Population stratification: a fundamental instrument used for population health management in Spain: good practice brief

POPULATION STRATIFICATION: A fundamental instrument used for population health management in Spain GOOD PRACTICE BRIEF Motivation and summary José Cerezo Cerezo1 Carmen Arias López 2 Adjusted morbidity groups Changing health services from a disease-centred to a patient-centred approach was one of objectives of the Spanish Strategy for Approaching Chronicity in the National Health System (2012). A strategic priority for facilitating this transformation was considered to be identifi cation of the health needs of every patient, so that interventions could be tailored. In the framework of the project “Stratifi cation of the population of the National Health System”, a locally developed and tested “population grouper”, Adjusted Morbidity Groups (AMG), was used in the majority of the Spanish regions to stratify patients’ risks according to morbidity and complexity (Ministry of Health, Social Services and Equality, 2018). Risk stratifi cation is widely used in population health management, health service planning and clinical management. Stratifi cation of the health risks of people with chronic diseases has been adopted in many European countries to strengthen population health management and provide better-tailored services. Some countries have purchased or adapted existing software, and others, like Spain, have developed novel, country-specifi c population tools for grouping and health risk assessment (Dueñas-Espín et al., 2016; Nalin et al., 2016). These practices are aligned with the European framework for action on integrated health services delivery as one of the key strategies for moving towards people-centred health services (WHO Regional Offi ce for Europe, 2016). AMG were set up in the Catalan Health Service by the Catalan Health Institute and the TicSalut Foundation as part of the Catalan Prevention and Chronic Care Programme. Later, the Spanish Ministry of Health, Social Services and Equality promoted two consecutive collaboration agreements with the TicSalut Foundation (Catalan Health Service), which enabled the extension of the AMG from Catalonia to the vast majority of the Spanish regions. By 2015, 38 million people had been grouped (Monterde et al., 2016; Ministry of Health, Social Services and Equality, 2018). 1 Consultant, WHO Barcelona Office for Health Systems Strengthening, Division of Health Systems and Public Health. 2 Technical officer of the Strategy for Addressing Chronicity in the National Health System, Subdivision of Quality and Innovation, Ministry of Health, Social Services and Equality. Key Messages • Risk stratifi cation tools such as the Adjusted Morbidity groups (AMG) can assist health systems in progressing from disease-centred to patient-centred care. • The AMG can be used to estimate current and future risks for mortality, morbidity and various indicators of health service utilization, enhancing health care management. • The AMG are particularly relevant for addressing patients with chronic comorbid conditions from both a system-wide and a clinical approach and allow benchmarking at various levels. • The AMG have proved to be fl exible and transferable among regions. • To develop and put into practice a tool of this nature, reliable, up-to-date, systematized, homogeneous, computerized primary health care records are indispensable. Process Fig.1. Stratifi cation pyramid based on complexity index Grouping by morbidity: Individuals are classifi ed into one of seven morbidity groups by their assigned international diagnostic codes, as follows: healthy population, pregnancy and/or labour, acute disease, chronic disease in one system, chronic disease in two or three systems, chronic disease in four or more systems and cancer. Grouping by complexity: Each morbidity group (except the healthy population) is divided into fi ve subgroups of complexity, the level of which is determined by analysis of a set of resource use variables, such as primary care visits, pharmaceutical prescriptions, mortality and risk for hospital admission. The complexity calculation was based on information on the population of Catalonia in 2011 (7.5 million). Combining morbidity and complexity resulted in 31 AMG. Individual clinical labels and complexity index: The AMG includes two additional kinds of information on each patient. First, an individual clinical label for the most relevant and/or prevalent disease is selected from a list of 80 agreed, prioritized health problems. Secondly, a numerical complexity index is calculated that allows pyramidal risk stratifi cation, in which each patient is allocated to a risk level or stratum (Fig. 1). AMG is a tool for population grouping and risk stratification that takes into account two factors: multimorbidity and complexity. The process requires the codified diagnostic codes of users’ morbidity, the date of diagnosis and, as the principal source of information, data collected in electronic primary health care records. Acute diagnoses are taken into consideration only if they were made during the study period (usually one year), while chronic diagnoses are considered regardless of the date. Nature of AMG The AMG is helpful for better understanding the distribution of health risks in the population. The Spanish health system is coordinated nationally but is decentralized to the 17 Spanish regions, many of which have used the AMG for diff erent analytical purposes. Some examples are given below. High-risk population: people with an individual complexity value greater than that of the 95th percentile of the population with chronic disease Moderate-risk population: people with an individual complexity value between the 80th and 95th percentiles of that of the