To investigate the nationwide prevalence, incidence, and sociodemographics of systemic lupus erythematosus (SLE) and lupus nephritis among children in the US Medicaid beneficiary population.
To investigate the nationwide prevalence, incidence, and sociodemographics of systemic lupus erythematosus (SLE) and lupus nephritis among children in the US Medicaid beneficiary population.
Children ages 3 years to <18 years with a diagnosis of SLE (defined as ≥3 claims with an International Classification of Diseases, Ninth Revision [ICD-9] code of 710.0 for SLE, each >30 days apart) were identified from the US Medicaid Analytic eXtract database from 2000 to 2004. This database contains all inpatient and outpatient Medicaid claims for 47 US states and the District of Columbia. Lupus nephritis was identified from ≥2 ICD-9 billing codes for glomerulonephritis, proteinuria, or renal failure, each recorded >30 days apart. The prevalence and incidence of SLE and lupus nephritis were calculated among Medicaid-enrolled children overall and within sociodemographic groups.
Of the 30,420,597 Medicaid-enrolled children during these years, 2,959 were identified as having SLE. The prevalence of SLE was 9.73 (95% confidence interval [95% CI] 9.38–10.08) per 100,000 Medicaid-enrolled children. Among the children with SLE, 84% were female, 40% were African American, 25% were Hispanic, 21% were White, and 42% resided in the South region of the US. Moreover, of the children with SLE, 1,106 (37%) had lupus nephritis, representing a prevalence of 3.64 (95% CI 3.43–3.86) per 100,000 children. The average annual incidence of SLE was 2.22 cases (95% CI 2.05–2.40) and that of lupus nephritis was 0.72 cases (95% CI 0.63–0.83) per 100,000 Medicaid enrollees per year. The prevalence and incidence rates of SLE and lupus nephritis increased with age, were higher in girls than in boys, and were higher in all non-White racial/ethnic groups.
In the current study, the prevalence and incidence rates of SLE among Medicaid-enrolled children in the US are high compared to studies in other populations. In addition, these data represent the first population-based estimates of the prevalence and incidence of lupus nephritis in the US to date.
Systemic lupus erythematosus (SLE) is a condition that can be seen in individuals throughout the age spectrum, and one of its most severe manifestations, lupus nephritis, may have devastating consequences at any age (1–6). SLE presenting in childhood accounts for 15–20% of all cases, and nephritis affects 20–75% of pediatric patients with SLE (7–9).
Children are among the poorest and most vulnerable populations in the US healthcare system, facing some of the highest barriers to obtaining appropriate healthcare (10, 11). Comparisons of relatively small pediatric and adult cohorts of SLE patients have shown that children and adolescents have more active lupus, in particular lupus nephritis, at presentation and over time when compared to adults (2, 12). In comparison with adult patients with lupus, children receive more intensive drug therapy and accrue more end-organ damage, often related to steroid toxicity (2). In the LUpus in MInorities, NAture versus nurture (LUMINA) multiethnic cohort, young age was an important independent predictor of new or worsening proteinuria on routine screening, and onset of SLE during adolescence may result in more aggressive disease and worse outcomes (12, 13). The observed high morbidity and mortality attributed to lupus nephritis in past studies of SLE in children may be due to delays in diagnosis and treatment (1, 3, 4).
To date, only a small number of studies have examined the prevalence or incidence of SLE among children. Reported SLE prevalence rates among children have ranged from 3.3 to 24.0 per 100,000 children (6, 14–17). The incidence of SLE among children referred to pediatric rheumatology specialists/centers has been studied very infrequently, and has been reported in 2 past studies to be 0.28 and 0.48 per 100,000 children per year (14, 17). The prevalence and incidence rates of lupus nephritis among children have not been determined for large pediatric populations nationwide.
Medicaid is a federal-state entitlement program for low-income adults, their children, the aged, and people with certain disabilities. Little is known about the prevalence, incidence, or sociodemographic characteristics of SLE or lupus nephritis among children enrolled in US Medicaid. We aimed to determine recent prevalence and incidence rates for SLE and lupus nephritis among US children receiving Medicaid health insurance. Determining the recent nationwide prevalence is an important first step to understanding the public health impact, and how poverty and access to healthcare may influence outcomes in children with SLE. We also aimed to compare incidence rates between girls and boys, stratified according to age and racial/ethnic groups. Having a better understanding of the demographics of pediatric SLE could provide insight into the underlying biologic processes.
