OUP user menu

Health indicators in the European regions—ISARE II

John Wilkinson, Luc Berghmans, Fréderic Imbert, Bernard Ledésert, André Ochoa
DOI: http://dx.doi.org/10.1093/eurpub/ckm088 178-183 First published online: 30 August 2007

Abstract

Background: Most comparisons of health data in Europe take place at the national level. However, there is increased interest in looking at health data at a sub-national level. This is because of the increased importance in many European countries, of regions and devolved powers to them. This study aimed to establish the availability of health data at a regional level and to construct an experimental database. Methods: Using a network of country correspondents, data were collected on a series of topics from all the regions of that country. In addition, a supplementary list of data was collected from one region of each country. Results: Out of the then 15 Member States of the European Union (EU), 14 countries participated in the study. Thirteen countries were able to supply data. Where data were available, using the criteria we developed, these were of relatively good quality. Data on mortality was most readily available, but data on the important public health topics such as obesity was much more difficult to obtain, and absent in many cases. Conclusions: It is possible to construct a database and a resultant set of indicators for relevant sub-national areas of Member States in the EU. This is not likely to be achieved through current routine data collection systems unless significant changes are made to the data collection processes such as those undertaken by Eurostat. There is, also, an urgent need to introduce comprehensive sub-national data collections on important public health topics such as obesity and smoking.

  • European Union
  • health indicators
  • regions

Introduction

In June 1997, the European Parliament adopted a plan of action for public health across the European community. This was called the Health Monitoring Programme (HMP).1 This had the remit of contributing to the creation of a community-wide system for monitoring health in order to measure health status, trends and health determinants throughout the community, facilitate planning, monitoring and evaluation of community programmes and actions. It also had the remit to provide Member States with appropriate health information to be able to make comparisons and to support their national health policies.

Furthermore, the importance of public health was strengthened by the Amsterdam Treaty which introduced a competency of public health for the first time in the European Union (EU).2

To achieve the objectives of the HMP, three types of action have been undertaken: the creation of community health indicators; the development of a community-wide network for sharing health data and analyses and work to improve the quality of health reporting on health in the EU e.g. PHR-EVA and PHR-PIA projects.3,,4

A number of projects have been undertaken in the countries of the EU. Some have been concerned with defining a set of common indicators that enable data concerned with health and methods for investigating health issues to be compared e.g. European Community Health Indicators (ECHI) project.5 However, much of the work carried out prior and during the HMP is concerned with comparisons at a national level. Although this approach has great value, it does not take into account the fact that a substantial amount of analysis and decisions regarding health issues are now taken at sub-national level. It is in this context that the National Federation of Regional Health Observatories, France (FNORS) established the Indicateurs Sante Régionaux d’Europe (ISARE) project to focus on the health of regions within the EU.

An initial project (ISARE I), was conducted during 1999–2001. This project identified (in the then 15 countries of EU) the sub-national level, which was most appropriate for the production and the comparison of indicators (the ISARE health regions). In addition, this work assessed data availability and the data sources at these sub-national levels.6,,7 In ISARE I, 300 health regions across 13 countries were identified with an average population of 1.2 million (but with considerable variation). In 10 out of the 13 countries, the recommended regions corresponded to a level of local democracy and in nine countries they corresponded exactly to one of the levels of the Nomenclature of Territorial Units for Statistics (NUTS) classification.

This first project had allowed the building of a network of country representatives, which formed the basis of the planning for the Phase 2 of the project, together with the comments and suggestions made by the European Commission.

Background

Local government is developing rapidly, and the importance of the regions as units of political and administrative management is increasing in Europe. In Spain, for instance, regional autonomous communities have acquired a high level of autonomy. In France, health care planning is already performed by regions, which are in charge of allocating budgets to hospitals. The tendency towards increasing decision making at regional level requires information at this level, such as needs assessment and analysis of service utilization.

Sharing such regional information across Europe allows health professionals and decision makers to understand the characteristics of their own region in the wider context of all European regions. Similarities and differences may raise questions and stimulate interaction about the approaches chosen for solving public health problems. The EU has also increasingly focused its activity on regions, with particular development activity (and funding) being channelled through the regions.

