There is an urgent need to develop new care models for patients with long-term and complex needs. Our goal is services that are seamless, pro-active and person-oriented.

Intervention:

The Patient-centered health care team (PACT) is a service model for frail elderly patients with multiple long term conditions. PACT is a seamless and proactive health service model that aims to ensure safe early discharge and prevent hospital admissions for elderly frail patients. The four pivotal elements are are supported by theoretical and empirical underpinnings: 1) Goal-oriented person centered approach 2) Inter-disciplinary comprehensive geriatric team: 3) Pro-active care plan.

Study Objective:

1) To investigate whether the PACT model improves health-related quality of life and patient generated goals 2) carry out a health economic evaluation of PACT.

Design:

The evaluation model for complex interventions is our guiding evaluation framework. This is a non-randomized, matched control, before after study. The intervention unit, is the care organization, including four hospitals - municipality dyads. Two intervention dyads and two control dyads. Index patients from intervention hospitals will be matched age, sex and number of chronic conditions.

Intervention group: Patients with emergency admission to the UNN internal medicine department in Tromsø and Harstad who are age > 65, have three or more long-term conditions, an emergency admission within the last year, and an informed consent is given by patient or next of kin. (Approved by ethics board 07.05.2015). The exclusion criteria are: Language barriers, and less than 3 months life expectancy.

Controls: Recruited from the Nordlanssykehuset (Bodø) and UNN-Narvik internal medicine departments, matched to the index patient's sex, age, number of long-term conditions. Control patients will be subject to the same data-collection as intervention patients.

Data collections: All patient data will be collected at baseline, 6 and 12 months. Outcomes: The primary outcome is the adjusted differences in the change of Quality of Life, measured by Short Form-36 (SF-36), physical health dimension between intervention and control groups at 6 months after inclusion in study.

Paradoxically, the very success of our health care system causes the number of patients with multiple are on the rise. Studies of European populations above 75 years of age show LTC-prevalences of more than 70%, and rates of multimorbidity around 40-50%(8-10). Patients with LTC-conditions accounted for 3/4ths of health care spending in an early US report, and patients with 4 or more co-morbidities consumed 5 times as much health services as those with no LTC in a recent Irish report. The current care system is designed to deliver disease specific, urgent/ episodic care, and is poorly suited to the needs this growing patient group

Best practice models for patients with Long Term conditions (LTCs) and complex needs have been developed. In medicine, the underlying models are often implicit, which can cause the misunderstanding that medical research is practical, factual and not theory driven. In complex interventions, underpinning theoretical models are considered essential for study design.. In the planning and design of this intervention, the investigators have chosen to use the Chronic Care Model (CCM). Other models of chronic care are either slightly overlapping or in alignment with the CCM, but none of them cover all the areas of the CCM. It is the only model with both a systems- a clinical- and a patient perspective. Furthermore it has a growing evidence base for its effects on both care-processes, health outcomes and cost-effectiveness. It continues to inspire care reform both in large international care organizations, and in our local northern Norwegian context. The CCM builds on two pillars: "The informed active patient" and "The pro-active prepared health care team", which are equal partners engaging in "productive interactions" for "health and functional outcomes". Evidence Based Medicine (EBM) is the basis for identification of appropriate actions. Integrated team care delivery is then tailored to the patients' needs, values and preferences. Both health management support and use of clinical information systems are included in the model as key supporting factors(33).

A recent report documents that the health care delivery in Helse Nord is far from the ideal CCM-care. The main challenges were a lack of attention to the patient's personal context and priorities, and a fragmented care delivery. A large-scale CCM inspired project which answers these challenges, is under development at the university Hospital of North Norway (UNN). Its dual focus is coordination of care through two core components: 1) the proactive, prepared interdisciplinary teams with personnel from both hospitals and the municipalities to address the current fragmented care delivery and 2) the informed active patient approach to care planning. The initiative, which will be funded by redeploying existing staff to work in the teams. The team will facilitate patient centered and integrated care by conducting a structured person centered need assessments, develop individual evidence based care plans, and provide service integration across levels of care.

Research questions, hypotheses and methods The CCM is currently established as best practice for LTC-care delivery. Yet, the evidence of effectiveness of the CCM in terms of health and functional outcomes in several systematic reviews remain inconsistent. All CCM interventions, must tailor the CCM to local historical, cultural and regulatory context. Thus the heterogeneity of CCM-interventions is large, which may explain the observed inconsistencies. In our review, the investigators noted that many CCM-interventions seemed to lose the link to the underlying theory in the operationalization process. Active two-way patient-provider dialogues and practical skills development in self-management are central for better outcomes. However "the informed active patient" and "self-management support" were often translated into passive one-way educational interventions. Furthermore, the content of many of the CCM-interventions was often inadequately described, making it difficult to ascertain their fidelity to the underlying theory.

