A reengineered hospital discharge program to decrease rehospitalization: a randomized trial

Abstract

Background: Emergency department visits and rehospitalization are common after hospital discharge.

Objective: To test the effects of an intervention designed to minimize hospital utilization after discharge.

Design: Randomized trial using block randomization of 6 and 8. Randomly arranged index cards were placed in opaque envelopes labeled consecutively with study numbers, and participants were assigned a study group by revealing the index card.

Setting: General medical service at an urban, academic, safety-net hospital.

Patients: 749 English-speaking hospitalized adults (mean age, 49.9 years).

Intervention: A nurse discharge advocate worked with patients during their hospital stay to arrange follow-up appointments, confirm medication reconciliation, and conduct patient education with an individualized instruction booklet that was sent to their primary care provider. A clinical pharmacist called patients 2 to 4 days after discharge to reinforce the discharge plan and review medications. Participants and providers were not blinded to treatment assignment.

Measurements: Primary outcomes were emergency department visits and hospitalizations within 30 days of discharge. Secondary outcomes were self-reported preparedness for discharge and frequency of primary care providers' follow-up within 30 days of discharge. Research staff doing follow-up were blinded to study group assignment.

Results: Participants in the intervention group (n = 370) had a lower rate of hospital utilization than those receiving usual care (n = 368) (0.314 vs. 0.451 visit per person per month; incidence rate ratio, 0.695 [95% CI, 0.515 to 0.937]; P = 0.009). The intervention was most effective among participants with hospital utilization in the 6 months before index admission (P = 0.014). Adverse events were not assessed; these data were collected but are still being analyzed.

Limitation: This was a single-center study in which not all potentially eligible patients could be enrolled, and outcome assessment sometimes relied on participant report.

Conclusion: A package of discharge services reduced hospital utilization within 30 days of discharge.

Funding: Agency for Healthcare Research and Quality and National Heart, Lung, and Blood Institute, National Institutes of Health.

Trial registration: ClinicalTrials.gov NCT00252057.

Figures

Figure 1. Study flow diagram

* Patients did not meet inclusion criteria if they were admitted from or planned discharge to an institutional setting (n = 74), planned hospitalization (n = 3) or discharge to a non-U.S. community (n = 5), were transferred to different hospital service (n = 8), did not speak English (n = 371) or have a telephone (n = 71), were on hospital precautions (n = 274) or suicide watch with a sitter (n = 10), were unable to consent (n = 181), had sickle cell disease as the admitting diagnosis (n = 38), had privacy status (n = 8), were deaf or blind (n = 2), or other (n = 4). † Usual care participants did not meet eligibility criteria if they were discharged to a nursing facility (n = 28), were transferred to another hospital service (n = 1), were previously enrolled (n = 1), died during index admission (n = 2), requested to be removed (n = 5), or other (n = 3). ‡ Intervention participants did not meet eligibility criteria if they were discharged to a nursing facility (n = 21), were transferred to another hospital service (n = 6), died during index admission (n = 1), requested to be removed (n = 2), or other (n = 8). § 107 intervention participants did not receive a reinforcement call because they could not be reached by telephone (n = 93), they were readmitted the same or next day (n = 2), there was no staffing coverage (n = 8), or other (n = 4).

Figure 2. Cumulative hazard rate of hospital utilization for 30 days after index hospital discharge

* The denominators for the events were 433 for usual care and 397 for intervention. This represents the number of discharges for each group, which includes index discharges and discharges from all subsequent admissions. At each discharge, the participant is returned to the risk pool. The denominator is thus constant during the entire 30 days. † Two events for the usual care group and 6 events for the intervention group were removed from this analysis because the date of admission was missing.