Effects of a Multimodal Transitional Care Intervention in Patients at High Risk of Readmission

Key Points Question Could a transitional care intervention targeting higher-risk medical patients reduce the risk of 30-day unplanned hospital readmission or death? Findings In this randomized clinical trial including 1386 patients at a high risk of unplanned readmissions across 4 hospitals in Switzerland, no statistical difference in the composite outcome of 30-day unplanned readmission or death between the intervention and control groups was found. There was no evidence of any intervention effects on postdischarge health care use, patient satisfaction with the quality of their care transition, or readmission costs. Meaning Results of this study suggest that the difficulties in preventing hospital readmissions continue, even when using multimodal interventions targeting higher-risk patients.


Study category and Rationale
Clinical trial. Risk category A: Clinical trials with interventions that are neither a therapeutic product nor a transplant product, nor a transplant. The investigator randomly assigned ("prospectively assigned") patients discharged home from the hospital to receive either standard intervention to reduce the risk of hospital readmission. The intervention include patient education, medication reconciliation, follow-up appointment planning, follow-up phone calls. A teaching and follow-up call intervention is not a medicinal product or device, a transplant or transplant product, a gene therapy, or a pathogenic organism.

Background and Rationale:
Hospital readmissions within 30 days are frequent. They represent a burden for the patients, but also for the entire health care system. Is it therefore recognized as important to improve the quality of the transition of care period in order to avoid as much as possible potentially avoidable readmissions. There are however still several gaps in current knowledge. First, most trials to reduce hospital readmission have been performed on specific patient populations such as patients with diabetes or heart failure, but less to an overall population of medical patients. Second, the most effective transition care interventions were also the most complex ones, but trials failed to target the medical patients who are most likely to benefit for better effectiveness. To identify those high-risk patients, no trials used one of the best validated, accurate, and easy-to-use tools, such as the "HOSPITAL" score. Finally, No interventional studies to reduce readmission have been previously performed in general medical patients in Switzerland.

Objective(s):
To evaluate in a randomized controlled trial the effect of a transitional care intervention prioritized to higher-risk medical patients (using the HOSPITAL score) on the composite of 30-day unplanned readmissions and death.

Outcome(s):
Primary outcome: unplanned readmission or death within 1 month after hospital discharge. Secondary outcomes: time to readmission/death, main cause of readmission, post-discharge care utilization, medication adherence, patient's perspective on quality of transition of care.

Study design:
Single-blinded randomized controlled trial.

Inclusion / Exclusion criteria:
All adult patients discharged home or nursing home after a hospital stay of one day or more in the department of medicine of the 4 participating hospitals will be screened. Patients already enrolled in the current study, not living in Switzerland, not speaking French/German, not having a phone number to be reached at, , or with cognitive troubles will be excluded. Only patients with a high risk of 30-day readmission estimated with the simplified "HOSPITAL" score will be invited to participate.

Measurements and procedures:
All patients will have their risk of readmission estimated according to the simplified "HOSPITAL" score, and have demographics, lab values, and diagnoses captured at inclusion. The patients randomized to the intervention group will receive a pre-discharge intervention with patient education, medication reconciliation, and follow-up appointment organization. They will then receive 2 short follow-up phone calls at 3 and 14 days after discharge to ensure medication adherence and follow-up by the primary care physician. Outcomes will be collected at 1 month by phone call interview by a blinded study nurse to capture outcomes.

Study Product / Intervention:
The intervention doesn't involve any medical drug or device, but: 1) Patient education on the specific diseases of the patients, including information leaflets about the most frequent diseases; 2) Medication reconciliation: the study nurse will review the list of medication with the patient and seek for any missing medication in the list; 3) Organization of a follow-up appointment by the primary care physician within 7 days after discharge; 4) Insure that a discharge summary is sent quickly to the primary care physician. 5) Two follow-up phone calls at day 3 and 14 after discharge to ensure medication and appointment adherence.

Control Intervention (if applicable):
The group control will receive usual care without additional intervention. Only an information sheet about the study participation will be given to the participant of the group control.

Number of Participants with Rationale:
We plan to include around 700 patients in each arm (total of 1,400 patients).

Study Duration:
Start

Considerations:
The primary outcome will be reported as a proportion of patients in each group and compared between the groups by a risk difference with 95% CI and a chi squared test. The primary analysis will be a comparison between two groups according to the intention-to-treat principle, i.e. all randomized patients will be analyzed in the allocated group regardless of any protocol violations such as cross-overs or early discontinuations of the intervention.
Because we will target patients at higher risk of readmission, we hypothesize that the intervention could reduce readmission by 25%. The sponsor designed the study, obtained the research funding, and wrote the protocol. He will supervise the data collection, manage the study nurses at each site, supervise the analysis done by the Clinical Trial Unit of the Bern University, interprete the study results with the help of the biostatistician, and write the report and final manuscript. The local investigator will help the Sponsor to implement the study at each site, and give local logistical

Laboratory
Not applicable

Data Safety Monitoring Committee
Due to the nature of the intervention, a DSMC is not needed. No change in the therapy will be made. The intervention doesn't included any medicinal product or device, transplant or transplant product, gene therapy, nor pathogenic organism.

Any other relevant Committee, Person, Organisation, Institution
Not applicable.

ETHICAL AND REGULATORY ASPECTS
The decision of the CEC concerning the conduct of the study will be made in writing to the Sponsor before commencement of this study. The clinical study can only begin once approval from all required authorities has been received. Any additional requirements imposed by the authorities shall be implemented.

Study registration
The study will be registered under the SNCTP Portal after the ethical committee approval (BASEC). We will also register the study under Clinicaltrials.gov before the inclusion of the first patients.

Categorisation of study
The study is a clinical trial. The study involve living persons over a certain period of time, to investigate the effects of a health-related intervention. To investigate the effects, the principal investigator exposes individuals who have been selected specifically for the trial to the intervention to be investigated, and then measures its effects on health.
The risk category is A: the clinical trial intervention include patient education, medication reconciliation, follow-up appointment planning, follow-up phone calls. This intervention is not a medicinal product or device, a transplant or transplant product, a gene therapy, or a pathogenic organism.

Competent Ethics Committee (CEC)
The project plan, the proposed participant information and consent form as well as other project-specific documents will be submitted to a competent Ethics Committee (CEC). No changes will be made to the protocol without prior Sponsor and CEC approval, except where necessary to eliminate apparent immediate hazards to study participants.
Premature study end or interruption of the study will be reported within 15 days. The regular end of the study will be reported to the CEC within 90 days. The project does not involve investigation with radioactive sources. The final study report will be submitted within one year after study end.

Competent Authorities (CA)
Not applicable.

Ethical Conduct of the Study
The study will be carried out in accordance to the protocol and with principles enunciated in the current version of the Declaration of Helsinki, the guidelines of Good Clinical Practice (GCP) issued by ICH, the Swiss Law and Swiss regulatory authority's requirements. The CEC will receive annual safety and interim reports and be informed about study stop/end in agreement with local requirements.

Declaration of interest
The sponsor has no existing conflict of interest.

Patient Information and Informed Consent
The investigators, or one of his/her delegated research team member will explain to each potential participant the nature of the study, its purpose, the procedures involved, the expected duration, the potential risks and benefits and any discomfort it may entail. Each subject will be informed that the participation in the study is voluntary and that he/she may withdraw from the study at any time and that withdrawal of consent will not affect his/her subsequent medical assistance and treatment.
The subject will be informed that his/her medical records may be examined by authorised individuals other than their treating physician.
All potential study participants will be provided with a patient information sheet and a consent form describing the study and providing sufficient information and time for the subject to make an informed decision about their participation in the study.
The patient information sheet and the consent form will be submitted to the CEC to be reviewed and approved. The formal consent of a subject, using the approved consent form, will be obtained before the subject is submitted to any study procedure.
The subject should read and consider the statement before signing and dating the informed consent form, and should be given a copy of the signed document. The consent form will also be signed and dated by the investigator (or his designee, e.g. a study nurse or coordinator) and it will be retained as part of the study records.

Participant privacy and confidentiality
The investigator affirms and upholds the principle of the participant's right to privacy and that they shall comply with applicable privacy laws. Especially, anonymity of the participants shall be guaranteed when presenting the data at scientific meetings or publishing them in scientific journals.
Individual subject medical information obtained as a result of this study is considered confidential and disclosure to third parties is prohibited. Subject confidentiality will be further ensured by utilising subject identification code numbers to correspond to treatment data in the computer files.
For data verification purposes, authorised representatives of the Sponsor (-Investigator), a competent authority (e.g. Swissmedic), or an ethics committee may require direct access to parts of the medical records relevant to the study, including participants' medical history.

Early termination of the study
Due to the intervention type of the study and the minimal risk for the patients, we are not foreseeing a possible early termination of the project. However in such case, the premature end or interruption of the research project will be reported to the CEC within 15 days.

Protocol amendments
Substantial amendments are only implemented after approval of the CEC.
Under emergency circumstances, deviations from the protocol to protect the rights, safety and wellbeing of human subjects may proceed without prior approval of the sponsor and the CEC. Such deviations shall be documented and reported to the sponsor and the CEC as soon as possible.
All non-substantial amendments are communicated to the CEC within the Annual Safety Report (ASR).

