A Two-Year Analysis of Ocular ED Visits Using Logistic Regression Models

Abstract

Purpose : To report statistical analyses and predictive models created with two years of ocular ED (emergency department) visits at a level 1 Trauma center.

Methods : A single-center, two-year retrospective chart review examined 6849 ocular emergency department visits. The criteria for inclusion involved an ICD-10 ocular diagnosis at one of the highest volume Level 1 trauma centers in the country. Utilizing Python, specifically Numpy, Pandas and SciPy libraries, statistical analysis was performed to explore differences between the pediatric and adult populations with regard to age, repeat visits, ocular diagnosis, insurance provider, geographic location. Utilizing Scikit-learn, logistic regression models were created to both identify risk factors for ocular trauma and conjunctivitis.

Results : Conjunctivitis (54%) and Trauma of Eye and Orbit (11%) comprised the majority of diagnoses given. In the pediatric population, conjunctivitis was the diagnosis for 82.9% all visits which was dramatically higher than the 49% of visits in adults. Patients in both the pediatric and adult populations only had one ocular related ED visit (1.11 vs. 1.16 visits); however, patients with Medicaid plans were more likely to have recurrent visits (17% vs. 10.9%). Age (OR .9613; p-value <0.001), gender (Female; OR 1.229; p-value <0.001), geographic location (Bronx; OR 1.218; p-value<0.032) and insurance provider (Metroplus insurance; OR 1.2; p-value 0.005) were significant predictors of conjunctivitis with a logistic regression model. Age > 18 years was a statistically significant predictor for increased ocular trauma (OR: 1.843; p-value <0.001) while female gender (OR: .6925; p<0.001) and insurance provider (OR: 0.6234; p-value<0.001) were significant as negative predictors of ocular trauma.

Conclusions : We identified age, gender, insurance, and geographic location were significant predictors for conjunctivitis and ocular trauma. Insights into ocular ED visits can help optimize both ED providers' workflows and ophthalmology consult services. These optimizations will lead to improved healthcare delivery.

This is a 2020 ARVO Annual Meeting abstract.