.1). LOS estimates for SLE hospitalizations were higher than the average LOS of 3.6 days (95%CI 3.6-3.7) for those without SLE, which was also stable over time.

Discussion

This retrospective serial cross-sectional analysis provides national estimates of US inpatient healthcare utilization trends for pediatric SLE over a decade. Using administrative data from the nationally representative KID, we identified approximately 27,000 US hospitalizations for pediatric SLE over a 4-year period. The characteristics of this national inpatient pediatric SLE cohort are consistent with previously reported age, race and sex demographics of pediatric SLE populations [1, 12, 19, 22]. It is notable that almost half of the cohort was 18-20 years of age, which likely reflects the onset of pediatric SLE during adolescence. Over half (57%) of the SLE hospitalizations were for those with SLE nephritis, which is not unexpected given the significant morbidity and mortality association with SLE renal disease, and we found that this proportion was stable over time. Our study extends the work of Tanzer et al, who found similar demographics and a high proportion of inpatients with SLE nephritis, using KID data from 2000 to 2006 [12].

We present previously unreported national US data on inpatient hospitalization and mortality rates for pediatric SLE. We found an average annual rate of 8.6 hospitalizations for pediatric SLE per 100,000 population. Given the estimated prevalence of childhood-onset SLE at 3 to 9 per 100,000 population [1], this estimated hospitalization rate indicates at least one hospitalization for every 2 patients with pediatric SLE every year. Although not reflective of the rate per individual, it represents a substantial utilization burden from the healthcare system perspective. Our estimate for average all-cause mortality rate in SLE patients was 1%, and 1.4% for those with SLE nephritis, both higher than the rate of 0.3% for those without SLE. Encouragingly, we found that inpatient SLE deaths decreased significantly over time from 1% to 0.6%, paralleled by a decrease in those with SLE nephritis and a minor decrease for those without SLE. This downward trend may represent fluctuation over time, but we think it likely represents a real decrease in mortality given the similar trend in both SLE groups. We speculate that this observed decrease in inpatient SLE mortality may be attributable to several reasons. Pediatric care for acute SLE flares may be improving due to: better access to outpatient and inpatient rheumatology and other subspecialty care; earlier diagnosis and treatment leading to less severe flares and improved long-term disease control; increased availability of newer and less toxic immunosuppressive treatments, such as mycophenolate mofetil and Rituximab therapy with less cyclophosphamide exposure [8-11]; and overall improved pediatric healthcare technology and intensive care.

While the above factors may be influencing the decreasing inpatient mortality rate for pediatric SLE, LOS appears to be unaffected and has not changed significantly over time. The longer LOS for pediatric SLE hospitalizations at 5.9 days compared to 3.6 days for those without SLE is indicative of high healthcare utilization for children with SLE. Furthermore, the longer LOS is more pronounced for those with SLE nephritis at 6.7 days. This confirms the work of Tanzer et al, who also reported high healthcare utilization for pediatric SLE, especially in those with kidney disease, in comparison to other pediatric hospitalizations [12]. In contrast, however, we did not find a statistically significant increase the LOS over time. The high resource utilization for inpatient pediatric SLE patients is likely due to the need for more invasive procedures, multi-specialty and intensive care management for SLE, and particularly SLE nephritis, in comparison to other pediatric hospitalizations. This is supported by our findings that medication infusions, dialysis, blood transfusions, vascular catheterizations and renal biopsies were the most common procedures for SLE inpatients. We also found a statistically significant increase in the rates of dialysis, blood transfusions and vascular catheterization procedures over the study period. Thus there appears to be an increased volume of procedures occurring over time, perhaps due to increased procedure availability and improved patient survival resulting in higher procedure rates for continued chronic care. Although the direct impact of therapeutic advances on outcomes is currently unclear, enhanced resource utilization may be leading to decreased inpatient mortality and improved overall survival for pediatric SLE patients. If this trend eventually translates into decreases in pediatric SLE hospitalization rate and LOS, then the resulting overall effect may be a decrease in healthcare utilization. We did not observe this effect over the decade of our study, however, and it is possible that this effect may not be seen until later in the disease course.