Importance of readmission
Each year, more than 1.4 million acute adult hospitalizations arise in Switzerland, among which about 12-15% are 30-day readmissions. 1 Transitions of care are indeed high-risk periods for patients, especially for those with multiple chronic diseases. 2 Many complications can occur as patients are discharged from hospitals to the ambulatory setting, which lead to unnecessary distress for the patients and costly readmissions. 3,4 Each hospital admission is moreover associated with a risk of new complications. 5,6 For these reasons, readmissions have been lately of high interest in health policies not only in the United States with the "Patient Protection and Affordable Care Act", 7 but also in Switzerland with the promotion of coordination of care between the different settings being one of the aim of the comprehensive strategy entitled "Gesundheit2020" (Health 2020), a total of 36 measures across all areas of the health system. 8 Some financial incentives have been also put in places in order to promote a reduction in hospital readmission in Switzerland with the new SwissDRG system payment.

Preventability of readmission and interventions
It is estimated that 30% of the readmissions are preventable, 9 and about 50% potentially preventable. 10 In order to avoid these preventable readmissions, several transitional care interventions have shown promising results. [11][12][13][14] It includes telephone follow-up, medication reconciliation, patient education, home visits, etc. A recent systematic review and meta-analysis showed that interventions are overall effective at reducing readmission. 15 However, the interventions that were effective are complex to implement and resources intensive. Furthermore, most of these interventions studies (27 out of 47) have been performed on specific patient populations such as patients with heart failure or chronic lung disease only, which is not transposable to the majority of medical patients who frequently have multiple chronic conditions. Among the 17 studies that included general medical patients, half targeted elderly patients only. We know however that age is not a risk factor for readmission, 16,17 and therefore by targeting specifically only the older patients' interventions may miss the group of patients with an actual high-risk of readmission. Among the 8 remaining randomized control studies in medical patients, 3 studies focused on medication interventions only. But more importantly, none tried to target high-risk patients for readmission. Although all patients deserve a high-quality discharge process that includes, for example, timely hand-off and follow-up appointments, more complex and costly interventions like postdischarge phone calls or home-based transitions coaching should be targeted to the patients who are most likely to benefit. Therefore, efficiently improving transitions in care requires hospitals to target discharge interventions at those patients at higher risk of potentially avoidable readmission. Recently, in a recent randomized controlled trial, Dhalla et al. did try to target high risk patients using the LACE index. 18 However, about 90% of the patients were identified as high-risk using the LACE index. This suggests that the threshold used was too low for the population, or that the LACE score failed to identify the real high-risk patients, which therefore may have contributed to the negative findings.

Identification of patients at high risk for readmission
To identify which are the patients at risk for these readmissions is challenging. It has been showed that clinicians (including nurses, case manager, medical residents, or attendings) are not able to accurately identify which patients are at high risk for readmission. 19 A systematic review in 2011 pointed out the lack of performance and validity of the existing prediction models in readmission. 20 However, we previously developed a prediction model to identify the patients at higher risk for potentially avoidable readmission that has shown good performance in several settings, and is now the most valid and easy to use prediction model (see paragraph 2.2 below).

Gaps in current knowledge
In conclusion, the current gaps in transitional care intervention knowledge are: • In Switzerland, little is known about the quality of the hospital discharge process, and what are the characteristics of the readmissions, and to our best knowledge none are currently underway. • The effectiveness of transitional care interventions to reduce readmissions has shown inconsistency. • We know that the most effective transition care interventions are also the most complex ones, but trials failed to target the medical patients who are most likely to benefit for better effectiveness. To identify those high-risk patients, no trials used one of the best validated, accurate, and easy-touse tools, such as the "HOSPITAL" score (see paragraph 2.2 below). • Most of the trials have been performed on specific patient populations, such as patients with either diabetes, chronic obstructive pulmonary disease (COPD), or heart failure. But the practicability of such interventions is limited in real-life because of the high prevalence of patients with multiple chronic medical conditions. • No interventional studies to reduce readmission have been previously performed in general medical patients in Switzerland. • The rate and type of post hospital-discharge events is not well-known in Switzerland, as well as the quality of the discharge process, and how it varies according to social factors such as urbanrural environments. • No simple tool to predict long-term mortality by hospitalized patients

Identification of high-risk patients for potentially avoidable readmission: the HOSPITAL score
We recently derived a prediction model to identify high-risk medical patients for 30-day potentially avoidable readmission to target transitional care interventions accordingly. This clinical decision tool is named "HOSPITAL" and includes variables readily available before discharge (Table 3.1). 21 The score has been now also validated in an international multicenter study with more than 117,000 patients in 4 different countries, including Switzerland, and showed good performance: an excellent overall accuracy (Brier score 0.08), a good discrimination power (C-statistic of 0.72), and an excellent calibration (p value of Pearson goodness-of-fit test of 0.89). 22 A retrospective validation in 3 academic hospitals in Switzerland (Inselspital, CHUV, and HUG) showed good performance with a C-statistic of 0.68. 23 A prospective validation study at the Fribourg Cantonal Hospital showed a C-statistic of 0.70. 24 We also further assessed the performance of the HOSPITAL score specifically for frequent diseases such as pneumonia, heart failure, COPD exacerbation, and myocardial infarction. 25 The performance remained stable with a C-statistic ranging from 0.68 to 0.71 according to the disease. We finally validated a simplified version of the HOSPITAL score, that includes 6 of the 7 variables (number of procedures being left out), and in which the criteria of "discharge from an oncology unit" has been replaced by "discharge from an oncology unit or diagnosis of active cancer". 26 In the simplified version, the risk of readmission was categorized into 2 groups: unlikely and likely to be readmitted, according to the total number of score points (threshold at 4 points). This simplified version showed the exact same performance as the original HOSPITAL score, and therefore can be considered to be used according to the setting. We further validated the simplified score in the phase 1 of the Target-READ (BASEC 2017-00858). Based on preliminary analysis, the simplified score showed overall similar discriminatory power (C-statistic 0.68) in comparison to the original version of the HOSPITAL score (C-statistic 0.66). Table 3.1: Simplified "HOSPITAL" score to identify the patients at high risk for 30-day readmission * Adapted and validated in Switzerland. 23,24,26 These studies demonstrate the generalizability and validity of the HOSPITAL score. 27 It is currently not only the best validated prediction model for hospital readmission so far, both internationally (USA, Canada, Israel, Denmark) and more locally in Switzerland, but is also an easy-to-use tool. Its performance is higher than the performance of scores previously described in a systematic review. 20 The HOSPITAL score has been also recently shown to be significantly better than the LACE score in a head-to-head comparison. 28 It also performs better than other currently clinical scores used in daily practice, such as those to evaluate the risk of bleeding under oral anticoagulants. 29,30

Context
The main objective of the study is to evaluate the effect of a transitional care intervention prioritized to higher-risk medical patients on the composite of 30-day unplanned readmissions and death. Readmissions to the hospital are frequent and costly, and represent a major health issue. Financial incentives have been put in place in several countries including Switzerland to improve the discharge process and reduce readmission rate. In Switzerland, the promotion of coordination of care between the different settings is one of the aim of The Federal Council's health policy priorities entitled "Gesundheit2020" (Health 2020). 8 While the issue of readmission and quality of discharge process has been identified and incentives started, the methods to reach such aims are still unclear, and innovative high-quality studies with effective interventions are still lacking.

Specific aims and hypothesis
Interventional study among patients at higher-risk for readmission Primary aim: 1) To evaluate in a randomized controlled trial the effect of a transitional care intervention prioritized to higher-risk medical patients (simplified HOSPITAL score ≥4) on the composite of 30-day unplanned readmissions and death.
Hypothesis: by providing a care transition intervention prioritized to the patients at higher-risk for readmission according to a validated clinical decision tool (HOSPITAL score), the quality of transition of care and the post-discharge adverse events such as readmission will be reduced in this population.

Secondary aims:
• To evaluate the effect of a targeted transitional care intervention to higher-risk medical patients on: hospital utilization, medication adherence, patient's perspective on quality of transition of care, time to readmission or death (whatever comes first). • To evaluate the effect of a targeted transitional care intervention on the primary outcome in subgroup population of patients with intermediate (simplified HOSPITAL score of 4 or 5) compared to high risk of readmission (simplified HOSPITAL score ≥6), or who suffer from common chronic diseases (diabetes, chronic heart failure, or COPD).