population with chronic disease Low-risk population: people with an individual complexity value lower than that of the 80th percentile of the population with chronic disease Population with no chronic disease Temporal and geographical distribution of morbidity: Fig. 2 shows the proportions of the population in the seven morbidity groups in one region: 68% had at least one chronic disease and 44% experienced multimorbidity. About 15% of the patients with multimorbidity had chronic diseases that aff ected four or more organ systems. Such simple fi gures can be used for analysis over time and by geographical area for better planning decisions, for example. Towards population health management A B C D Document number: WHO/EURO:2018-3032-42790-59709 © World Health Organization 2018 Source: Ministerio de Sanidad, Servicios Sociales e Igualdad, 2018 Source: Ministerio de Sanidad, Servicios Sociales e Igualdad, 2018 Fig. 2. Population distribution by morbidity group Acute disease Pregnancy and labour Chronic disease in four or more systems Chronic disease in two or three systems Chronic disease in one system Healthy population Active cancer Fig. 3. Mortality and resource use by risk stratum Resource needs by health risk. Fig. 3 shows the results in another region, where pyramidal risk stratifi cation was used to calculate the values for a set of variables of resource use and cost analysis for every risk stratum. The highest risk stratum, which includes only 5% of the population, consumed the highest share of health resources, with a clear gradient across the risk spectrum. Changes over time and by geographical area can be useful for ensuring that resource allocation patterns meet needs. Life-course and gender distribution of health risk. In a third region, patients stratifi ed according to their complexity index were distributed on the population pyramid as depicted in Fig. 4. This revealed the numbers of high-risk individuals by sex and age group. As observed in this example, the higher the age group, the more patients are in the high-risk stratum. This becomes especially relevant for those over 65 years, who are most likely to have multimorbidity. High-risk patients are also seen in the age group 50-60 years, with a direct impact on the labour and economic sectors. This information can be particularly helpful for designing interventions. High- risk population Moderate-risk population Low-risk population Baseline-risk population Morbidity groups % Healthy population Pregnancy and labour Acute disease Chronic disease in one system Chronic disease in two or three systems Chronic disease in four or more systems Active cancer 13.4 1.1 16.1 24 29.2 14.9 1.3 Population (%) Mortality rate (x 100) Visits to primary care (mean) Emergency admission rate (x 100) Emergency visit rate (x 100) Dispensed drugs (mean) Health care expenditure (mean) 5 16.6 22.2 58.1 160.8 13.4 7067€ 15 1.1 12.4 7.5 72.5 8.0 2121€ 30 0.2 7.0 2.9 51.9 3.6 779€ 50 0.1 2.0 0.6 17.3 1.0 164€ In a survey carried out in 2017 by the Ministry of Health, Social Services, and Equality, Spanish regions reported a wide range of applications for the AMG for better health management and resource planning. Population health management and case fi nding: The most widespread use is in fi nding cases in primary and secondary health care in order to include them in the regional programmes for complex or advanced chronic patients. For example, in Madrid, the risk calculated in the AMG is used, with other variables and with a clinician’s validation, to refer patients to such programmes. The strength of the concordance between the AMG risk levels (high, medium, low) and the intervention levels assigned by physicians (high, medium, low) was assessed as moderate to good (González González et al., 2017). Impact and uses of the AMG Fig. 5. Frequency (%) of the main chronic diseases by risk stratum Chronic conditions and health risk: The proportion of high-risk patients in each of the most relevant noncommunicable diseases can be also determined by the AMG. Fig. 5 shows the results for a fourth region of Spain, with the proportions of the population at high risk shown in red. The proportion of patients in the highest complexity category depends on the disease. Whereas the vast majority of people with high blood pressure were in the moderate or low risk strata, a high proportion of patients with heart failure were classified as high-risk patients. This information is valuable for forecasting use of health resources according to the noncommunicable disease profile of the population. Fig. 4. Population distribution by age, sex and risk stratum Source: Ministerio de Sanidad, Servicios Sociales e Igualdad, 2018 Source: Ministerio de Sanidad, Servicios Sociales e Igualdad, 2018 • The information provided by risk stratifi cation tools such as the AMG can assist health systems in progressing from disease-centred to patient-centred care. Through better patient health profi ling, health services can respond more accurately and comprehensively to the actual health needs of both groups and individuals. • The AMG can be used to estimate current and future risks for mortality, morbidity and various indicators of health service utilization, enhancing health care management by the introduction of transparent, evidence-based criteria in decision-making about health programmes, policies and resource allocation. • The AMG are particularly relevant for addressing patients with chronic comorbid conditions from both a system-wide and a clinical approach. Such patients are readily identifi ed through the AMG and can be included in case management programmes for patients with complex chronic disease (with physician validation). In addition, inclusion in the AMG of information from electronic health records