Study population and recruitment
Community-based HIV/AIDS organizations in Canada provide a number of programs and services to people living with or affected by HIV, which may include prevention initiatives, individual or group counseling/support, and community outreach and/or education. In addition, organizations in Canada are situated in diverse geographic settings ranging from dense urban settings to rural, northern, and/or remote settings, with some focused on specific at-risk populations and/or cultural or ethnic groups.
We will draw our sample from those organizations affiliated with the Canadian AIDS Society and from relevant provincial HIV/AIDS networks (e.g., the Ontario AIDS Network), and send an organizational invitation to the executive director and management team (if applicable). The invitation will indicate that if they are interested in having their organization participate, access to SHARE will be provided to all interested staff. Given that SHARE is currently only provided in English, we will exclude organizations that do not have at least one key decision maker who is comfortable participating and corresponding in English.
To ensure clarity in our study recruitment, we will outline that consent from the executive director is required for the organization to participate. We will also indicate that we require one key organizational decision maker to fill out a brief survey measuring their intention to use research evidence (see the Outcomes section for more detail on the survey) on behalf of their organization at baseline and again at the completion of the trial. We will request that the executive director complete the survey, but will indicate that they can delegate to another manager provided the manager has a decision-making role about programs, services, and advocacy, and provided the manager does not include the conduct of research among their core responsibilities. Because the overall intent of the intervention is to support the use of research evidence in decisions about CBOs' programs, services, and advocacy, we deemed it most appropriate for the executive director (or another manager) to complete the survey because they would have the most impact on whether research evidence is used to inform decisions.
Based on the membership list provided by the Canadian AIDS Society on their website, there are 120 CBOs available to draw the sample from. Drawing on previous experience with this sector, we expect to achieve an approximate response rate of 70%. To increase our response rate, the Canadian AIDS Society will send out an email to all its members, encouraging them to participate by highlighting the importance of the trial. We will provide additional incentive to enroll in the trial by holding a draw where we will select three organizations to receive prizes (gift cards) worth $500, $250 and $100.
We will run a two-arm RCT with a 'full-serve' evidence service (SHARE) as the intervention arm and a 'self-serve' version as the control arm. The components of each version of SHARE are outlined in Table and described below.
Components of the 'full-serve' and 'self-serve' evidence service
Intervention arm: 'full serve' evidence service
Organizations allocated to this study arm will receive access to a 'full-serve' version of SHARE, which provides:
1. an online searchable database of HIV-relevant systematic reviews (retrievable based on a taxonomy of topics related to HIV/AIDS and open text search - see Additional file 1
: Appendix 1 for the taxonomy of topics);
2. periodic email updates (at least one per month), which will profile the types of new reviews recently added to the database (e.g., the number of Cochrane reviews) and provide a brief overview of the range of topics addressed by the new reviews;
3. access to user-friendly summaries produced by us or by others (when available);
4. links to scientific abstracts;
5. peer relevance assessments, which involves periodic requests (contained in the single record for each review) to complete a brief assessment of how useful the information in the newly added review is (one question with a five-point scale - see Additional file 2
: Appendix 2 for additional details) with the average score posted once an assessment is completed;
6. an interface for participants to leave comments (up to 250 characters in length) in the records of systematic reviews in the database (e.g., if a participant wants to leave a comment indicating the review was useful and why);
7. links to full-text articles (when publicly available); and
8. access to worksheets that help CBOs find and use research evidence
To provide access to user-friendly summaries (see component three above) we will provide links to user-friendly summaries produced by nine groups (when available) from around the world: Australasian Cochrane Centre (AAC) Policy Liaison Initiative, Database of Abstracts of Review of Effects (DARE), Effective Healthcare Research Programme Consortium, Evidence AID, Health Knowledge Network, Health-Evidence.ca, Reproductive Health Library, Rx for Change, and Supporting Policy Relevant Reviews and Trials (SUPPORT)
Organizations allocated to the control group will only be provided website access to a listing of systematic reviews that are organized by year of publication with links to the record on PubMed (or another publicly available source when not available on PubMed) and access to worksheets that help CBOs find and use research evidence.
After consenting to participate in the trial, we will use simple randomization to assign organizations to receive either the 'full-serve' or the 'self-serve' evidence service. The list of participating organizations will be sent to a statistician (TB) who will assign a unique ID number to each organization, conduct the randomization, and keep both the key linking the organizations to their ID and the randomization log in a secure password protected folder at the Ontario HIV Treatment Network to provide a clear audit trail. We will perform simple randomization sampling using the SAS SELECTSURVEY procedure to assign equal numbers of organizations to the 'full-serve' and the 'self-serve' groups. The procedure will be performed with a fixed seed so that the sampling can be replicated if needed. The statistician will then provide the list of unique IDs with the results of the randomization to the SHARE database administrator at the Ontario HIV Treatment Network (external to the research team) who will provide individuals from each participating organization with access to the 'full-serve' or 'self-serve' versions of SHARE. This will require the SHARE database administrator to have access to the key linking the unique IDs to the organizations but it will remain concealed from the research team.
