Family Conferences to Facilitate Deprescribing in Older Outpatients With Frailty and With Polypharmacy

Key Points Question Do general practitioner–led family conferences promoting deprescribing in older adults with frailty and polypharmacy result in fewer hospitalizations? Findings In this cluster randomized trial of 521 community-dwelling older adults with frailty and polypharmacy, the number of hospitalizations over 12 months did not differ significantly among those who received a maximum of 3 family conferences. The number of potentially inappropriate medications decreased significantly in the intervention group after 6 months, but the reduction was not retained at 12 months. Meaning The findings of this trial suggest that family conferences for shared decision-making can successfully initiate the process of discontinuing medication, but no clinical benefit in terms of hospitalization was found.


Background of the study
The study focuses on geriatric frailty syndrome. This refers to a state of physical frailty with increased predisposition to the development of dependency and morbidity in older persons. Frailty is often the result of the presence of multiple diseases or multiple functional limitations and includes the following characteristic features: Weight loss, loss of muscle mass, fatigue, exhaustion, and reduced physical activity. Frailty is associated with an increased risk for the occurrence of falls and confusional states, as well as hospitalization and mortality. Geriatric frailty syndrome can be improved with interventions such as increased physical activity.
Polypharmacy is an independent risk factor for the occurrence of falls, hospitalization, and increased mortality, and is considered a significant pathogenetic factor for the occurrence of geriatric frailty syndrome. However, polypharmacy is also a risk factor for overuse of inadequate medications. Reducing polypharmacy (deprescribing) thus represents a promising intervention for improving or stabilizing the health status of frailty patients.
Care for patients with frailty syndrome in the outpatient sector is often difficult due to the complexity of the problems. Lack of care coordination in turn promotes unplanned hospital admissions as a result of fall events, pain syndromes, confusion or weakness. Since hospitalization is costly for these vulnerable patients and is itself a high risk, reducing hospitalization rates is an important goal of primary care case management in geriatric frailty syndrome. This should be achieved by strengthening the empowerment of patients and their relatives. The challenge of patients with frailty syndrome for family physicians comprises not only a better organization of primary care, but above all the prioritization of possible therapeutic measures based on the individual preferences of the patient together with the patient, the family caregivers and, if necessary, the professional carer. Following on from this, family conferences

Intervention
The intervention will be conducted in two phases: Phase 1 (preparation of the intervention): The participating primary care physicians of the intervention group will receive three consecutive training sessions, in which medical knowledge as well as competencies for the implementation of structured family conferences will be trained by means of communication training (2 obligatory face-to-face sessions, 1 optional training session, possibly as a webinar). In this context, physicians are provided with evidence-based specialist information as well as written materials to support the discussion with patients and relatives, including a toolkit with relevant non-pharmacological interventions. Obligatory content is a medication check with joint prioritization of drug therapy in the context of the family conference. Family physicians can optionally request an individual medication review by a clinical pharmacologist via a hotline. Other optional elements of the toolkit include: a) Consultation on measures to maintain mobility, fall prevention and pain therapy; b) Joint preparation of an emergency plan, if necessary taking into account a living will; c) Discussion of any existing need for nursing and psychosocial support, arrangement of nursing care counseling by a local nursing care support center.
Phase 2 (implementation of the intervention in the family practices): Conducting three family conferences per patient at baseline and at 3 and 9 months for 30-45 minutes each during a home visit involving family caregivers and/or care providers. The joint discussion should include: a) first, the patient's general preferences regarding his or her health care are worked out (e.g., regarding possible indications for hospitalization), b) a medication check is drawn up, in which the preferences initially worked out are applied, c) further topics from the above-mentioned toolkit are worked on as required. This is followed by a summary with documentation by the GP and an agreement on the further procedure, such as follow-up appointments.
The study period per patient is 12 months with data collection at T0 (baseline), T1 (6 months) and T2 (12 months).

