Methods

The project would use a variety of different sources of data in order to build as complete a picture as possible in the different countries and sub-regions. This is because of the dangers of using only one data source where information may be incomplete. Each source of data has both advantages and risks associated with it.

1. Collection of aggregate level statistics on mortality and infant mortality

These offer information in general about patterns of health in the region, although their method of collection may differ and they may not be very accurate. For example, there are different ways of recording death and these may vary by region or by country. Some data may be missing altogether. The project would combine and standardize aggregate level data to provide a general picture of health. Mortality statistics complement the data about life styles obtained from respondents who will have, at the time of interview, a greater or shorter life expectancy. They offer census type comprehensive information about causes of death, thus avoiding sampling error although still subject to other types of error, especially for trend analysis starting in Soviet times. If we conceive of mortality statistics as registering the end health state of all respondents over a period extending beyond the year 2060, then there are good a priori reasons to expect current behaviors of the living to produce future changes in mortality statistics, whether for better or for worse. By interviewing the living, we are able to come to anticipate this "final" statistics hic et nunc.

2. Survey information about living conditions, lifestyles and health.

To fully answer the question of what lifestyle factors might be influencing mortality and morbidity we need individual level data. A sample of 4000 is envisaged in Russia, 2500 in Ukraine and 2000 in each of the other countries. This would provide adequate representation of different population groups. Normally, there is a problem of finding reliable survey organisations in some of these countries. However, the project co-ordinatin team have had many years of experience of conducting surveys in post-communist countries including some of the research partners and has consulted a variety of other organisations carrying out international surveys, such as the United States Information Agency. The project co-ordinator has been developing and testing these links through a series of conferences, which took place between March 1998 and August 1999 bringing together relevant survey partners and questioning them about the feasibility of the research and the methods of research. In this way a team as reliable as possible has been developed covering a wide range of countries. However, in some countries, despite considerable investigation, it was decided that research partners could not be found who could adequately provide reliable results. This is why some countries have been excluded from the research.

The survey data would be collected using standardised random techniques (multi-stage clustered sampling) and face-to-face interviews so that a representative sample of the population can be drawn. The survey data would be analysed using the SPSS statistical software package, with which all teams are familiar. The data base would be developed in each country using a standardised "mask" so that the same data with the same variable names is produced and these would then be pooled into a combined data set for comparative analysis. The consortium members have developed expertise since the early 1990s in doing this.

However, rigorously conducted, survey information can only collect information according to pre-set questions and tells us only a certain amount about how lifestyle and culture mediate health. For this reason a qualitative part to the project has also been designed.

3. The qualitative in-depth interviews and focus groups

These would provide subjective details about lifestyles and health which would complement the survey. However, their numbers are necessarily smaller than those covered in the survey and this is why we have concentrated upon particular regions in particular countries. The usual problem with qualitative research is the fact that it is difficult to incorporate inductive insights (i.e. insights derived from the data collection and interpretation) in a systematic way and to compare qualitative results cross-nationally. This is why we have developed the "framework" analysis using WinMax software to enable collection, recording and analysis of these interviews in a standardised form. The in-depth interviews would be used also as a way of collecting medical data via health checks and medical histories using the assistance of the interface-level health practitioners interviewed in another part of the research.

4. Interviews with health professionals and collection of environmental data

There are too many examples in the post-Communist world of unrealistic policy prescriptions proposed by flown-in Western "experts" who fly out again without evaluating the consequences of their advice. We would seek to avoid such mistakes. In order to incorporate the views of the producers as well as the consumers of services, we are proposing interviews in the 10 selected sub-regions (3 sub-regions in Russia, Ukraine and Kazakhstan plus Chernobyl region) with people at different levels of the health system and to discuss with them the aggregate level and survey data. This incorporates a "Delphi" style element to the project and provides feedback from users of the project at an early stage (see Gibson 1998, Normand et al. 1998 for uses of the Delphi method in health contexts). This is also an element in the dissemination strategy. It is important to produce recommendations for policy which are relevant and sensitive to the real issues facing interface level practitioners. This would be combined with the collection of environmental data about the region including pollution and types of employment as relevant factors shaping health outcomes in interaction with other factors. The study would thus seek to forge links between scientists and health professionals at a local level.

5. Multi-method, multi-level analysis

Relying only upon aggregate level data or only upon survey data or only upon qualitative interviews can lead to erroneous conclusions. Most studies are married to a particular method which can provide only a partial and limited view of health problems. Given that health issues are tied to economic, lifestyle, environmental, gender and other factors in complex ways, such perspectives can only be partial. This project aims to link a variety of different data sources and a variety of sources of expertise amongst its partners in order to provide a view of the issues from many different angles.