The transition from hospital to home is a high-risk period in a patient's illness. Poor communication between healthcare providers at hospital discharge is common and contributes to adverse events affecting patients after discharge. The importance of good communication at discharge will increase as more primary care providers delegate inpatient care to hospitalists. Any process that improves information transfer among providers at discharge might improve the health and safety of patients discharged from U.S. hospitals each year, and to appreciably reduce unnecessary healthcare expenditures. Information transfer among healthcare providers and their patients can be undermined because of inaccuracies, omissions, illegibility, logistical failure (e.g., information is never delivered), and delays in generation (i.e., dictation or transcription) or transmission. Root causes include recall error, increased physician workloads, interface failures (e.g., physician-clerical) and poor training of physicians in the discharge process. Many of the deficiencies in the current process of information transfer at hospital discharge could be effectively addressed by the application of information technology. The proposed study will measure the value of a software application to facilitate information transfer at hospital discharge. The study is designed to compare the benefits of discharge health information technology versus usual care in high-risk patients recently discharged from acute care hospitalization. The design is a randomized, single-blind, controlled trial. The outcomes are readmission within 6 months, adverse events, and effectiveness and satisfaction with the discharge process from the patient and physician perspectives. The cost outcome is the physician time required to use the discharge software.

Objectives: The study is designed to compare the benefits of discharge health information technology versus usual care in high-risk patients recently discharged from acute care hospitalization.

STUDY HYPOTHESES:

The primary efficacy endpoint is the proportion of patients readmitted at least once within 6 months after the index admission. Readmission is for any reason and includes observation status and full admission status.

Primary hypothesis: Among high-risk patients recently discharged from acute care hospitalization, there is a significant decrease in the primary efficacy endpoint for patients who receive discharge health information technology versus usual care discharge instructions.

Secondary hypothesis 1A: In the same patient population, the time to first readmission is greater for patients who receive discharge health information technology versus usual care discharge instructions.

Secondary hypothesis 1B: In the same patient population, the mean number of hospital days per patient within 6 months after index hospital discharge is lower for patients who receive discharge health information technology versus usual care discharge instructions.

Secondary hypothesis 2: In the same patient population, the mean score for effectiveness and satisfaction with discharge process is greater for patients who receive discharge health information technology versus usual care discharge instructions.

Secondary hypothesis 3: In the same patient population, the proportion of patients who report their pharmacist needed to clarify the discharge prescription is lower for patients who receive discharge health information technology versus usual care discharge instructions.

Secondary hypothesis 4: In the same patient population, the proportion of patients with at least one adverse event within 4 weeks after hospital discharge is lower for patients who receive discharge health information technology versus usual care discharge instructions.

Secondary hypothesis 5: In the same population, the mean satisfaction score with drug information will be higher for patients who receive discharge health information technology versus usual care discharge instructions.

Secondary hypothesis 6: Among primary care physicians who provide post-discharge care to high-risk patients, the mean score for discharge process effectiveness and satisfaction will be greater for patients who receive discharge health information technology versus usual care discharge instructions.

Secondary hypothesis 7: Among hospitalist physicians who discharge high-risk patients, the mean score for physicians' satisfaction with the discharge process will be greater for physicians assigned to discharge health information technology versus usual care discharge instructions.

METHODS: The trial design is a randomized cluster, single-blind (outcome assessors blind), controlled trial. The study design conforms to recent guidelines for randomized controlled trials. The test intervention is discharge application of health information technology. The control intervention is usual care (hand-written discharge instructions) described below. Each patient will remain in the study for 6 months. Enrollment in the study will last approximately 18 months. There will be no interim analysis.

Research personnel will obtain informed consent from potentially eligible inpatients. Informed consent from patients will occur during the screening visit.

Screening visit: Investigators will train research personnel to perform screening and informed consent. The screening visit may occur within 2 days of the planned discharge. After obtaining informed consent, research personnel will record items in the baseline assessment. Research personnel will ask patients about self-rated health, coronary artery disease (including angina pectoris myocardial infarction), diabetes mellitus in past year, hospitalization in past year, number of doctor visits in past year, presence of an informal caregiver able to care for the patient for several days, age, and gender. The screening questionnaire was validated. Research personnel will calculate a PRA score during the screening visit. PRA scores 0.5 and above define high-risk patients who have a 50% probability of being admitted to a hospital two or more times within 4 years. When the PRA score is applied to Medicaid beneficiaries followed for one year, 57% of patients with PRA 0.5 and above will have at least one hospital admission or 0.99 +/- 0.24 hospital admissions per person-year survived (mean +/- SE). Research personnel will offer informed consent to patients with PRA score 0.4 and above.

Research personnel will record limited information for patients who are ineligible or who refuse consent.

Baseline Assessment: The baseline assessment will occur after informed consent and before discharge. The ten-point clock test will be used as the screening instrument for orientation. Research personnel will record patient's name, address, age, stated race, gender, and discharge medication prescription. Research personnel will record patient contact information and alternate contact information in order to perform post-hospital telephone interviews required by the protocol.