Data collection
In all patients (of both study arms), several parameters will be collected or survey instruments will be used before the beginning (baseline), during (6 months) and at the end of the intervention (12 months): Basic data (once before start and if necessary during the course) 1. age 2. gender 3. educational level 4. marital status 5. living situation 6. degree of care 7. degree of disability

At baseline (T0), during the intervention (T1), and after completing the intervention (T2).
• Average number and duration of hospitalizations per patient (to determine hospitalization rate). • Relevant diagnoses • weight history • physical weakness (hand grip strength with dynamometer) • subjective assessment of the degree of fatigue • walking speed • degree of physical activity • cognition (episodic memory, executive function CERAD test) • Number of medications/active ingredients taken, among others, to determine potentially inadequate medications • Psychological status (Geriatric Depression Scale (GDS)) • Health-related quality of life (EQ-5D, VAS scale) • frequency of falls • Activities in daily living (Barthel index) • blood pressure • pulse • Cardiovascular morbidity (myocardial infarction, apoplex) • mortality • Costs (incremental cost-effectiveness ratio)

Study participants
The recruitment procedure is multi-step. One after the other, first GPs and then patients will be recruited. A total of 676 patients will be included by 136 GPs (5 patients per GP). During recruitment by the GP, the patient will be asked to name a relative as the main reference person, who will be present at the family conferences. If applicable, a representative of the ambulatory care service will also be invited to participate.

Inclusion and exclusion criteria:
GPs: can participate if they are registered as specialists in general medicine or specialists in internal medicine practicing general practice with the responsible Association of Statutory Health Insurance Physicians. Patients: Inclusion is pssible if all of the following criteria are met: 1) Positive screening for geriatric frailty syndrome.
2) Age 70 years 3) Regular intake of 5 agents (polypharmacy). 4) Need for nursing care (nursing degree 1 or higher) 5) Need for Nursing care in their own home by family members and/or outpatient nursing service Exclusion occurs when, in the judgment of the primary care physician, one or more of the following criteria are present: 1) moderate or severe dementia 2) reduced life expectancy of 6 months (palliative care) 3) nursing home residents 4) insufficient language skills in German of patient and family caregivers or nonavailability of an interpreter Relatives or trusted persons and professional carers: can participate if specified by the patient.Tthe patient will also be asked whether a professional carer should be involved in the family conferences.
Please justify the extent to which the selected gender distribution is appropriate for identifying possible gender differences (e.g., in the efficacy of the treatment, the safety of the medical device). Female patients as well as male patients are consecutively included in the study without stratification to sex. In the evaluation, gender is considered throughout as one of the relevant factors that may influence the results. In particular, for questions related to the outcome of care, potential differences for men and women will be described explicitly.

Process evaluation:
In order to be able to understand the change process and for the generalizability of the study results, a comprehensive collection of process data accompanying the pragmatic intervention study is essential. Process evaluation is conducted according to international recommendations for process evaluations of complex interventions on the topics: Implementation, Mechanisms of Action, and Context. It focuses on the following questions: what is implemented and how is it implemented (fidelity, dose, adaptation, reach)? How does the administered intervention lead to change (particpant response, mediators, unanticipated pathways and consequences)? How does context influence implementation, mechanisms, and outcomes?
Therefore, different process parameters on cluster (practice) and individual level (participants) will be collected with qualitative and quantitative methods (cf. Tab. 1). To investigate the feasibility and to optimize the intervention, semi-standardized interviews with family physicians and short telephone interviews with patients will be conducted after piloting the family conferences. The recruitment process, including reasons for nonparticipation, will be documented at both the cluster and individual levels. Contextual factors will be collected taking into account important structural and process-related aspects (e.g. socioeconomic and sociodemographic characteristics of GPs and patients) at baseline. To determine intervention fidelity, a structured documentation form will be completed by both school attendees and primary care physicians after each training session. In addition, the family conferences will be evaluated using a semi-standardized questionnaire at t0, after three months, and after six months, to be completed by primary care physicians immediately following the family conference.
The attitudes towards the COFRAIL intervention and the GPs' experiences with it will be discussed in four focus groups with a convenient sample of six to twelve participants at the end of the study. In addition, semi-standardized individual interviews will be conducted with a random sample of approximately 10 GPs per region. In a subsample of 10%, families' experiences (e.g., consideration of preferences; changes in physician-patient communication; barriers and facilitators) will be assessed after 12 months using semi-standardized individual interviews. The process evaluation is exploratory only. All quantitative data are analyzed descriptively; all qualitative data are analyzed using qualitative content analysis according to Mayring. (All process evaluation documents regarding informed consent can be found in Annex 5-7).

