216 A new mortality prediction tool for Stevens-Johnson syndrome/toxic epidermal necrolysis RG Micheletti and MH Noe Department of Dermatology, University of Pennsylvania, Phil- adelphia, PA Stevens-Johnson Syndrome/toxic epidermal necrolysis (SJS/TEN) is a severe mucocutaneous drug reaction associated with significant morbidity and mortality. SCORTEN is a SJS/TEN- specific severity of illness score to predict in-hospital mortality. The objective of this study is to determine the accuracy of SCORTEN in large, US cohort, examine additional risk factors, and create an updated severity of illness score to predict in-hospital mortality. Data were collected from 18 academic medical centers in the United States from January 1, 2000 - June 1, 2015. Individuals  18 years of age with a diagnosis of SJS/TEN confirmed by a derma- tologist were included. 370 patients had information available for analysis. 54 patients (15.14%) did not survive to hospital discharge. SCORTEN predicted a mortality rate of 20.0% (Standardized Mortality Ratio: 0.73). Using factors from the original SCORTEN model as well as additional potential prognostic factors, 6 covariates were independent predictors of in- hospital mortality: age 50 years (OR: 2.99, 95% CI: 1.42-6.31), BSA 40% on admission (OR: 2.94, 95% CI: 1.41e6.09), malignancy (OR: 3.91, 95% CI: 1.72e8.88), dialysis (OR: 11.22, 95% CI: 2.72e46.28), serum BUN >10 mmol/L (OR:3.46, 95% CI: 1.72e6.94) and serum bicarbonate < 20mmol/L (OR: 2.37, 95% CI: 1.19e4.73). An updated prediction score assigns a value of 1 for the presence of each predictor, with the exception of dialysis, which receives 2 points. Calibration of the updated score demonstrated excellent agreement between predicted (15.0%) and actual mortality (15.1%) (Brier Score 0.1039). In conclusion, SCORTEN overestimated mortality in this large US cohort of SJS/TEN patients. Changing the age to 50 years, increasing the BSA cutoff to 40%, and adding dialysis to the model improved prognostication. Future use of a more accurate prediction model can provide improved prognostic information to patients and physicians and help stratify patients for clinical trials. 217 Epidemiology and management of dermatologic conditions in the pediatric oncology population H Song 1 , S Robinson 2 and J Huang 1 1 Boston Children’s Hospital, Boston, MA and 2 Harvard Combined Dermatology Residency Program, Boston, MA There have been tremendous advancements in oncologic therapy over the past few decades. The development of novel therapies and recognition of associated cutaneous reactions have highlighted the importance of dermatologic care in the pediatric oncology population. However, the epidemiology of dermatologic conditions and their management in this pop- ulation are unknown. The objectives of this retrospective study were to describe baseline characteristics of outpatient pediatric oncology patients seen by dermatology, and to deter- mine whether referral to dermatology changes management of dermatologic issues. The medical records of patients with a Dana-Farber Cancer Institute (DFCI) medical record number and a dermatology visit at Boston Children’s Hospital between 2008-2015 were reviewed for demographic and medical information. We identified 523 unique pediatric oncology patients seen by dermatology between 2008-2015. 88.1% received chemotherapy, 49.2% received radiation, and 37.9% underwent a hematopoietic stem cell transplant. Of those 59.5% (250/420) referred by oncology, 55.6% (139/250) initial visits resulted in a change in diagnosis, 28.4% (71/250) had no change in diagnosis, and 16% (40/250) were routine skin checks. Among dermatology visits resulting in a change in diagnosis, derma- tology diagnosed a skin reaction to therapy in 9.4% (13/139) and cutaneous malignancy in 4.3% (6/139) of patients. In those with no change in diagnosis, there was still a change in management in 73.2% (52/71) of patients, with topical therapies recommended in 92.3% (48/52) of patients. Among 20 patients where there was oncologic concern for an atypical nevus, malignancy, or GVHD, dermatology provided an alternate benign diagnosis in 100% (20/20) of patients. Of 93 visits where a biopsy was performed, 18 resulted in a histopath- ologic diagnosis of malignancy. Referral to dermatology impacts the care of pediatric oncology patients, resulting in changes in both diagnosis and management of these patients. 218 Clinical and pathologic correlation in biopsied alopecia patients J Zampela and J Alhariri Johns Hopkins, Baltimore, MD Differentiating between scarring and non-scarring alopecia has posed a diagnostic dilemma for clinicians. Histopathology is often used as the gold-standard for distinguishing between these entities. Herein wein we perform a retrospective analysis using pathology reports to assess the correlation betwen clinical differential diagnosis to histopathologic diagnosis on scalp biopsies. A search for “alopecia” yielded 694 adult patients and after exclusion of non- scalp biopsies and reports without a clinical differential diagnosis, 458 reports that were included in the analysis. Average age of the patients was 48  16 years for scarring and 46  19 years for non-scarring. A total of 20 different diagnoses were made across all biopsy reports with forms of lupus representing the greatest fraction. We show that clinical exam agrees with pathology in 80% of cases. The kappa correlation coefficient is 0.59 (p < .0001), indicating moderate agreement, which did vary by race and sex. However, clinical and histopathologic diagnosis only agree in 60% of cases. These results suggest that clinical exam is moderately reliable in distinguishing between scarring and non-scarring alopecia, but may not reliably provide a diagnosis. As such, clinical exam must be used judiciously in com- bination with histologic evaluation for diagnosis. 