D. Steinmetz1, I. Solsky1, B. Rapkin2, M. Parides1, J. M. Leider3, H. In1,2 1Montefiore Medical Center/Albert Einstein College of Medicine,Department Of Surgery,Bronx, NY, USA 2Montefiore Medical Center/Albert Einstein College of Medicine,Department Of Epidemiology And Population Health,Bronx, NY, USA 3Jacobi Medical Center/ Albert Einstein College of Medicine,Internal Medicine,Bronx, NY, USA
Introduction: Diagnosis of a gastrointestinal (GI) cancer via the emergency department (EDdx) is considered a marker of poor outcomes. Understanding patterns of health care usage of EDdx patients may illuminate targets for intervention to improve patient outcomes.
Methods: Cancer registry and pre-diagnosis administrative data were collected for patients diagnosed with GI cancers (esophagus, stomach, pancreas, colon, rectal and anal cancers) in 2010-2014 at a single academic institution serving one of the poorest urban regions of the country. Persons with an ED visit in the month prior to cancer diagnosis were considered to have EDdx. Descriptive statistics were performed including classification of patients by patterns of healthcare utilization in the year prior to diagnosis (outpatient only, ED with or without outpatient visits (EDuser), no visits, and prior hospitalizations). Logistic regression identified predictors of EDdx. Kaplan-Meier method and Cox proportional hazards regression assessed the influence of patterns of healthcare utilization on survival for EDdx and non-EDdx patients.
Results: Of 3,174 patients analyzed, 13% were EDdx patients. EDdx patients were more commonly non-white (82.5% vs. 77.8%), with non-private insurance (72.4% vs. 66%), more comorbidities (≥3: 74% vs. 67%), upper GI cancers (52% vs. 39%) and late cancer stage (56% vs. 40%). Patients with prior hospitalizations were more likely to be EDdx (OR 1.76, 95%CI: 1.21-2.54); however no difference in ED diagnosis rates was observed for other care patterns. EDdx was associated with improved survival for EDuser and worse survival for all other care patterns (Figure 1). Significant interaction was found between EDdx and care patterns for overall mortality (p = <0.05). After controlling for demographics and cancer factors, EDusers had improved mortality regardless of location of diagnosis, with markedly improved outcomes when diagnosed in the ED (EDuser & non-EDdx HR 0.74 95% CI: 0.58-0.94, EDuser & EDdx HR 0.42 95% CI: 0.25-0.71) compared to patients who were outpatient only and non-EDdx. As expected, EDdx patients with no prior visits or with prior hospitalizations had the worst outcomes (no visits & EDdx HR: 1.20 95% CI 1.65-2.94, prior hospitalization & EDdx HR: 1.52 95% CI 1.11-2.06).
Conclusion: Diagnosis in the ED was not uniformly a predictor of poor outcomes. Patients in this inner city hospital who rely on routine ED healthcare experienced better than expected outcomes and the diagnosis of cancer in the ED resulted in improved outcomes for these patients. Results suggest the need to understand benefits as well as harms before implementing broad-brush policies to divert routine ED care.
