Introduction: Care of frail and dependent elders with multiple chronic conditions is a major challenge for health care systems. The objective of this study is to evaluate the effect of coordinating the existing structures in the private and public sector for the care of frail and dependent persons over age 60, and susceptible of presenting complex bio-psycho-social issues. This approach is aimed at improving the coordination, continuity, quality and efficacy of care in this population, which presents a high risk of hospitalization, emergency room visits, institutionalization and mortality.

Methods: Three-year cluster randomised controlled trial. A control group receiving usual care (follow up by primary care physician and home nursing service) will be compared to an intervention group that will be provided, in addition, in-home multidimensional geriatric assessment with access to a 24h/7 day a week call service, and coordinated long-term follow-up. Survival analyses will be conducted to compare the outcomes between groups.

Primary outcome:

- Hospitalizations: rates of first hospitalization, number, cause and length of stay.

Secondary outcomes:

• Emergency room visits: rates of first visit, number and cause
• Institutionalization: number of patients
• Mortality: rate and number of deaths and place of death (home versus hospital)

Background Since the second half of the twentieth century, a major prolongation in life expectancy has led to a rapid increase in the number of older adults in Switzerland, with a high prevalence of multiple chronic diseases, functional incapacities and cognitive deficits.

In general, care of the frail elderly is characterized by fragmentation and discontinuity within and between sectors, resulting in frequent and inappropriate hospitalizations and premature nursing home placement with an increase in health care costs. Furthermore, many older patients with progressive chronic diseases prefer to die at home, but more than 50 percent will do so in hospital, with multiple hospitalizations in their final year. This is mostly related to the availability of health resources in the community.

Care-giver burden with increased morbidity in this population is an often hidden consequence of caring for dependent older adults.

A study performed in 2003 at the University Hospital of Geneva emergency room5 showed that among 255 patients (mean age 79 years) evaluated during a three-month period for inability to stay at home, 59 percent had no acute medical problem but rather a breakdown of their informal support system. Ninety per cent were frail and dependent, and 85 percent had not been evaluated by their primary care physician previous to their arrival to the emergency room. This study suggests that many needs and disabilities of the frail elderly are not identified and that the lack of coordination between the different actors of the community primary health care system does not enable these patients to stay at home.

Consequently, there has been an increasing interest in the ability of integrated models of care to improve health outcomes and shift service utilization from the institutional setting into the community.

Integrated models of care vary greatly among countries and depend on local health policy, financing, as well as professional and sociocultural values. Broadly, three different levels of integration have been described - linkage, coordination and full integration - depending on the level of integration between the different actors of the administrative, financial and service sectors with a continuum in the degree of collaboration, ranging from informal to those structured with all the services included under a single roof.

The first level or linkage is the organizational model that best applies to Geneva. Patient care takes place between the primary care physician and existing structures depending on patient needs. Communication between professionals involved is encouraged in order to promote continuity of care. However each structure functions under its own legal, financial and operational environment with no cost shifting between structures.

No distinct model of integrated care for the frail elderly is in place in the Canton of Geneva, as collaboration depends on the informal linkage between the actors in the community.

This study will evaluate the implementation of the second level of integration by coordinating existing community resources to improve the care of frail elderly patients presenting with complex medical and psychosocial issues in the real-life clinical setting.

Design A prospective randomized controlled cluster trial with two intervention arms and two control arms will be conducted. Each arm corresponds to a nursing team in one of two home nursing service health care centers (Onex and Bernex).

Because of the difficulty in randomizing frail complex patients individually, patients were randomized by nursing teams.

The nursing personnel of each center are divided in two teams of similar size (nine nurses per team in Onex and six nurses per team in Bernex) and composition. Each team is responsible for following patients in a predetermined sector, by patient address.

The research team initially began recruitment in July 2009 in the Onex home nursing service health center. The Bernex center was added in June 2011 in order to reach the recruitment target. Two nursing teams were identified in each center.

Patients are assigned to a nursing team depending on their place of residence. The allocation of the nursing teams to the intervention or to the control arms will be performed by use of a random number generator. Each nursing team will be designated by a number (1 to 4) and the first four random numbers generated, allocated to the corresponding team. It was decided before the process that the highest number in each site would be allocated to the control arm and the other to the intervention arm. Once a patient is attributed to a team, he or she will remain so for the entire follow-up period with no transfer of patients between teams. The recruitment of patients will not be influenced by their health status, since eligibility will be determined only after the patient is attributed to one group or another.