219 Natural history of chronic wounds in patients with recessive dystrophic epidermolysis bullosa D Solis 1 , J Nazaroff 2 , Y Dutt-Singkh 2 , S Choi 3 , M Barriga 1 , I Bailey-Healy 1 , M Marinkovich 4 and JY Tang 2 1 Stanford University, Redwood City, CA, 2 Stanford University, Stanford, CA, 3 Stanford University School of Medicine, Redwood City, CA and 4 Stanford University School of Medicine, Stanford, CA Patients with recessive dystrophic epidermolysis bullosa (RDEB) have chronic non-healing wounds and an increased risk for SCC. There is limited data about the natural history of RDEB wounds, and these data are needed as comparison for any new wound healing treatments for RDEB. We implemented an IRB approved study to evaluate self-reported history of chronic wounds and scars in 32 patients with RDEB. Chronic wound disease burden was also verified by clinical evaluation in a subset of 22 out of 32 patients (69%). Patients were recruited from the Epidermolysis Bullosa Clinical Research Consortium database, a North American registry for the clinical characterization of EB, and the Stanford EB clinic. Survey response rate was 32/52 (62%), and 65% of respondents were male with a mean age of 21 (range, 12-32). We defined chronic wounds as wounds that were open for months vs. chronic scars as skin areas that healed but frequently re-blistered. The majority of chronic wounds and scars were located on the trunk and extremities. The mean chronic wound size was 76 cm 2 (range, 5-600) and wounds were unhealed for a mean of 7 years (range 2 - 17). For chronic scars, the mean size was 54 cm 2 (range, 15-140), and the mean time to re-blister was 5 weeks (range 1-8). Many patients reported using allografts for treatment of open chronic wounds, but the mean duration of wound healing was 6 weeks. In summary, patients with RDEB have chronic wounds and chronic scars, with chronic wounds lasting for years and chronic scars frequently healing and reblistering within weeks. This natural history study creates a baseline compar- ison to measure wound healing in future clinical trials for chronic wounds and scars in patients with RDEB. 220 Predictors and cost of hospitalization for atopic dermatitis in US adults and children S Narla 1 , D Hsu 1 , J Thyssen 2 and JI Silverberg 1 1 Northwestern University Feinberg School of Medicine, Chicago, IL and 2 Department of Dermatology and Allergy, Herlev-Gentofte University Hospital, University of Copenhagen, Hellerup, Denmark Little is known about the inpatient burden of atopic dermatitis (AD). Wesought to determine the risk factors and burden of hospitalizations for AD in the US. Datawere analyzed from the 2002-2012 National Inpatient Sample, including of a 20% representative sample of all hospitalizations in the US. The prevalence of hospitalization for AD ranged from 419-519 and 66-72 per million in hospitalized children and adults, respectively. Hospitalization rates for AD were highest in the northeast during the winter and south during the summer. Geometric mean cost of care [95% CI] was lower for a primary diagnosis of AD vs. no AD in adults ($3,502 [$3,360-$3,651] vs. $6,849 [$6,775-$6,925]; P¼0.002) and children ($2,716 [$2,542-$2,903] vs. $4,488 [$4,302-$4,682]; P¼0.002). However, the high prevalence of hospitalization resulted in total inpatient costs of $8,288,083 per-year for adults and $3,333,868 per-year for children. Both adults and children who were admitted for AD were more likely to have non-white race/ethnicity, lowest-quartile annual household income, Medicaid or no insurance and had fewer chronic conditions. In conclusion, there are racial and health care disparities in hospitalization for AD, stressing the need for improved access to adequate outpatient and inpatient dermatologic care for all patients. 221 Does atopic dermatitis remain a disease of the advantaged in adulthood? A national birth cohort study K Abuabara 1 , LL Kohn 2 and SM Langan 3 1 Program in Clinical Research, Department of Derm, UCSF, San Francisco, CA, 2 UCSF, San Francisco, CA and 3 London School of Hygiene and Tropical Medicine, London, United Kingdom Studies examining the association between atopic dermatitis (AD) and socioeconomic status (SES) focus on disease onset or cross-sectional snapshots of prevalence in childhood and have not addressed the episodic and chronic nature of AD. Therefore we aimed to evaluate the relationship between SES and AD prevalence among individuals over time. We performed a longitudinal cohort study using data from the 1970 British Cohort Study, a nationally repre- sentative cohort of individuals followed from birth until age 42 in the UK (N¼13,038). Period prevalence of AD was measured at ages 5,10, 26, 30, 34, and 42 years and ranged from 5-11%. The mean age of onset was 12 years (SD 10 years). Similar proportions of individuals had AD during childhood only and adulthood only (12.4% and 11%), and 5.1% had AD during both time periods. We developed logistic regression models to examine the associa- tion between SES (using the Registrar General’s designation of social class by occupation) and AD disease course (no AD in childhood or adulthood, AD in childhood only, AD in adult- hood only, and AD in childhood and adulthood (termed persistent AD). Controlling for sex, higher social class during childhood was associated with higher AD prevalence in childhood only (OR 1.3, 95% CI 1.1-1.6), however neither childhood social class or adult social class was significantly associated with adulthood only AD (OR 1.2, 95% CI 0.98-1.5 and 1.2, 0.99- 1.6 respectively). Higher childhood SES was associated with persistent AD (OR 1.6, 95% CI 1.2-2.2). In this large birth cohort, adult onset AD was as common as childhood onset dis- ease. SES was predictive of childhood AD and of persistent AD into adulthood, but not new onset adult AD. These data can be used to target prevention and treatment efforts. Additional research examining factors associated with SES should consider how they affect the onset and persistence of AD at different age periods. Clinical Research: Epidemiology of Skin Diseases | ABSTRACTS www.jidonline.org S37