Background: Specialty wards (e.g., cardiology or oncology wards) group clinically similar hospitalized patients in efforts to improve outcomes and costs. When these wards exceed capacity, subsequent patients overflow to “geographically dispersed” alternate wards. Geographic dispersion has been associated with care inefficiency and worse outcomes on specialty wards, but has not been studied in a large, diverse sample of hospitalized medicine service patients.
Methods: We performed a retrospective cohort study in 3 diverse University of Pennsylvania Health System (UPHS) hospitals. We included all medicine, family medicine, and geriatric service patients admitted from emergency departments (EDs), intensive care units (ICUs), or other wards in 2014 and 2015. Data were obtained from Penn Data Store, a UPHS clinical data warehouse that includes all payers. The primary exposure was placement on non-medicine wards, i.e., “geographical dispersion.” We excluded patients who switched from medicine to non-medicine wards and vice-versa (n=427 [2%]). The primary outcome was hospital length of stay (LOS). The secondary outcomes were hospital discharge to home and skilled nursing facilities (SNFs) and in-hospital mortality.
We performed mixed effects linear and logistic regression models with a random effect to account for clustering by hospital. We first performed univariate analyses of geographical dispersion and each outcome separately. We then performed separate multivariable regression models, adjusting for age, gender, race, ethnicity, insurance type, body mass index, number of Elixhauser comorbidities, days spent in ICUs, number of ward transfers during hospitalization, number of previous inpatient, ICU, and ED visits during the prior 12 months, numbers of unique medications and procedures during hospitalization, the Centers for Medicare and Medicaid Services four-level severity risk adjustment (mild, moderate, major, severe), and hospital admission source.
Results: The study population included 19,032 visits among 18,956 patients in 33 wards. Median age was 60 years, (IQR 45-74), 55% were female, and 61% were black (Table). Thirty-three percent of patients were geographically dispersed.
In multivariable regression, geographically dispersed patients had longer LOS (1.1 days, 95% CI 1.1-1.2, p<0.0005) and higher odds of discharge to SNF (OR=1.1, 95% CI 1.0-1.2, p=0.04). Geographically dispersed patients had lower odds of discharge to home but this was not statistically significant (OR=0.9, 95% CI 0.9-1.0, p=0.08). In-hospital mortality did not differ between the groups (OR=1.0, 95% CI 0.6-1.5, p=0.9).
Conclusions: Geographically dispersed hospitalized medicine patients had significantly increased LOS (>1 day longer) and higher odds of discharge to SNF. Future studies are needed to confirm these findings and explore underlying mechanisms of these associations.
.jpg)