The Medicaid Analytic eXtract (MAX) data system contains demographic information and all billing claims for Medicaid enrollees from 47 states and the District of Columbia. (Arizona, Tennessee, and Maine do not provide statewide data, and were thus excluded from these analyses.) Using the MAX data from January 1, 2000 to December 31, 2004, we identified all individuals between the ages of 3 years and <18 years with a diagnosis of SLE, defined as having ≥3 hospital discharge or physician visit claims with a recorded International Classification of Diseases, Ninth revision (ICD-9) code for SLE of 710.0, each recorded at least 30 days apart. Among the children with SLE, we then identified those with lupus nephritis, defined as having ≥2 physician and/or inpatient claims for nephritis, proteinuria, and/or renal failure (ICD-9 codes 580–588, 630–640, and 791.0) occurring on or after the SLE billing diagnoses, and each recorded at least 30 days apart. We have previously demonstrated that this algorithm has a positive predictive value of 80% for the identification of lupus nephritis among adults in a Medicaid population (18).
From the MAX administrative data, in addition to physician and hospital claims, we obtained the following demographic data on all Medicaid enrollees ages 3 years to <18 years: date of birth, sex, race (White, black, Hispanic or Latino, Asian, Native Hawaiian or other Pacific Islander, American Indian or Alaskan Native, other, or more than one race), and zip code of residence. Utilizing the age at the time of the first SLE or lupus nephritis claim, we categorized age into 4 categories (ages 3 to <6 years, 6 to <9 years, 9 to <12 years, 12 to <15 years, and 15 to <18 years), recategorized the racial/ethnic groups into 5 categories (White, African American, Asian, Hispanic, Native American, and other), and categorized US region of residence (based on zip code) as Northeast, South, Midwest, and West.
The prevalence, with 95% confidence intervals (95% CIs), of SLE and lupus nephritis per 100,000 Medicaid-eligible children was calculated from all identified cases and all Medicaid enrollees ages 3 years to <18 years in the MAX database between 2000 and 2004. Prevalence was calculated for the total cohort and stratified by sex, race, US region, and age category, and by cross-classified sex and age category and sex and racial/ethnic group category.
The annual incidence rates, with 95% CIs, for SLE and lupus nephritis were calculated by identifying cases of newly diagnosed SLE and lupus nephritis, meeting the definitions above, in children who had at least 24 months of Medicaid enrollment without any SLE (or lupus nephritis) claims. The time of SLE onset was defined as the date of the first SLE or lupus nephritis claim. Subjects were included in the person-months denominator once they had 24 months of Medicaid enrollment, and were censored at the time of their first SLE or lupus nephritis claim. Average annual incidence rates (per 12 months of enrollment) for these 3 years were calculated for the total cohort and in stratified groups based on sex, race/ethnicity, US region, and age category, and by sex-and-age category. Prevalence rate ratios and incidence rate ratios, with 95% CIs, were calculated using Poisson techniques to demonstrate differences in incidence rates among different sociodemographic groups.
As Medicaid medical coverage was discontinuous for many children during the period of analyses, we conducted a sensitivity analysis for incidence rates by restricting the analysis to those with at least 24 months of continuous Medicaid enrollment without any SLE or lupus nephritis claims. All analyses were conducted using SAS, version 9.2. Data were obtained from the US Center for Medicare and Medicaid Services through an approved data use agreement, and our results are presented in accordance with their policies (data on 10 or fewer individuals have been suppressed, along with other potentially identifying data). Our institution, Partners Healthcare, waived institutional review board approval for use of human subjects in this study.