Other reasons draw on the epidemiological importance of sharing regional health information. First, producing health indicators at a sub-national level allows the identification of epidemiological patterns, otherwise hidden by national averages. Linked to this argument, is the well-known fact that public health problems do not respect national boundaries.8 Currently, Eurostat collects and publishes a very limited range of health indicators at regional level, but has been pointed out in our previous report, this relates to NUTS 29 regions, which is not always relevant to decision-making processes in Member States

The main objective of the second ISARE project was to assess the feasibility of collecting data at the defined sub-national levels in the countries of the EU.

Methods

The methods for this study had three strands: the creation of a partnership with the representatives (or correspondents) of the different countries of the EU; the development of a survey instrument to collect the selected data and the building of an experimental database.

Representatives were contacted in all the participating countries (all the then countries of the EU) who acted as the main suppliers of data from each country to the project. These representatives were identified through the Working Party Committees of the HMP. As a result, the representatives, including a range of individuals, all working at senior level in information and/or public health in each country. We were unable to identify a representative from Denmark who was able to participate. These representatives were in the most part, the same representative who had participated in the ISARE I project, thus giving useful continuity in the project.

Key aspects of the project were: to determine the two lists of the data to be collected at the regional level; to develop clear definitions for these data and the year to be collected; to update the data on health regions collected in the first ISARE project, to assess sources of regional-level data and to comment on its quality.

Two lists of indicators were constructed, a main list and a supplementary list. Indicators were selected taking into account the first version of the ECHI list.5 We started with a list of potentially 80 indicators, which had been identified in the ISARE I project and added to by ECHI. The steering group determined the final list of indicators taking into account relevance and availability (for example, we decided that to be considered for inclusion, indicators must be available in at least seven countries—out of the 13 in the study). We asked the correspondents of the ISARE project in each country to assess the availability of those indicators, those most available were selected to build the main list, and a second set of less available indicators were selected to build the supplementary list. Data on screening coverage was rarely available at regional level; however, the number of new cases of breast cancer was available, but not generally for cervical cancer.

The lists of indicators are shown in table 1. The main list was collected for all the regions of the country and the supplementary list of data was collected from one region in each country.

View this table:
Table 1

The main and supplementary lists of data collected

Main list of data collected in each region of each country
Health care professionals
    (1) Number of physicians
    (2) Number of nurses (including midwives)
    (3) Number of nurses (excluding midwives)
    (4) Number of midwives
Health care services
    (5) Number of acute care hospital beds
    (6) Number of hospital in-patients admissions
Demographic and socio-economic data
    (7) Mid-year population estimate
    (8) Number of live births
    (9) Number of deaths
    (10) Percentage of unemployed persons (15–64-years old)
Mortality data
    (11) Number of perinatal deaths
    (12) Number of stillbirths
    (13) Age/sex breakdown of deaths by cause
Morbidity data
    (14) Number of new cases of AIDS patients
Risk factors
    (15) Distribution of BMI in the population
    (16) Percentage of regular smokers aged 15 years or more
Living and working conditions at work
    (17) Number of persons injured or killed in road traffic accidents
Supplementary list of data collected in a selected region for each country
Health care professionals
    (18) Number of general practitioners
    (19) Number of dentists
    (20) Number of pharmacists
Health care services
    (21) Number of hospital beds, acute care, gynaecology or obstetrics or maternity beds
    (22) Number of bed days, acute care/year
    (23) Number of bed days, acute care/year, gynaecology or obstetrics or maternity
    (24) Number of hospital in-patients admissions, gynaecology or obstetrics or maternity
    (25) Number of caesarean sections
    (26) Number of cataract operations
    (27) Number of hip replacements
    (28) Number of induced abortions
Demographic and socio-economic data
    (29) Percentage of the adult population (25–64-years old) completed upper secondary education
    (30) Percentage of the adult population (25–64-years old) completed tertiary education
Morbidity data
    (31) Number of tuberculosis cases
    (32) Number of female breast cancers
Living and working conditions
    (33) Number of cases of accidents related to work
Prevention data
    (34) Percentage of infants vaccinated against diphtheria
    (35) Percentage of infants vaccinated against tetanus
    (36) Percentage of infants vaccinated against pertussis
    (37) Percentage of infants vaccinated against poliomyelitis
    (38) Percentage of infants vaccinated against measles

Definitions were established by the steering group, which were based on existing definitions from the WHO Health for all Database,10 or alternatively OECD11 and the ILO.12 For the data collection during this phase of the project, 1999 was chosen as reference year. When data were not available for 1999, data for the closest year was requested (before or after 1999).