The control arm will consists of two clusters of patients over 60 years of age followed by a primary care physician and routinely evaluated by the home nursing service with the resident assessment instrument-home care (RAI-HC) and identified as frail by pre-defined clinical criteria.

The intervention arm will consist of two clusters of patients over 60 years of age followed by a primary care physician and routinely evaluated by the home nursing service with the RAI-HC and identified as frail by predefined clinical criteria that will receive in addition an in-home multidimensional geriatric assessment, access 24 hours a day, 7 days a week to a call service provided by the Community Geriatrics Unit and coordinated long-term follow-up.

Because of the real-life clinical setting, blinding will not be feasible.

Target population The target population will be patients over 60 years of age followed by a primary care physician who prescribes a home intervention by the home nursing service. These patients, who will be routinely evaluated with the Resident assessment instrument-Home care (RAI-HC), will be identified as frail if they fulfill predefined clinical criteria (see inclusion criteria).

Home nursing service evaluations are prescribed by the primary care physician usually after an acute event (fall) or decompensation of a physical or psychological illness.

The target population already presents a variable degree of frailty and dependencies that require in-home interventions by a home nursing service and for which the Community Geriatrics Unit may be requested by the primary care physician to evaluate the most complex situations. This, however, is not usual practice. Patients identified were not previously known to the Community Geriatrics Unit.

Geographical setting The Canton of Geneva has a population of 466,918 inhabitants, 16 percent of which are 65 and over. Most of the elderly population live at home, 14 percent of which live in nursing homes.

This study was carried out in two of the 45 communes of the Canton of Geneva: Onex with à population of 17,942 and Bernex, with a population of 9,772 inhabitants. These communes were chosen because of their size and geographical proximity to the Community Geriatrics Unit site at Loex and of the good collaboration with the primary care physicians of the region. Each commune has a home nursing service health center.

The socioeconomic status is similar and homogeneous in this population and should not present a confounding factor.

Follow-up in the intervention group

• Each patient will be routinely administered the RAI-HC by the assigned home care nurse. If inclusion criteria are present, the nurse will present the protocol orally and in writing to the patient and/or his or her caregiver. It is clearly specified in the written information that the patient will not incur any extra out-of-pocket expenses since in-home geriatric assessment is reimbursed by the health insurance. The visit is 90 %reimbursed by the patient health insurance (private and obligatory). The remaining 10 percent deductible will be waived by the Community Geriatrics Unit.
• If consent is obtained, patient information will be transmitted to the research nurse for CHIP score calculation based on the RAI-HC data and inclusion in the study.
• The Community Geriatrics Unit doctor will inform the primary care physician of his or her patient's inclusion in the study.
• The Community Geriatrics Unit doctor will contact the patient to schedule an in-home multidimensional geriatric assessment. If necessary, an evaluation will be performed by another member of the Community Geriatrics Unit team, such as a physical therapist, occupational therapist, psychologist, dietician or social worker.
• The results and recommendations of this evaluation will be transmitted to the primary care physician and to the home care nurse.
• Coordinated follow-up between the primary care physician, the home nursing service nurse and the Community Geriatrics Unit will be established.
• Patients will be receive written instructions (information will be available in their home log book) about whom to contact in case of an acute health problem:
• The first person to contact is the primary care physician;
• If not available, the home nursing service nurse (imad) should be contacted;
• The nurse or the patient should contact the call service of the Community Geriatrics Unit that is available 24 hours a day, 7 days a week, for an urgent intervention to be carried out by telephone and/or a home visit;
• Team meetings between the primary care physician, home nursing service and Community Geriatrics Unit will be scheduled as needed to discuss complex situations.

Follow-up in the control group During the same period, patients who are followed by the nursing team assigned to the control arm will be administered the RAI-HC. If frailty criteria are detected, patient information will be transmitted to the research nurse for CHIP score calculation and inclusion in the study. Patients will continue receiving usual care and follow-up by the primary care physician and the home nursing service.

Sample size The planned sample size of 300 (150 per arm) was calculated based on a 5 percent alpha error (two-sided) and a power of 80 percent to detect a difference in hospitalizations of 12 percent. This difference was based on a previous meta-analysis of controlled studies that utilized comprehensive geriatric assessment.24 The expected hospitalization rate in frail elders is approximately 30 percent.

In addition, 150 patients will be recruited sequentially, randomized by cluster arm.

