Strategy for proactive integrated care for high-risk, high-cost patients/Estrategia de atención proactiva integrada a pacientes en riesgo de alto consumo de recursos

Introduction Healthcare models are evolving from healthcare fragmented across centres and levels of care, towards an increasing integration of services. This new model makes it necessary to identify the specific needs of target populations and, based on this, diversify the services provided to better meet the requirements of various population groups and even specific individuals [1]. In recent years, several centres of excellence working in the field of integrated care (Kaiser Permanente, the Veterans Health Administration, and the British NHS, among others) have been dedicating considerable and growing efforts to the identification of the most complex segment of the population by analysing already existing data and the introduction of proactive care models [2, 3]. The Baix Empordà Integrated Health Service (SSIBE Serveis de Salut Integrats Baix Empordà) is an integrated health organisation, responsible for the public healthcare services for a catchment population of 125,000. It has a single healthcare structure and also a single electronic medical record database, both of which are clear facilitators for the integration of healthcare [4]. Since 2001, SSIBE has focused a line of research and development on the analysis of morbidity and expenditure considering individuals. We have grouped patients on an ongoing basis by morbidity using Clinical Risk Groups (CRGs) and analysed the relationship of these groups with resource use; it was found that 1% of patients used 22% of the healthcare budget [1, 5]. Given this, we took on the challenge of prioritising healthcare and resources to complex chronic patients. In this context, we developed a predictive model to identify the target population, based on the observed morbidity with high total health expenditure as the dependent variable. This pilot study allowed us to identify factors related to chronicity associated with a greater use of resources [6]. Our objective is to be able to define a proactive health-care strategy for potentially high-cost patients identified using a predictive model. Once defined, this strategy should allow us to divert resources to and tailor interventions for this type of patient. A second objective is to assess whether the interventions adopted are able to decrease the morbidity compared to the expected rates. The strategy is divided into two parts: first, we should define the intervention project, the predictive model and its validation. Secondly, we should design a specific intervention that allows us to assess results by performing a clinical trial. The current paper focuses on the first part of …


Introduction
Healthcare models are evolving from healthcare fragmented across centres and levels of care, towards an increasing integration of services. This new model makes it necessary to identify the specific needs of target populations and, based on this, diversify the services provided to better meet the requirements of various population groups and even specific individuals [1].
In recent years, several centres of excellence working in the field of integrated care (Kaiser Permanente, the Veterans Health Administration, and the British NHS, among others) have been dedicating considerable and growing efforts to the identification of the most complex segment of the population by analysing already existing data and the introduction of proactive care models [2,3].
The Baix Empordà Integrated Health Service (SSIBE Serveis de Salut Integrats Baix Empordà) is an integrated health organisation, responsible for the public healthcare services for a catchment population of 125,000. It has a single healthcare structure and also a single electronic medical record database, both of which are clear facilitators for the integration of healthcare [4]. Since 2001, SSIBE has focused a line of research and development on the analysis of morbidity and expenditure considering individuals. We have grouped patients International Journal of Integrated Care -Volume 12, 29 May -URN:NBN:NL:UI:10-1-112963 / ijic2012-28 -http://www.ijic.org/ on an ongoing basis by morbidity using Clinical Risk Groups (CRGs) and analysed the relationship of these groups with resource use; it was found that 1% of patients used 22% of the healthcare budget [1,5].
Given this, we took on the challenge of prioritising healthcare and resources to complex chronic patients. In this context, we developed a predictive model to identify the target population, based on the observed morbidity with high total health expenditure as the dependent variable. This pilot study allowed us to identify factors related to chronicity associated with a greater use of resources [6].
Our objective is to be able to define a proactive healthcare strategy for potentially high-cost patients identified using a predictive model. Once defined, this strategy should allow us to divert resources to and tailor interventions for this type of patient. A second objective is to assess whether the interventions adopted are able to decrease the morbidity compared to the expected rates.
The strategy is divided into two parts: first, we should define the intervention project, the predictive model and its validation. Secondly, we should design a specific intervention that allows us to assess results by performing a clinical trial. The current paper focuses on the first part of the project.

Methods
The project is being developed in four health areas (HAs), geographical zones for the administration of health in our region, with a total catchment population of 93,233 (16.3% >65 years of age). Two workshops have been held with the key stakeholders involved: primary care doctors, internists, nurses, geriatricians, emergency doctors, and social workers. These workshops were aimed and structured to promote discussion and reach conclusions by consensus. On the basis of these sessions, the interventions to be carried out and corresponding lines of action were established ( Table 1).
We built a predictive model for high-cost patients on the basis of age, sex and morbidity according to the CRGs, expenditure on medications >p95, use of hospital drugs in outpatient settings, and hospitalisation.
The intervention project was then tested in one of the HAs and its usefulness was assessed.

Results
These are obtained from three phases: Identification of 4833 high-risk, high-cost patients; 1.
80% being in CRGs 5, 6 and 7 (single dominant or moderate chronic; multiple significant chronic; or three or more dominant chronic conditions). Their previous use of resources compared to other patients is reported in Table 2.
Setting up of an alert identifying complex chronic 2.
patients in the shared electronic medical records and distribution of these patients by primary care doctor. Table 1. Definition of the lines of action for complex chronic patients Line 1: Identification of the target population Line 2: Accessibility and coordination of care Establish a coordinated care pathway for use on discharge to ensure that patient care is well coordinated in primary care Define and implement emergency care protocol for complex chronic patients Line 3: Proactive actions for patient monitoring Line 4: Organisational culture Make available a guide to social services that defines the procedures to be followed Create forums to make all healthcare professionals involved aware of the different strategies one year after the introduction of the programme. For this purpose, we plan to carry out a study with the design of a clinical trial to control for the variable "intervention" and quantify its effect.
It is already clear, however, that this strategy must be complemented by other strategies in the areas of the emergency services, hospitalisation and social care, in order to adapt care to meet needs of this type of patient.
Specification of the interventions to be carried out 3.
and their validation using a pilot, carried out in one of the HAs with the involvement of nine healthcare units and a group of 94 patients. The common actions that should be carried out included reviewing and updating of: prescriptions, and assessment of the level of adherence to medications, attendance to primary care/hospital appointments and more specific issues (such as use of inhalers, dietary control, and social risk). We set aside time in the working day of the professionals for them to carry out these interventions. The programmed interventions included: appointments, home visits and/or telephone consultations. Outcomes can be seen in Table 3.

Discussion
We set up an overall strategy using a predictive model to identify high-risk, high-cost patients. This identification of such patients facilitates the task of care of primary care doctors/nurses and continuity with the other levels of care. The relatively small number of patients selected and assigned to each primary care doctors/ nurses enables personalised analysis and intervention. The intervention defined has been found to be feasible and has made it possible to identify areas of improvement in the monitoring and control of chronic patients by primary care professionals.
In order to act proactively, it is now necessary to define specific interventions. We aim to evaluate the results
Nos planteamos el reto de priorizar la atención y los recursos al servicio de los pacientes crónicos complejos. En este marco hemos desarrollado un modelo predictivo para la identificación de la población diana, basado en la morbilidad atendida y en que la variable dependiente es el gasto sanitario total elevado. Esta experiencia nos permite identificar los factores asociados a un mayor consumo de recursos en relación a la cronicidad [6].
Se ha validado el proyecto de intervención en un ABS y se ha valorado su adecuación.
La intervención definida ha sido factible y permite identificar aspectos de mejora en el seguimiento y control de los pacientes crónicos por parte de los profesionales de atención primaria.