Health Policy Monitor
Skip Navigation

Predictive model for emergency admissions

Partner Institute: 
University of Barcelona
Survey no: 
(15) 2010
Contel JC (1), Rajmil L (2), Lopez S (2), Farre J (1), Corbella X (1)
Health Policy Issues: 
Long term care, System Organisation/ Integration, Political Context, Funding / Pooling, Quality Improvement, Access, Responsiveness
Current Process Stages
Idea Pilot Policy Paper Legislation Implementation Evaluation Change
Implemented in this survey? yes yes no no no yes no


Predictive modeling is currently tested in a pilot project in one area in Catalonia. A distinctive feature is that good quality data is available both from hospital and primary care information systems. The aim is to assess the likelihood of unplanned admissions and readmissions to hospitals. The Catalonian Health Evaluation Agency has conducted a first evaluation, showing similar results reported in the international literature. The pilot project will be extended to other areas in Catalonia.

Purpose of health policy or idea

Predictive modeling is an area of increasing interest and study. Well known Thomas Bodenheimer recently introduced both a policy paper and an article in the New England Journal of Medicine (NEJM) about the importance of complex chronic patients' identification and population stratification in the field in order to design different clinical strategies to cope with different segments of chronic care in a care management approach. Both new predictive models to identify populations at risk and proactive management of these patients should be combined to offer a comprehensive care management approach.

Main points

Main objectives

To identify patients with chronic conditions who are at risk of emergency hospitalization in order to give a proactive care management to these patients.

Type of incentives


Groups affected

Chronic Patients, Providers: Hospitals and Primary Health Care Centres, Catsalut (Commissioning body)

 Search help

Characteristics of this policy

Degree of Innovation traditional innovative innovative
Degree of Controversy consensual rather consensual highly controversial
Structural or Systemic Impact marginal rather fundamental fundamental
Public Visibility very low high very high
Transferability strongly system-dependent system-dependent system-neutral

Degree of Innovation: Predictive modeling in Catalonia is very innovative because it integrates primary care data. The quantity and quality of data from primary care centres is better than just three years ago, since incentives to identify chronic patients in clinical records was introduced recently. 

Transferability: Earlier foreign experiences are based on analyzing data from hospital utilization, just in some cases some primary care data are incorporated. In Catalonia, over 80% of primary care centres have the same primary care clinical record and information system. That is a very good opportunity to analyse data coming both from hospital and primary care. 

Political and economic background

Predictive modelling is a policy stimulated by integrated health organizations. In the case of Catalonia there is increasing interest based on the literature.

Change of government

There is increasing interest in exploring and experiencing first steps in this new approach.

Change based on an overall national health policy statement

It could become a national policy after experiencing and evaluating predictive modellling pilot projects.

Purpose and process analysis

Current Process Stages

Idea Pilot Policy Paper Legislation Implementation Evaluation Change
Implemented in this survey? yes yes no no no yes no

Origins of health policy idea

The idea to conduct a pilot project on predictive modeling was generated by new supporting literature. This has been discussed within the Primary Health Care Innovation Plan commissioned by Catalonian Department of Health in 2007. Afterwards, the Primary Care Innovation Plan contracted the Catalonian Health Agency to develop this new pilot project and to publish a report.

The idea of predictive modeling is based on identifying patients who could be at risk for emergency hospitalizations and on managing them proactively. This is very usual in insurance, weather forecasting and other sectors. Till now home health services are delivering care as a demand of a patient or his / her family. This new approach allows to identify patients at risk by analyzing data which is available in both primary care and hospital information systems.

Predictive modeling has been promoted in some pilot projects. The Department of Health will evaluate the pilots to consider the extension as a policy. 

Initiators of idea/main actors

  • Government
  • Providers
  • Payers: Catalan Health Service: public payer for health services in Catalonia
  • Patients, Consumers
  • Others

Approach of idea

The approach of the idea is described as:
new: It is vey new to the Catalan health system.

Innovation or pilot project

Local level - It is a pilot project related to almost 200.000 inhabitants.

Stakeholder positions

Predictive modeling is a very innovative concept in Spain. No earlier Spanish publications are available on this issue. However, the importance to identify people who are at risk of emergency admissions is increaingly acknowledged as this segment of population causes very high costs to the health system.

Forthcoming evidence generated in the pilot projects could encourage policymakers to introduce this health policy strategy. There are difficulties to share available data coming from hospital and primary care to collaborate in predictive modelling projects but it is desirable to be overcome.

Evidence from predictive modeling is very important for Commissioners (responsible for planning and contracting health providers) in order to encourage providers to be more proactive for specific vulnerable population segments. The Public Health Agency should be more interested in introducing this kind of population orientated approach.