Of 30,420,597 children covered by Medicaid from 2000 to 2004, 2,959 were identified as having SLE. The overall prevalence of SLE was 9.73 (95% CI 9.38–10.08) per 100,000 children. Of the 2,959 children with SLE, 83.8% were female, 39.6% were African American, 25.4% were Hispanic, and 20.6% were White. Moreover, 41.9% of the children with SLE resided in the South and 23.0% in the West. The calculated prevalence of SLE overall and for specific demographic groups of children are shown in Table 1.
|No. of subjects†||SLE||Lupus nephritis|
|No. of cases (% of total)||Prevalence||95% CI||No. of cases (% of total)||Prevalence||95% CI|
|Total||30,420,597||2,959 (100)||9.73||9.38–10.08||1,106 (100)||3.64||3.43–3.86|
|Female||15,087,004||2,480 (83.8)||16.44||15.80–17.10||901 (81.5)||5.97||5.59–6.37|
|Male||15,333,593||479 (16.2)||3.12||2.86–3.42||205 (18.5)||1.34||1.17–1.53|
|3 to <6 years||11,365,324||55 (1.9)||0.48||0.37–0.63||19 (1.7)||0.17||0.11–0.26|
|6 to <9 years||9,969,015||140 (4.7)||1.40||1.19–1.66||47 (4.3)||0.47||0.35–0.63|
|9 to <12 years||9,303,079||453 (15.3)||4.87||4.44–5.34||159 (14.4)||1.71||1.46–2.00|
|Female||4,537,477||370 (12.5)||8.15||7.36–9.03||134 (12.1)||2.95||2.49–3.50|
|Male||4,765,602||83 (2.8)||1.74||1.40–2.16||25 (2.3)||0.52||0.35–0.78|
|12 to <15 years||8,268,263||923 (31.2)||11.16||10.47–11.91||344 (31.1)||4.16||3.74–4.62|
|Female||4,049,181||769 (2.6)||18.99||17.70–20.38||283 (25.6)||6.99||6.22–7.85|
|Male||4,219,082||154 (5.2)||3.65||3.12–4.27||61 (5.5)||1.45||1.12–1.86|
|15 to <18 years||7,335,965||1,388 (46.9)||18.92||17.95–19.94||537 (48.6)||7.32||6.73–7.97|
|Female||3,830,632||1,188 (4.0)||31.01||29.30–32.83||431 (39.0)||11.25||10.24–12.37|
|Male||3,505,333||200 (6.8)||5.71||4.97–6.55||106 (9.6)||3.02||2.50–3.66|
|African American||8,329,751||1,173 (39.6)||14.08||13.30–14.91||482 (43.6)||5.79||5.29–6.33|
|Female||4,165,315||993 (3.4)||23.84||22.40–25.37||406 (36.7)||9.75||8.84–10.74|
|Male||4,164,436||180 (6.1)||4.32||3.73–5.00||76 (6.9)||1.82||1.46–2.29|
|Hispanic||6,527,326||751 (25.4)||11.51||10.71–12.36||281 (25.4)||4.30||3.83–4.84|
|Female||3,257,913||626 (21.2)||19.21||17.77–20.78||221 (20.0)||6.78||5.95–7.74|
|Male||3,269,413||125 (42.2)||3.82||3.21–4.56||60 (5.4)||1.84||1.42–2.36|
|White||12,505,440||608 (20.6)||4.86||4.49–5.26||149 (13.5)||1.19||1.01–1.40|
|Female||6,181,429||511 (17.3)||8.27||7.58–9.02||120 (10.8)||1.94||1.62–2.32|
|Male||6,324,011||97 (3.3)||1.53||1.26–1.87||29 (2.7)||0.46||0.32–0.66|
|Asian||874,234||208 (7.0)||23.79||20.77–27.26||98 (8.9)||11.21||9.20–13.66|
|Female||426,835||175 (5.9)||41.00||35.35–47.55||82 (7.4)||19.21||15.47–23.85|
|Male||447,399||33 (1.1)||7.38||5.24–10.38||16 (1.5)||3.58||2.19–5.84|
|Native American||455,991||61 (2.1)||13.38||10.41–17.19||28 (2.5)||6.14||4.24–8.89|
|Other||1,727,855||158 (5.3)||9.14||7.82–10.69||68 (6.2)||3.94||3.10–4.99|
|Female||829,097||132 (4.5)||15.92||13.42–18.88||54 (4.9)||6.51||4.99–8.50|
|Male||898,758||26 (0.8)||2.89||1.97–4.25||14 (1.3)||1.56||0.92–2.63|
|South||12,386,156||1,239 (41.9)||10.00||9.46–10.58||468 (42.3)||3.78||3.45–4.14|
|West||6,706,321||681 (23.0)||10.15||9.42–10.95||258 (23.3)||3.85||3.41–4.35|
|Northeast||5,347,669||559 (18.9)||10.45||9.62–11.36||215 (19.4)||4.02||3.52–4.60|
|Midwest||5,980,451||480 (16.2)||8.03||7.34–8.78||165 (14.9)||2.76||2.37–3.21|
The prevalence of SLE was >5 times higher among girls than among boys. Prevalence was highest among girls ages 15 to <18 years (31.01 per 100,000 Medicaid-enrolled children, 95% CI 29.30–32.83). The prevalence of SLE among Asian children (23.79 per 100,000 Medicaid-enrolled children) exceeded that in the other racial and ethnic groups, and the prevalence was higher among African Americans (14.08 per 100,000 Medicaid-enrolled children), Hispanics (11.51 per 100,000 Medicaid-enrolled children), and Native Americans (13.38 per 100,000 Medicaid-enrolled children) than among Whites (4.86 per 100,000 Medicaid-enrolled children). The SLE prevalence rates were lowest in the Midwest (8.03 per 100,000 Medicaid-enrolled children) compared to the other regions of the country.