Intervention allocation: The time of random treatment allocation will be after the baseline assessment and before discharge. Patients will not receive study treatment if they fail to consent or if they fail the inclusion/exclusion criteria. Treatment assignment will be in a 1:1 ratio to either discharge application of health information technology or usual care discharge instructions. The unit of randomization will be the hospitalist physician who performs the discharge process. The randomization process is designed to assure random allocation by cluster with the cluster determined by the discharging physician. Allocation concealment is not possible since all the enrolled patients who are discharged by the hospitalist physician will receive the same study intervention.

Dispense patient logbook: The purpose of the patient logbook is to promote ascertainment of study endpoints.

Patient telephone interview: discharge process effectiveness and satisfaction The purpose of the first telephone interview is to acquire data to measure secondary endpoints 2, 3, and 5. One week (5 to 9 days) after the hospital discharge date, research personnel will perform a telephone interview with the patient. Interviewers will instruct the patient to avoid mentioning the random intervention assignment. To address secondary hypothesis 2, interview questions will follow the PREPARED text. The PREPARED instrument surveys four key process domains: information exchange (community services and equipment), medication management, preparation for coping after discharge and control of discharge circumstances. The questions in PREPARED measure the patient's overall satisfaction with discharge, whether equipment and community service needs were met, and use of health services and health related costs post-discharge. The telephone interviewers will ask patients if their pharmacist had to call the doctor when attempting to fill the discharge prescriptions. The purpose of the question about pharmacists is to address secondary hypothesis 3. The telephone interviewers will ask questions from the Satisfaction with Information about Medicines Scale (SIMS). The SIMS is a 17-item survey with internal consistency and test-retest reliability. The SIMS survey instrument addresses secondary hypothesis 5.

Primary care physician questionnaire: discharge process effectiveness and satisfaction. The primary care physician questionnaire addresses secondary hypothesis 6. Within 10 to 18 days after the hospital discharge date, research personnel will contact the primary care physician to perform a survey.

Patient interview: adverse event assessment. The purpose of the second patient interview is to address secondary hypothesis 4. Approximately 4 weeks (20 to 40 days) after the index hospital discharge date, research physician personnel will perform a telephone interview with the patient. Physicians trained to assess adverse events will perform the telephone interview. Interviewers will instruct the patient to avoid mentioning the random intervention assignment. The interview tool is a modification of a validated survey instrument.

Hospitalist (discharging) physician questionnaire: The purpose of the survey is to address secondary hypothesis 7.

Patient interview: readmission assessment. The purpose of the third patient interview is to ascertain the primary endpoint, secondary endpoints (1A, 1B), and tertiary endpoints. Approximately 6 months (170 to 190 days) after the hospital discharge date, research personnel who are blinded to intervention assignment will perform a telephone interview with the patient. Interviewers will instruct the patient to avoid mentioning the random intervention assignment. Interviewers will ask the patient to consult their patient logbook while answering questions. Interviewers will record the admissions to the hospital, dates of admission, duration of hospital stay, number of outpatient physician visits, and number of emergency department visits that did not result in hospital admission.

Guess treatment assignment by blinded observers: The purpose of the guess is to measure the effectiveness of the blind.

Conditions for Early Withdrawal of Treatment: Patients may terminate study intervention at any time and return to the standard care if they withdraw their consent. If a patient withdraws from the study for any reason, then research personnel will conduct an end-of-study visit.

Sample size determination: The primary analysis is the difference in proportion of patients in the two study groups who achieve the primary efficacy endpoint of readmission within 6 months of discharge. The estimated event rate in the standard therapy group is 37%, which is the control group event rate from a systematic review of randomized controlled trials of discharge interventions. The minimum clinically relevant difference, 13%, corresponds to a standardized increment of 28.2% and is the empirical boundary for quantitative significance.

The required sample size for the primary analysis is 275 patients in the group assigned to discharge application of health information technology and 275 patients in the group assigned to control (usual care) therapy. In a previous study of discharge planning, the investigators enrolled 28% (363/1296) of potentially eligible patients. In the same study, 72% (262/363) of enrolled patients completed the 6-month assessment. Our hospitalist service discharges 297 patients per month. We estimate we will screen 5456 patients within 18.37 months. We estimate 50% of screened patients will be potentially eligible according to the Pra criteria. Among potentially eligible patients, we estimate 28% will consent to study enrollment. Therefore, the number of enrolled patients will be 5456 x 50% x 28% = 764. We estimate 72% (550/764) of enrolled patients will continue in the study until the 6-month assessment.

After 3 months of patient enrollment, we found the rate of enrollment was too low to achieve the required sample size. In 2005, we requested and received approval from Agency Healthcare Research Quality and Institutional Review Board to lower patient inclusion criterion, probability or repeat admission (PRA), from 0.50 to 0.40.

Inclusion Criteria:

• Inpatients at OSF Saint Francis Medical Center
• Discharged by the hospitalist service or other inpatient services
• High risk for poor post-discharge outcomes defined as probability of readmission (PRA) 0.4 or above

Exclusion Criteria:

• Less than 18 years old
• Unwilling or unable to provide written consent
• Life expectancy less than 6 months
• Will receive outpatient care from a primary care physician who is the same as the discharging physician
• Do not speak English or Spanish
• Not alert and oriented when admitted
• Do not have telephone for post-discharge contact
• Do not reside in Central Illinois
• Will be discharged to a nursing home
• Previously enrolled as subjects in the trial

6 months

Required medication reconciliation in both groups might have reduced the adverse event rates in both groups.