A Reengineered Hospital Discharge Program to Decrease ...

1 A Reengineered Hospital Discharge Program to DecreaseRehospitalizationA Randomized TrialBrian W. Jack, MD; Veerappa K. Chetty, PhD; David Anthony, MD, MSc; Jeffrey L. Greenwald, MD; Gail M. Sanchez, PharmD, BCPS;Anna E. Johnson, RN; Shaula R. Forsythe, MA, MPH; Julie K. O Donnell, MPH; Michael K. Paasche-Orlow, MD, MA, MPH;Christopher Manasseh, MD; Stephen Martin, MD, MEd; and Larry Culpepper, MD, MPHB ackground:Emergency department visits and rehospitalization arecommon after Hospital :To test the effects of an intervention designed to min-imize Hospital utilization after :Randomized trial using block randomization of 6 and arranged index cards were placed in opaque envelopeslabeled consecutively with study numbers, and participants wereassigned a study group by revealing the index :General medical service at an urban, academic, :749 English-speaking hospitalized adults (mean age, ).Intervention:A nurse Discharge advocate worked with patientsduring their Hospital stay to arrange follow-up appointments, con-firm medication reconciliation, and conduct patient education withan individualized instruction booklet that was sent to their primarycare provider.

2 A clinical pharmacist called patients 2 to 4 days afterdischarge to reinforce the Discharge plan and review and providers were not blinded to treatment :Primary outcomes were emergency departmentvisits and hospitalizations within 30 days of Discharge . Secondaryoutcomes were self-reported preparedness for Discharge and fre-quency of primary care providers follow-up within 30 days ofdischarge. Research staff doing follow-up were blinded to studygroup :Participants in the intervention group (n 370) had alower rate of Hospital utilization than those receiving usual care(n 368) ( vs. visit per person per month; incidencerate ratio, [95% CI, to ];P ). The inter-vention was most effective among participants with Hospital utili-zation in the 6 months before index admission (P ). Ad-verse events were not assessed; these data were collected but arestill being :This was a single-center study in which not all poten-tially eligible patients could be enrolled, and outcome assessmentsometimes relied on participant :A package of Discharge services reduced Hospital uti-lization within 30 days of :Agency for Healthcare Research and Quality and NationalHeart, Lung, and Blood Institute, National Institutes of Intern ;150 author affiliations, see end of registration number: in 5 hospitalizations is complicated by postdis-charge adverse events (1, 2), some of which may leadto preventable emergency department visits or readmis-sions.

3 Despite this finding, Hospital Discharge procedureshave not been standardized (3). In addition, the decliningpresence of primary care providers (PCPs) in hospitals hasnot been adequately accompanied by systems to ensurethat patient data are transferred to subsequent caregivers(4, 5). For example, Discharge summaries frequently lackcritical data and are not sent to the PCP in a timely fashion(6, 7), resulting in outpatient clinicians being unaware oftest results that were pending at Discharge (8) and evalua-tions that were scheduled to be done after Discharge notbeing completed (9). Similarly, patients are often left un-prepared at Discharge ; many do not understand their dis-charge medications and cannot recall their chief diagnoses(10). With more than 32 million adult discharges in theUnited States each year (11), these deficiencies in the tran-sition of care increase illness, unnecessary Hospital utiliza-tion, and peridischarge interventions have shown a reduc-tion in Hospital readmission rates and cost (12 14), emer-gency department visits (15), and postdischarge adverseevents (16), whereas some have shown little or no effect(17 20).

4 Peridischarge interventions have also shown im-proved PCP follow-up and outpatient work-ups (21) andhigher patient satisfaction (15). Most of these studies havefocused on specific diagnoses (14, 22, 23) or highly se-lected populations, such as geriatric adults (12, 13, 19, 24).Some have focused on specific aspects of the Discharge ,such as increasing access to primary care follow-up (25),connecting with transitional nursing services (26), or im-proving patients ability to advocate for themselves afterdischarge (12). To date, no study has evaluated a standard-See also:PrintEditors TableConversion of graphics into slidesAnnals of Internal MedicineImproving Patient Care178 2009 American College of Physiciansized Discharge intervention that includes patient education,comprehensive Discharge planning, and postdischarge tele-phone reinforcement in a general medical 2004, we began an in-depth examination of hospi-tal Discharge , for which we designed a package of servicesto minimize Discharge failures a process calledreengi-neered Discharge (RED) (Table 1) (3, 27).