The prevalence of lupus nephritis showed similar sociodemographic variation (Table 1). Among the 2,959 children with SLE, 1,106 (37%) were identified as having lupus nephritis. Overall, the prevalence of lupus nephritis was 3.64 cases per 100,000 children. The prevalence was 4.46 times higher among girls than among boys, was highest among girls ages 15 to <18 years, and was highest among Asians, followed by similarly high rates among African Americans, Hispanics, and Native Americans. The prevalence of lupus nephritis was slightly lower in the Midwest than in the other regions of the country. Figure 1 displays the prevalence rate ratios for SLE and lupus nephritis among girls and boys in each non-White racial/ethnic group relative to Whites. The prevalence rate ratios were highest for lupus nephritis among Asian girls and Asian and Native American boys compared to White children of the same sex.
The average annual incidence rates of SLE and lupus nephritis per 100,000 Medicaid-enrolled children per year in the US from 2002 to 2004 are displayed in Table 2. Overall, the average annual incidence rate of SLE was 2.22, and that for lupus nephritis was 0.72, per 100,000 children per year. Incidence rates increased dramatically with age, and were again highest in all non-White racial/ethnic groups compared to Whites. The Midwest had significantly lower incidence rates of both SLE cases and lupus nephritis cases per 100,000 children in the population enrolled in Medicaid per year.
|Person-years||No. of cases (% of total)||Incidence rate||95% CI||Person-years||No. of cases (% of total)||Incidence rate||95% CI|
|Total||28,007,765||621 (100)||2.22||2.05–2.40||28,009,598||203 (100)||0.72||0.63–0.83|
|Female||13,667,244||527 (84.9)||3.86||3.54–4.20||13,668,718||171 (84.2)||1.25||1.08–1.45|
|Male||14,340,521||94 (15.1)||0.66||0.54–0.80||14,340,880||32 (15.8)||0.22||0.16–0.32|
|3 to <6 years||NR†||NR†||0.13||0.07–0.26||NR†||NR†||0.06||0.02–0.16|
|6 to <9 years||5,961,603||27 (4.3)||0.45||0.31–0.66||NR†||NR†||0.17||0.09–0.31|
|9 to <12 years||5,838,248||112 (18.0)||1.92||1.59–2.31||5,838,495||30 (14.8)||0.51||0.36–0.73|
|12 to <15 years||5,299,359||203 (32.7)||3.83||3.34–4.40||5,299,917||63 (31.0)||1.19||0.93–1.52|
|15 to <18 years||4,219,162||270 (4.3)||6.40||5.68–7.21||4,220,055||96 (47.3)||2.27||1.86–2.78|
|Female||2,097,260||231 (3.7)||11.01||9.68–12.53||2,098,016||81 (39.9)||3.86||3.11–4.80|
|Male||2,121,902||39 (6.3)||1.84||1.34–2.52||2,122,039||15 (7.9)||0.71||0.43–1.17|
|African American||9,311,413||254 (40.9)||2.73||2.41–3.08||9,312,092||81 (39.9)||0.87||0.70–1.08|
|Female||4,616,319||213 (34.3)||4.61||4.03–5.28||4,616,853||69 (34.0)||1.49||1.18–1.89|
|Male||4,695,094||41 (6.6)||0.87||0.64–1.19||4,695,239||12 (5.9)||0.26||0.15–0.45|
|Hispanic||6,003,144||147 (23.7)||2.45||2.08–2.88||6,003,611||51 (25.1)||0.85||0.65–1.12|
|White||10,320,599||137 (22.1)||1.33||1.12–1.57||10,321,012||31 (15.3)||0.30||0.21–0.43|
|Asian||914,530||38 (6.1)||4.16||3.02–5.71||914,638||19 (9.4)||2.08||1.33–3.26|
|Other||1,084,581||39 (6.3)||3.60||2.63–4.92||1,084,726||15 (7.4)||1.38||0.83–2.29|
|South||9,218,017||208 (33.5)||2.26||1.97–2.58||9,218,682||72 (35.5)||0.78||0.62–0.98|
|Northeast||5,684,974||114 (18.4)||2.01||1.67–2.41||5,685,320||41 (20.2)||0.72||0.53–0.98|
|Midwest||6,501,099||120 (19.3)||1.85||1.54–2.21||6,501,416||32 (15.8)||0.49||0.35–0.70|