The first ISARE project identified appropriate regions for comparison6 for the 13 participating countries out the 15 EU Member States. All recommended levels had responsibilities in the field of health promotion and all but one performed the function of public health reporting. Ten out of the 13 recommended regions corresponded to a level of local democracy and nine corresponded exactly to one of the levels of the NUTS9 classification. No recommendation for a regional level could be made for Finland or Greece. In ISARE II, it was necessary to determine regions for Finland and Greece. For Greece, the new health regions defined in autumn 2001 were used because they corresponded to a political level. For Finland, ‘Hospital districts’, a level of organization of hospital care, were chosen.

We also took account of reforms that had occurred in England. Health authorities were abolished in 2002. We took the decision to use government office regions for the second phase of ISARE.

We looked at the data under the following headings: availability, conformity to the definitions, use and comparability.

To analyse availability, conformity to the proposed definition and the quality of each of the data items, we established a number of criteria to assess these. First, a score was attributed to each of the data items to judge its availability. The availability score was calculated as follows:

Availability score = Number of countries for which data were fully available (all regions, 1999) + 0.5 × number of countries from which the data is fully available, but only for certain regions and/or for a year other than 1999.

In table 2, availability of the data and conformity to the definitions has been ascribed a category of ++, +, +/−, −, −− in a consistent way related to the calculated score. Details of calculation of these scores are provided in the full report of the project.13

View this table:
Table 2

Summary table of sub-national data collected in the ISARE II study

Number of regionsAvailabilityConformity to the definitionUseComparabilityTo include in the database
Health Professionals
Data collected for all regionsPhysicians+++++++Yes
Nurses (including midwives)+/−+++++No
Nurses (excluding midwives)+++++Yes
Midwives+/−++++++No
Data collected in one regionGeneral practitioners+++++++Yes
Dentists++++++++Yes
Pharmacists+++++++Yes
Health Services
Data collected for all regionsNumber of beds for acute care++++++++Yes
Hospital in-patients admissions+++++++Yes
Data collected in one regionHospital beds in gynaecology, obstetrics and maternity+++++++Yes
Bed days, acute care/year++++++++Yes
Bed days/year in gynaecology, obstetrics and maternity++++++Yes
Hospital in-patients admissions/year in gynaecology, obstetrics and maternity++++++Yes
Caesarean sections++++++++Yes
Cataract operation++++++++Yes
Hip replacements+++++++Yes
Induced abortions+++++++Yes
Demographic and socio-economic data
Data collected for all regionsMid-year population estimate+++++++Yes
Live births++++++++Yes
Deaths++++++++Yes
Percentage of unemployed in range 15–64years++++++Yes
Data collected in one regionPercentage of the adult population (25–64years) having completed secondary education++/−+++No
Percentage of the adult population (25–64years) having completed higher education++++No
Mortality data
Data collected for all regionsPerinatal deaths+++++++Yes
Stillbirth+++++++Yes
Deaths by age, sex and cause+++++++Yes
Morbidity data
Data collected for all regionsNew cases of AIDS+/−+/−++++No
Data collected in one regionNew cases of tuberculosis++++++++Yes
New cases of breast cancer+/−++++No
Risk factors
Data collected for all regionsDistribution of BMI+/−++++No
Percentage of smokers aged over 15 years+/−++++No
Conditions of life and work
Data collected for all regionsPersons injured or killed in road traffic accidents+++++++Yes
Data collected in one regionAccidents at work+++++++Yes
Prevention data
Data collected in one regionPercentage of children immunized against
Diphtheria+/−++++No
Tetanus−−+++No
Pertussis++++No
Poliomyelitis+/−++++No
Measles++/−++++Yes

Data quality was assessed in two ways. First, by asking the country representative to provide some judgement and second, by comparing the reports of other reports from the HMP. To provide some objective assessment the following questions were asked to the country representatives:

  • Is the data used in the region?

  • From your point of view, is the data quality good enough to perform comparisons between the regions of your country?

  • From your point of view, is the data quality good enough to perform time comparisons in your regions?

We then ascribed a score, shown in table 2 of the proportion of countries with a high degree of confidence in the data using the same categories as above (++, +, +/−, −, −−).

Results

The results of the data collection process are shown in table 2. Of the 14 countries participating in the ISARE II project, 13 countries provided data (Austria, Belgium, England, Finland, France, Germany, Greece, Ireland, Luxembourg, the Netherlands, Portugal, Spain and Sweden) for the section of the questionnaire relating to all regions. Finland and the Netherlands were only able to provide information on a limited number of topics, and Spain, was only able to provide information for a limited number of regions.

