Overall performance of the protocol

To assess the ability of the manual to assist national investigators to obtain answers to the key questions, it was necessary to test it in diverse environments and cultures. Specifically, the following different systems were selected:

All procedures in the manual were applied by research teams in the five study sites. All team leaders (or main investigators) had backgrounds in health professions (nutrition, nursing, community medicine, public health). In several cases they also had anthropological training. Regardless of background and previous experience, the investigators agreed unanimously that the initial training workshop was a valuable preparation for carrying out the study.

The teams completed the studies within the projected time frame of six to eight weeks. In two areas (China and Peru) the manual was translated into the national language; in the other three areas, the teams worked directly from the English manual, using local translators as needed to maintain literal translations of concepts.

The investigators and their field teams generally found it easy to apply the specific techniques that they attributed to the detailed, step-by-step instructions in the manual. A number of modifications in the instructions have been made as a consequence of the thoughtful suggestions from the teams, and the post-field test revisions are based on extensive discussions with the investigators. For example, the field teams made suggestions for identifying the most appropriate time to conduct the studies and provided creative suggestions concerning the use of specific examples in the local culture to help respondents understand questions.

In some cases the teams found it advisable to adapt the dietary assessment modules - Modules 4 and 5 (Part II-C) in the manual. In the Philippines, respondents were asked to give a 24-hour recall before they were questioned with the food frequency procedure. The investigators instituted this change in order to orient the respondents to the concept of daily quantities. In India the team experienced some difficulty in recording portion size and preferred not to use the 24-hour recall in this manner. One investigator suggested that the food frequency procedure developed by IVACG (IVACG, 1989) could be substituted for Module 5, provided it was modified to use the food list developed from the other modules. Thus, the field test experiences suggested that it is possible to modify the dietary assessment section of the protocol to fit local circumstances, as long as the data yielded by the procedures can be used to rank-order the key-foods in the market survey.

The community food system data tables were successfully applied in all the test sites. The tables are also useful in clearly highlighting where further research on food composition and species identification is required. In the study among the Aetas, the list of foods contained 128 items of which twenty-seven had no scientific identification and forty-two had no recorded values for carotene or retinol. Food items for which no composition data could be found also occurred in the Cajamarca study and in Filingué, Niger. Complete data were reported for all of the food items in the community food system data tables of Doumen, China and Sheriguda, India.

To determine the ability of the protocol to document seasonal variation in food availability, and cultural perceptions of food selection as these relate to seasonality, the investigators in the Philippines repeated the study in both wet and dry seasons among the Aetas. The team found that in order to describe the variety of local vitamin A-containing food, it was necessary to have data from the wet season. They suggested that conditions in the dry season could have been obtained through interviews, without observation, by carefully interviewing key-informants. The data on cultural perceptions about food attributes and qualities remained stable. Thus, they concluded that a single study, carried out in the wet season, would have obtained all the necessary information.

To determine how sensitive the protocol is to differences in food availability and selection in rural and periurban communities within the same geographic area, the investigators in Peru conducted studies in two locations. The team documented substantial differences between the two sites, with greater use of animal foods and wild plants in the more rural site. Families in the periurban communities relied primarily on purchased foods, and the food list was much longer. On the other hand, there were many similarities in food items and food beliefs, although food beliefs appeared to be more diverse in the periurban area, in part because of in-migrants from other regions.