allows health professionals (physicians, nurses) to forecast a patient´s prognosis and tailor clinical interventions accordingly. • The AMG allow benchmarking at various levels. Health managers can identify and compare areas with larger health demands and resource consumption with better-performing areas. In addition, physicians and nurses can compare patients according to their complexity index and with average rates in their health area. • The AMG have proved to be fl exible and transferable among regions, as shown by their use in 13 of the 17 Spanish autonomous health systems. • In order to develop and put into practice a tool of this nature, reliable, up-to-date, systematized, homogeneous, computerized primary health care records are indispensable. • In decentralized health systems such as that in Spain, successful regional initiatives can be identifi ed and scaled up if there are adequate mechanisms for selecting good practices and eff ective collaboration agreements. Lessons learned Proactive case management of high-risk patients in primary care: The AMG are also used as routine health indicators in individual primary health care records for proactive clinical decision- making. In some regions, such as Catalonia, the AMG risk score is listed in the patient’s electronic health records and is thus accessible to health care professionals (physicians, family nurses, case managers etc.). They can then draw up a list of their most complex patients by combining the information provided by the stratifi cation tool with other clinical variables and therefore compare a patient with the rest of their assigned population. Resource planning: In other regions, the AMG are used in macro-management to estimate current or future health care costs and resource utilization in order to allocate health resources accordingly. For example, in the Balearic Islands, the AMG are used to calculate the annual pharmaceutical budget of primary health care family physicians. The pharmaceutical expenditure and the complexity indexes of the patients assigned to the family practitioner in the previous year are used to estimate the next year’s budget, and the estimates are adjusted by the practitioner’s actual pharmaceutical expenditure to determine the fi nal annual budget. Strategic purchasing: The AMG are used in Catalonia with other variables to adjust the annual per capita payment to primary health care teams. In the Madrid region, the AMG are used to calculate the capitative prescription budget of primary health care centres. Health workforce planning: The AMG can contribute to optimizing health workforce planning and allocation. In Catalonia, the Nursing Council has proposed a new model for establishing the minimum number of nurses required on primary health care teams to ensure the quality of health care. In this model, morbidity measured by the AMG was part of the allocation formula. Research and decision-making in public health: Research uses may include identifi cation of vulnerable groups, analysis of population morbidity or selection of controls for epidemiological studies. One region used the AMG complexity index to prioritize individuals for eligibility for infl uenza vaccination and for alerting them by SMS. Performance assessment: The AMG are also used to adjust many indicators of effi ciency and quality in primary and emergency health care. This brief is part of our work programme on strengthening the health system response to NCDs. For more information, check out our website at http://www.euro.who.int/en/health-systems-response-to-NCDs. © World Health Organization 2018 Dueñas-Espín I, Vela E, Pauws S, Bescos C, Cano I, Cleries M et al. Proposals for enhanced health risk assessment and stratification in an integrated care scenario. BMJ Open 2016;6(4):e010301. González González AI, Miquel Gómez AM, Rodríguez Morales D, Hernández Pascual M, Sánchez Perruca L, Mediavilla Herrera I et al. (2017) Concordancia y utilidad de un sistema de estratificación para la toma de decisiones clínicas. [Concordance and usefulness of a stratification system for clinical decision making.] Aten Primaria 2017;49(4):240–7. Ministerio de Sanidad, Servicios Sociales e Igualdad. Informe del proyecto de Estratificación de la población por Grupos de Morbilidad Ajustados (GMA) en el Sistema Nacional de Salud (2014–2016). [Report on the project for population stratification by Adjusted Morbidity Groups (AMG) in the National Health System (2014–2016).] Madrid; 2018 (http://www.msssi.gob.es/organizacion/ sns/planCalidadSNS/pdf/informeEstratificacionGMASNS_2014-2016.pdf ). Monterde D, Vela E, Clèries M. Los grupos de morbilidad ajustados: nuevo agrupador de morbilidad poblacional de utilidad en el ámbito de la atención primaria. [Adjusted morbidity groups: a new method for grouping population morbidity for use in the field of primary care.] Aten Primaria 2016;48(10):674– 82. Nalin M, Bedbrook A, Baroni I, Romano M, Bousquet J, editors. White paper on deployment of stratification methods. Brussels: European Commission; 2016. WHO Regional Office for Europe. Strengthening people-centred health systems in the WHO European Region: framework for action on integrated health services delivery. Copenhagen; 2016 (http://www.euro.who.int/__ data/assets/pdf_file/0004/315787/66wd15e_FFA_IHSD_160535.pdf?ua=1 ). We acknowledge the contribution of the representatives of the Autonomous Regions in the Committee of the Strategy for Addressing Chronicity in the Spanish National Health System and their information systems managers who participated in this project. We also thank all the health professionals who register relevant clinical data in their daily work, which is essential for the AMG. Acknowledgments References Contact us Document number: WHO/EURO:2018-3032-42790-59709