For the data from one region alone, 12 countries responded: Austria, Belgium, England, Finland, France, Germany, Greece, Ireland, Luxembourg, Portugal, Spain and Sweden. We did not attempt to ascertain whether these regions were representative of the countries concerned, as the main objective was to obtain the data. Most of these regions related to the base regions of the country representatives

Availability of data was generally good (16 data items) or very high (12 data items) and was never considered ‘very bad’. However, availability was never perfect insofar as no data item was provided for every one of the regions studied and for the year requested (1999). The birth and death data were found to be most readily available along with the number of acute hospital beds. Smoking and body mass index (BMI) data were the data least likely to be available in the main list and in the supplementary list the vaccination uptake rate for pertussis or tetanus was least likely to be available.

Conformity of the data to the proposed definitions based on those of WHO Europe and OECD, is better. So when data were available, conformity in a very large majority of cases, is very high (21 data items, considered by over 80% of our correspondents to conform to the definition) or high (nine data items, considered by 71–80% of our correspondents to conform to the definition). For three data items conformity could only be considered moderate, for four data ‘bad’ and for one data item as ‘very poor’. Data on mortality by age, sex and cause was generally readily available at sub-national-level on a consistent basis.

To shed more light on the quality of the data, two questions, on comparability, were posed to the country correspondents:

  • Do you consider the quality of the data sufficient to effect comparisons between regions of your country?

  • Do you consider the quality of the data sufficient to effect comparisons within your region of your country?

From the results, it appears that comparability is always ‘very high’ (31 data items) or ‘high’ (six data items). Only one item is in the ‘poor’ category. The few reservations expressed concerned comparability over time, changes in definitions, methods of gathering data and more rarely sub-national comparisons.

On the basis of this survey, we suggest that a large part of the data studied (27 out of 38 items) could immediately be incorporated into a sub-national European database, given its availability and quality. Data on health services, mortality and working conditions have been retained in the database. Their conformity and comparability was always high or very high. Their availability is also satisfactory or very satisfactory with few exceptions. On the other hand, data on health professionals, vaccination coverage for children, morbidity and the data concerning demographic and socio-economic risk factors pose more problems. Amongst data on health professionals, we recommend inclusion of data relating to nurses, but not midwives because of the availability and definitional problems. All the other data from this group can be incorporated. Both the demographic and socio-economic data related to the study levels could not be included in the experimental database for data collected in one region because of poor conformity to the definitions and poor availability. It was possible to include the four other data items for all regions in this category. For the data relating to morbidity, it appeared that availability and conformity to the definition was too poor for the number of new breast cancers to be included for the time being. As regards vaccination, of five data items studied, we suggest that only the percentage of children vaccinated against measles could be incorporated into a European database. Effectively, the rates of vaccination against diphtheria, tetanus, pertussis and poliomyelitis do not routinely conform to the definitions, nor are they better than moderately or poorly available. Finally, the two data items concerning risk factors (distribution of BMI and percentage of smokers aged over 15) could not be included either, as on the one hand a high number of correspondents considered them not to be comparable either in time or place, and also their availability was poor.

Case study infant mortality

This concerns the health indicator, infant mortality. Of the 214 ISARE health regions in the 14 countries participating in the ISARE project, data were collected from 156 regions to calculate infant mortality (73% of regions). The average level of infant mortality over all the regions studied is 4.8 deaths under 1 year per 1000 live births.

The range of the country mean was between 3.7 and 5.7; however, the range within all the regions was 0.9–11.7. Variability within each country was always greater than the variability found between countries.

Figure 1 shows these data presented as a map. Data is available from the website.

Figure 1

Infant mortality rates in the ‘Isare Health Regions’ in Europe 1999

Discussion

ISARE I project painted a very optimistic picture of data availability at the regional level in Europe. However, ISARE II only partially confirmed this. According to ISARE I, demographic and socio-demographic data should have been available in all the health regions, but this was not verified in ISARE II. It was notable that the data, which are required to support some of the most important public health challenges facing the EU, such as smoking and obesity, were not readily available at regional level.