Attribute Value Points
Low Hemoglobin level at discharge (< 120 g/l) yes 1 Discharge from an Oncology service or active cancer diagnosis* yes 2 Low Sodium level at discharge ( < 135 mmol/l) yes 1 Index admission Type: urgent or emergent (nonelective) yes 1 Number of hospital Admission(s) during the previous year 0-1 0 Length of stay ≥ 8 days* yes 2

Investigational Product (treatment, device) and Indication
The study is not testing any product. The intervention consist of patient education, patient information, and follow-up planning.
The intervention group will receive a standardized transition care intervention (called "TARGET intervention") by a trained nurse as described below (pre-discharge component + 2 post-discharge follow-up phone calls at day 3 and day 14). The pre-discharge component includes patient information, medication reconciliation, patient education, and planning of a first post-discharge primary care physician visit (Table 3.2a). Two follow-up phone calls will be made by a nurse, at D3 (+/-1 day) and D14 (+/-1 day), and will include a reinforcement of the patient education, the verification of timely primary care physician visit and that a timely discharge summary is sent to the primary care physician, and review with the patient of the medication list with assessment of potential adverse drug events (Table 3.2b). Adherence to the intervention will be collected by collecting which of the 3 components the patient indeed received (i.e. pre-discharge component, first follow-up phone call, second follow-up phone call). The contact information of the patient will be necessary in order to reach the patient for these 2 follow-up phone calls. -Provide standard patient information leaflet of the following disease when relevant: heart failure, coronary heart disease, atrial fibrillation, peripheral artery disease, thromboembolism, stroke, chronic obstructive pulmonary disease, gastrointestinal bleeding, cirrhosis, renal failure, diabetes.
-Understanding what to do in the occurrence of new danger signs based on their diseases, and what to do if they occur Medication reconciliation -Compare medication list at admission and discharge (medication reconciliation), ask patient about any missing medication from home. Inform treating hospital physician if any discrepancy found.
-Review medication's purpose with the patient. Hand out the discharge medication list to the patient. Teach back.
-If the patient has an aerosol therapy for COPD, ask the patient to show how to use it.
-Provide general explanation about the importance to follow the medication therapy at home as prescribed.
General patient education about lifestyle -Patient education about general health recommendation: stop smoking and/or alcohol, healthy food, regular physical activity, etc.
-Hand out the patient information leaflet about general lifestyle recommendations.
Dependence level -Check the Activity of Daily living of the patient (using the Katz score). If the score is >6 points inform the medical team.
-Ask about possible support at home by next of skin, or nurse at home.
Follow-up care planning -Organize medical support: identify the primary care physician, and make appointment for a follow-up visit as soon as possible (ideally within 7 days following discharge). If the patient doesn't have a primary care physician, suggest a few primary care physicians. Hand out an appointment sheet with: name and phone number of the primary care physician.
-Insure a timely discharge summary sent to the primary care physician shortly after hospital discharge. Additionally, to the discharge summary, the primary care physician will be notified by a separate sheet on the higher risk of readmission identified by her/his patient.
Barriers -Review any barriers to a safe discharge, including patient's ability to carry out the discharge plan, willingness and ability to pick up their medications and to take them; and willingness and ability to keep their follow-up appointments. Take actions to reduce these barriers and to communicate these barriers to the medical team.
-Give the opportunity to the patient to ask for questions, and rely them to the medical team if needed. Table 3.2b: Post-discharge follow-up phone call of the TARGET intervention (day 3 and 14) 1. Health condition: Ask whether condition improved, worsened, or recurred. Ask about any reduced walking distance that is possible since hospital discharge. Ask about capacity to prepare meals. Ask for weight trend over the last days/weeks, blood pressure, glucose monitoring data when relevant.

2.
Primary care physician visits: Check when was or is scheduled an appointment to the primary care physician. In case of health condition or medication issue, the patient will be encouraged to contact his primary care physician.

Medications:
Review list of current medications and ask about any discrepancy. Monitor significant adverse events for high alert medicines (anticoagulants, narcotics and opiates, insulin, and sedatives because they represent areas of greatest harm and greatest opportunity for improvement). The most common types of harm associated with these medications include hypotension, bleeding, hypoglycemia, delirium, lethargy, and oversedation. Underline the importance of following the medication as prescribed, and to seek for primary care physician advice in case of any question or side effects. 4. Patient education: Refresh about general health recommendations (importance of medication compliance, healthy food, regular physical activity, etc.), and more specific but simple recommendation about the diseases of the patients if part of the patient info leaflets (heart failure, coronary heart disease, atrial fibrillation, peripheral artery disease, thromboembolism, stroke, chronic obstructive pulmonary disease, gastrointestinal bleeding, cirrhosis, renal failure, diabetes).

Preclinical Evidence
Not applicable.

Clinical Evidence to Date
Not applicable. See point 3.1 Background and rationale for existing evidence about interventions to reduce hospital readmission (under "Preventability of readmission and interventions)."

Dose Rationale / Medical Device: Rationale for the intended purpose in study (pre-market MD)
Not applicable.

Explanation for choice of comparator (or placebo)
The comparator will be usual care. The purpose of this study is to evaluate effects on clinical outcomes of the above described intervention in comparison to standard care. Therefore, we compare a standardize intervention to the local usual care. A hospital would start such intervention when it has been shown to improve care in comparison to what is currently done. Any type of "shame" intervention may influence in an undetermined way the risk of readmission, and it will be then difficult to answer the study question.

Potential benefits
Readmissions to the hospital are frequent and costly, and represent a major health issue. Although incentives have been already put in place in Switzerland (as in the US), few has been concretely made in the field to improve the transition of care of the patients in Switzerland. Transition of care between the hospital and the ambulant setting is a known period at risk for adverse events, and has recognized potential for quality improvement. In Switzerland, the promotion of coordination of care between the different settings is one of the aim of The Federal Council's health policy priorities entitled "Gesundheit2020" (Health 2020). While the issue of readmission and quality of discharge process has been identified, the methods to reach such aims are still unclear, and innovative high-quality studies with effective interventions are still lacking. Efficiently improving transitions in care requires hospitals to target discharge interventions at those patients at high-risk of potentially avoidable readmission. This study aims primarily to evaluate the effect of a transitional care intervention prioritized to higher-risk medical patients according to the simplified HOSPITAL score on the composite of 30-day unplanned readmissions and death.
This research project is innovative and promising in several ways to help address many gaps in current knowledge. 1) we will include all medical patients, regardless of the cause of admission, and will include many patients with multiple chronic conditions as in real-life, which will provide more generalizable results; 2) we will target the higher risk patients using one of the most accurate and validated tool, the HOSPITAL score in order to better allocate transition care resources and to improve effectiveness. This score has been now widely validated, including in Switzerland; 3) we will perform the first randomized controlled trial to test a transition care intervention in medical patients in Switzerland; 4) finally, we will look at the patient's perspective about transition care, and also the causes of readmission in Switzerland, but also evaluate how the readmissions varies according to rural vs urban environment, and according to the type of health insurance.
The main implications of the study are the following: 1) We may demonstrate the efficacy in Switzerland of a set of interventions that separately showed some promising results in other countries (geographical transportability). Although information from existing studies in other countries might be valuable, our health care system and our population is different, and effectiveness of such intervention must be tested before to be widely implemented.
One of the innovative part of the study is to target intervention to the patients at higher risk for readmission using a validated tool. 2) We will validate the clinical impact of the simplified HOSPITAL score, i.e. its effectiveness to effectively help reducing the readmission in high-risk patients when implemented together with a discharge intervention. This would give to the HOSPITAL score the highest degree of evidence (level I out of IV) in validation of prediction models. 31 3) To reduce post hospital-discharge adverse events, such as 30-day readmission or death, in a large population of medical patients, who have typically multiple chronic conditions. 4) To better allocate of transition care interventions to those who are most likely to benefit, i.e. the patients at higher risk for readmission, using a validated tool. 5) To improve the discharge process experienced by the patients, by providing a better follow-up after discharge and a better coordination between the hospital and the primary care settings for better treatment continuity.
All the main actors of the health care system may benefit from the study findings: 1) The patients may benefit from a better quality of the transition of care and a reduced risk of post-discharge adverse events such as readmissions; 2) The hospitals may use the intervention in order to improve their quality of care process, and reduce the financial penalties due to costs of readmissions; 3) The primary care providers may be helped to avoid losing contact with their high risk patients in the post-hospital discharge period, in order for them to intervene before an adverse event arise; 4) The decision makers in Switzerland will be informed on the quality of transition of care, and see the potential benefits of a transition of care intervention between settings, with a better care coordination.

Potential risks
All the component of the intervention are known procedures already provided sometimes to the patients.
The study here is grouping in a standard package the most promising interventions together. There is therefore no known risks of such intervention. The main innovative part of the study is to target a standard package of patient support to the patients who are most likely to benefit, in order to make the intervention overall more successful. In conclusion, we are not expecting any risks for the patients, other than the confidentiality breach inerrant to any clinical study. To avoid any confidentiality breach, the data will be collected directly and stored in a web-secured database (Redcap). Each personal study will be provided with a personal access and identifiable login to Redcap. The signed informed consents and the discharge letters from the hospital will be stored in a binder and securely locked in a closet closed with a key at each hospital.

Justification of choice of study population
The study population will consist of adult inpatients discharged alive home (or nursing home). Only patients discharged from a medical department will be recruited, since the simplified HOSPITAL score used to identify the higher risk patient for readmission has been developed and validated for this population only. Among those, the project is to focus only on the patients with the highest risk of 30-day readmission based on the HOSPITAL score. The HOSPITAL score and its simplified version have been widely and internationally validated and are currently the most valid tool to identify the patients at high risk of readmission. Potential patients will be identified through active screening by dedicated trial staff in each centre.
The risk of readmission exist only if a patient has been: 1) hospitalized; 2) discharged alive. This explain the choice of the study population. Each year, more than 1.4 million acute adult hospitalizations arise in Switzerland, among which about 12-15% are 30-day readmissions. 32 Many complications can occur as patients are discharged from hospitals to the ambulatory setting, which lead to unnecessary distress for the patients and costly readmissions. 3,4 Because patients who are transferred to another acute care facility, or a rehabilitation center have are in a structure with a certain level of medical care available, these patients will have their risk of readmission artificially modified. Moreover, the intervention aims mainly to involve the patients in their own care and health recovery, which obviously is not something that can be done or the effect measured as long as they are in a medical facility. Therefore, patients transferred to another acute care facility or rehabilitation center will be excluded from the study.
Due to the nature of the intervention that emphasize in great part the involvement of the patients in their health care, patients who have cognitive troubles such as dementia will not be included. Because the intervention is about patient education the nurse must be able to communicate in her own language to the patient. Therefore only patient speaking French will be included at the site of Fribourg, Lausanne and Neuchâtel, and patients speaking French or German at the site of Biel. A translator is not feasible because the time between the screening of the patient (medical decision of discharge) and the discharge of the patients is most often limited to a few hours and therefore doesn't allow to contact external translators.