Protection and safety of the study participants.
3.1 and 3.2 Description of the procedure for recruiting of study participants and obtaining informed consent for study participation from study participants, parents or legal guardians, if applicable.

Recruitment method:
The recruitment procedure is multi-step. At two locations (Düsseldorf, Rostock) as many GPs will be contacted step by step until the necessary number is reached. In addition, GPs will be invited by mail to participate in a training course or in the study. In the next step, all interested GPs will be informed about the study verbally or in writing. After expressing interest, the signed consent form is obtained from the GPs. Subsequently, the general practitioner draws up a list of patients who are basically eligible for participation in the study on the basis of the billing code for geriatric diseases in the last three months (EBM code 03362). The general practitioner and the study assistant employed in the project jointly check which patients are potentially suitable for study participation on the basis of the inclusion and exclusion criteria (see 2.3.2). Subsequently, patients are contacted consecutively by the GP and their willingness to participate is requested until the number of five patients per GP is reached or the list is completed. The written consent of the patients is obtained by the GPs (consent form), after which the contact details are passed on to the local study center. After baseline data collection (t0), randomization into control and intervention group will be performed. The patients in the intervention group will be asked to name a caregiver/relative and, if necessary, a representative of the nursing service. The other participants will also be informed about the study.
All recruitment documents for patients and GPs can be found in the attachment to this application. These include cover and info letters for GP and patient, patient inclusion checklist or study flow chart (GP) and informed consent form (patient and GP). Furthermore, the information letters for caregivers/relatives and representatives of the nursing service are attached (see Annex 1-4).

Informed consent:
The GPs will be informed about the study in both written and verbal form by the local study center and will be requested to give written consent. Patients will be informed and educated in detail about the study and data protection regulations by the GPs, then if necessary by the study staff. Study participation and data collection are only permitted after the patients have given their consent. All patient information materials (cover letters, patient information letters and consent forms) and physician information are attached to this application (see Attachments 1 and 2).

Duration of recruitment (enclose materials such as advertisements, leaflets, etc.).
Recruitment of GPs will be done after submission of ethics vote from 01/12/2018 and will be completed within 6 months. From 01/04/2019, recruitment of patients will be done by GPs within four months. Participating patients will be cared for by their GP before, during and after the study as usual.

Particularly vulnerable persons:
In case of withdrawal of consent by the patient, their treatment by their GP and, if necessary, nursing care will be assured in the same way as before.

Do the participating persons receive an reimbursement of expenses or a payment?
All participating physicians receive 200€ reimbursement per patient, regardless of their group assignment (intervention / control). The patients and their confidant or representative from the ambulatory care service do not receive any payment for their participation.