Actors and positions

Description of actors and their positions
Department of Healthvery supportivesupportive strongly opposed
Welfare Departmentvery supportiveneutral strongly opposed
Primary care providersvery supportiveneutral strongly opposed
Hospitalvery supportivesupportive strongly opposed
Social services providersvery supportiveneutral strongly opposed
Catalan Health Servicevery supportivesupportive strongly opposed
Patients, Consumers
Patientsvery supportiveneutral strongly opposed
Family carersvery supportiveneutral strongly opposed
HTA agencyvery supportivevery supportive strongly opposed

Influences in policy making and legislation

This innovative approach is more introduced in companies which have the need of client segmentation, and in insurance companies to identify people who are at risk to be excluded or to recalculate premiums.

The Department of Health and the Catalan Health Service are increasingly interested in introducing this kind of care management approach. It could be extended as a health policy but it is not expected to be legislated or regulated.

Nevertheless, the National health service scheme, predictive modeling could be a very good opportunity to identify people at risk, manage them proactively to reduce cost and get better health outcomes.

Legislative outcome


Actors and influence

Description of actors and their influence

Department of Healthvery strongvery strong none
Welfare Departmentvery strongstrong none
Primary care providersvery strongstrong none
Hospitalvery strongstrong none
Social services providersvery strongneutral none
Catalan Health Servicevery strongvery strong none
Patients, Consumers
Patientsvery strongstrong none
Family carersvery strongstrong none
HTA agencyvery strongstrong none
HTA agencyHospitalDepartment of Health, Catalan Health ServiceSocial services providersWelfare Department, Primary care providers, Patients, Family carers

Positions and Influences at a glance

Graphical actors vs. influence map representing the above actors vs. influences table.

Adoption and implementation

The Catalan Health Technology Assessment Agency will validate an instrument to be expanded in different areas of the country. The Department of Health and Catalonian Health Service could offer this stratification tool to other regions.

There are no side effects related to this approach where little has been done till now in this area regarding proactive care management. Separately funded providers (hospital and primary care centres) should conceive the advantage of sharing data to analyse and return them to identify different population segments to be managed in a more appropriate way.

Monitoring and evaluation

First results confirmed results seen in former literature. The previous history of unplanned admissions, the presence of some chronic conditions and co-morbidities are the most important factors to be considered in order to identify people at risk.

Some countries with well developed predictive modeling strategies are advancing and introducing new data to enrich the model in order to better identify vulnerable populations.

Enhancing the database should be considered a very progressive pathway. Integrating data from primary care information systems, like it is done in the pilot projects in Catalonia, will enrich the value and validity of predictive modeling. The segmentation of the population into different (vulnerable) groups will be more effective.

Review mechanisms

Mid-term review or evaluation, Final evaluation (external)

Dimensions of evaluation

Structure, Process, Outcome

Results of evaluation

Major factors have been detected that predict a higher, if preventable use of emergency hospitalization:

Having had more than two unplanned admissions in the previous year (OR = 37.3, CI 95% = 25,6-54,3) is the most predictive cause of intake, together with age, sex, and chronic conditions such as diabetes, heart failure, ischemic heart disease, emphysema, COPD, asthma, more than 5 concurrent diseases and polipharmacy

Expected outcome

It is desirable to combine both strategies: to identify people at risk and to provide upstream care for these people using a variety of strategies, like proactive home care services, patient self-management and case management initiatives.

Impact of this policy

Quality of Health Care Services marginal rather fundamental fundamental
Level of Equity system less equitable four system more equitable
Cost Efficiency very low high very high

Predictive modeling of emergency admissions could be a starting point to reorient services for chronic patients who are at higher risk.

A health system that is able to identify people at risk and give them appropriate services is more equitable than others which are more reactive. Patients do not have the same level of need. Having innovative tools to identify them and act proactively is more equitable.


Sources of Information

Catalonian Health Evaluation Agency report:

Bodenheimer T, Berry-Millett R. Care Management of patients with complex health care needs. Princeton: Robert Wood Johnson Foundation, 2009.

Ham C.The ten characteristics of a high-performing chronic care system. Health Econ Policy Law. 2009: 7:1-20.

Health Services Management Centre (HSMC). Evidence for transforming community services. Review: Services for long term conditions. Birmingham: University of Birmingham, 2009.

Author/s and/or contributors to this survey

Contel JC (1), Rajmil L (2), Lopez S (2), Farre J (1), Corbella X (1)

(1) Institut Catala Salut (2) Agencia Avaluació Tecnologia i Recerca Mèdica

Suggested citation for this online article

Contel JC (1), Rajmil L (2), Lopez S (2), Farre J (1), Corbella X (1). "Predictive model for emergency admissions". Health Policy Monitor, April 2010. Available at