2.1

Study setting and design

This is a retrospective cohort study that evaluates IHT from a tertiary, academic medical center (AMC) to three community hospitals within a large integrated healthcare system in Massachusetts between April 1 to December 31, 2021.The AMC ED has an annual census of around 60,000 ED visits in 2021, with 37,992 annual inpatient discharges. The three community hospital sites had between 9928 and 19,586 annual inpatient discharges and were located between 3.4 miles and 19.2 miles from the tertiary hospital. The study was approved by the Institutional Review Board (Protocol 2022P000696) and performed according to STROBE guidelines [ ].

All ED patients who were admitted to the general medical service (GMS) and were >18-years-old were potential candidates for our IHT program. To initiate consideration for IHT, the operational process ( Fig. 1 ) begins with the Emergency Medicine (EM) clinician selecting on the electronic bed request form if the patient meets any preset clinical criteria that would make them ineligible for transfer. These criteria included consulting service involvement, new malignancy, need for IR or ERCP, discharge <72 h ago from the same hospital, unstable for transport, and “other” as a catch all (See Appendix 1 for full list of exclusion criteria). The admitting office at the AMC then reviews the admission request. If there are no exclusion criteria and there are available beds at one of the three community hospitals in the integrated healthcare system, an admitting representative would approach the patient with a standardized script to ask for consent to transfer. If the patient or the patient’s healthcare proxy consents, the case is forwarded to the community hospitalists and nursing supervisors for approval, after which provider and nursing handoffs occur and an ambulance transfer is arranged. If any of the above conditions were not met, the transfer process is cancelled, and the patient would queue as usual for the next available general medicine bed at the AMC. There is no financial penalty for patients who decline transfer.

Flow chart of patient movement through transfer process.

2.2

Outcome measures and data collection

Data was obtained from the institutional EHR (Epic). Patient demographics for the cohort was extracted, including age, sex, race, ethnicity, language, insurance status, education level, and zip code. The zip code was further categorized as to whether it was within one of the AMC’s five priority neighborhoods, which are the focus of the institution’s required commitments under the Affordable Care Act [ ]. These five communities account for 69% of the city’s Black population. Demographic information was self-reported by patients to registration during their ED stay. We investigated differences in population demographics at two decision points in the process. The first was whether the patient was made eligible for transfer by ED clinicians, and the second was whether the patient consented to transfer when approached by registration staff. We then assessed whether the transfer was completed.

2.3

Data and statistical analysis

The unit of analysis was at the patient encounter level. Statistical analyses were performed in STATA 15 (StataCorp, College Station, TX). We performed Chi-square analysis followed by post-hoc multiple comparisons with Bonferroni correction to determine whether certain groups were over-represented at each stage of the transfer process. We then performed logistic regression to assess which demographic variables led to a greater likelihood of moving to the next stage of the transfer process.