What type of documentation is planned? (Attach documentation forms if necessary).
The outcome variable of the primary research hypothesis is the reduction in the average number of hospitalizations per patient (hospitalization rate) within 12 months of the start of the intervention. The secondary research hypothesis addresses the improvement of the patient's health status in terms of frailty syndrome expression (weight progression, physical weakness, fatigue level, walking speed, physical activity), cognition, number of medications taken, psychological status, health-related quality of life, fall frequency, activities of daily living, and costs.
In addition, relevant diagnoses as well as the safety parameters blood pressure, pulse, cardiovascular morbidity and prognosis related to mortality will be recorded.
The following table describes where and how the target parameters will be collected. Baseline data collection and data entry into a database (ORACLE database, see also section 4.4) will be performed by the study nurses of the local study centers (t0). Six months and 12 months after baseline data collection, follow-up data collection (t1, t2) will be performed by the GPs and study nurses, and data entry into the database will be performed by the study nurses.

Is cooperation with a statistician planned?
Yes: Dipl.-Math. Birgitt Wiese, Institute of General Medicine, Hannover Medical School (MHH)

Measures provided for the protection of the collected data
The data will be entered into a central ORACLE database (Oracle 9i) at the local study centers (Düsseldorf and Rostock) using a web-based data entry system secuTrial®. Access to the internal database and the web server will be controlled by two firewalls connected in series.
The data transfer will be done by 128 bit SSL encryption. There will not be a local data storage. secuTrial® fulfils the requirements of the FDA (21 CFR Part 11) and complies with the guidelines for Good Clinical Practice (GCP). The data will be stored with a pseudonym without identification data. The members of the study group will have access to the electronic data entry system according to a detailed user role and rights concept. An audit trail ensures an automatic protocol of all data entries, changes and deletions. Various data quality assurance measures will be implemented: These include online plausibility checks during data entry, data checks after data entry and user training before data entry. In addition, all those involved in data collection and entry will be trained by the data management team (study assistants, student assistants) and will receive a manual describing and explaining all functions.

Have discontinuation criteria been defined?
• for the individual study participants: Yes o Participation can be withdrawn by individual patients at any time without giving reasons.
• for the termination of the entire study: Yes o The criterion for termination of the study is defined as an increase in the hospitalisation rate of > 30% in the control and/or intervention group.

Harm and risks
What are the types of potential risks, harms, burdens or other disadvantages for the persons participating in the study?
The intervention aims to improve communication between GPs, patients and relatives. Medical risks are thus not to be expected, as the study does not directly intervene in medical and pharmacological treatment. It is assumed that there is no risk or harm to be expected for the patients participating in the study. Nevertheless, in addition to the hospitalization rate, blood pressure, pulse and mortality are defined as safety parameters, which will be regularly analyzed by the project administration and by the cooperation partners. For data collection purposes, patients in both the intervention and control groups will be interviewed by trained study assistants. This will only be done with the explicit consent of the patients. In addition, the study assistants will be trained by the study team so that they are prepared to interact with older people. The patients will be given sufficient time to answer the questions or carry out the tests, as these could be perceived as strenuous. Breaks/interruptions of the interviews will be offered if needed.

To what extent, if any, do study-related measures deviate from the usual routine treatment?
All patients will receive the usual routine treatment within the framework of normal care by their participating doctors. Nursing services will continue to be provided by the outpatient nursing service as before.
In the intervention group we additionally offer: Conducting three family conferences per patient at the beginning of the study as well as after 3 and 9 months for 30-45 minutes each in the context of a home visit involving the family caregiver and/or a professional caregiver. The joint conversation should include the following: -First, the patient's general preferences regarding his or her health care will be identified (e.g. possible indications for hospitalisation), -a medication check will be conducted, in which the initially elaborated preferences will be applied, -If appropriate, further topics from the toolbox (non-pharmacological interventions to secure mobility, emergency planning, provision of psychosocial support) are dealt with and a summary of the further procedure is prepared by the GP. (see paragraph 2.2.1 Intervention).