|West||6,603,676||179 (28.8)||2.71||2.34–3.14||6,604,180||58 (28.6)||0.88||0.68–1.14|
The incident rate ratios comparing the incidence rates of SLE and lupus nephritis in girls relative to boys were lowest among children ages 3 years to <6 years (incident rate ratio 3.68, 95% CI 0.76–17.72) and highest among those ages 6 years to <9 years (incident rate ratio 8.45, 95% CI 2.54–28.06). For girls ages 15 years to <18 years compared to boys of the same age, the incident rate ratio was 5.99 (95% CI 4.27–8.41). In comparison, our sensitivity analysis yielded an average annual incidence rate of 2.25 for SLE and 0.74 for lupus nephritis per 100,000 children per year, with an incident rate ratio of 6.33 for SLE and lupus nephritis in girls compared to boys in the youngest age category, and an incident rate ratio of 6.70 in girls compared to boys in the oldest age category (Table 3).
|No. of cases||Person-years||Incidence rate||95% CI||No. of cases||Person-years||Incidence rate||95% CI|
Employing US nationwide billing data from 5 recent years for children enrolled in Medicaid, the federal-state program that provides health insurance to the poor, we have estimated the overall prevalence of SLE to be ∼10 per 100,00 children. The prevalence of SLE is 5.3 times higher in girls than in boys. More than one-third of the children (37%) had lupus nephritis. The prevalence of lupus nephritis was estimated to be 3.6 per 100,000 children, and was 4.5 times higher in girls than in boys. As has been shown in several previous studies in other populations, there was a higher prevalence of both SLE and lupus nephritis among all non-White children compared to White children. Rates of SLE and lupus nephritis were significantly lower in the Midwest section of the country than in other regions, likely due to the racial and ethnic population composition there (lower proportions of non-Whites). We identified only 2 past studies of SLE incidence among children, one in Canada in 1991–1993 (14) and one conducted in Austria in 1997–1998 (17). In these studies, the incidence rates were estimated to be 0.28 and 0.48 per 100,000 children per year, respectively, and both rates are lower than our incidence estimates in US Medicaid–enrolled children. As in the few past studies that have examined the female-to-male ratio for new-onset pediatric SLE and lupus nephritis, we have found that the incidence rate ratios (girls-to-boys) were lowest in the 3 years to <6 years age group. The same pattern was found in a nationwide French study involving 156 pediatric SLE cases (19).
The current rate estimates among Medicaid-enrolled children in the US demonstrate the substantial demographic variation that exists in this disease (Tables 1 and 2). This range may represent true differences across populations, as has been observed in the past, with higher prevalence rates among African American, Hispanic, Native American, and Asian populations compared with non-Hispanic White populations. However, these differences may also be due to the challenges of determining prevalence, including variation in case definition and case ascertainment between studies (16, 20), as well as a lack of a consensus age definition for pediatric SLE (with past studies utilizing different age cutoff values, ranging from age 16 years to age 21 years) (Table 4).
