Article Title

Patterns of Multiple Emergency Department Visits: A Discretionary Care-Seeking Behavior?

Publication Date



emergency department, claims data


Background/Aims: Identifying an effective strategy to reduce emergency department (ED) visits remains challenging. This study seeks to empirically identify patient characteristics and utilization patterns that are significantly associated with multiple ED visits by patients within a 1-year period. The rationale is that a single ER visit is likely to be subject to random variation that is inherently difficult to predict. This study focuses only on Geisinger Health Plan members who had at least one ED visit in a 2-year period (2013–2014) across all plan types for which Geisinger Health Plan was “at risk” (i.e. Medicaid, Medicare, commercial, the Exchange) and identifies patient characteristics and other care utilizations (e.g. high-end diagnostic imaging, outpatient surgery, physician office visits) that are associated with multiple ED visits within a 12-month period.

Methods: The sample was restricted to adult Geisinger Health Plan members (≥ 18 years old) and included 67,997 unique members. Two multivariate logistic regression models were estimated: 1) a model that includes primary care physician (PCP) fixed effects, and 2) another model that does not include PCP fixed effects. The PCP fixed effects were used to examine if the PCP influences the likelihood of multiple ED visits.

Results: More frequent physician office visits were associated with greater likelihood of multiple ED visits (P < 0.001). Multiple use of high-end diagnostic imaging (computed tomography and/or magnetic resonance imaging) and multiple outpatient surgeries also were both associated with greater likelihood of multiple ED visits (P < 0.001). These associations remain significant even after controlling for age (older patients were less likely to make multiple ED visits), gender, plan types (Medicaid members were more likely to have multiple ED visits than members with other plan types) and chronic conditions. The odds ratios obtained with and without PCP fixed effects were similar, indicating that PCPs do not contribute significantly to patients’ likelihoods of multiple ED visits.

Conclusion: These patterns suggest that some patients may be seeking outpatient care that is not explained by their clinical and demographical characteristics alone. More studies are necessary to explore the underlying mechanism that explains such an apparent care-seeking behavior