Prior to the start of the trial, all organizations will be requested to provide a list of emails of management and staff interested in receiving access to SHARE, which will be provided to the SHARE database administrator at the Ontario HIV Treatment Network. We will then send bi-monthly emails to the executive director (or another delegated staff member for correspondence) to identify any staff that have either joined or left the organization in order to accurately track usage at the organizational level. The SHARE database administrator at the Ontario HIV Treatment Network will send the updates to individuals affiliated with organizations with access to the 'full-serve' version of SHARE (the updates will be written by MGW and checked by the co-investigators). The statistician (TB) is a member of the study team but will only be involved with randomization at the start of the trial and the data analysis upon completion of the trial. Therefore, participants and all investigators except the statistician (TB) and the SHARE database administrator will be blinded to group assignment.
Measuring the impact of knowledge transfer and exchange (KTE) interventions, such as the evidence service proposed here, poses significant challenges as there is a long chain of factors between a KTE intervention such as SHARE and the health status of clients of CBOs or of broader populations [10
]. For example, it has been demonstrated that assessing the impact of KTE interventions on the practice of physicians poses challenges due to the fact that many factors other than the practice guidelines or recommendations that were disseminated may influence how practices are changed [36
Given these constraints, our primary and secondary outcomes for the trial are proxy measures for research use. The primary outcome will be a measure of utilization that is similar to what Haynes et al
. (2006) used in their trial of the McMaster Premium Literature Updating Service (PLUS) [39
]. Specifically, we will track utilization at the organizational level by calculating the mean number of logins/month/organization (the total organizational logins/month will be averaged across the number of users from each organization) across trial groups during each of the baseline period, intervention period, and crossover period. We will also provide related descriptive measures such as the mean number of logins/month for different types of positions within the organization (executive director, management and staff), the range of logins/month within the organization, the proportion of organizations with at least one user accessing the 'full serve' and 'self-serve' versions of SHARE each month, the frequency with which systematic review records and related resources are accessed (e.g.
, URLs to abstracts, user-friendly summaries, and/or full-text), and the number of times the email updates to the 'full-serve' group are forwarded.
Each version of the evidence service will be hosted on the Ontario HIV Treatment Network server and for the duration of the trial will require a user login that will be used to link each participant's identification with their usage of the evidence service website and to their organization. SHARE is a new database that is not yet publicly available (it will be upon completion of the trial), which allows us to evaluate it without participants being able to gain access from a publicly available site. In addition, requiring a user login will help protect against contamination of the intervention and control group. However, we cannot protect fully against the possibility of participants from the organizations sharing information given that many may collaborate with each other.
For the secondary outcome, we will use the theory of planned behaviour to measure participants' intention to use research. The theory of planned behaviour proposes a model about how human action is guided [40
] and consists of three variables -- attitudes (i.e.
, beliefs and judgments), subjective norms (i.e.
, normative beliefs and judgments about those beliefs), and perceived behavioural control (i.e.
, the perceived ability to enact the behaviour) -- that shape the behaviour intentions of people, which is subsequently a strong predictor of future behaviour [41
]. In Figure , we outline the model of the theory of planned behaviour and map how different elements of the evidence service may affect each of the three variables.
Linkages among the intervention, contextual developments, and theory of planned behaviour constructs.
The theory of planned behaviour has been extensively used and tested in the fields of psychology and healthcare. Systematic reviews conducted in the psychology field have demonstrated that the theory explains about 39% of the variance in intention and about 27% of the variance in behaviour [42
]. A number of studies have demonstrated the feasibility of producing valid and reliable measures of the key theory of planned behaviour constructs for use with healthcare professionals [44
]. A systematic review suggests that the proportion of the variance in healthcare professionals' behaviour explained by intention was similar in magnitude to that found in the broader literature [47
]. With the successful transfer of the theory from assessments of individuals to assessments of healthcare professionals involved in an agency relationship with their patients, we are confident in its further transfer to key decision makers in CBOs in agency relationships with other decision makers and staff in their organization.
