Emergency department utilization among skin cancer patients: a retrospective study of the national emergency department sample 2013–2015

Skin cancer-related emergency department (ED) visits are among the most common cancer-related ED visits in the USA. However, ED utilization among skin cancer patients has not been evaluated. To assess overall utilization of EDs among skin cancer patients, reasons for skin cancer-related visits, and factors associated with inpatient admission. This was a retrospective, cross-sectional study of adults with skin cancer presenting to EDs using years 2013–2015 of the Nationwide Emergency Department Sample. In total, 693,835 of the 223,329,909 weighted ED visits were made by patients with skin cancer. Visits among this population were frequently due to age-related comorbidities and skin cancer treatment-specific adverse events. Melanoma accounted for the minority of skin cancer-related visits (27.58%), but over half of subsequent inpatient admissions (51.18%) and was associated with greater odds of inpatient admission compared to keratinocyte carcinoma (OR 1.278, 95% CI 1.264–1.293). Treatment and staging codes were not available, and thus, differences in ED utilization among skin cancer stages and treatment approaches could not be assessed. Ultimately, these findings are important in improving anticipatory outpatient care for patients with skin cancer and in guiding appropriate management of this unique population in the ED.


Introduction
While the treatment of skin cancer and its associated complications is predominantly managed in the outpatient setting, the use of emergency departments (EDs) for acute care also occurs, albeit to a lesser degree. In a 2017 study by Rivera et al., "melanoma and other skin cancers" was identified as the 8th most common category of cancer among adult cancer-related ED visits [1]. A systematic review of ED utilization among cancer patients identified the most common presenting symptoms as febrile neutropenia, infection, pain, fever, and dyspnea [2]. While ED visits among skin cancer patients likely share some overlapping presentations with the general adult cancer patient population, skin cancerspecific complications may also present to the ED. Given the frequency of ED visits related to skin cancer, a better understanding of the reasons for visits and outcomes of the visits is needed. Insight into these visits can guide efforts to prevent or manage cancer-related complications in the dermatology clinic and reduce utilization of the emergency department. Identification of common ED presentations of skin cancer patients is important in improving management. Therefore, the goal of our study was to determine the overall utilization of EDs among melanoma and keratinocyte carcinoma patients, the reasons for skin cancer-related visits, and the factors associated with inpatient admission.

Methods
This study examined ED visits among skin cancer patients using 2013 to 2015 data from the Nationwide Emergency Department Sample (NEDS). The NEDS is part of the Healthcare Cost and Utilization Project (HCUP), sponsored by the Agency for Healthcare Research and Quality. The NEDS is the largest all-payer ED database in the USA [3]. For 2014, the NEDS contains information from 31 million ED visits at 945 hospitals that approximate a 20-percent stratified sample of US hospital-owned EDs. Weighted, it estimates 138 million ED visits in 2014 [3]. NEDS is representative of US hospital-owned EDs by stratifying based on hospital geographic region, trauma center designation, urban-rural location of the hospital, teaching hospitals, and for-profit status. NEDS was compiled using two databases: the HCUP State Emergency Department Databases (SEDD) and the State Inpatient Databases (SID) [3]. The SSED capture discharge information on all ED visits that do not result in same-hospital admission, and the SID capture discharge information on all ED visits that result in a same-hospital admission. Information on geographic characteristics, hospital characteristics, patient characteristics, and the nature of visits is included. The NEDS contains encounter-level data, not patient-level records, and therefore contains no direct patient identifiers.
Skin cancer-related ED visits (ages ≥ 18 years) were identified using the HCUP Clinical Classification Software (CCS) codes. The CCS is a software tool that collapses the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9 CM) into a smaller number of categories. The HCUP NEDS database includes up to 15 CCS diagnosis codes for each visit. Skin cancer-related ED visits will be identified as any visit in which the CCS codes 22 and 23, which code for melanomas of the skin and other nonepithelial cancer of skin, respectively, were recorded. Visits in which there was both a CCS code for melanoma and a CCS for nonmelanoma were excluded from the analysis. The primary reason for the ED visit will be identified as the first listed non-skin cancer CCS code among all reported CCS codes. Skin cancer visits due to complications of systemic therapy or radiation were identified using the ICD-9 E-codes E933.1 and E872.9 for abnormal reaction to systemic therapy and abnormal reaction to radiotherapy, respectively.
Skin cancer-related ED visits were assessed by patient, hospital, and visit characteristics. The following patient demographics were included: age, gender, insurance status, residence (urban vs. rural), and annual median household income estimated using residential zip code. Hospital characteristics such as geographical region (Northeast, Midwest, South, and West), location in metropolitan or non-metropolitan area, and whether the hospital is teaching vs. nonteaching were included as well. Visit characteristics included day of the week and disposition status.
All statistical analyses were performed using SAS version 9.4 (SAS corporation, Cary, NC, USA). Survey procedures were used to account for the HCUP-NEDS sampling design. Weighted national frequencies of skin cancer-related ED visits were calculated for each study year (2013,2014,2015). Descriptive analyses were performed to characterize patient demographics, visit characteristics, and hospital characteristics listed previously. Rao-Scott X 2 tests were used to compare skin cancer-related ED visits to non-skin cancerrelated ED visits. The top reasons for skin cancer-related ED visits and their disposition status (admitted or discharged) were determined. Multivariable logistic regression analysis was used to examine predictors of inpatient admission for skin cancer-related ED visits. P < 0.05 (double-sided) was used to determine significance for all statistical tests.

