NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
McCormack L, Sheridan S, Lewis M, et al. Communication and Dissemination Strategies to Facilitate the Use of Health-Related Evidence. Rockville (MD): Agency for Healthcare Research and Quality (US); 2013 Nov. (Evidence Reports/Technology Assessments, No. 213.)
This publication is provided for historical reference only and the information may be out of date.
Communication and Dissemination Strategies to Facilitate the Use of Health-Related Evidence.
Show detailsResults of Literature Searches for Key Questions 1, 2, and 3
We identified 4,152 articles from all sources (after removing duplicates) for all three Key Questions (see Figure 2). Two independent reviewers examined each abstract and applied our inclusion and exclusion criteria. Based on that process, we retained 445 articles for full text review. The majority of the full-text articles were classified to one or more Key Questions (KQ)—106 articles pertained to KQ 1; 163 articles pertained to KQ 2; 84 articles pertained to KQ 3, and 98 articles were classified as overlapping. Each overlapping article potentially applied to two or more KQs and was not classified into one KQ category.
Figure 2
PRISMA.
Of the full-text articles, we excluded 386, leaving 61 articles for data abstraction. Nine articles (representing seven studies) are relevant to KQ 1; 42 articles (representing 38 studies) are relevant to KQ 2; and ten articles (representing nine studies) are relevant to KQ 3. Appendix C lists all articles excluded at the full-text review stage and the reason for exclusion.
This section presents the results for KQ 1: the effect of various communication strategies on both intermediate and distal outcomes. For KQs 2 and 3, we provide more information in the two sections that follow.
Introduction
For this KQ, we examined comparative studies of the following communication strategies: tailoring messages to individuals; targeting messages to audience segments; using narratives to convey messages; and using framing to convey messages to various end-users. For this KQ, as we noted in our methods, we included only randomized controlled trials (RCTs).
Some trials compared two strategies directly with each other (e.g., targeting vs. tailoring); others used a combination of strategies (e.g., targeting and tailoring vs. targeting). The tables below describe individual trials and their results and document our SOE grades. Detailed evidence tables for KQ 1 studies are in Appendixes D. As noted above, we retained nine articles after full-text review that met inclusion criteria,72–80 which report on seven unique trials about communication strategies.
Description of Included Studies
Of the seven included trials, we assessed two as low risk of bias76,77,80 and five as moderate risk of bias.72–75,78,79 None was assessed as high risk of bias. All trials reported on the effects of interventions on various behaviors. Five of the trials reported on the effects of intervention messages and materials on adherence to guidelines about screening (four on breast or cervical cancer and one on colon cancer). One trial reported on obtaining influenza and flu shots. One trial reported on the effects of the intervention materials on dietary behaviors.
Because of the diversity of communication strategies and potential interactive effects, we graded the SOE for each communication or combination of communication strategies separately. For instance, we graded the SOE for one trial comparing framing versus narratives by itself, but we graded the SOE for two trials comparing framing versus targeting together even though they examined different behavioral outcomes.
The investigators tested these interventions in study populations in the United States and Hong Kong. Sample sizes ranged from 174 participants to 5,500 participants. Five of the trials used convenience samples that were drawn from various populations including community health clinics, a public housing unit, university classrooms, and a California county. Two studies drew patients from large clinical practices.
Key Question 1. Communicating Evidence to Patients and Clinicians
Key Points
- Framing (gain/loss) versus narratives (yes/no). Loss framed messages used in conjunction with narratives were more persuasive (i.e., convincing) than either loss framed messages in conjunction with statistical information alone or gain framed messages in conjunction with either narratives or statistical information (one trial; insufficient SOE).
- Framing (gain/loss) versus targeting (targeted/not). The loss-framed message used in combination with non-targeting (i.e., a more broad appeal either culturally or societally, such as a collectivist appeal) was more persuasive relative to any other combination of framing and targeting, but the results held only in the short-term for one of the trials and the targeting was done on different factors across the trials (two trials; insufficient SOE).