Overall Objective
The overall objective of this RCT is to evaluate whether a standardized intervention targeted to the patients at higher risk of readmission will lead to a reduction in the risk of 30-day unplanned readmission or death.

Primary Objective
Primary aim: 1) To evaluate in a randomized controlled trial the effect of a transitional care intervention prioritized to higher-risk medical patients on the composite of 30-day unplanned readmissions and death.
Hypothesis: by providing a care transition intervention prioritized to the patients at higher-risk for readmission according to a validated clinical decision tool (simplified HOSPITAL score), the quality of transition of care and the post-discharge adverse events such as readmission will be reduced in this population.

Secondary Objectives
Secondary aims: • To evaluate the effect of a targeted transitional care intervention to higher-risk medical patients on: hospital utilization, medication adherence, patient's perspective on quality of transition of care, time to readmission or death (whatever comes first). • To evaluate the effect of a targeted transitional care intervention on the primary outcome in subgroup population of patients with intermediate (simplified HOSPITAL score of 4 or 5) compared to high risk of readmission (simplified HOSPITAL score ≥6), or who suffer from common chronic diseases (diabetes, chronic heart failure, or COPD).

Safety Objectives
Not applicable.

Primary Outcome
The primary study outcome will be the number of patients who have an unplanned readmission or die within 30 days after discharge.
The primary outcome will be the composite of first unplanned readmission or death within 30 days after discharge of index admission (inclusion time). The unplanned readmission will be defined as a nonelective hospitalization that happens within 30 days after discharge from the index hospital discharge to any division of any acute care hospital (i.e. not only the hospital where the patient was originally discharged). Elective hospitalization will be defined as a non-urgent hospitalization scheduled at least 1 day prior to the admission day. Death will be defined as any death occurring within 30 days after discharge. We will further collect the cause of readmission or death. Death is the worst outcome after a hospital discharge. It is common practice in this domain to include both readmission and death as a composite outcome, because patients who died may well have been readmitted if they hadn't died. 18,[33][34][35] In order to collect thoroughly the primary outcome and avoid missing outcomes, the outcomes will be assessed through 5 steps (see below under section 9.2.1).

Secondary Outcomes
Secondary outcomes: • Individual components of the primary composite outcome: 30-day unplanned readmission, or 30day mortality. Medication adherence will be evaluated using the Morisky 4-item Medication Adherence Scale (MMAS-4). [36][37][38] This instrument measures patients' self-reported adherence to their medications over the last 4 weeks through 4 simple questions and has been widely validated. These questions are the following: 1) Do you ever forget to take your medicine? 2) Are you careless at times about taking your medicine? 3) When you feel better do you sometimes stop taking your medicine? 4) Sometimes, if you feel worse when you take your medicine, do you stop taking it? Each question can be answered by "yes" (1 point) or "no" (0 point). A patient is considered to have a high adherence to the medications if the total score is zero, a medium adherence for score between 1 and 2, and a low adherence for score of 3 or 4. We will estimate the proportion of patients with high adherence, i.e. with a score of 0. In order to collect thoroughly the outcomes and avoid missing data, the secondary outcomes will be assessed at the same time and through the 5 same steps as the primary outcome. However, the medication adherence (MMAS-4) and patient's perspective (CTM-3) will only be collected from the patient her/himself.

Other Outcomes of Interest
Not applicable.

Safety Outcomes
Not applicable.

General study design and justification of design
The Target-READ is a national multicenter single-blinded randomized controlled trial design among adult inpatients discharged home from the medical department of one of the 4 participating centers (Biel, Fribourg, Lausanne, Neuchâtel).
All medical patients will be screened for eligibility as soon as their hospital discharge is planned. After a baseline evaluation for all inclusion and exclusion criteria, each enrolled participant of all sites will be randomized centrally to the intervention or usual care group using a computer-generated randomization list (allocation ratio 1:1) stratified by discharge site and by readmission risk category according to the simplified HOSPITAL score (intermediate vs high risk, i.e. a HOPSITAL score of 4 or 5 vs ≥6), with randomly varying block sizes of 2, 4, and 6.
Given the nature of the intervention, it will be not possible to blind patients and study nurses. However, treatment allocation in the database will be coded (1/2) and the study nurses collecting the outcomes or working on data cleaning and the statistician performing the analysis will be blinded to the group allocation. Patients will be specifically asked not to mention their group allocation in the final assessment. The statistician will be using the coded variable to prepare a blinded analysis report. Only after discussion of the results with the PI and closing of the database, the code will be broken and a final unblinded analysis report will be generated.
The number of participants to be included will be at least 690 per arm (see below under point 7.1 and 11.2 for more details).
The study will follow the CONSORT statement, 40 and the intervention will be described according to the TIDieR checklist and guide. 41

Methods of minimising bias
Selection bias is considered the most relevant biases in this trial. Screening of all consecutive adult patients about to be discharged home will be performed to avoid selection bias.
Patients will be centrally randomized into the intervention or control group (see section 6.1 above). Given the nature of the intervention, it will be not possible to blind patients and study nurses (see section 6.1 for details).

Randomisation
See section 6.1 for details about randomization.

Blinding procedures
See section 6.1 for details about blinding procedures.

Other methods of minimising bias
A limitation of the study design is the potential for selection bias. In particular, patients who are refusing to participate might have a different risk of readmission than the patients who are willing to participate. To avoid a selection bias, a careful monitoring of recruitment is essential. Not to be able to compare some basic characteristics between the patients who refused to participate and those who agreed to participate is a limitation that would reduce the scientific validity of the entire study.
Some baseline characteristics are thus recorded in the database for all ca. 18'000 subjects of the target population irrespective of whether they will participate in the study or not. This includes the following data: subject ID, screening date, first three inclusion criteria, age. If a subject fulfils the first three inclusion criteria, the following additional data will be recorded: single variables of the HOSPITAL score (i.e. hemoglobin and sodium levels, elective vs. non-elective admission, cancer diagnosis, number of previous hospitalizations in the last 12 months, length of stay). If, according to the HOSPITAL score, the subject is at high risk for readmission, in addition, the exclusion criteria will be recorded. Such data will not be collected from subjects with documented refusal for further use of their data.
Recording these (routine) data in the database is considered as "further use" according to Article 33 (HRA).

Unblinding Procedures (Code break)
See section 6.1 for details about unblinding procedures.

Eligibility criteria
Inclusion criteria: • Adult patients planned to be discharged home or nursing home.
• Discharge alive from a medical department of one of the four participating centers (Biel, Fribourg, Lausanne, Neuchâtel). • Hospital stay of at least 24 hours.
• Patient at higher risk of 30-day readmission based on the simplified HOSPITAL score.
Exclusion criteria: • Previous enrolment into the current study (Target-READ phase 2).
• Not living in the country in the next 30 days.
• No phone to be reached at.
• Not speaking French or German (depending on the site).
• Refusal to participate, or unable to give consent (including cognitive troubles defined as a disorientation in time and space).

Recruitment and screening
All patients announced to be discharged soon by the medical team of each participating center will be screened for eligibility. Screening logs will be kept at each site and basic coded information (gender, age, risk of readmission based on HOSPITAL score, etc.) regarding each screened patient will be entered in the database. Variable required to calculate the HOSPITAL score (hemoglobin level, sodium level, diagnosis of cancer, number of hospitalization in the last 12 months, mode of admission, and length of stay) will be entered in the database in order to automatically and therefore reliably calculate the HOSPITAL score. Because of the great expenditure approaching 18'000 subjects for consent to collect these data, and given the importance collecting these data to minimize any selection bias (see section 6.2.3), referring to HRA Art. 34, we will refrain from asking these subjects for consent to collect these data. Such data will not be collected from patients with documented refusal collecting such Screening data will be evaluated on a regular basis by an independent person to detect possible selection bias.
Only patients at high risk for readmission will be included. The determination of the risk of readmission will be based on the valid prediction model "HOSPITAL" score. This score requires the collection of 7 variables (see section 3.1, Table 3.1). All the information needed to calculate the score will be drawn directly from the electronic health record, no additional measurement outside usual care will be made (the score will be calculated only with available usual care data available at time of screening). No compensation nor payment will be made to the participants.

Assignment to study groups
See section 6.1 for randomization.

Criteria for withdrawal / discontinuation of participants
Patients must be withdrawn from the study for any of the following reasons: • Patient refuses to continue • Patient is lost to follow-up Patients may be withdrawn from the study at the discretion of the Investigator based on his or her clinical judgment, but only in exceptional situations. The eCRF should be completed accordingly, providing the primary reason for discontinuing the study prematurely.

Identity of Investigational Products (treatment / medical device)
The intervention will take place in the initial hospital admission (Index hospitalization) defined as the hospitalization in which the patient inclusion took place, and will be pursued after patient hospital discharge. The TARGET intervention has 2 components, one during the hospitalization before the hospital discharge (pre-discharge component), and one after the hospital discharge (post-discharge component).
The pre-discharge component of the TARGET intervention consists of the six following steps: Every study nurses will be receiving the same training for the TARGET intervention. The training will be led by the sponsor, individually and in group, as well as theoretically and practically.
Contact information including name, address and phone number of the patient will be necessary to contact the patient for the 2 follow-up phone call that are part of the intervention, and to collect the outcome at 30 days.

