Study design and planning

With the proposals for the three ARV studies approved, a series of workshops was planned to run throughout 2005, in order to provide training and support to the country teams who would conduct the research. Regular e-mail contact was also maintained between country teams and facilitators in order to ensure a free sharing of ideas and information. The first of these workshops took place over five days in Bagamoyo, Tanzania, in February 2005. Ten participants attended from four countries - Tanzania, Uganda, Botswana and South Africa - including pharmacists, medical doctors and social scientists. Three staff members from the University of Amsterdam acted as facilitators.

The primary objectives of the Bagamoyo workshop were (i) to review each of the country proposals and to define common ground for a comparative study; (ii) to develop common and/or country-specific research objectives and methodologies; and (iii) to work out detailed country plans (including the selection and training of fieldworkers and pre-testing research instruments) and budgets. The workshop was designed to be fully participatory, and it opened with participants presenting peer reviews of each other's proposals. Based on the discussions arising from these critiques, a shared research objective was formulated: To identify factors that facilitate or constrain adherence to ART among adults through a pre-intervention study. The objective was to work towards the design and subsequent implementation of interventions that would promote optimal adherence to ART. On this basis, the participants expressed their willingness to make the country studies as comparative as possible, while still keeping country-specific issues in mind.

Two further guiding principles for the studies were also agreed upon: (i) teams would use a mix of quantitative and qualitative research methods; and (ii) they would adopt a multi-level perspective, meaning that they would investigate issues of ARV adherence from several different points of view: individual ARV users; the community; health facilities (including all cadres of staff); and policy-makers and other national level actors.

From this starting point, the remainder of the workshop involved formulating more detailed research questions, identifying relevant variables (for the quantitative tools) and themes (for the qualitative tools), and then designing appropriate research instruments. Research tools used in the 2004 course, questionnaires from a Horizons ARV adherence survey conducted in Mombasa, Kenya, as well as other literature (Horizons, 2005; Farmer, 1999; Vitolins et al., 2000; Sankar et al., 2002) were handed out to provide additional input. Eleven draft research instruments were produced by the end of the workshop, from which individual countries could select and adapt as necessary (http://mednet3.who.int/prduc/Bagamoyo/Bagamoyo.htm).

Initial planning of fieldwork was also undertaken, for example in relation to sampling frames. This included deciding upon the categories of health facility to be investigated - e.g. NGO, private-for-profit, government - and in which areas of the country they would be. This was not always a straightforward process: some areas of at least one of the participating countries were 'over-researched', and it was hard for this team to find a facility that had not been previously subjected to (usually clinical or epidemiological) investigation.

With respect to ARV users, the number of respondents was considered, as well as the importance of systematic selection criteria in order to reduce bias to a minimum. It was also stressed that selection should be conducted in such a way as to ensure that people would not be stigmatized as a result of their participation in the studies. Finally, there was discussion about data management as an ongoing process throughout the fieldwork period.

During the evaluation of the Bagamoyo workshop, the participants conceded that it was only at this stage that they could fully appreciate the extent of the work that lay ahead, and - in relation to this - how small their budgets were. However, by the end of the workshop a shared research objective had been defined; high quality research instruments had been produced; and the revised country plans and budgets were well on their way to being finalized.

Fieldwork

Fieldwork took place in Uganda, Tanzania and Botswana between April and July 2005. Between the Bagamoyo workshop and the start of fieldwork, each country selected the research instruments they wanted to use from the 11 draft versions that had been produced, and then adapted these to local conditions. This adaptation included translation to the vernacular, pilot testing, and subsequent fine-tuning.

Each of the country teams was composed of people from very different backgrounds, including pharmacists, doctors, social scientists and administrators. This combination of backgrounds, levels and disciplines produced a dynamic synergy of complementary skills. In addition, data collection personnel were hired and trained in each of the three countries. These were either social science graduates or social workers based at the various health facilities. The general feeling from each of the three country teams was that health workers, community members and ARV users were highly cooperative and willing to participate, but it was not always so easy to access those working at the national level.

Data collectors receiving training in Tanzania.

The main constraints faced by the teams were financial and logistical. The budget was extremely tight for all of them, and since a number of people were involved in the various studies, the researchers sometimes had to use their own personal funds for accommodation, meals and transport. While undoubtedly causing some hardship, this is nonetheless an indication of the high level of motivation and commitment by members of the three teams. In addition, it was a major challenge for people who already had full-time jobs to meet - in some cases, hundreds of kilometres from where they lived - in order to organize and conduct the work.

Contact between the teams was maintained on their own initiative, with regular e-mail exchanges, text messages, and occasional phone calls. These discussions concerned both data collection and initial analysis. In addition, the team members from the University of Amsterdam and WHO also kept in touch with the teams via e-mail throughout the fieldwork.

Analysis and writing up

Analysis workshops were held in each of the three countries in July 2005, attended by the principal investigators from each country (all of whom had attended the Bagamoyo workshop) and their national research colleagues. Direct assistance was provided by two medical anthropology Ph.D. candidates from the University of Amsterdam, one visiting Botswana, the other visiting Uganda and Tanzania. This was a key moment for each of the three country studies, since these visits prompted the national participants to come together to consider what they had found during their fieldwork. In view of their work commitments, it was not always easy for the team members to coordinate meetings for analysis and writing. Furthermore, few of them had any training in either quantitative or qualitative data analysis, so this was an occasion for hands-on 'learning by doing' in the company of a colleague who had well-developed analytical skills.