Using a manual to support health researchers who want to construct measures based on the theory [41
], our colleagues have developed and sought preliminary feedback on a data-collection instrument by first assessing face validity through interviews with key informants and then pilot testing it with 28 policymakers and researchers from 20 low- and middle-income countries who completed it after participating in a KT intervention [48
]. In addition, Boyko et al
. (2010) found moderate test-retest reliability of the instrument using Generalizability Theory (G = 0.50) [49
] when scores from a sample of 37 health system policymakers, managers, professionals, citizens/consumers, and researchers participating in stakeholder dialogues convened by the McMaster Health Forum were generalized across a single administration, and even stronger reliability (G = 0.9) when scores were generalized across the average of two administrations of the tool [48
]. In the reliability assessment by Boyko et al
. (2010), the first administration of the tool immediately followed a McMaster Health Forum stakeholder dialogue, which may have promoted enthusiasm for using research evidence among participants. This likely produced higher measures of intention on the first administration of the tool as compared to the second, resulting in the lower G-score. Given that we won't be administering the tool in a similar atmosphere of enthusiasm for using research evidence, it is likely that the level of reliability of the tool will be sufficient without two administrations at both baseline and follow-up.
We have slightly modified the wording in each of the questions of the tool to reflect the different intervention being tested (SHARE) and the target audience (CBOs) (see Additional file 3
: Appendix 3). We will administer the instrument to one key decision maker from each organization during the baseline period, as well as at the end of the six-month intervention period, through a brief online survey that takes approximately 10 minutes to complete. We will use unique identifiers for each participant to ensure their responses to the previous survey are linked for calculations of before-and-after changes in their intention to use research evidence. We will follow up with participants who do not complete the survey once per week for three weeks to minimize the number of participants lost to follow up.
Data management and analysis
Data will be entered into SPSS 16.0 using unique identifiers that link each participant to their respective organizational identifier assigned during the randomization process. Analyses will be conducted by two members of the team (MGW and TB) and, during the analysis, all investigators -- except for one of us who is involved in the both the analysis and randomization (TB) -- will be blinded to the key linking the organizations to their unique identifiers.
We will treat both outcome measures as continuous variables and analyze the change in these measures over time using a two-way mixed effects linear repeated measures analysis of variance (ANOVA), which will assess the effects within groups, between groups, and over time with the latter as the main feature of interest. In addition, we will control for four variables -- province the organization is located in, size of organization (as measured by number full-time equivalent staff in the organization), number of participants that participated from each organization, and the number of clients served each year by the organization -- using analysis of covariance. For the analysis of the secondary outcome, we will also control for whether the key decision maker is full-time or part-time, and whether they have had research training in the past. Each of these variables may at least partially explain research use (e.g., the amount of staff support an executive director or manager has may determine the extent to which they can spend time finding and using research evidence), and therefore adjusting for them will allow for a better assessment of the effects of the intervention. Moreover, as part of a secondary analysis, we will assess whether there is an interaction between each (entered as fixed factors) and the outcome measures. Given the likelihood that the distribution of the outcomes will be skewed, we will transform the data where necessary and possible, which may include adjusting the time period for which we calculate the mean number of logins/month/organization (e.g., calculating the mean over two months) if the number of logins is low and there are insufficient data for analysis. We will also qualitatively compare the number of participants in the intervention and control groups that do not complete the follow-up survey, and attempt to assess if there are reasons for why they dropped out based on their baseline characteristics.
For all analyses, we will use the intention to treat principle, report 95% confidence intervals, and consider p-values equal to or less than 0.05 (two-tailed) to be significant. For the primary outcome measure (mean logins/month/organization), missing data are irrelevant as they are a naturalistic measure. For the secondary outcome measure (obtained through the survey), missing data can be taken into account through the use of a mixed-effects model.
Given a fixed sample size of approximately 85 organizations (70% of 120 organizations) a sample size calculation is not relevant. Instead, we have calculated the level of statistical precision that we can expect given our fixed sample size. To calculate the expected statistical precision in the trial, an estimation of intra-class correlation coefficient (ICC) of measurements for individuals over time for the primary outcome is required. However, we have no mechanism to estimate the ICC due to the fact that no similar study with this population has been conducted (at least to our knowledge). Therefore, we calculated estimates of statistical precision for ICCs of 0.2, 0.3, 0.5, 0.7 and 0.8 based on a six-month trial period with 80% power, an estimated standard deviation of 1.0, significance of 0.05 (two-sided test), and 42 organizations per study group (total n = 85) [50
]. Assuming the primary outcome data will be collected from all 85 organizations during the intervention period at six follow-up points (one per month), the time-averaged detectible differences (in standard deviation units) between the two groups is at best 0.35 (for ICC = 0.2), which increases with successively greater ICCs to 0.39 (for ICC = 0.3), 0.47 (for ICC = 0.5), 0.53 (for ICC = 0.7), and 0.56 (for ICC = 0.8).