Experimental Intervention (treatment / medical device)
Not applicable.

Control Intervention (standard/routine/comparator treatment / medical device)
Participants randomized in the control group will receive information and organization of the postdischarge care in accordance with local usual care.

Packaging, Labelling and Supply (re-supply)
Not applicable.

Storage Conditions
Not applicable.

Experimental Intervention
The intervention will be conducted before hospital discharge. Inclusion criteria regarding patient availability have to be met in order to ensure feasibility of the intervention. The whole interventional process is described under section 3.2 above.

Control Intervention
Patients being assigned to the control arm will be treated in accordance with standard care. They will receive in addition to the patient information consent an information sheet stating that they have been selected in the control group.

Dose / Device modifications
Not applicable.

Compliance with study intervention
During the intervention the discussion with the patient is expected to enhance compliance. We will measure which part of the intervention each participant were exposed to (pre-discharge component, 1 st and 2 nd post-discharge components).

Data Collection and Follow-up for withdrawn participants
A study patient can withdraw his/her consent to participate in the trial at any time and without any declaration of reason. Data collected until the time point of withdrawal will remain in the database and will be used for analysis.

Trial specific preventive measures
Not applicable.

Study Drug / Medical Device Accountability
Not applicable.

Return or Destruction of Study Drug / Medical Device
Not applicable.

Study flow chart(s) / table of study procedures and assessments
Index hospitalization 1) Follow-up 2) (post-discharge) 2) Assessment by phone call Laboratory parameters: the simplified "HOSPITAL" score includes 2 lab values (haemoglobin and sodium level). These values are part of routine care. Only available data in the patient records will be used. The study participants will never get an extra blood draw solely for the purpose of the study to measure these parameters if they would not be available in the patient records at time of discharge.

Assessment of primary outcome
The primary outcome will be collected at least 30 days after inclusion via phone interview by a blinded study nurse outside the participating site. The blinded study nurse will have only access to: 1) the contact information from an Excel file; 2) the outcome collection form in Redcap. In that way the study nurse will be unaware of the group allocation of the participant. The participant will be asked in the patient informed consent and at the beginning of the phone call to not divulgate her/his group allocation.
In order to collect thoroughly the primary outcome and avoid missing outcomes, the outcomes will be assessed through 5 steps: 1. Phone interview with the study participant (3 phone calls attempts on at least 2 different days).
2. If study participant her/himself is not reachable: phone interview with a next of kin of the study participant (2 phone calls attempts on at least 2 different days). 3. If study participant her/himself or next of kin are not reachable: phone interview with the primary care physician of the study participant. 4. If information is still missing, the nurse will call the study nurse of the local site to look for any outcome event that would be reported in the electronic health system. 5. In any case of a readmission: the medical report of the readmission will be collected to assess the exact place, time, length, and cause of hospital readmission.

Assessment of secondary outcomes
The secondary outcomes will be collected by phone interview at the same time of the primary outcome and with the same procedure.
Number of unplanned readmission: if the patient, the next of kin or the primary physician report a readmission, the date of readmission and the length of stay will be asked. In any case, the medical report of the readmission will be collected to double check the validity of the information and obtain the causes of readmission.
Number of days of hospitalization within 30 days: if the patient, the next of kin or the primary physician report one or more readmission, the total number of days spent at the hospital will be measured through the electronic health system for the participating site, or by calling the hospital if a readmission occurred outside the 4 participating sites.
Diagnoses at readmission or cause of death: if the patient, the next of kin or the primary physician report a readmission, the medical report will be collected. If a death is identified, the medical report will be collected if the death occurred at the hospital, otherwise the cause will be obtain from the primary care physician, or the certificate of death.
Number of emergency visits within 30 days: the patient, the next of kin or the primary physician will be asked about the number of visit to any emergency room (any clinic or hospital).
Number of primary care physician visits: the patient, the next of kin or the primary physician will be asked about the number of visit to any treating physician (generalist of specialist). Costs of readmission: the costs of the readmission will be collected at the hospital site.

Assessment of other outcomes of interest
Not applicable.

Assessment of safety outcomes
Not applicable.

Adverse events
Not applicable.

Laboratory parameters
Not applicable.

Vital signs
Not applicable.

Assessments in participants who prematurely stop the study
Each participant will be asked for the reason of premature withdrawn of the study participation.

Procedures at each visit 9.3.1 Screening (Visit 0)
The following activities and assessments have to be performed at the screening visit (before inclusion): • Assess eligibility • Document demographics • Assess the risk of 30-day readmission based on the simplified HOSPITAL score

Visit 1: Patient inclusion and pre-discharge intervention
The following activities and assessments have to be performed at the first visit: • Confirm the eligibility • Obtain the informed consent • Randomize the participant • Obtain contact information needed for visits 2, 3, and 4 (collected separately from all other variables). • Obtain the baseline characteristics • Document medical history • Document medication • Obtain level of autonomy if available • Provide the participant with the pre-discharge intervention (include mainly organization of a follow-up visit by the primary care physician, patient education). • Provide the participant with the disease info leaflet when relevant (available for the following diseases: heart failure, coronary heart disease, atrial fibrillation, peripheral artery disease, thromboembolism, stroke, chronic obstructive pulmonary disease, gastrointestinal bleeding, cirrhosis, renal failure, diabetes).

Visit 2: Post-discharge follow-up phone call 1 (Day 3 ±1)
The following activities and assessments have to be performed at the second visit only for the participants allocated to the intervention group: • Ask the patient about her/his general health condition in comparison to hospital discharge.
• Compare the current medication list with the medication list from hospital discharge.
• Document any suspicion of frequent drug adverse events such as: hypotension, bleeding, hypoglycemia, delirium, lethargy, and oversedation. • Re-inforce patient education and importance of the treatment adherence.
• Assess if a follow-up visit by the primary care physician happened.

Visit 3: Post-discharge follow-up phone call 2 (Day 14 ±1)
The following activities and assessments have to be performed at the third visit (same as visit 2) only for the participants allocated to the intervention group: • Ask the patient about her/his general health condition in comparison to hospital discharge.
• Compare the current medication list with the medication list from hospital discharge.
• Document any suspicion of frequent drug adverse events such as: hypotension, bleeding, hypoglycemia, delirium, lethargy, and oversedation. • Re-inforce patient education and importance of the treatment adherence.
• Assess if a follow-up visit by the primary care physician happened. The following activities and assessments have to be performed at the fourth visit for all included patients:

Visit 4: Outcome collection (Day 30 up to day 45)
• Document any 30-day unplanned readmission or death • Document the number of primary care physician visits, the number of emergency department visits.
• Conduct questionnaire for medication adherence (4-MMAS), and patient's perspective on the quality and effectiveness of the transition or discharge from an acute care hospital (CTM-3).
• Collect discharge letter (causes of readmission) and costs of hospitalization of any first unplanned 30-day readmission.

Drug studies
Not applicable.

Medical Device Category C studies
Not applicable.

Medical Device Category A studies
Not applicable.

Other clinical studies
The Sponsor's SOPs provide more detail on safety reporting.
During the entire duration of the study, all adverse events will be fully investigated and assessed for seriousness and potential relationship with the TARGET intervention. If a serious adverse event (SAEs) occurs and it cannot be excluded that the event is attributable to the intervention, it is fully documented in source documents and in an SAE report form. Study duration encompasses the time from when the participant signs the informed consent form until the last protocol-specific procedure has been completed.

Definition and assessment of serious adverse events and other safety related events
An Adverse Event (AE) is any untoward medical occurrence in a clinical investigation participant administered an intervention and which does not necessarily have a causal relationship with the intervention. An AE can therefore be any unfavourable and unintended sign (including an abnormal laboratory finding), symptom, or disease temporally associated with the intervention, whether or not related to the intervention. [ICH E6 1.

2, adapted]
A Serious Adverse Event (SAE) is classified as any untoward medical occurrence that: • requires inpatient treatment not envisaged in the protocol or extends a current hospital stay; • result in permanent or significant incapacity or disability; • is life-threatening or results in death; or • causes a congenital anomaly or birth defect.
In addition, important medical events that may not be immediately life-threatening or result in death, or require hospitalisation, but may jeopardise the patient or may require intervention to prevent one of the other outcomes listed above should also usually be considered serious.
SAEs should be followed until resolution or stabilisation. Participants with ongoing SAEs at study termination will be further followed up until recovery or until stabilisation of the disease after termination.

Assessment of Causality
Both Investigator and Sponsor make a causality assessment of the event to the TARGET intervention. A causal relationship towards the intervention should be rated as follows: • Not related: The event is definitely not associated with the intervention; a relationship can be ruled out. • Possibly related: The relationship between the intervention and the event is possible, but other causes cannot definitely be ruled out. • Related: The event is definitely associated with the intervention. Assessment of Severity SAE severity will be classified as either mild, moderate, or severe.

Reporting of SAEs
If, in the course of the trial, an SAE occurs and it cannot be excluded that the event is attributable to the intervention under investigation, the event must be reported immediately and within a maximum of 24 hours of learning of its occurrence to the Sponsor using the SAE report form provided by the Sponsor.
The form should be sent via email to the following address: Jacques.donze@insel.ch The Sponsor will re-evaluate the SAE and return the form to the site.
All of the above events are reported to the coordinating and local Ethics committee(s) (as applicable) within 15 days.

