Bridging the gaps between research, policy and practice in low- and middle-income countries: a survey of researchers

Study participants

We surveyed researchers in 10 low- and middle-income countries (China, Ghana, India, Iran, Kazakhstan, Laos, Mexico, Pakistan, Senegal and Tanzania) who conducted research in one of the four clinical areas described earlier in the article. Within each area, a particular emphasis was placed on an intervention that was supported by strong evidence from international and local research: insecticide-treated materials to prevent malaria; intrauterine devices for family planning; oral rehydration therapy to prevent dehydration in children with diarrhea; and the DOTS strategy (directly observed treatment, short course) to control tuberculosis.

We purposively sampled countries to achieve breadth in levels of economic and health care systems development, political systems and geographic locations. In addition, all country teams had to have demonstrated (through collaborations with the World Health Organization [WHO]) a strong interest in bridging the gaps between research, policy and practice and in further developing the capacity to evaluate such efforts. The selected countries differ in their health status indicators, rates of coverage for the interventions under study, and access to the Internet or computers with a CD-ROM (Appendix 1, available at www.cmaj.ca/cgi/content/full/cmaj.081164/DC1).

We identified researchers using three sources: lists of authors identified through a search of MEDLINE and ISI Web of Science for each combination of country and topic; lists of researchers identified by WHO staff; and lists of researchers identified by country teams. We defined researchers as those who spent at least 10% of their time doing research, including the production, synthesis and sharing of research. Our definition could include clinician scientists, university professors, research managers in nongovernmental organizations and civil servants with program-evaluation responsibilities.

We did not calculate the sample sizes needed because we sought to survey, whenever possible, all researchers involved in conducting research on each topic in each country. In only four countries — Ghana, Iran, Pakistan and Senegal — was a sample selected because the number of eligible researchers was significantly more than 25.

Development of the questionnaire

We developed a self-administered questionnaire comprised of two main sets of questions: one set addressed researchers’ activities in bridging the gaps between research, policy and practice; the other addressed potential system-level, organizational and individual correlates of researchers’ engagement in these bridging activities. For the first set of questions, we drew on the WHO’s “World Report on Knowledge for Better Health” 11 and four existing questionnaires 4^,5^,12^,13 to identify conceptual domains to be covered by the questionnaire. We retained the wording and order of the questions whenever possible.

We grouped the domains into three broad categories: (a) “producer-push” efforts (what is “pushed,” or communicated, to target audiences outside the research community; to whom; by whom; how; and with what effect); (b) efforts to facilitate “user pull” (i.e., what strategies are used to provide access to research and to develop target audiences’ capacity to use research); and (c) exchange efforts (i.e., how target audiences are involved in research and bridging activities). The development and testing of our questionnaire are described elsewhere. 14 WHO’s translation service translated the questionnaire for China, Kazakhstan, Mexico and Senegal; country teams translated the questionnaire for Iran and Laos.

Survey administration

We mailed the questionnaire or used a drop-off and pick-up approach in all countries except China (for one of its two topics [family planning] and Iran (for its one topic [tuberculosis treatment]), where for each topic we administered the questionnaire at a meeting where all eligible researchers were expected to be present. We used several approaches to increase the response rate: personalized letters, follow-up of contacts and provision of a set of WHO publications as an incentive. 15 Survey work was completed in all 10 countries between April 2004 and April 2005.

Statistical analysis

We checked, coded and managed all data centrally in order to permit the analysis of pooled data from all 10 countries. We calculated proportions for most surveyed researchers’ characteristics, bridging activities and potential correlates of engagement in these bridging activities. We combined the top two categories whenever an ordinal scale was used (e.g., frequently or always undertaking an activity, agreeing or strongly agreeing with a statement).

For the dependent variables in the logistic regression models, we selected three bridging activities, one from each broad category of the organizing framework, that offer particular promise. For one of the three bridging activities, we selected “providing systematic reviews” as a promising producer-push effort because reviews can reduce bias and the play of chance in estimating effects and can save time for those those who can draw on them rather than have them identify, select, appraise and synthesize the research literature on their own. 16 We selected “providing access to a searchable database of research products about the topic” as a promising bridging activity to facilitate user pull because timeliness was one of only two factors identified with some consistency in a systematic review of factors that increase the prospects for use of research in policy-making. 17 For the third bridging activity, we selected “establishing or maintaining long-term partnerships related to the topic with target audience representatives” as an exchange effort because interaction was the second of the two factors that emerged in the systematic review. 17

Based on a combination of the research literature and our own knowledge of the field and the contexts in which researchers function, we selected 20 potential system-level, organizational and individual correlates to examine in each logistic regression model. For missing values, we used multiple imputation, whereby each missing value was replaced by the mean of 10 different estimates. We adjusted standard errors according to Rubin’s rules. 18 We excluded observations when the dependent variable was missing. We estimated all models using Stata/SE 9.2 for Macintosh with robust variances. 19

We present results by topic rather than by country so that sample sizes would be larger and because the Millennium Development Goals, and most national health goals, focus on topic-specific bridging activities.

Limitations

Our study had four limitations worth noting. First, the potential for social desirability bias (researchers may have reported higher levels of engagement in bridging activities than might actually be the case) exists. Second, the questionnaire included questions that applied to all target audiences (even though some activities, such as disseminating clinical practice guidelines, may have been relevant only to a target audience such as health care providers) and required researchers to provide a single response even if some responses may have varied by target audience. Third, linguistic or cultural differences may have affected respondents’ interpretation of select questions. Fourth, the missing or imprecise responses to an open-ended question about institutional affiliations precluded us from adjusting for clustering (at the organizational level) in the regression models; however, this would have affected the standard errors but not the point estimates.

Conclusion

Our findings indicate that researchers in low- and middle-income countries report frequent engagement in only some types of promising bridging activities. Future initatives could focus on supporting those bridging strategies targeted at health care providers that have been found to be effective in some contexts. They could also focus on addressing those factors that appear to increase the prospects for using research in policy-making. In light of 53 ministerial delegations calling for renewed efforts to bridge gaps between research, policy and practice, 3 our survey provides a baseline against which these future initiatives can be measured.