These analysis workshops followed the principles of technical assistance that have been developed over the last decade by the University of Amsterdam during collaborative studies on a number of health topics, including immunization, medicine use, malnutrition and reproductive health.

Given that most of the researchers did not come from a research background, it was important to provide guidance in analysis technique. A number of issues were therefore stressed throughout the three analysis workshops. With the multi-level perspective adopted in each of the three studies, for example, it was important to try to understand respondents' 'emic' perspectives. This assumes that 'there is no one correct view', which is a helpful premise when considering the thoughts and opinions of ARV users, providers and the community as a whole, since it permits a nuanced interpretation of what has been reported by a wide variety of respondents.

The research teams were also advised (i) to think critically about data validity, and to look out for reporting bias; (ii) to think of how the quantitative and qualitative data could best be triangulated and used together; and (iii) always to bear in mind the ultimate objective: designing workable intervention/s to improve ARV adherence.

The analysis approach taken with the qualitative data is given in Table 2. Quantitative data analysis was facilitated by a system devised by the South African team who had attended the Bagamoyo workshop, and assistance was received from them over the phone during the course of the analysis workshops.

The workshops ended with the production of suggested report outlines to be worked on in the run-up to the final writing workshop in Jinja, Uganda, in November 2005. Individual team members were given specific tasks and deadlines, and as the country teams developed their reports, they distributed them via the Internet and received technical and editorial input from each other as well as from the team members at the University of Amsterdam and WHO. Thus each country report went through several reviews and revisions before a final report writing workshop.

Report writing workshop

The objectives of this five-day workshop, held in Jinja, Uganda, in November 2005, were (i) to discuss and work towards finalizing the draft country reports; (ii) to write a short 500-word policy brief for each country; and (iii) to start work on a synthesis article for publication. The workshop was attended by eight full-time participants from the three countries, and was facilitated by five technical advisers (three from the University of Amsterdam and two from WHO).

The workshop started with a series of peer review sessions, during which members from two of the participating country teams would review the draft report of the other country team, with additional input from the course advisers. This peer review process focused on the following: (i) adequacy and clarity of the presentation of data; (ii) appropriateness of analysis and conclusions (iii) reflection on differences and similarities in findings between the three countries; and (iv) formulating recommendations for improvements.

A WHO editor - one of the two WHO advisers - provided additional advice on presentation and style. Some of the 'key tips' provided are given in Table 3 below. The importance of the reports having a clear 'red thread' was also stressed, linking the data with the interpretation of findings, and subsequently with the recommendations for future interventions and/or policy changes.

A second important activity at the workshop was the initial drafting of a 500-word policy brief, based on the essential 'take home' message that the country teams wanted to stress. A role play exercise, entitled 'Four floors with the minister', was conducted to encourage the teams to distil their message. The idea was that they were in a lift with the Minister of Health, and would be travelling up four floors together: what was the message they wanted to put across in that short time?

The hands-on work then began on the reports, with team members from each country sitting together alongside one of the course advisers. Each of the teams restructured parts of their report according to the suggested outline provided by the WHO editor. As soon as a chapter or section was completed, the editor was then able to start work. This work continued throughout the final three days of the workshop. The initial draft of a synthesis paper outlining the main collective findings of the reports was also produced. The workshop concluded with the production of a time-line for the finalization of reports.

Intervention development

Since this was a 'pre-intervention' study aimed at designing interventions to promote ARV adherence, a session was held during the Jinja workshop to define three possible intervention approaches for each of the country teams. It was recognized that a mix of interventions may work synergistically, ensuring that the whole is greater than the sum of the parts. A ranking exercise, which had already been introduced to the participants during the PRDUC course, was suggested for prioritizing the various suggested interventions. Six additional criteria were defined that could assist in the prioritization of interventions, including: (i) anticipated cost; (ii) likely impact; (iii) feasibility; (iv) sustainability; (v) acceptability at various levels; and (vi) likelihood of harm and/or unintended consequences. See Table 4 for a list of the suggested interventions.

Table 4: Interventions suggested during the Jinja report writing workshop

It was also agreed that country teams would report back to the various stakeholders who had participated in the research, and present their proposals for future interventions. Limited funds were made available from HealthLink to facilitate this activity. While the initial recommendations for future interventions were derived from the data collected during the studies, it was recognized that additional and potentially important issues regarding the interventions would emerge during such consultations.

Funding for these interventions and their evaluations has not yet been obtained, and country teams were encouraged to seek funds from Embassies, PEPFAR, the Centers for Disease Control and Prevention, USA, and other funding organizations.

Publications

Output from the ARV studies included the following:

• Individual study reports, for in-country distribution to stakeholders at local and national levels.

• A synthesis article, based on the findings from all three countries.

• The current publication.

• Conference papers, including abstracts from each of the three countries, submitted to the 2006 International AIDS Conference in Toronto.