Reporting of Safety Signals
All suspected new risks and events that require safety-related measures, i.e. so called safety signals, must be reported to the Sponsor within 24 hours. The Sponsor must immediately inform all participating Investigators about all safety signals.
The Sponsor must report the safety signals to the coordinating and local Ethics committee(s) (as applicable) within 7 days.

Reporting and Handling of Pregnancies
Any pregnancy must be reported to the sponsor at once.

Periodic reporting of safety
An annual safety report is submitted by the Sponsor once a year to the responsible Ethics committee and is also provided to all participating Investigators.

Follow up of (Serious) Adverse Events
Participants with any reported ongoing SAE at the last scheduled study contact will be followed until resolution of the event or a stabilized condition of the subject has been achieved or until the subject is lost to follow-up.

STATISTICAL METHODS
The statistical analysis of the trial will be done at Clinical Trials Unit (CTU) Bern by a statistician blinded to the allocation. This process is defined in standard operating procedures. After start of the trial but before recruitment ends, a statistical analysis plan will be written. The plan will determine all necessary data preparation steps (e.g. additional validations, generation of new variables), definitions (e.g. analysis sets), and statistical analyses (e.g. models and outputs such as tables and graphs). All statistical analyses will be presented as effect measure plus 95% confidence interval (CI). A significance level of 5% will be used.

Hypothesis
The statistical null hypothesis is that there is no difference in the proportion of 30-day unplanned hospital readmissions and death between the intervention and control group. The alternative hypothesis is that there is a difference in the proportion of unplanned hospital readmissions between both groups. Therefore, the statistical framework within which this trial is conducted is that of a superiority trial.

Determination of Sample Size
The intervention phase (randomized controlled trial) will be restricted to the eligible patients who are at intermediate to high risk of 30-day readmission according to the simplified HOSPITAL score, i.e. ≥4/13 points (Figure 3.1). We estimate to have around 18,000 patients discharged during the 20-month study period, 30% of which (5,400) will be at higher-risk for a 30-day readmission or death.
Because we will target patients at higher risk of readmission, we hypothesize that the intervention could reduce readmission by 25%, i.e. more than the 18% reduction found in a recent meta-analysis where patients were mostly not at high risk for readmission. 15 Based on previous findings, the expected 30day readmission and death rate for patients at intermediate or high risk according to the simplified HOSPITAL score is around 27%. Allowing for 10% loss to follow-up, we determine that we will need 1,380 patients for the study to have 80% power, i.e. 690 in each arm. According to a conservative estimation, the sample size is expected to be reached within the study time frame.

Statistical criteria of termination of trial
There is no interim analysis planned, i.e. there are no stopping rules on the individual or trial level.

Datasets to be analysed, analysis populations
The main analysis set will include all randomized patients according to the intention-to-treat principle. A secondary analysis set will include patients treated per-protocol, i.e. all patients that received the predischarge component plus at least one of the 2 post-discharge follow-up phone calls of the intervention and completed the study according to protocol. A third analysis set will exclude all patients with readmissions within the first 24 hours.

Primary Analysis
The primary analysis will be a comparison between two groups according to the intention-to-treat principle, i.e. all randomized patients will be analyzed in the allocated group regardless of any protocol violations such as cross-overs or early discontinuations of the intervention.
The primary outcome will be reported as a proportion of patients in each group and compared between the groups by a risk difference with 95% CI and a chi squared test. Thirty-day deaths, dichotomized patient satisfaction and medication adherence will be analyzed in the same way. The proportion of 30 day unplanned readmissions in each group will be estimated using the cumulative incidence function with death as competing event and compared by a risk difference with 95% CI. The competing events 30-day deaths with and without previous readmission will also be estimated using the cumulative incidence function and compared by a risk difference with 95% CI. Count outcomes (number of days of hospitalization, number of emergency room visits and primary care provider visits) will be analyzed by Poisson or negative binomial regression with the time of observation as an offset. Effects will be reported as incidence rate ratios with 95% CI. Time to readmission or death (whatever comes first) will be analyzed by Cox regression and reported as hazard ratio with 95% CI.

Secondary Analyses
For all outcomes a secondary analysis will be done on the per-protocol set, i.e. all patients that received the pre-discharge component plus at least one of the 2 post-discharge follow-up phone calls of the intervention and completed the study according to protocol.
In a sensitivity analysis for all outcomes, all patients with readmissions within the first 24 hours will be excluded.
Furthermore, for the primary outcome subgroups defined by 1) risk for readmission (HOSPITAL score of 5 and 6 vs ≥7), and whether the patient suffers from 2) diabetes, 3) chronic heart failure, or 4) COPD will be analyzed using logistic regression with treatment group, subgroup indicator and their interaction as covariates. An odds ratio with 95% confidence interval will be reported for each subgroup together with a p-value for interaction.

Interim analyses
There is no interim analysis planned.

Safety analysis
Since the primary outcome is related to the SAE we will not run any specific analysis.

Deviation(s) from the original statistical plan
The SAP is a version-controlled document which allows to tracking of all changes to the statistical analysis. Any post-hoc changes will be transparently reported in all publications.

Handling of missing data and drop-outs
Missing data may occur due to drop-outs or deaths. The former will lead the absence of all outcome information as outcomes are only assessed once and will be accounted for by multiple imputation based on all available baseline information. The latter will not lead to missing data for the primary outcome, 30-day deaths and time to primary outcome. For 30-day unplanned readmission, death is a competing event and the cumulative incidence at 30 days will be used. Count outcomes will be handled by including the observation time as offset in the Poisson regression. For patient satisfaction and medication adherence multiple imputations will be used. In a sensitivity analysis, the worst possible outcome will be assigned to deaths (i.e. low adherence with MMAS-4 score of 4 and a negative respond to all question of the CTM-3).

Data handling and record keeping / archiving
The Investigators will maintain appropriate medical and research records for this trial, in compliance regulatory and institutional requirements for the protection of confidentiality of subjects. The Principal Investigator, Sub-Investigator, and Clinical Research Nurses or Coordinators will have access to the records.
The Principal Investigators will permit authorized representatives of the Co-Sponsor and regulatory agencies to examine (and when required by applicable law, to copy) clinical records for the purposes of quality assurance reviews, audits, and evaluation of the study safety and progress.

Case Report Forms
For each enrolled trial participant, an eCRF will be maintained. All data will be entered in the eCRF at the local site. eCRFs must be kept current to reflect subject status at each phase during the course of the trial and are part of the central acquisition by the CTU Bern. Coded identification for each patient will be as follow: [trial-ID]-[site-no]-consecutive number. Authorized to enter data into the eCRF are the local trial team staff according to the authorization list. Authorized persons will be identified by their usernames. The local PI is responsible for proper training and instruction of the trial personnel filling data into the eCRF.

Specification of source documents
Source data must be available at the site to document the existence of the trial participants. The following documents are source documents and can only be found at the local trial site: • Demographics and administrative data • completed questionnaires during hospital stay as well as follow-up • Informed consent form • Patient screening, enrolment and randomization log • Patient records and medical history including reports from treating physician • Laboratory reports • Documentation of the patients medical treatment • Nurse report including autonomy level of the patient if available.
• Results of relevant examinations • Any other relevant record to document SAEs

Record keeping / archiving
All trial data must be archived at the participating site for a minimum of 10 years or according to national requirements after trial termination or premature termination of the clinical trial. In addition, the electronic data will be kept for the mandatory period.

Data management
Describe plans for data entry, coding, security, and storage, including any related processes to promote data quality (e.g., double data entry; range checks for data values). In case electronic data capture systems are used, this chapter shall include a description of procedures for verification, validation and securing the database.
Reference to where details of data management procedures can be found, if not in the protocol.

Data Management System
The CRFs in this trial are implemented electronically using a dedicated electronic data capturing (EDC) system (REDCap, https://www.project-redcap.org/). The EDC system is activated for the trial only after successfully passing a formal test procedure. All data entered in the CRFs are stored on a Linux server in a dedicated mySQL database.
Responsibility for hosting the EDC system and the database lies with CTU Bern.

Data security, access and back-up
The server hosting the EDC system and the database is kept in a locked server-room. Only the system administrators have direct access to the server and back-up tapes. A role concept with personal passwords (site investigator, statistician, monitor, administrator etc.) regulates permission for each user to use the system and database as he/she requires.
All data entered into the CRFs are transferred to the database using Secure Sockets Layer (SSL) encryption. Each data point has attributes attached to it identifying the user who entered it with the exact time and date. Retrospective alterations of data in the database are recorded in an audit table. Time, table, data field and altered value, and the person are recorded (audit trail).
A multi-level back-up system is implemented. Back-ups of the whole system including the database are run internally several times per day and on external tapes once a day. The back-up tapes are stored in a secure place in a different building.
The personally identifiable information (contact information of the participants) will be stored separately from all other study variables. Contact information are necessary in order to contact the participants randomized into the intervention group for the follow-up phone calls at day 3 and day 14 (visit 2 and 3), as well as for the outcome collection at day 30 (visit 4). While the study variables are stored in Redcap, the contact info will be stored in an Excel spreadsheet in a different secured cloud storage (htpps://docs.zoho.eu). The login access to the cloud is nominative and protected by a personal password. The access to the contact information will be strictly restricted only to the study nurses who need to contact the participants for the visit 2, 3, and 4. The cloud allows versioning, and audit trails to identify which change has been made by which study nurse. Only the necessary contact information will be collected at time of inclusion, such as the name and phone numbers of the participants as well as their proxy, the name and phone number of their primary care physician. No medical information will be stored in the contact information spreadsheet, since they are exclusively stored in Redcap. At the end of the study, once the outcomes have been collected for all included participants, the contact information file will be permanently deleted.