We found that several patient-specific risk factors were associated with worse outcomes. Patients with SLE nephritis had increased inpatient mortality and LOS, which is consistent with expected higher morbidity and mortality in this group. Older age was associated with increased inpatient mortality likely due to increasing complications of SLE and its treatment with age. Similarly the increased mortality risk and LOS for patients of non-White race is consistent with prior findings of more severe SLE in non-White populations [19, 20]. Medicaid patients had an increased LOS, suggesting a role for lower socioeconomic status in the risk for worse outcomes, but we did not find an association with SLE mortality as previously described [23]. For hospital-specific factors, location in the South region was associated with increased mortality, for reasons that are unclear. Although there was a higher proportion of non-White and Medicaid patients in this region (data not shown) with potentially higher disease severity, location in the South region was an independent risk factor, and this may be due to other unmeasured factors related to SLE disease severity.

Our study presents the first estimates to our knowledge of inpatient hospitalization and mortality rates for pediatric SLE in the United States, as well as relevant healthcare utilization trends. The strength of our study lies in the large number of observations representative of the national pediatric SLE population. Despite this, a potential limitation is the lack of validation for the single ICD-9-CM code of 710.0 for identification of SLE using administrative data, which could lead to inaccurate outcome estimates; however, the code has been validated in coding algorithms for identification of SLE nephritis patients using administrative data [13]. We importantly identified those with nephritis as a significant proportion of the SLE inpatients with higher healthcare utilization and mortality rates, and it is possible that the prevalence of nephritis may actually be higher if patients were admitted for another reason, and the recorded ICD-9-CM codes did not capture a history of nephritis. Unfortunately, we were unable to characterize other clinical manifestations and disease severity; the KID does not contain clinical details, and severity measures, such as the All Patient Refined Diagnosis Related Groups (APR-DRG), were not available for all years of interest, and were not reliable as coding algorithms changed over time.

Our analysis of healthcare utilization trends over a decade provides valuable insight into the burden of SLE care from both the patient and healthcare system perspectives. Overall resource utilization for SLE patients, and especially those with nephritis, has remained high; however, we were unable to determine individual patient burden as patients could not be tracked over time due to the lack of unique patient identifiers in the KID, and some patients may be represented more than once in the dataset. We thus think that our estimates are especially meaningful from an overall healthcare burden perspective, and we indeed found a high average LOS for SLE patients who are likely to have repeated hospitalizations during the course of their chronic disease, compared to those without SLE. Our finding of increasing rates of common therapeutic procedures also reflects the high overall healthcare burden for pediatric SLE. Coding practices may have affected the completeness of ICD-9-CM procedure codes, as the recorded codes may be more representative of invasive procedures (such as dialysis) than non-invasive procedures (such as MRI), but our data still sheds light on the frequency of invasive and often more costly procedures performed. Further evaluation of national healthcare costs for pediatric SLE would be a helpful additional measure of healthcare utilization; however, we were unable to accurately evaluate costs in the KID due to unavailability, missingness and imprecision of cost to charge ratios for the years of study. Given the potential impact of mycophenolate mofetil and other new treatment regimens, detailed analysis of the relationship between trends in specific treatments and the outcomes of interest would also be informative, but we were unable to address this due to the lack of itemized billing information in the KID.

In conclusion, we present novel data that raises interesting questions relevant to the patient and healthcare system burden of inpatient care, as well as survival for pediatric SLE patients. While our findings indicate that the US national hospitalization rate and LOS have remained stable for pediatric SLE, they represent a significant disease burden compared to those without SLE. The rates of common therapeutic procedures are increasing, while inpatient mortality is decreasing. Therefore overall outcomes may be improving for pediatric SLE patients, but certain groups remain at high risk for increased healthcare utilization and death. More research is needed to understand the role of advances in treatment and healthcare technology in the changing course of pediatric SLE healthcare utilization and survival.