Our research questions are: What is the effect of a theory driven CCM-implementation on:

• Patient health and functional outcomes?
• System outcomes: Length of stay and emergency admissions in primary and/ or secondary care institutions?

Material and methods:

Design: The evaluation model for complex interventions is our guiding evaluation framework . This is a non-randomized, matched control, before after study(46). The investigators recognize that randomization is the most robust method of avoiding systematic bias between comparison groups. However, our intervention aims to change usual geriatric care, in terms of both structures and routines at the organizational level of two municipal-hospital dyads. A patient level randomization will require the organization to switch between old "usual care" and a new team-based pro-active routines on a patient-by-patient basis, is unlikely to be successful. As there are only four hospitals involved in the study, randomization at the organizational level would not really control for bias either. The chosen design improves comparability between the comparison groups by matching them on factors known to be important for outcomes (age, sex and number of chronic conditions), to adjust for baseline differences between the populations with a before-after design, and to adjust for known confounders by the propensity score method(47, 48). The design is approved by Cochrane to be included in systematic reviews of interventions (46).

Setting: Trials are often performed on selected patients without comorbidities making studies suffer from limited external validity. Van Royen et al (2014) argue that there is a driving demand for real-word clinical practice data. In this project, the team members will develop the intervention as part of the daily activity at the hospitals including "normal" medical patients > 67, making the trial setting naturalistic. The real-world setting and normal patient caseload makes the design resemble usual care, thus increasing the generalizability from the trial to other patients in regular practice. A naturalistic setting is viewed to be the gold standard for economic evaluations, which is also an objective for the project.

The intervention: PACT is a seamless and proactive health service model that aims to ensure safe early discharge and prevent hospital admissions for elderly frail patients. The four pivotal elements are all part of the CCM-model, and are supported both by theoretical and empirical underpinnings: 1) Inter-disciplinary comprehensive geriatric team: Frail elderly patients are often multi-morbid, and in need of multiple simultaneous competencies. The team will push for early discharge to minimize the deconditioning and iatrogenic risks associated with hospitalization(. The team will identify eligible patient's, make an integrated assessment of needs upon admission, marshal the necessary resources to resolve current clinical issues, and prepare for early discharge supported by home based services such as home -monitoring, and -care. The team will develop and use different intervention options such as case management, discharge- and follow-up protocols, an array of telehealth options (telephone support, telemedicine, telemonitoring and smart home solutions) combined with in-person home visits when needed. Patients will remain a team responsibility for a period of 3-5 days after discharge 2) Involving patients in care and self-management: When the clinical situation is stable, the patient will be invited to make a systematic assessment of needs, values and preferences with a care-team member. The patient's wishes are then translated into realistic care goals in a shared decision making process. Most importantly, the team can avoid undesired care, which is especially important in a palliative phase. The teams should also provide opportunities for patient education, social- and physical adaptions, and/or skills training to improve self-management. Patient involvement and engagement in care has been shown to improve health and functional outcomes. 3) Pro-active care plan: CCM promotes a systematic early identification of functional decline. The team should discuss the most probable scenarios of clinical deterioration, and provide action plans for them, which are available to both patients and team. An example is to provide patients with Chronic Obstructive Pulmonary Disease with a drug kit, which the patient can use under guidance from care personnel. The pro-active care plan should also include a road map towards the health goals identified together with the patient. 4) One point of contact: To avoid unnecessary delays and prompt action to early signs of clinical problems, patients should not have to "hunt" down the right professional. A 24/ 7 call-center will be their one-point- of contact. The call center will have the resources to mobilize the necessary clinical responses, including single members of- or the completely comprehensive geriatric team.

Inclusion Criteria:

• Patients who either received treatment in PACT team for > 24 hours (Treat group) or an emergency admission to the UNN (Control group) in Tromsø/ Harstad/ Narvik/ Bodø. Age > 60. Have complex long-term needs.

Exclusion Criteria:

Live outside of designated municipalities. Receive < 24 hours of care from PACT team.

Due to data privacy legislation regarding sensitive data in Norway we are not at liberty to share original data.

This is a registry based study design. The Primary outcome is extracted from routine data recorded by clinical and administrative personell who are unaware of study allocation in the electronic health care record.

The questionnaire outcome data can not be masked as the patient and study personell will be aware of treatment allocation group. To avoid preferential recording of these outcomes, study personell assist in data collection in either treatment or control groups - never both, so that they could not compare or favor one group above the other.