|Study location (ref.)||Date||Study design||Maximum age||No. of subjects||Prevalence per 100,000 children|
|Saskatchewan, Canada (30)||1981–2004||Clinic-based||≤18 years||225||17.7*|
|Taiwan (15)||1999||Population-based, national health insurance||<16 years||365||6.3|
|Hawaii (6)||1993–2002||Clinic-based||<19 years||50||24.0|
|British Columbia, Canada (16)||2004||Clinic-based||<19 years||40||3.3 for non–Native Americans, 8.8 for Native Americans|
|US Medicaid (present study)||2000–2004||Administrative data||<18 years||2,959||9.73 overall, 14.08 for African Americans, 11.51 for Hispanics, 4.86 for Whites, 23.79 for Asians, 13.38 for Native Americans|
Variation in prevalence and incidence estimates may also reflect differences in the source populations. The vast majority of past studies have been hospital- or clinic-based, likely representing patients with more severe disease. When using such data, it is often difficult to calculate population-based prevalence estimates, since assumptions regarding referral patterns must be made in order to define the population that gave rise to the cases seen in hospitals and clinics. Administrative healthcare databases have been used for research purposes, including surveillance, outcomes research, and quality assessment. However, these sources, too, have their limitations. Complete case identification may be difficult to ensure, particularly since validation of the case definition is not always feasible.
In this study, we investigated the prevalence, incidence, and sociodemographic characteristics of SLE and lupus nephritis among children whose healthcare was covered by Medicaid in the US during 2000–2004. The US Medicaid population is distinct from the general population in socioeconomic terms, by definition; to be eligible for Medicaid in most states, individuals must be living in poverty according to the federal annual income threshold (21). This socioeconomically disadvantaged population likely includes a higher proportion of, and more severe cases of, SLE than the more affluent remainder of the US population. By most measures, the incidence and severity of lupus are higher among individuals with lower socioeconomic status compared with individuals with higher socioeconomic status. Mortality from lupus is also higher among those with lower socioeconomic status (22–25). Significant predictors of poor outcomes and disease progression in the LUMINA cohort have included poverty, lack of education, and lack of social support (22, 26). Adult lupus patients in LUMINA with incomes below the federal poverty level were 4 times more likely to die than were those with higher incomes (27), and poverty was a stronger predictor of mortality than was ethnicity (28).
The identification of patients with SLE and lupus nephritis with the use of administrative billing data has been previously applied in adult populations. Algorithms employing 2 separate billing claims, separated in time, for a diagnosis of SLE had excellent performance in a study using administrative billing data from Quebec (29). We increased this minimum to at least 3 billing diagnoses of SLE, as Medicaid does not uniformly code subspecialty (thus, we could not examine rheumatologists' visits separately), and this allowed us to exclude those who were seen for only one “rule-out” SLE visit and followup. In a past study, we found that the range of renal disease billing codes employed had an 80% positive predictive value for lupus nephritis in an adult Medicaid population (18). However, neither of these methods of case identification has been validated for pediatric patients. In addition, many children were not continuously covered by Medicaid during the period of study. For the calculations of incidence rates, we limited all observations to Medicaid-enrolled time periods only (denominator population as well as numerator case numbers). We also performed sensitivity analyses limited to the 80% of children who were continuously enrolled in Medicaid for a minimum of 24 months during these years, and the results were very similar (Table 3).
This is the largest and only nationwide study to date of a pediatric SLE population comprising >30 million US Medicaid–enrolled children. We analyzed national data collected over 5 years in an administrative data set that captures all claims in the Medicaid medical insurance system. Within this large population, we were able to investigate potential variation in SLE and lupus nephritis prevalence and incidence rates in children across the country and by age and sociodemographic group. The elevated numbers of identified pediatric SLE cases highlight the public health significance of the disease. These data provide an important indication of the burden of this disease according to age, sex, race/ethnicity, and region of the US, and underscore the extent to which SLE is disproportionately distributed within society. The implications of this unequal distribution are important at the individual, family, and societal levels.
These results should allow further study of healthcare access and outcomes among pediatric patients with SLE and lupus nephritis in the Medicaid- beneficiary population. The estimated prevalence and incidence rates and demographic descriptors may also provide guidance to clinicians and policymakers in the development of strategies to reduce barriers and improve care for children with SLE.
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. Hiraki 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. Hiraki, Feldman, Liu, Winkelmayer, Costenbader.
Acquisition of data. Hiraki, Liu, Fischer, Costenbader.
Analysis and interpretation of data. Hiraki, Feldman, Liu, Alarcón, Fischer, Winkelmayer, Costenbader.