373 Retrospective algorithmic application of a dermatological complexity tool

Background: Current tools to assess patient dermatological complexity are labor intensive and not easily applicable on a large scale. Objective: Create an automated algorithm from previously implemented manual tool using electronic health record data. Methods: Dermatology outpatient encounters 1/1/2019 through 4/30/2021 were retrieved refined include top 20 primary associated 10 secondary ICD-10 coded visits. Provider reported stability scores also retrieved. Complexity calculated the following Components: 1) number of diagnoses or treatment options, 2) amount and/or data be reviewed, 3) risk complications, morbidity, mortality. Forty ("Test Cohort”) randomly selected cohort (n=5,863) for scoring (gold-standard) comparison score. Results: McNemar’s symmetry test revealed similar between methods (p=0.906) Test Cohort. There was statistically significant fair agreement when including (weighted kappa= 0.36, 95% CI 0.15-0.57, p<0.001) excluding kappa=0.30, 0.09-0.52, p=0.004) scores. Sensitivity analysis components following: moderate Component 1 kappa = 0.53, 0.37-0.70, p<0.001), 2 0.35, 0.13-0.56, p=0.005), 3 0.47, 0.25-0.69, p<0.001). Conclusion: Compared gold standard, our relatively precise in capturing encounter complexity.