Benefits and opportunities
What is the expected benefit? Individual Benefit: Improving drug therapy safety (AMTS) with a reduction in the number of potentially inappropriate medicines (PIM) prescribed and with a reduction in drug-related side effects (fall rate, hospitalisations) contributes to patient safety. In addition, an improvement of the health status compared to the control group is intended through a reduction of the expression of the frailty syndrome in the course. Furthermore, the participation and self-determination of patients with frailty syndrome will be improved by strengthening shared prioritisation and participatory decision-making by participating in the family conference.
Benefits for care: The aim of the family conference is to improve the communicative skills of GPs in dealing with complex treatment problems. This method will improve communication and cooperation between the actors involved. This enables a more structured understanding of problems and the targeted search for solutions. Patients who do not directly participate in the study may also benefit from this. External benefits: The knowledge gained could be used to discuss the use of the family conference instrument for other complex areas of primary care. If the evaluation is positive, these results can be incorporated into the development of didactic formats for continius medical education and undergraduate training of physicians.

Please justify why you consider the potential harm to be justifiable in relation to the expected benefits.
In the context of improving patient safety, harm can almost be excluded. Patient participation is voluntary and consent can be withdrawn at any time without any disadvantages. By connecting the competences of the professional groups caring for the patient, a more efficient solution can be ensured in the in case of a problem arising.

Signatures and statement
6.1 Statement and signature of the director of the clinic or institute (with stamp) I agree to the conduct of the study mentioned under 1.1. I confirm that the resources (number and qualification of staff, infrastructure, equipment and facilities) for a successful implementation are available. I also confirm that the recruitment of study participants will not be compromised by competing studies.
(Stamp) Rostock, Date Signature (Prof. Dr. Attila Altiner) Annex 1 -Recruitment documents GPs (cover letter, study information, fax response, consent form). 2 -Recruitment documents patients (cover letter, study information, consent forms) 3 -Study information to caregiver/relatives, 4 -Study information for nursing staff 5 -Process evaluation documents for general practitioners 6 -Documents process evaluation patients 7 -Process evaluation documents for caregivers/ relatives

Introduction
The COFRAIL study is a multicenter cluster-randomized controlled intervention trial to increase patient safety through promoting family conferences in primary care.

Data source
676 patients will be recruited in 138 general practices, with the practices randomised into the intervention and control groups in a ratio of 1:1, so that 338 patients will be assigned to the intervention group and 338 patients to the control group. Data will be collected at baseline T0, at T1 (after 6 months) and at T2 (after 12 months). In addition, in the intervention group, data are collected on three family conferences that take place between T0 and T1 or between T1 and T2. Sociodemographic data, data on quality of life (EQ-5D-5L), memory tests (CERAD), physical functionality, geriatric depression scale, medication, hospital stays, doctor contacts, history of falls, as well as other data relevant to health economics will be collected. The data will be entered into a central ORACLE database in the web-based EDC system secuTrial®. In order to achieve high data quality, extensive plausibility checks are implemented in the input system; in addition, further data checks will be conducted after the data has been entered.

