Transfer boarding delays care more in low-volume rural emergency departments: A cohort study

Abstract

Purpose: Emergency department (ED) crowding is increasing and is associated with adverse patient outcomes. The objective of this study was to measure the relative impact of ED boarding on timeliness of early ED care for new patient arrivals, with a focus on the differential impact in low-volume rural hospitals.

Methods: A retrospective cohort of all patients presenting to a Veterans Health Administration (VHA) ED between 2011 and 2014. The primary exposure was the number of patients in the ED at the time of ED registration, stratified by disposition (admit, discharge, or transfer) and mental health diagnosis. The primary outcome was time-to-provider evaluation, and secondary outcomes included time-to-EKG, time-to-laboratory testing, time-to-radiography, and total ED length-of-stay. Rurality was measured using the Rural-Urban Commuting Areas.

Findings: A total of 5,912,368 patients were included from all 123 VHA EDs. Adjusting for acuity, new patients had longer time-to-provider when more patients were in the ED, and patients awaiting transfer for nonmental health conditions impacted time-to-provider for new patients (16.6 min delays, 95% CI: 12.3-20.7 min) more than other patient types. Rural patients saw a greater impact of crowding on care timeliness than nonrural patients (additional 5.3 min in time-to-provider per additional patient in ED, 95% CI: 4.3-6.4), and the impact of additional patients in all categories was most pronounced in the lowest-volume EDs.

Conclusions: Patients seen in EDs with more crowding have small, but additive, delays in early elements of ED care, and transferring patients with nonmental health diagnoses from rural facilities were associated with the greatest impact.

Keywords: crowding; emergency service; hospitals; patient transfer; rural.

Figure 1.

Flow diagram of Veterans Health Administration (VHA) emergency department (ED) study patients, 2012–2014

Figure 2.. Delays in care associated with each additional patient boarding in the emergency department (ED) at the time of ED registration.

Estimates are the impact of one additional patient in each category, measured as a percentage change in the time (from the log-transformed multivariable hierarchical linear regression). Estimates are adjusted for number of ED beds(continuous variable), CCS diagnosis category (Level I categorical variable), disposition of index patient (categorical variable: admit, discharge, transfer), arrival during nights (5p-8a) or weekends (Sat, Sun) (dichotomous variable), emergency severity index (categorical 5-level variable), presence of mental health unit (dichotomous variable), and rural residence (dichotomous variable). Grey bars indicate the impact on rural patients, and the black bars indicate the impact on urban patients. The top row measures the delays associated with each patient waiting for local admission. The middle row measures the delays associated with each patient waiting for inter-hospitals transfer. The bottom row measures the delays associated with each patient waiting for discharge. The left column measures the delays by disposition category of those whose primary diagnosis is not a mental health diagnosis. The right column measures the delays by disposition category of those whose primary diagnosis was a mental health diagnosis. Patients being discharged without a mental health diagnosis were excluded from this model for collinearity.

Figure 3.. Association between boarding patients and delays in time-to-provider.

A. Time-to-provider is reported per additional patient in the ED in each category, stratified by tertile of ED volume. Tertile 1 corresponds with the lowest volume EDs and tertile 3 corresponds to the largest volume of EDs. B. Time-to-provider is reported per additional patient in the ED in each category, stratified by whether the facility had inpatient mental health available (dichotomous variable). Note that even when inpatient mental health is not available, some patients with mental health diagnoses may be admitted to a non-mental health unit. All error bars represent the 95%CI.