Calendar Recruitment began in July 2009. The study was initially programmed for 18 months. Nevertheless, due to circumstances related to the real-life clinical setting, there were many challenges. Recruitment was delayed in 2010 during the five-month maternity leave of the research nurse, who was difficult to replace. Furthermore also in 2010, the home nursing service health care center of Onex moved its offices, causing recruitment delays. Finally in 2011, the Bernex home nursing service team had to deal with a period of transition to a computerized record system, and a new research nurse had to be hired.

The recruitment target was reached on July 1, 2012 with follow up of outcomes until December 15, 2012. Data analysis and manuscript redaction will be carried out in 2013-2014.

Ethical considerations The study was approved by the Ethics Committee of the Departments of Internal Medicine and Community Medicine of the University Hospital of Geneva on August 26, 2008 (Protocol No. 08-128) as well as the Ethics Committee of the Geneva Medical Association for clinical ambulatory research on October 24, 2008 (Protocol number: 08-06).

Signed consent was obtained after allocation to the intervention group, since integrated care is not routinely part of usual care. Consent for the control group was waived by the ethics committees, given that all patients were already being followed by the home nursing service that routinely obtains consent to use collected data.

Measures of intervention efficacy

Primary outcome:

- Hospitalizations: rates of first hospitalization, number, cause and length of stay

Secondary outcomes:

• Emergency room visits: rates of first visit, number and cause
• Institutionalization: number of patients
• Mortality: rate and number of deaths and place of death (home versus hospital)

Data collection In both groups, the research nurse will collect data from patients' computerized hospital records, computerized home nursing service records and records from the Community Geriatrics Unit.

Data collected in all patients will include the following:

• Alarms;
• Demographic data;
• Medical diagnosis;
• Medications;
• Number of consultations at home for an acute problem (by telephone or a home visit) by the Community Geriatrics Unit physician, the primary care physician or other emergency services.

Statistical analyses Descriptive results will be reported by counts and percentages or by means and standard deviations. Characteristics between groups will be compared using exact Mann-Whitney, Chi-squared or Fisher tests. The main and secondary outcomes will be compared in intention to treat between intervention and control groups by using survival analyses. For mortality, the Kaplan-Meier survival estimator, log-rank test and Cox regression model will be used. As death is a competing event for the other outcomes (institutionalization, first hospitalization and first emergency room visit), competing risk models will be used. The cumulated incidences and the hazard ratio will be obtained with the package cmprsk for R version 3.0.1. All reported hazard ratios will be adjusted by health care center and the tests for comparison of cumulated incidence over follow-up period will stratified by health care center. Cumulated incidence at one, two and three years will be compared using a Wald test. Two-sided p-values less than 0.05 are considered significant. All analyses will be performed using S-Plus 8.0 for Windows and R version 3.0.1.

Funding Funding of swiss francs 94,500 ($104,241) was obtained from the Quality Bureau of the University Hospital of Geneva in December 2008. This amount was used to hire a part-time (50 percent) research nurse assistant to coordinate the study, collect data and manage the database from July 2009. Additional funding of swiss francs 66,000 ($72,800) was obtained from the Edmond J. Safra Fund in order to continue the study in October 2011.

Discussion Care of the frail and dependent elderly is usually characterized by frequent and unnecessary hospitalizations, emergency room visits and institutionalization. Most patients express the desire to die at home but most will do so in hospital, in many cases due either by the absence of a caregiver or by lack of coordinated care.

The investigators postulate that these outcomes can be achieved by implementing an integrated care network, with close coordination between the different actors as well as access to a specialized community geriatrics unit that includes a round-the-clock call service.

The authors of the study will also seek to demonstrate that the establishment of such a network is feasible and will provide effective and efficient health care for this population. If the results are positive, it is hoped that political decision-makers in the health care sector will be encouraged to further implement and support such programs.

Inclusion Criteria:

• Patients followed by the home nursing service and primary care physician
• Over 60 years of age
• Able to give consent (patient or caregiver)
• Able to speak French
• Meeting frailty criteria as determined by one of the following alarms detected by the RAI-HC:
• Alarm 8: cognition
• Alarm 12: social role and function
• Alarm 15: falls
• Alarm 22: frailty of the informal care giver system.

Exclusion Criteria:

• Age under 60 years
• Unable (patient or caregiver) to give informed consent
• Unable to communicate adequately in French
• Patients' not receiving nursing visits (only receiving practical aid or services from occupational therapists).