Analysis and archiving
Data files will be extracted from the database into statistical packages to be analyzed. The status of the database at this time is recorded in special archive tables.
The study database with all archive tables will be securely stored by CTU Bern for at least 10 years. The sponsor also keeps the Trial Master File and final reports for at least 10 years.

Electronic and central data validation
Data is checked by the EDC system for completeness and plausibility. Furthermore, selected data points are cross-checked for plausibility with previously entered data for that participant. If applicable: In addition, central data reviews will be performed on a regular basis to ensure completeness of the data collected and accuracy of the primary outcome data.
Before database lock the PI will validate the collected data with his signature.

Coding/Pseudonymization
Study-related data of the patient will be collected in a coded manner. A code (unique, consecutive numbered per center) will be attributed to each patient registered.
The name, address, and contact information will be collected in order to reach the patient for the 2 follow-up calls that are part of the intervention, and for the phone interview at 30 days to collect the outcomes. The 2 follow-up phone calls at day 3 and 14 (part of the intervention) will be performed by the local study nurse from the enrollment site. The phone call at day 30 for outcome collection purposes will be performed by a blinded study nurse from one of the other study site. All identified data (first name, last name, home address, phone number, name of next of kin, phone number of next of kin, name and contact information of the primary care physician) will be collected at inclusion into a file outside the Redcap CRF with very limited access (only the study nurse in charge of the intervention at each participating site will have access, and the blinded study nurse in charge of the outcome collection). This file will be stored in a secured cloud storage as specified under pint 12.2.2. The blinded study nurse for outcome collection will have a restricted access in Redcap only to the to the outcome data collection form and end of study form. In that way, the study nurse will not be able to be aware of the group allocation of the participants.

Monitoring
For quality control of study conduct and data retrieval, at all participating sites, remote and on-site monitoring visits will be performed during study conduct by an appropriately trained and qualified monitor of CTU Bern. The focus of monitoring will be on the implemented study processes, including the informed consent procedure, patient eligibility, primary endpoint, and on study documentation (essential documents). Any findings and comments will be documented in site visit report and communicated to the sponsor as applicable. Prior to study start (first participant enrolled) a monitoring plan detailing all monitoring-related procedures, including time points of visits, will be developed. Study site personnel will support the monitor in his/her activities. All source data and relevant documents will be accessible to monitors, and questions of monitors are answered during site visits.
In addition to on-site monitoring, central data monitoring will be performed to ensure completeness and consistency of the data in the eCRF.

Audits and Inspections
Source data/documents must be available to audits by the Co-Sponsor, the Coordinating PI or designee or to inspections by competent authorities.
A competent authority or CEC may wish to conduct an inspection (during the study or after its completion). If an inspection is requested, the local PI must inform the Coordinating PI and the local Co-sponsor immediately. The Investigators at the participating sites will give full cooperation to the inspectors in their activities.
The Coordinating PI may exclude participating sites/Investigators from further participation in the study because of fraud or non-compliance with the study protocol, ICH-GCP guidelines, or applicable laws.
Study sites or PIs may stop recruiting patients to this study when the Investigator or the Co-sponsor deems inclusion of patients into this study to be no longer ethical for medical or organizational reasons. In this case, the PI should give detailed reasons to the Coordinating PI.

Confidentiality, Data Protection
Trial-related data of the patient will be collected in a coded manner. A unique patient identification Page 48 of 54 number will be attributed to each patient registered into the trial (for more details see section 12.1.1). Identification of patients must be guaranteed at each site using the patient screening, enrolment and identification list. In order to avoid identification errors, patients identification number and the year of birth have to be provided on the eCRF. Patient confidentiality will be maintained according to applicable legislation.
Direct access to source documents will be permitted for purposes of monitoring, audits and inspections. The monitoring institution (i.e. CTU Bern) and the local authorities will have access to all information necessary for such tasks during and after the study.

Storage of biological material and related health data
All health related data will be stored at the local recruiting site at a secure location.

PUBLICATION AND DISSEMINATION POLICY
We plan to stimulate the implementation of our study results in a variety of ways, both locally in Switzerland and internationally: 1. We plan to disseminate our findings by publishing our study results in peer-reviewed medical journals, but also by presenting our study at national and international meetings (annual meetings of the Swiss Society of General Internal Medicine, of the Society of Hospital Medicine, of the Society of General Internal Medicine, and/or the International Society for Quality in Health Care).
2. We plan to disseminate the results through media press releases, that includes newspaper articles and radio interviews. This will insure to reach a large audience, including the patients themselves.
3. We will create a « TARGET-READ toolkit » to provide to the hospital all required information in order to replicate the discharge intervention used in our study. The packet will contain a guide that takes users step-by-step through the process of implementing and refining change processes at their own institutions, but also all patient's documents developed for the TARGET-READ trial. This toolkit will be made available online on the study website.
4. We will get the support from the ANQ (National Association for the development of Quality in Hospitals and Clinics in Switzerland). This association is founded by all Swiss hospitals and clinics in order to promote the quality of care. If shown successful, the ANQ is willing to help disseminate the TARGET-READ transition of care intervention to all hospitals and clinics in Switzerland, and also support the dissemination of the TARGET-READ toolkit.
5. We will work with the Agency for Healthcare Research and Quality (AHRQ) and the AHRQ Office of Communication and Knowledge Transfer to disseminate our findings through their Innovations Exchange platform. This platform includes high quality study results that are believed to change the clinical practice (https://innovations.ahrq.gov). The AHRQ contacted me after the publication of the development of the HOSPITAL score, and were willing to include the HOSPITAL score as soon as it would have been tested along with an intervention (the TARGET-READ trial fit to this criterion). The AHRQ Innovation Exchange offers an excellent opportunity to disseminate largely and internationally our study results.
Together, these activities will ensure that we share with many stakeholders what we learn from this study, and therefore may improve the quality of transition of care, and reduce complications in the posthospital discharge period.

Funding
This project is funded entirely by a SNSF-professorship from the Swiss National Science Foundation (PP00P3_170656). The funders reviewed and approved the project, but have no role in the study design, interpretation of the data, or manuscript writing.
There is no conflict of interest to declare.

Other Support
No other support.

INSURANCE
This research project is of risk category A with minimal risks and burdens and is therefore exempt of insurance requirement.

Overview
• This analysis is based on study protocol version 1.1 (13.02.2018) and statistical analysis plan (SAP) version 2.0 (22.04.2020).
• Data was exported and frozen on August 05, 2020.
• The report was generated on January 11, 2023.

Outcomes
The primary outcome is the number of patients who have an unplanned readmission or die within 30 days after discharge.
Secondary outcomes are: • Costs of readmission at 30 days.

Patients sets
The full analysis set (FAS) includes all randomized subjects who were discharged alive. Following the intent-to-treat (ITT) principle, subjects will be analyzed according to the treatment they are assigned to at randomization regardless of the treatment actually received. • Blindness breach

Analysis
• Statistical methods are described in detail in the SAP.
• Continuous variables are presented with mean and standard deviation (sd), or median, lower quartile (lq) and upper quartile (uq).
• Categorical variables are presented with number (n) and percentage (%) of patients.
• If there are missing data, the number of non-missing observations is shown.
• For all outcomes, only events within 30 days were considered.
• Unless explicitly mentioned, all analyses were adjusted for the stratification factors used in randomization, i.e. discharge site and readmission risk (intermediate risk with a simplified HOSPITAL score of 4 or 5 vs high risk with a simplified HOSPITAL score of 6 or higher).

Primary analysis
• Primary analysis was based on the FAS.
• The primary outcome and the binary secondary outcomes deaths within 30 days and patient satisfaction were analyzed using a Mantel-Haenszel risk difference stratified for the stratification factors used in randomization (discharge site and readmission risk, i.e. simplified HOSPITAL score of 4 or 5 vs 6 or higher) with a Klingenberg two-sided 95% confidence interval (CI) [3] and a stratified Cochran-Mantel-Haenszel test (using emh in Stata).
• The risk of 30 day unplanned readmissions in each group was estimated using the cumulative incidence function with death as competing event calculated from flexible parametric survival models (using stpm2 followed by standsurv in Stata) [4,5]. Groups were compared using the cumulative incidence difference with 95% CI and a z-test, based on delta method standard errors and a normal approximation. We also report the cumulative incidence of the competing event (death without prior readmission) for each group and the risk difference between groups.
Page 4/36 • Time to unplanned readmission or death was compared between groups using a log-rank test stratified for the stratification factors used in randomization. As an effect measure, we estimated the restricted mean survival time truncated at 30 days using flexible parametric survival models with group and sratification factors as covariates (using stpm2 followed by standsurv in Stata) [6]. We report the restricted mean survival time for each group and the difference between groups with 95% CI and a p-value.
• Count outcomes (number of unplanned and planned hospital readmissions, number of unplanned and planned days of hospitalization, number of emergency room visits, number of primary care provider visits) are presented with number of patients, person-time and incidence per 30 persondays with 95% CI. Groups were compared using a negative binomial regression with robust standard errors (using nbreg with vce(robust) in Stata) with the group and the stratification factors as covariates and the observation time as offset. There was considerable zero inflation for the number of unplanned days of hospitalization and it was modeled using zero-inflated negative binomial regression with robust standard errors (using zinb with option vce(robust) in Stata). We report incidence rate ratio with 95% CI.
• The costs of readmission was only analyzed for patients who had a readmission using linear regression with the treatment group and the stratification factors used in randomization as covariates.
The treatment effect is presented as mean difference with 95% CI and a p-value. Since the cost data were skewed we used robust standard errors and included a further sensitivity analysis based on a generelaized linear model with a gamma distribution and a log link. The results is presented as mean ratio with 95% CI and a p-value.
• For the primary outcome and the risk of unplanned readmission we also report stratum-specific estimates because there was evidence that the effect of the intervention was not homogeneous over the strata.