Results
There were 693,835 weighted visits with a skin cancer diagnosis out of 223,329,909 total weighted ED visits between 2013 and 2015 (0.31%). Melanoma and keratinocyte carcinoma accounted for 27.58% (191,374/693,835) and 72.42% (488,448/693,835) of these visits, respectively. Patients with skin cancer presenting to the ED were most often males (53.22%) ages 65-84 (49.89%) from fringe or metropolitan cities with < 1 million in population (57.58%). They were most often from the highest-income quartile (28.51%) and most commonly insured through Medicare (71.18%). The majority of patients presented on weekdays (72.93%) and were most frequently treated and released without inpatient admission (51.05%). The hospitals with the most frequent visits related to skin cancer were most often located in the south (37.09%). 45.41% of skin cancer-related ED visits resulted in hospital admission, compared to 16.39% of nonskin cancer-related ED visits (Table 1).
Controlling for patient and visit characteristics, skin cancer-related ED visits were more likely to occur among males (female adjusted OR (aOR) 0.662 [95% CI 0.659-0.665], P < 0.0001) from fringe or metropolitan locations less than < 1 million in population ( Nonspecific chest pain, cardiac dysrhythmias, pneumonia, skin and subcutaneous tissue infections, and septicemia were the most common reasons for ED visits among skin cancer patients ( Table 2). Complications of surgical procedures of medical care was the 8th most common diagnosis. The visits most commonly resulting in inpatient admission included septicemia (97.48%), pneumonia (82.68%), and cardiac dysrhythmias (63.48%). Of the 6,293 visits due to complications of chemotherapy or radiation, 90.43% of patients were admitted. 51.18% of melanoma patients and 46.61% of nonmelanoma skin cancer patients were admitted to the hospital (Table 3). On multivariable analysis, demographic factors associated with increased odds of inpatient admission included being male, older, from the 25th-50th-income quartile, and having a melanoma diagnosis. In fact, the odds of being admitted to the hospital from the ED for melanomarelated admissions were 1.278 (P < 0.0001, 95% CI = [1.264, 1.293]) times higher than for those visiting the ED without melanoma-related concerns. Those with a primary payer other than Medicare and those living in a location smaller than a metropolitan with ≥ 1 million in population were less likely to be admitted (Table 4).