Objective, endpoints and statistical analyses
The main objective criterion -the primary outcome -is the number of hospitalisations per patient within the observation period of 12 months.
The secondary outcomes are the following: -Medication: number of medications per patient, Drug Burden Index (DBI) defined as number of anticholinergic or sedative medications and prevalence of potentially inadequate medication. -Hand force measurement -Cognition: scores of the CERAD tests "word list" and "naming animals -Sum score of the Geriatric Depression Scale (GDS) -Mobility measured by the Timed Up & Go -Test -Health-related quality of life (EQ-5D-5L) -Activities of daily living measured by the Barthel Index -Weight The statistical analyses will include detailed descriptive statistics (frequencies, position and variance parameters) for the parameters (total and separately for the intervention and the control group) to describe the study population (age, gender, education according to CASMIN, frailty index, etc.) as well as to assess the primary outcome and the secondary endpoints. A mixed regression model will be used for the analysis of the primary outcome. As a random effect, the practice affiliation characterized by the practice ID will be included in the model. In addition to the group (control or intervention), the factors age, gender and co-morbidities are included as fixed effects for adjustment. A linear mixed model with the target variable "number of hospitalizations" is envisaged; according to the actual distribution of the target variables, an ordinal logistic or a Poisson model can also be applied.
The secondary outcomes will be analyzed using mixed linear regression models or mixed logistic regression models (binary or ordinal) depending on the distribution. The practice affiliation is included in the model as a random effect, fixed effects are group affiliation, age, gender and comorbidities.
As an alternative to the classical regression methods, the methods of Classification Tree Analysis (CART) will be applied. This method focuses on subgroups of cases that are as homogeneous as possible with regard to the distribution of the response. The characterisation of homogeneous subgroups is done with the help of a binary tree: first, the entire sample will be split into two subspaces. Each subspace created in this way and all further subspaces can be successively split further so that a hierarchically structured division will be obtained overall. The CART analysis is carried out with 10-fold cross-validation. The cut-points of the predictors are determined by the procedure; missing values in the predictors are replaced by surrogate variables. Target parameters in the respective CART analyses are the outcomes (primary and secondary); predictors are group membership (control or intervention), age, gender, comorbidities and other possible influencing factors. Advantages of these methods are the independence of the distribution of the predictors as well as the easily interpretable presentation of the results as a decision rule.
The evaluation of the safety parameters (in particular mortality and hospitalizations) will be provided to the Data Safety Monitoring Board (DSMB) at regular intervals descriptively as frequencies blinded (i.e. it is not recognizable which group represents the intervention and which the control).

Aim, outcomes and statistical analyses of the health economic evaluation
The aim of the health economic evaluation is to determine the efficiency of the intervention by comparing the costs and outcomes of the intervention group with the costs and outcomes of the control group (care as usual). All costs related to the intervention as well as to the utilisation of health care services  from the perspective of social insurance in Germany (health insurance, long-term care insurance and pension insurance) will be considered.
To determine the efficiency of the intervention, a cost-effectiveness analysis and a cost-utility analysis will be performed. The incremental cost-effectiveness ratio (ICER: additional costs for each additional hospital admission avoided) and the incremental cost-utility ratio (ICUR: additional costs for each additional quality-adjusted life year (QALY) gained) will be calculated as a quotient of the cost and benefit differences between the intervention group and the control group. QALYs are utility values and are calculated by recording health-related quality of life with an established preference-based quality of life instrument based on the EQ-5D-5L (Herdman et al., 2011) and economically valued with a German health cost table (Greiner et al., 2005) to generate utility values. 95% confidence intervals for outcome and costs will be determined non-parametrically based on distributional characteristics using bootstrap procedures (Briggs, 1997). Univariate and probabilistic sensitivity analyses will be performed to account for uncertainty, and costeffectiveness acceptability curves will be constructed (Fenwick, 2004).

Analysis population
The evaluation of the primary outcome measure will follow the intention to treat (ITT) principle. This means that all patients who are included in the study, i.e. all intended patients in the intervention group and all patients in the control group who give their study consent, will be included in the analysis. The patients remain in the original group for the analysis, even if, for example, the family conferences are not carried out for a patient in the intervention group. If the study is terminated before completion, the number of hospitalisations up to the time of termination is used as the primary endpoint, i.e. the Last Observation Carried Forward (LOCF) principle is applied.
As a sensitivity analysis, a per protocol (PP) analysis will be performed, i.e. all patients who properly completed the study will be included in this analysis.

Data processing
Missing values will not be replaced a priori. Some of the procedures mentioned above (CART, RPA) can handle missing values, i.e. missing values are replaced by so-called surrogate variables only for the split criteria. Sensitivity analyses will be carried out to investigate possible biases due to missing values that are not missing by chance.

Software used
The analyses will be carried out with the software packages SPSS, STATA, SAS, and CART.  Krauth, C. et al. (2005). Empirical standard costs for health economic evaluation in Germanya proposal by the working group methods in health economic evaluation. Gesundheitswesen, 67(10), 736-746.