(1996; 157 pages)
It can be concluded that the two most important strengths of this research project's methodology were the combination of qualitative and quantitative data collection methods; and the flexible research design which allowed for modifications according to local conditions. Formulation of common injection practices indicators and two universal tracer conditions helped to provide cross-country comparable data. All indicators developed in this research, with the exception of the hypothetical tracer conditions, can be considered feasible.
With respect to the user-oriented methods applied in the research, household surveys provide excellent quantitative data on the extent and prevalence of injection use. A prerequisite is that the sampling frame be straightforward ensuring that representative data is collected. No major problems were encountered with the cluster sampling method used. It seems an unnecessary limitation that only households with children under five were sampled. Although it is justified to focus on children under five as a particularly vulnerable group, it is not necessary to exclude families without pre-school children. Prevalence of injection use was established retrospectively, using a two week recall period. In Uganda, the households were visited a second time fourteen days after the original interview. This follow-up visit proved to be a rather precise instrument in measuring the two-week prevalence of injection use as it reduces memory bias in respondents.
Allowing flexibility in questionnaire design, as opposed to a blue print design, has the great advantage that specific questions relevant to the local situation can be included. This has proven to be very useful, for example in Uganda where a question on home possession of syringes and needles could be included. The same question would have been quite inappropriate in the Indonesian context where injection technology has not been domesticated to the same extent. To ensure comparability, however, and to avoid that vital information is accidentally excluded from the questionnaire, it is advisable in a multi-country study such as this one, that the questionnaires be compared and revised before they are tested and applied in the field.
It is important to try to measure for which common health problems injections are preferred and used. In addition to universal tracer conditions, the inclusion of country-specific tracer conditions has the advantage of covering the most relevant diseases in that country. However, it proved to be rather difficult to find tracer conditions which meet the criteria (a self-limiting ailment for which other forms of treatment than injections are appropriate). For example, for symptoms such as fever or severe vomiting, injections could be medically justified in some cases. Therefore it seems necessary to include the degree of severity in the definition of the tracer conditions. The defined tracer conditions proved useful for the recording of actual illness cases but many methodological and conceptual difficulties were not foreseen. Because combinations of tracer conditions are common, it is advisable that the interviewers in the field record all symptoms and classify them afterwards, using exclusive categories.
With respect to the provider-oriented methods applied in the research, the problems encountered in studying doctors and other providers of injections were numerous. Non-response and refusal to cooperate occurred often in both countries, affecting the reliability of the results. In addition, under-representation of (certain) private and non-formal or untrained providers is a serious limitation in both country studies.
Key aspects of the rapid assessment methodology
• Identification of variables to be measured and key indicators.
• Standardized systematic sampling procedures to cover a variety of health care settings.
• User-oriented methods:
• Provider-oriented methods:
• Analysis using pre-defined indicators:
• Strengths of research methodology: