Selection of research methods

To gather information on how obligations of justice are implemented in international research, a case study research methodology was selected. A case study is an empirical inquiry that investigates a contemporary phenomenon in depth and within its real-life context. It is an appropriate methodology to employ when one's research question seeks to explain how or why a complex social phenomena works, particularly when that understanding is encompassed in critical contextual conditions.23

Case study research generally (but not always) uses qualitative methods.23 For our project, a triangulation approach was employed that relied on in-depth interviews, direct observation and document analysis. In-depth interviews were conducted with stakeholders involved in the trial being studied because they would bring the experiences and perspectives of the various trial stakeholders into account in significant detail. The interviews were supplemented by an examination of trial-related documents such as the grant proposal, ethics application and protocol and by direct observation at community advisory board meetings and trial sites. Ethical approval for the study was obtained from Mahidol, Oxford and Monash universities.

Selection of a case study

To identify an international research group that consciously designs its trials to fulfil obligations of justice, we relied on biomedical researchers known to us working in the field to recommend a group and facilitate an introduction. SMRU, a field research site of the Bangkok-based Mahidol-Oxford Tropical Medicine Research Unit, was contacted on our behalf. SMRU is located in Mae Sot (Thailand) and since 1985 has been conducting operational research that is designed to improve the health of the Karen and Burmese refugees, migrants and displaced persons living on the Thai-Burmese border.24 (The Karen are the second largest ethnic group in Burma and have a long history of persecution by the Burmese military government.) It also functions as a health provider, running several clinics that service this border population. SMRU was interested in collaborating with us on the project. Further discussion confirmed that its research fit our main criterion for a case study, ensured that the project was feasible from SMRU's end and determined the actions that needed to be taken before data collection could commence. SMRU then selected its ongoing VHX trial to be the subject of the case study. The VHX trial is funded by the Wellcome Trust. It seeks to describe the epidemiology and compare the efficacy of three treatments for vivax malaria—chloroquine/primaquine, chloroquine and artesunate (http://clinicaltrials.gov/ct2/show/NCT01074905). Trial sites are five SMRU clinics—Mae La, Wang Pha, Mawker Thai, Mun Ru Chai and Mae Kon Ken (each located within an hour of Mae Sot)—and participants are drawn from the border population. There were roughly 410 participants at the time of this case study research.

Sampling and recruitment strategies for in-depth interviews

Participants in the case study were selected from the following categories of VHX trial stakeholders:

• principal and co-investigators,

• trial participants,

• members of the Tak Province Border Community Advisory Board (T-CAB) and

• funder representatives (from the Wellcome Trust's Science Funding Division).

The method of selecting participants was purposive, as it was based on their involvement in the VHX trial.23 Selection criteria for participants were as follows:

1. being members of one of the aforementioned four groups,

2. being 18 years of age or older and

3. speaking English or Burmese fluently.

The VHX trial principal and co-investigators were identified using the trial protocol and recruited in person at SMRU. Snowball sampling techniques were employed to identify science portfolio advisors at the Wellcome Trust because they were more difficult to access. T-CAB members were identified and recruited for interview by a T-CAB coordinator. Five of fourteen T-CAB members were identified as appropriate interview subjects because they lived in villages near the VHX trial sites and would, therefore, be most able to describe the impact of the trial on their community.

Trial participants who met the project selection criteria were identified by the SMRU medics who were in charge of the VHX trial at the two trial sites where interviews were conducted. The Mawker Thai clinic and the Mae La clinic were selected as interview recruitment sites because together they captured the diversity of the VHX trial participants. Mawker Thai clinic is used by migrant workers and displaced persons living south of Mae Sot (Thailand) and Mae La clinic is used by refugees living in Mae La camp, which is located 1 h north of Mae Sot.

At the two clinics, trial participants were recruited opportunistically for interviews by the medic in charge of the VHX trial. Trial participants had to wait for 45 min to 1 h for their blood test results as part of their follow-up visits for the trial, so taking part in a 45-minute interview did not considerably extend their clinic visits. Trial participants were compensated by SMRU for transport and a day's work for each follow-up visit as part of the VHX trial. Ultimately, the aim was to sample a range of trial participants, spanning all three treatment groups and who had spent varying lengths of time in the VHX trial.

For all four types of stakeholders, recruitment for interviews continued until every (consenting) member of the stakeholder group had been interviewed or data saturation was reached.