5 We did a ran-domized, controlled trial to evaluate the clinical effect ofimplementing RED among patients admitted to a generalmedical and ParticipantsWe conducted a 2-group, randomized, controlled trialof English-speaking patients 18 years of age or older whowere admitted to the medical teaching service of BostonMedical Center, Boston, Massachusetts a large, urban,safety-net Hospital with an ethnically diverse patient pop-ulation. Patients had to have a telephone, be able to com-prehend study details and the consent process in English,and have plans to be discharged to a community. Wedid not enroll patients if they were admitted from a skillednursing facility or other Hospital , transferred to a differenthospital service before enrollment, admitted for a plannedhospitalization, were on Hospital precautions or suicidewatch, or were deaf or blind. Boston University s institu-tional review board approved all study morning, a list of admitted patients was reviewedfor initial eligibility ( Hospital location, age, date and timeof admission, and previous enrollment).

6 Last names of po-tential participants were ranked by using a random-num-ber sequence to determine the order in which to approachpatients for enrollment. A trained research assistant thenapproached each patient and further determined eligibilityaccording to inclusion and exclusion criteria (Figure 1).By using block randomization (28) with varying blocksizes of 6 and 8, we randomly arranged index cards indi-cating either the usual care or intervention group. Weplaced the cards in opaque envelopes labeled consecutivelywith study numbers. We assigned eligible participants whoconsented to enrollment to a study group by revealing theconcealed index card. This process continued until 2 par-ticipants were enrolled each day of the week (or 3 partici-pants if the first 2 participants were randomly assigned tothe usual care group). This protocol ensured that researchassistants could not selectively choose potential participantsfor enrollment or predict assignment.

7 Participants ran-domly assigned to usual care received no further interven-tion. There were 40 participants in the usual care groupand 38 in the intervention group who were enrolled but nolonger met inclusion criteria at Discharge (most commonlybecause they were discharged to a nursing facility). Becausethe primary analysis was by intention to treat, we includedthese participants in the analysis, with the exception ofthose who died before index Discharge , requested to beremoved, or were previously enrolled (Figure 1).InterventionsNurse Discharge advocates (DAs) carried out all aspectsof the in- Hospital intervention. We hired 6 part-time DAsto work with intervention participants to ensure coverageby 1 DA 7 days a week, 5 hours a day. We trained all DAsto deliver the RED intervention by using a manual con-taining detailed scripts, observation of relevant clinical in-teractions, and simulated practice sessions. The primarygoals of the DA were to coordinate the Discharge plan withthe Hospital team and educate and prepare the participantfor Discharge .

8 At admission, the DA completed the REDintervention components outlined inTable 1. Additionalinformation about the DA training manual is publishedelsewhere (3) and can be found on our Web site ( ).With information collected from the Hospital teamand the participant, the DA created the after- Hospital careplan (AHCP), which contained medical provider contactinformation, dates for appointments and tests, an appoint-ment calendar, a color-coded medication schedule, a list oftests with pending results at Discharge , an illustrated de-scription of the Discharge diagnosis, and information aboutwhat to do if a problem arises. Information for the AHCPwas manually entered into a Microsoft Word (Microsoft,Redmond, Washington) template, printed, and spiral-bound to produce an individualized, color booklet de-signed to be accessible to individuals with limited healthliteracy. By using scripts from the training manual, the DAused a teach-back methodology (29) to review the contentsof the AHCP with the participant.

9 On the day of dis-charge, the AHCP and Discharge summary were faxed tothe department visits and rehospitalizations arecommon after Hospital trial demonstrated that a nurse Discharge advocateand clinical pharmacist working together to coordinatehospital Discharge , educate patients, and reconcile medica-tions led to fewer follow-up emergency visits and rehospi-talizations than usual care trial was conducted at a single center, and not alleligible patients were systematic approach to Hospital discharges can reduceunnecessary health service use. The EditorsImproving Patient CareA Reengineered Hospital Discharge Program to Decrease February 2009 annals of Internal MedicineVolume 150 Number 3179A clinical pharmacist telephoned the participants 2 to4 days after the index Discharge to reinforce the dischargeplan by using a scripted interview. The pharmacist hadaccess to the AHCP and Hospital Discharge summary and,over several days, made at least 3 attempts to reach eachparticipant.

10 The pharmacist asked participants to bringtheir medications to the telephone to review them andaddress medication-related problems; the pharmacist com-municated these issues to the PCP or Measures and Follow-upAt the time of recruitment, research assistants collectedbaseline data, including sociodemographic characteristics;the Short Form-12 Health Survey, Version 2 (30); thedepression subscale from the Patient Health Question-naire-9 (31); and the Rapid Estimate of Adult Literacy inMedicine (32). We calculated the Charlson ComorbidityIndex score by using primary and secondary diagnoses re-corded on the index admission Discharge summary (33).We determined the number of Hospital admissions andemergency department visits in the 6 months before indexadmission through medical record review (Boston MedicalCenter Hospital utilization) and participant report (allother Hospital utilization).The primary end point was the rate of Hospital utili-zation the total number of emergency department visitsand readmissions per participant within 30 days of theindex Discharge .