The ability of country representatives to supply the data was dependent on their individual ability (and willingness) to supply the data, but in general the people who were involved in ISARE I and II were by and large the same. In some countries (France and England), the data collection was relatively easy as the information was available from national databases. In other countries, it was necessary to use multiple sources.

This was particularly the case in Spain where federalism is very marked. The correspondent made the choice to use regional organizations to assemble the data, even though national databases exist. This was reported as being due to a better quality of the regional data. However, this meant that data was not returned for all regions. There were similar problems in Belgium.

ISARE II showed that in order to construct a European regional database of health indicators it is not only important to use national institutions that produce statistics, but also to access a network of regional structures. Resorting to local correspondents certainly complicates the collection of data, but seems indispensable. This allows access to data that are only available locally, because in some countries national institutions do not routinely collect or hold certain data.

The data were gathered at the level of health regions, as defined in ISARE I. In each country, a geographical level was selected with the correspondent according to various criteria, amongst which especially was correspondence with a level of local democracy and the existence, at this level, of competency in the field of public health. The use of these criteria does, however, call for some comments. The ISARE health regions are sometimes subject to change, as shown by examples in Greece or England where the local organization of health structures changed between ISARE I and II. Similar changes are expected to occur in Ireland and this will undoubtedly be a problem for the construction of a regional health database for Europe.

The non-correspondence of the ISARE health regions with NUTS 9 and the different levels of NUTS, pose several problems. On one hand, this non-correspondence results in some cases in the absence of data, as for example in Ireland, where estimates of population are not routinely available at the level of ISARE health regions. However, faced with similar situations our correspondents generally indicated that these data could be recalculated from their information systems. On the other hand, this non-correspondence poses problems relative to policies of standardization of data at the European level and the integration of regional health data into existing databases. In effect, in order to allow collection, setting up and dissemination of harmonized regional statistics within the community, NUTS has been officially recognized as the reference for territorial nomenclature by the European Parliament and Council.9,,14

The disparity in size of the ISARE health regions, already highlighted in the ISARE I report, poses methodological problems as seen in the example on infant mortality. To give an idea of the problem, there is a ratio of 1: 300 in terms of population between the smallest ISARE health region (Gotlands Laen in Sweden—57 684 inhabitants) and the largest (Nordrhein Westfalen in Germany—17 984 452 inhabitants). However, this problem will be exacerbated with the expansion of the EU to embrace countries with small populations (e.g. Malta).

All this reinforces on the recommendations made in ISARE I: the need to adopt a permissive approach by integrating different geographical levels into the European regional database.

A third phase of the ISARE project is now underway which will include the EU Member States, which joined the EU in 2004. Further work will be undertaken to update the health regions identified in ISARE 1 and to include all the home countries of the UK. This will build on the lessons learnt from ISARE III.

Acknowledgement

The ISARE II project was funded from the European Union Public Health Programme.

Conflicts of interest: None declared.

Key points

  • The construction of a health database at the level of the regions of Europe is possible.

  • The use of a network of correspondents working on health intelligence in the regions brings considerable added value.

  • Routine data collection systems e.g. Eurostat need to take account of the needs of the regions and to use boundaries that are meaningful.

  • More attention needs to be given to an European-wide system to collect data on major public health problems at a sub-national level such as obesity and smoking.

Appendix

Members of the ISARE II project

Gérard Badeyan (France), José Maria Berguristain Aranzasti (Spain), Luc Berghmans (Belgium), Luigi Bertinato (Italy), Helmut Brand (Germany), Danièle Fontaine (France), Richard Gisser (Austria), Susanne Holland (Sweden), Frédéric Imbert (France), Stamatios Kardasis (Greece), Seppo Koskinen (Finland), Dimitri Kounalakis (Greece), Pieter Kramers (The Netherlands), Isabel Larranga Padilla (Spain) Bernard Ledérsert (France) Christos Lionis (Greece), Tim McCarthy (Ireland), Antoni Montserrat (DG-Sanco European Commission), André Ochoa (France) Alexandre Pitard (France), Mady Roulleaux (Luxembourg), Frédéric Sicard (European Commission), Giuseppe Solomita (Italy), Fernando Tavares (Portugal), Alain Trugeon (France), Wendy Tse Yared (WHO—European Office) Harriet van Veldhuizen (The Netherlands), Guy Weber (Luxembourg), John Wilkinson (UK) and Laurens Zwakhals (The Netherlands).

Footnotes

  • *The members of the ISARE II project team are listed in the Appendix.

References

View Abstract