In-depth interview procedures

Nineteen in-depth (semi-structured) interviews were performed with VHX trial stakeholders: investigators (five interviews), T-CAB members (four interviews), trial participants (eight interviews), and Wellcome Trust science portfolio advisors (two interviews). Four interview guides were written, one for each category of VHX trial stakeholder. The interview question guides were developed over three stages:

1. ensuring alignment between the questions and the ethical constructs under investigation,

2. consultation with the Mahidol-Oxford Tropical Medicine Research Unit and SMRU staff in Bangkok and Mae Sot, and

3. pilot testing.

For investigators and funder representatives, a series of open-ended questions was designed such that interviewees were asked to describe, first, their roles and responsibilities during each phase of the VHX trial (funding, design, subject recruitment and data collection, analysis and post-trial) and, second, their perspective on the health impact of the VHX trial on participants and the border population during and after the trial. Follow-up questions probed specifically for information on the VHX trial's selection of research target (disease focus and research question), provision of ancillary care, post-trial benefits and research capacity strengthening (see box 1). Final question guides for T-CAB members and trial participants contained more targeted questions and fewer follow-up questions (see box 2). This was because SMRU staff did not think that trial participants and T-CAB members would know what to say in response to open-ended questions. In order to elicit as much narrative as possible and to avoid getting only yes or no answers in these interviews, the phrases ‘Why?’ or ‘Please provide an example.’ were added to the end of questions.

Pilot interviews were conducted with two researchers in Melbourne, Australia and two VHX trial participants at Mawker Thai clinic, and interview questions were modified accordingly. For interviews with trial participants and T-CAB members, three translators were used—an SMRU medic at Mawker Thai clinic, an SMRU medic at Mae La clinic and an SMRU doctor (for T-CAB interviews). These translators carried out the consent process with trial participants and T-CAB members in Burmese in the interviewer's presence. Each interviewee signed two consent forms, one for his/her records and one for ours. All interviews were recorded on a digital tape recorder and followed the same format: the interviewer would ask a question in English, the translator would ask the question in Burmese, the trial participant would respond in Burmese, and then the translator would briefly summarise his/her response to the interviewer in English. Occasionally, the interviewer would ask follow-up questions that were not on the interview guide.

Interviews with VHX trial investigators and Wellcome Trust science portfolio advisors were conducted (in English) as conversationally as possible in order to build rapport. Interviews were an average duration of 72 min, with trial participant interviews generally running much shorter (25–50 min).

Direct observation procedures

Direct observation was undertaken over a 5-week period at four VHX trial sites and at two T-CAB meetings in Mae Sot. We adopted the stance of the observer as participant. The observer as participant stance provides the most ethical approach to participant observation because the researchers' activities are known to the group being studied and the emphasis for the researcher is on collecting data rather than participating in the activities being observed.25

To collect data, BP travelled to SMRU clinics nearly every weekday over the 5-week period. The first day at each clinic, she performed a walk-through with an SMRU staff member. Based on that walk-through and subsequent observations, she drew a map of the clinic. Each day she visited a clinic, she would observe in the study room during clinic hours from 09:30 to 12:30. Initially, the language barrier made it very difficult to know precisely what was happening at the clinics. Eventually, the main observation strategy was to try and identify VHX trial participants based on the medical tests they were seen receiving, to confirm this with SMRU medics and then to continue observing those individuals in order to determine what study-related and ancillary care they received. Through direct observation, BP was also able to witness the research skills of the Karen clinic staff and observe their training (such as on-the-job training and lectures).

While observing, notes and sketches were initially jotted down and then expanded upon later on the same day in a separate notebook.26

Collection of trial-related documents

Trial-related documents were collected from VHX trial investigators and included: the grant proposal, ethics submissions, trial protocol, T-CAB meeting minutes related to the trial, and trial participant case report forms (CRFs). Information from CRFs was especially useful for identifying the ancillary care that was given to VHX trial participants because CRFs included a form for Concomitant Medications, which listed all medications a trial participant had received for conditions other than vivax and falciparum malaria. At four of the five trial sites, 50 CRFs (200 in total) were sampled to generate a picture of what non-malarial conditions were treated during the trial. CRF data were also used to confirm data from trial participant interviews about their treatment group allocation, the number of vivax recurrences they had experienced and what ancillary care they had been provided with during the VHX trial.

