How Hispanic patients access dermatologic care for skin diseases is unknown. This study aims to determine if differences exist in accessing the emergency department (ED), primary care, and outpatient dermatologic offices for skin diseases between Hispanic and non-Hispanic White patients. This cross-sectional study used nationally representative data from the Medical Panel Expenditure Survey (MEPS) from 2016-2019. A total of 109,337,668 (weighted) patients with any skin disease diagnosed at an ED, primary care, or dermatology visit were identified. Hispanics comprised 13.0% and non-Hispanic Whites comprised 68.8% of this subpopulation. Overall, 94.1% of Hispanic patients attended a primary care visit for their skin complaint, 5.8% saw a dermatologist, and 0.1% attended an ED visit. Compared to non-Hispanic Whites, Hispanics were more likely to attend a primary care visit (aOR 1.865; 95%CI, 1.640-2.122) and less likely to attend an outpatient dermatology visit (aOR 0.536; 95%CI, 0.471-0.610), after adjusting for insurance status, education, income, sex, age, and comorbidities. Our study suggests that, compared to non-Hispanic Whites, Hispanic patients access primary care more frequently and outpatient dermatologic offices less frequently for their skin conditions. Language barriers, less familiarity with the healthcare system, and lack of adequate health insurance may play roles in this observation.

Artificial intelligence models match or exceed dermatologists in melanoma image classification. Less is known about their robustness against real-world variations, and clinicians may incorrectly assume that a model with an acceptable area under the receiver operating characteristic curve or related performance metric is ready for clinical use. Here, we systematically assessed the performance of dermatologist-level convolutional neural networks (CNNs) on real-world non-curated images by applying computational "stress tests". Our goal was to create a proxy environment in which to comprehensively test the generalizability of off-the-shelf CNNs developed without training or evaluation protocols specific to individual clinics. We found inconsistent predictions on images captured repeatedly in the same setting or subjected to simple transformations (e.g., rotation). Such transformations resulted in false positive or negative predictions for 6.5-22% of skin lesions across test datasets. Our findings indicate that models meeting conventionally reported metrics need further validation with computational stress tests to assess clinic readiness.