Discussion
In this study, we used a nationally representative sample from 2013 through 2015 to assess ED utilization among skin cancer patients. Skin cancer-related visits shared similar presentations with previously reported overall cancer-related ED visits, but also had presentations unique to skin cancer. Specifically, compared to the 2017 study by Rivera et al. which reviewed all cancer-related ED visits [1], seven of the top ten most common reasons for ED visits were shared between skin cancer-related visits and all cancer-related visits. Many of these primary diagnoses were related to age-related comorbidities, including cardiac dysrhythmias and chronic obstructive pulmonary disease, which is in line with the idea that both skin cancer and cancer in general tend to primarily affect the elderly. However, these primary diagnoses appeared to be less severe in skin cancer-related visits than all cancer-related visits as the admission percentage were lower for skin cancer-related visits for all seven diagnoses. ED visits and inpatient admissions were more likely to occur among both males and elderly skin cancer patients. However, while patients from higher-income quartiles had greater utilization of the ED, they were less likely to be admitted to the hospital as a result of their ED visit. Similarly, relative to those with Medicare, those with private insurance had higher adjusted odds of having a skin cancer-related ED visit but decreased adjusted odds of being admitted to the hospital. The discrepancy in ED visits versus admissions and poorer outcomes may either be due to an access-to-care issue in that low socioeconomic status individuals predominantly visit the ED for severe reasons or that less access to care is associated with more late-stage disease due to less access to outpatient treatments and/or poorer early detection of skin cancer. This is further supported by a 2006 study using the SEER database by Reyes-Ortiz et al. which found that older individuals covered by Medicare in lower socioeconomic areas had poorer 5-year melanoma survival than older individuals covered by Medicare in higher socioeconomic areas [3].
Unique to skin cancer-related visits was the relatively high rate of skin and subcutaneous infections as well as complications of surgical procedures or medical care (4th and 8th most common primary diagnosis, respectively), which was absent in the top ten primary diagnoses for all cancers in the analysis performed by Rivera et al. Both diagnoses resulted in admission greater than 45% of the time. This was interesting in the setting of the widely reported low complication rates seen with skin cancer treatment [4]. However, this observation is limited by the unknown temporality between treatment and the ED visit, as well as the lack of detail in regard to the treatment modality. Future studies are needed to better understand the connection between serious infections and skin cancer. Notably, even though melanoma represents less than 5% of all skin cancers [5], it represented 27.58% of skin cancer-related ED visits and 51.18% of skin cancer-related inpatient admissions from the ED. Additionally, there were greater odds of admission for melanoma compared to a keratinocyte carcinoma in multivariable analysis. This is likely indicative of the higher severity of melanoma and a greater application of systemic interventions for its treatment. Irrespective of skin cancer type, we observed the rate of inpatient admissions to be much higher for patients with skin cancer compared to patients without skin cancer (45.04% versus 16.39%).
As with previous studies that utilize the NEDS database, our analysis is limited by the use of CCS codes. Since the NEDS database is only able to utilize up to 15 CCS codes, misclassification is possible if a skin cancer CCS code is not included. In addition, patients with both a melanoma and nonmelanoma skin cancer CCS code were excluded from the analysis. Both of these scenarios should only account for a minimal proportion of the skin cancer-related ED visits, Despite these limitations inherent to the NEDS database, our study provides a robust, baseline evaluation of ED utilization specific to skin cancer across the USA. The large size of the NEDS database with over 30 million ED visits annually allows our results to be highly generalizable. Additionally, the use of survey procedures to account for the HCUP-NEDS sampling design limited bias and improved the validity of our results.

Conclusion
Controlling for patient and visit characteristics, skin cancerrelated ED visits were more likely to occur among elderly patients, patients from higher-income quartiles, and those with private insurance. Melanoma was also associated with higher inpatient admission rates than nonmelanoma skin cancer. The high frequency of visits due to skin and subcutaneous infections was a unique feature of skin cancer and warrants further investigation. We recommend expansion of national databases to include staging and treatment to allow for future research into the preventability of some of the inpatient admissions and the best mechanisms to achieve those potential reductions. Overall, these findings are important in guiding ED triage decision-making and improving anticipatory outpatient care for patients with skin cancer.
Author contributions B.C. and E.R. were involved in conception and design. All authors were involved in analysis and interpretation of data. E.R. and Y.K. wrote the main manuscript text. All authors were involved in revising the article.
Funding RT is funded by the Brian Werbel Memorial Fund through the Case Comprehensive Cancer Center.