Table 1.

Factors to facilitate successful population health intervention studies in North–South Collaborations

Factor	Description	Examples/lessons from our study	Checklist of some of the challenges to consider
1) Build relationships and shared ownership	Global health research should be designed collaboratively among those with local knowledge, those with methodological expertise, and those in positions to implement findings (29). To do so requires building and maintaining trusting relationships, addressing disagreements to maximize collaborative performance. Commitment to work together, creating a common vision, frequent communication, understanding each other’s culture, and pre-planning are key determinants of successful collaborations (9, 10).	- Occupational health/infection control researchers had to build relationships with social scientists and clinicians - Researchers and practitioners of different races worked closely together - Unions had to be brought onside - Local university researchers, national and international researchers struggled to reach consensus and to establish shared ownership, creating ongoing tensions	- Varying interdisciplinary perspectives - North–South power dynamics - Racial and gender power relations - Labor–management power dynamics (especially in workplace studies) - Community -university differing priorities - Scalar differences (work unit, facility, province, nation)
2) Conduct feasibility study	Feasibility studies are small studies conducted before a main study, in order to improve the design of the main study; for example, to estimate sample size, willingness of participants, response rates to questionnaires, etc. They are useful to ascertain the priorities of the various stakeholders and attitudes toward the proposed RCT and/or its components (30, 31).	We began gathering data in 2007 in workshops and focus groups; then conducted a large baseline survey in three hospitals; then created a training program in which trainees gathered more data; we also conducted more situational analyses and further focus groups.	Researchers should continuously consider whether feasibility studies (qualitative and quantitative) would be helpful to identify stakeholder priorities and concerns, as well as to address challenges as they emerge (7).
3) Build receptor capacity	It is essential, particularly in North–South collaborations (32) that partner organizations understand not only the policies and practices that must be followed in the intervention but also the basics of the research process itself.	Our study involved several Northern students who spent many months with Southern partners; we also conducted many training sessions for local practitioners.	Considerations should include building capacity of: - local researchers and research trainees - Northern trainees (usually graduate students) - local practitioners who will be involved in operationalizing the protocols - local decision-makers
4) Create an information system to support the population health intervention	It is often necessary to create a dedicated database for large studies. However, data gathering systems for RCTs should ideally be sustainable beyond the RCT to assist in monitoring the sustainability of the intervention. Particular care to information technology issues in North–South technology transfer (33) as well as data privacy and confidentiality is needed.	We developed an information system – the Occupational Health and Safety Information System (OHASIS) – which we originally intended to have installed and used in all facilities. Technical challenges required us to fall back on paper-based forms that were faxed and data entered at the university.	-Micro considerations (do the staff entrusted with data gathering have the skills and time to do this well) - Meso considerations (does the organization support the data gathering and is providing the infrastructure – including space/computers – for this purpose) - Macro is the information technology support in place.
5) Conduct additional feasibility and pilot studies	A pilot study is a version of the main study that is run in miniature to test tools or components of the study or whether the various components can all work together (35, 36). It needs to have clearly articulated goals and procedures. It need not include randomization. Sometimes tools need to be pilot tested more than once before being used in a larger study.	- The challenges identified in the data collected in the first set of feasibility studies led to more qualitative assessments, discussions with stakeholders, and an additional pilot study, which, in turn, identified further challenges. The instruments and forms had to be re-revised many times.	- Is there a good basis to believe that the intervention will be successful (on theoretical grounds, if not previous observational studies?) If not, and new information arises during the course of the study that challenges original preconceptions, are mechanisms in place to take these into consideration?
6) Clearly articulate expectations from partner organizations, and get all stakeholders onside, with a view to scale-up from the start	A critique of population health intervention studies is that they are often not generalizable. Thinking about scale-up (37) is, therefore, needed from the start. Also, while clinical RCTs generally cover the full costs of an intervention, in population health interventions the danger of creating unsustainable processes looms even larger; engage these discussions early in the planning process.	- We were very successful in planning for scale-up, as the involvement of the Canadians began at scale (international), then proceeded to national scale, with the work at the provincial level always seen as leading to scale-up. The challenge was deciding what to fund; if the research funded all the local personnel training, it was felt that this would not be sustainable, so a balance had to be reached.	- If the study funds operational personnel to implement the intervention, will, and could, the health organization commit to maintaining such personnel should the study show the intervention to be successful? If not, have the consequences been considered?If the research funds do not fund operational personnel, and staffing levels decrease such that the study integrity is jeopardized, is there a contingency plan?
7) Develop and refine a detailed protocol	Excellent guidelines exist in his regard (6, 7); the challenge is getting to that point, and being prepared to revisit the elements of the protocol as needed.	- This has been an ongoing challenge, particularly because of different research cultures and disciplinary traditions regarding the ease and desirability of making changes along the way (which requires amendments to ethics approvals, etc.)	- Have the various partners been informed that challenges along the way may require revisiting the protocols?
8) Consider the ethical issues and obtain ethics approval	There is no algorithm to resolve conflicts among general moral considerations (31), for example, between privacy and justice, or between different conceptions of justice. The relationships built, in combination with informed institutional ethics reviews, are needed to develop the best protocol, taking the various ethical principles into account.	- Priory of ‘ethical imperatives’ differed within the team, for example, the ethical imperative to publish versus not to offend local institutions by showing problems in the system; the ethical imperative to make changes to maximize the likelihood of success of the interventions versus abiding by intended protocols.	- Has there been sufficient discussion among all parties of the benefits as well as risks from the research, not only for participants, but also at a systems level?