Secondary analysis
The secondary analysis for all outcomes was based on the per-protocol set (PPS) only considering patients without missing value for the respective outcome (complete cases).

Sensitivity analysis
We included five sensitivity analyses: • The primary endpoint was re-analyzed excluding early readmissions or death, i.e. patients who died or had a readmission up to the day after index discharge • All outcomes were analyzed in crude analyses not adjusting for stratification factors. Binary outcomes were be compared by chis-squared test, continuous outcomes by Student's t-test or the Wilcoxon-Mann-Whitney test, count outcomes using an exact Poisson-test and time-to-event outcomes by a log-rank test. Effects are presented as non-stratified risk difference, mean difference or Mann-Whitney statistic (i.e. the probability that a random patient from the intervention group will have a higher value than a random patient from the control group), incidence rate ratio, and restricted mean survival time difference, respectively, with 95% CI.
• The primary outcome was also estimated using survival methods, i.e. • Unplanned readmission at 30 days in the presence of the competing risk of death was also estimated using the non-parametric cumulative incidence function estimator with 95% CI according to [7] ignoring the stratification.
• Missing data for patient satisfaction due to death were imputed assuming the worst case, i.e. a negative respond to all question of the CTM-3.

Subgroup analysis
The primary outcome was analyzed for the following subgroups: • Risk for readmission (HOSPITAL score of 4 or 5 vs =6) • Clinical site

• Diabetes
• Chronic heart failure • COPD • Cancer • Living place (nursing home vs rest) • Living status (alone vs rest) • Health insurance (Semi-private and private vs rest) Subgroups were analyzed using a generalized linear model with a binomial distribution and an identity link. Effects for each subgroup were calculated from separate models. An interaction was analyzed comparing models with treatment group, subgroup and stratification factor with and without the interaction of treatment and subgroup using a likelihood ratio test. As a secondary analysis, subgroups were analyzed using Mantel-Haenszel methodology. A Mantel-Haenszel risk difference stratified for the stratification factor used at randomization was calculated for each subgroup with a two-sided 95% CI. An interaction was tested for using a Mantel-Haenszel test of homogeneity for the subgroup ignoring the stratification factors used at randomization.

Missing data
• The number of patients with non-missing observations is reported for each outcome • Drop outs were rare (less than 5%) and were assumed to have no readmission or death, censored for survival analyses, assumed to have no event and an offset of 1 hour for count outcomes, multiply imputed for patient satisfaction (see below).
• Deaths did not lead to missing data for the primary outcome, 30-day deaths and time to primary outcome. For death or readmission at the day of discharge, we used an event time of half a day.
• For the other outcomes deaths were: Bern, January 11, 2023, Version 3.0 TARGET-READ Phase 2 treated as competing event for 30-day unplanned readmission via the cumulative incidence.
included using the time of death as offset for count outcomes multiply imputed for patient satisfaction (see below).
• Multiple imputation for patient satisfaction were based on age, gender, nationality, living status, discharged destination, insurance, duration of index hospitalization, HOPSITAL score variables, death, readmission, number of PCP visits, source of revenue and the treatment group. Fifty imputed datasets were calculated and combined using Rubin's rule [8].
• The cost of readmission was only analyzed for patients with a documented readmission. Missing data was multiply imputed in a separate model with only those patients using the dame variables as for patient satisfaction (except for readmission).

Quality controls
• A second statistician reprogrammed the primary and secondary analysis of the primary outcome based on the exported data.
• The same results were obtained.

Changes from SAP
The analysis is consistent with principle features of the statistical methods described in the statistical analysis plan (SAP). The following deviations were made: • For the primary outcome and the risk of unplanned readmission, we also report stratum-specific estimates because there was some evidence that the effect of the intervention may not be homogeneous over the strata.
• For the analysis of count variables, we used robust standard errors for the negative binomial regression.
• For the analysis of the number of unplanned hospitalization days, we used zero-inflated negative binomial regression since there was evidence of zero-inflation.
• The cause of readmission or death is not yet included because it was only available as freetext and requires categorization.
• For the analysis of the cost of readmission we used robust standard errors as the data was skewed. In addition, we included a sensitivity analysis based on a generalized linear model with a gamma distribution and a log link.            Table 9: Stratum-specific analysis of the primary outcome (unplanned readmission or death within 30 days after discharge) and unplanned readmission within 30 days after discharge together with its competing event (death without prior readmission). This analysis was included because there was evidence that the homogeneity assumption might not hold for these outcomes (i.e. that the effect of the intervention depended on the stratum). A negative risk difference would indicate a benefit of the intervention.

Summary
• We did not find any evidence that the risk for unplanned readmission or death within 30 days after discharge was decreased in the intervention compared to the controls group (table 8). It was actually increased by 1.7% (95% CI -2.5 to 5.9%, p = 0.44).
• However, there was some evidence for heterogeneity between strata (i.e. site and readmission risk group) with a p-value of 0.05. The strata-specific analysis (table 9) indeed showed some heterogeneity with negative effects of the intervention on high risk patients in Neuchatel (and tendency-wise on intermediate risk patients in Biel), a tendency for some benefit on intermediate risk patients in Lausanne and not much evidence for any effect in the other strata.
• The risk for unplanned readmissions did not much differ between treatment groups (table 8). As for the primary outcome, some heterogeneity was observed with increased risk for unplanned readmission for high risk patients in Neuchatel but a reduced risk for intermediate risk patients in Lausanne.
• The competing event for unplanned readmissions-death without previous readmission-tended to be increased in the intervention group.
• We did not find evidence for an effect if the intervention on the time to unplanned readmission or death (table 10), the post-discharge health care utilization (table 11), patient satisfaction with the quality of transition of care between hospital and home (   Table 15: Strata-specific analysis of the primary outcome (unplanned readmission or death within 30 days after discharge) and unplanned readmission within 30 days after discharge together with its competing event (death without prior readmission) based on the per-protocol set. This analysis was included because there was evidence that the homogeneity assumption might not hold for these outcomes (i.e. that the effect of the intervention depended on the stratum). A negative risk difference would indicate a benefit of the intervention.  Table 15: Strata-specific analysis of the primary outcome (unplanned readmission or death within 30 days after discharge) and unplanned readmission within 30 days after discharge together with its competing event (death without prior readmission) based on the per-protocol set. This analysis was included because there was evidence that the homogeneity assumption might not hold for these outcomes (i.e. that the effect of the intervention depended on the stratum). A negative risk difference would indicate a benefit of the intervention. Table 17: Secondary analysis of post-discharge health care utilization within 30 days after discharge based on the per-protocol set. The incidence in each group is expressed per 30 person-days (i.e. over the observed period). An incidence rate ratio smaller than one would indicate a benefit of the intervention.
Intervention group (N = 631) Control group (N = 645) Incidence rate ratio P-value non-missing n (incidence, 95% CI) non-missing n (incidence, 95% CI) (95% CI) Number of unplanned hospital readmissions  Table 19: Secondary analysis based on the per-protocol set of the cost of readmission for patients with a readmission using complete cases. The mean difference is calculated from a linear regression with robust standard errors, the mean ratio from a gamma regression. A negative mean difference and a mean ratio smaller than one would indicate a benefit of the intervention.

Summary
• The secondary analysis on the per-protocol set confirmed the findings form the main analysis.
• There was no significant effect of the intervention on the primary outcome or on unplanned readmission but the intervention was associated with an increased risk for death.
• There was no effect on any of the other outcomes.   Table 21: Strata-specific sensitivity analysis of the primary outcome (unplanned readmission or death within 30 days after discharge) excluding early readmissions or deaths. This analysis was included because there was evidence that the homogeneity assumption might not hold (i.e. that the effect of the intervention depended on the stratum). A negative risk difference would indicate a benefit of the intervention. Table 22: Sensitivity analysis without adjusting for the stratification factors (crude analysis) of all outcome. Binary outcomes are presented with risks and risk difference (in time to primary outcome with the restricted mean survival time (RMST) truncated at 30 days and count outcomes with incidence (per 30 patient-days) and incidence rate ratio.

Summary
• The sensitivity analyses confirmed the findings form the main analysis.
• In particular, also crude analysis (table 22) and non-parametric survival analysis (tables 24 and 25.) confirmed that the intervention did not have an effect on the primary ouctcome but did increase the risk of death within 30 days .
• In crude analysis using an exact Poisson test, the number of planned hospitalization days appeared to be lower in the intervention group (

Summary
• The subgroup analyses showed the heterogeneity between risk group and sites (which has been seen before, section 6) although the interaction of risk group and site alone was not significant.
The intervention tended to have a negative effect on high risk patients and in Biel.
• The intervention increased the risk for unplanned readmission or death for patients with COPD and living in nursing homes.