Outputs of case study research methodology

All interviews were transcribed verbatim and translated from Burmese into English (where required). Data were then analysed according to the principles of thematic analysis, with co-coding performed independently by two researchers, in order to determine which obligations of justice were fulfilled by the VHX trial and how this was achieved.27 Thematic analysis is a method of identifying and reporting patterns (themes) in data.27 ,28 Once themes that pertained to an obligation of justice were identified, the final step was to assess whether the collated data extracts from each related theme provided evidence of how VHX trial stakeholders met the obligation.

As this paper focuses on methodology, our discussion of results will be brief. Significantly, our findings demonstrate that the case study approach we employed was able to generate the evidence needed to answer our research questions. Our results show that SMRU generally upheld its obligations of justice in the VHX trial. Trial investigators confirmed that vivax is now the most common form of malaria in the border population and SMRU has the data to support this assertion. Vivax can result in significant morbidity over time. As the vivax parasite has liver stages that can remain dormant for weeks, each infection is associated with multiple relapses, which can result in chronic anaemia. Nevertheless, vivax does not typically cause severe illness and, as a result, is not considered the top health concern of the border population. According to Investigator 01, Vivax is a problem but not a serious—I would not put that as the number one priority in terms of health in the population. It probably goes after respiratory infection, diarrhoeal disease and in terms of public health, tuberculosis is emerging as a big problem. Of course, if you look at just the sheer numbers, of course, we still treat many more cases of vivax than we treat tuberculosis, but it's difficult to compare because one is a disease that almost you would, you know, could compare as a flu, as a mild flu, except that in young children, in very young babies, then it can be dangerous and in pregnant women it's not very good, but in adults it's like a flu.

Trial investigators consider vivax to be a health priority but not the top health concern of the host communities. Similarly, T-CAB members and trial participants identify vivax as a health concern because it can affect their ability to work, deal with family matters and engage in social activities. However, many trial participants report concern for falciparum malaria because it can affect the brain and cause death. T-CAB members and trial participants also identify other illnesses such as cancer, dengue, diarrhoeal disease and tuberculosis as significant health concerns.

To measure the health needs and priorities of the border population, SMRU relies on four main strategies: epidemiological surveys, following prospective cohorts, clinic data collection systems and information from local clinic staff and T-CAB members.

Interview data, direction observation and CRF analysis also confirm that ancillary care beyond the disease under study is provided to VHX trial participants. However, ancillary care is not provided for all health conditions and is mainly limited to care for acute illnesses that are inexpensive to treat and that fall within the skill set of the SMRU medics and nurses, who are from the Karen and Burmese border population and provide the majority of care. Trial investigators are generally only called on to deal with complex cases. Since SMRU runs the clinics that serve as the VHX trial sites, this healthcare is provided to community members as well. Referral networks for other health conditions are already in place between the clinics and the nearest Thai and Burmese hospitals, but CRF data show that they have rarely needed to be used during the VHX trial.

Interviews and direct observation data show that the research capacity strengthening performed as part of the VHX trial was carried out mainly at the individual level. It consisted of training SMRU laboratory and clinic personnel from the border population to perform the assays and clinical tests required to conduct the trial. At some clinic sites, medics and nurses with limited research experience assumed management roles and gained leadership skills. The majority of the training was led by one of the VHX trial co-investigators, with assistance from other VHX trial co-investigators, who also served as her translators.

Finally, our data indicates that there is a high likelihood that post-trial benefits will be provided to VHX trial participants and their communities. Although the trial was still collecting data during our research, we were able to ask investigators how trial results would be translated into health benefits for the host communities post-trial. We discovered that, as the VHX trial was not testing a new drug, the most likely health impact would be a change in treatment practice for vivax at SMRU clinics and other medical NGOs on the Thai-Burmese border. Following a subsequent trial to optimise the treatment regimen for the border population, primaquine may be provided as the standard vivax treatment at SMRU clinics, with funding from the Global Fund. One investigator described how such changes would be implemented: If we change the treatment policy, we have the Malaria Handout and it is revised every year, every two years, so we need to revise that Malaria Handout and then all the clinics along the border use that Malaria Handout. So to revise the Malaria Handout it is the job of Andrew* and the other [SMRU] doctors who are treating the malaria… So we have the malaria meeting and malaria workshop every year, at least once a year. So in that workshop we invite all the NGOs, all the medics, so the Mao Tao clinic and other NGOs or some other associations and then they try it. So we distribute the handouts and then sometimes they advise me to go there and then give the training at the workshop. (*Name changed for confidentiality reasons)

