In designing the injection practices research it was considered important to strike a balance between various aspects of the study (Report of the First Informal Workshop, 1991:5-6):

In order for the country research projects to meet the needs of the country, it was seen as important that the objectives of the proposals be based on a country specific problem identification. Methods had to be adapted to local conditions. As a consequence, considerable differences exist between the country studies, for example in sampling procedures, sample size, focus of the research, use of specific research tools and in the analysis of the data collected. At the same time, by establishing common indicators measuring the prevalence and the appropriateness of injection use during the first two workshops, an attempt was made to ensure some level of comparability.

The research teams aimed at establishing the extent of injection use, in the general population via surveys both at the household level and in health facilities, either by the review of prescriptions (Uganda) or through exit interviews with patients (Indonesia). The teams have also assessed the medical appropriateness (rationality) of injection use. This has been researched using certain tracer conditions and 'hypothetical' illness cases which were presented in household surveys and focus group discussions. The appropriateness of injection administration procedures (hygiene) in health facilities was also studied by both teams. As to the popularity of injections, the Indonesian team has relied on focus group discussions with users and providers, and on structured questionnaires with injection users (patients in health facilities). In Uganda, the study formed part of a larger ethnographic research project into the practice of injections. In-depth interviews with key informants were conducted which have produced interesting insights into the popularity and specific conditions of injection use in the Ugandan context.

During the final workshop both the advantages and disadvantages of a flexible research design were assessed. While the research protocol allowed the country teams to take local conditions and needs into account, a disadvantage of this flexible approach is that comparison of the results between countries is only possible to a limited degree. Therefore, comparisons between both countries in this report will be made only when appropriate, for example while discussing the common indicators measuring the prevalence and the appropriateness of injection use. Most of the common indicators proved to be effective in order to measure the popularity of injection use in various settings and relating prevalence to various variables, such as type of facility where injection was given, urban or rural setting, etc.

The mixture of quantitative and qualitative methodologies in the injection practices research allowed for cross-validation of data. Household questionnaires proved an effective way to estimate prevalence rates of injection use in these two countries where injections are a common route of treatment. There were no major problems with non-response or sampling. A follow-up visit to the same household after fourteen days - which was performed in Uganda - can produce even more reliable results, but is, of course, more costly and time-consuming. The design of a proper questionnaire can be improved by using information gathered from in-depth interviews with key informants. The use of simple tracer conditions to investigate the extent of medically inappropriate injection use proved to be a complicated issue with many methodological pitfalls. Still, it seems the only way to categorize complaints and to calculate injection rates for different illness categories. Focus group discussions, using hypothetical illness cases to generate discussion on the subject of injections, proved to be a useful method if a quick understanding of local ideas and practices is desired.

The study of providers' practices required a great deal of creativity. Serious problems of non-response and lack of cooperation by certain types of providers, particularly those without formal training were identified. Still, in both countries researchers tried to study injection use in health facilities in a number of ways: through interviews with providers and patients, through observations and chart reviews. Although it is difficult to obtain consent, observation of actual practices and hygienic procedures seems paramount to estimate the extent of the problem of unsafe injections in developing countries.

Appendix 2 gives more details on the quantitative and qualitative methods used in the country studies. An assessment of the usefulness of each method is made. Following Van der Geest & Hardon (1988), a distinction is made between user-oriented methods (focusing on the consumers of injections) and provider-oriented methods (focusing on the providers of injections). All tools (questionnaires etc.) are presented in Appendix 3.