Where SMRU research results suggest that a study regimen is more effective than that which is used in current clinical practice, revisions are made to the Malaria Handout (ie, the treatment guidelines for malaria written by SMRU) by SMRU doctors. These doctors then train SMRU clinic staff to implement the changes in their daily medical practice. Other medical NGOs on the border are also informed and trained to implement the changes at an annual malaria meeting. Thus, the fulfilment of obligations of responsiveness, ancillary care and post-trial benefits is facilitated by SMRU's long history of combining health services and research with the active involvement of the border population (its staff).

Strengths and limitations

The methodology we describe in this paper has a number of strengths. Most critically, it was able to generate the evidence needed to address our research questions from multiple perspectives. By conducting in-depth interviews with not only funder representatives and investigators but also trial participants and T-CAB members from host communities, we obtained both emic and etic accounts of community health concerns and the benefits created by the VHX trial for participants and their communities. Spending five weeks on-site in Mae Sot, attending T-CAB meetings and going to SMRU clinics with trial investigators helped establish a pre-existing level of familiarity with the interviewer that was essential to building rapport and trust during interviews with trial stakeholders (which were largely conducted in the latter three weeks of the visit). This meant that interviews were better able to generate authentic insights into interviewees' perspectives on the VHX trial. Our methods were also sensitive to drawing out important contextual conditions. In-depth interview guides are continually revised as informants provide new information that researchers have not previously identified.29 We, therefore, incorporated new insights into our question guides throughout the interview process and re-interviewed trial stakeholders (using the added questions) during the final week of data collection in Mae Sot. In effect, we captured rich details (not discernable from the VHX trial protocol) on the complex processes that culminated in SMRU's adherence to ethical requirements and the contextual conditions underlying their achievement. Examination of trial stakeholders in their natural setting enhanced our comprehension of how the trial processes described in interviews worked in practice.

Direct observation over a prolonged period facilitated a broader understanding of SMRU's operations, the local health system and the relationship between the two. This was highly important to grasping how the VHX trial topic and outputs fit into SMRU's overall research strategy and would supplement existing health services. Triangulation of methods, data sources and analysts demonstrated that converging conclusions could be generated by different data collection methods, data sources within methods (eg, accounts from different trial stakeholders) and analysts who reviewed the findings.

Although it may be argued that an ethnographic methodology would provide greater insight and understanding of the VHX trial's fulfilment of ethical requirements, ethnography typically involves data collection over a lengthy period (ie, 1 year or more) and considerable time for data analysis. Developing practical guidance for research groups entails gathering data about different international clinical trials in a multitude of settings. Relying on ethnography would be an inefficient way to collect this data and would require a significant investment of resources for each trial studied. The case study research methods we describe can generate the required evidence in a much shorter time frame.

Despite the strengths discussed above, there were a number of limitations to the case study methodology we employed in our research. First, we did not interview members of one significant category of VHX trial stakeholder—the Karen and Burmese medics who were responsible for running the trial. Logistical issues (availability of translators) limited our ability to interview members of this group. As a result, we did not get first-hand perspectives on, for example, research skills learnt and ancillary care provided as part of the VHX trial. Second, the language barrier made conducting in-depth interviews and direct observation difficult. Some degree of miscommunication and misinterpretation of events was unavoidable, particularly in initial interviews with translators. Despite training, translators occasionally became confused by questions during interviews. Where this occurred, the interviewer would re-explain the question briefly during the interview and again more comprehensively after it finished. Third, as the VHX trial was ongoing, measurement of post-trial benefits was limited to T-CAB members and investigators' descriptions of what would happen if the trial was successful, rather than their description of actual post-trial processes that could be observed.

To address the first two of these limitations, it would have been useful to have a bi-cultural Karen-Australian research assistant (who could speak Burmese).30 This would have obviated the need to rely on busy SMRU staff as translators and lessened the difficulties associated with the language barrier in interviews and during direct observation. Measurement of post-trial benefits could be enhanced with a follow-up site visit after the VHX trial concludes.