- Targeting (yes/no) versus tailoring (yes/no). Findings were mixed; that is non-significant or counterintuitive for the three studies that compared targeting with tailoring. In all three studies, investigators hypothesized that the tailored version of the intervention would have a greater effect on the outcome than the targeted version. However, there were no significant differences in outcomes between those receiving the targeted or tailored version of the intervention in two studies. In a third study, the targeted version was associated with greater likelihood of self-reported screening relative to the tailored version in one study. The authors attributed this unexpected finding to a possible ‘boomerang effect’ (i.e., because the tailored letter may have been too alarming) and/or insufficient customization of tailored version. Across the three studies, investigators targeted and tailored the interventions based on different factors (three trials; insufficient SOE).
- Targeting (yes/no) and tailoring (yes/no) versus targeting only. Investigators found no statistically significant differences when they targeted an intervention to the subpopulation and personally tailored it to each study participant compared to a version of the intervention that was only targeted. They attributed the lack of differential impact to a possible ‘ceiling effect’ in the study population given the fairly high baseline screening rates (of about 80 percent) (one trial; low SOE).
Challenges in interpreting the current body of literature include:
- Use of multiple communication strategies simultaneously. In several cases, investigators used some combination of the four communication strategies when developing their interventions instead of comparing only a single strategy with another single strategy. Because comparisons were not one-to-one, it was more challenging to isolate the effects of each strategy.
- Combining communication strategies with channel variation. In one trial, investigators enhanced the communication strategy by also varying the communication channel for the intervention (i.e., using a lay health worker). While this tactic creates the potential for a more powerful effect, it is also complicates determining the effect of each strategy relative to the other.
- Variation in use of strategies for patients versus clinicians. None of the trials that met our review addressed using the four communication strategies with clinicians; therefore, we were unable to address KQ 1b.
Table 14 documents the strength of evidence grading for each of the five comparisons and gives the overall SOE grade.
Table 14
Strength of evidence of communication strategies.
Detailed Synthesis
Table 15 describes the seven trials and their results in detail. This information is presented in ways reflecting the conceptual framework and the orientation of analysis of the original investigators. In most cases, however, the review team had to calculate differences between groups (e.g., in mean values on a scale or percentages). Because numerous findings led to negative differences (because of the original choices about the directionality of comparison), the table indicates whether the difference was negative or positive and notes which group the findings favored. By favored, we mean which study group had the better result, namely a higher screening rate or better eating habit.
Table 15
Studies of communication strategies.
Framing (Gain, Loss) Versus Narratives (Yes/No) (Anecdotal/Statistical Evidence)
A small (N=174) trial examined the effect of experimental advertisements that differed in terms of how consequences of getting screened for breast cancer (with mammography) were framed (gain or loss) and how the evidence was presented (narrative/anecdotal vs. non-narrative/statistical) (Table 15).73 In this trial, investigators randomly assigned 116 women to one of four groups with different message combinations: (1) control; (2) gain frame and non-narrative/statistical; (3) loss frame and non-narrative/statistical; (4) gain frame and narrative/anecdotal, and (5) loss frame and narrative/anecdotal. Gain-framed messages focused on the potential for screening to save lives (“…they are less likely to die of breast cancer”); loss-framed messages focused on the possibility of death from not being screened (“…they are more likely to die of breast cancer”). The narrative/anecdotal approach involved personal narrative stories (“Doctors were able to detect her breast cancer at an early, treatable stage, and now Sara can look forward to a long life, watching her grandson, Jeffrey, grow up”); non-narrative/statistical approach had a numerical emphasis (“Doctors are able to detect their tumors at an early, treatable stage, and they [women] are 30 percent less likely to die of breast cancer”).
The effect of the different approaches varied. Women who received the narrative/anecdotal and loss-framed message (Group 5) reported the highest mean likelihood of getting a mammogram (Group 5; 5.54 on a 7-point Likert scale). Likelihood values dropped off as follows: non-narrative/statistical and gain frame (Group 2; 5.48); non-narrative/statistical and loss frame (Group 3; 4.37); and finally narrative/anecdotal and gain frame (Group 4; 4.07). Framing and use of narratives had an interactive effect on subjects’ predictions of their own mammography behavior. The effects of framing were moderated by how the evidence was presented; specifically, among those exposed to narrative/anecdotal evidence, the loss-framed messages were more persuasive. By contrast, among those receiving non-narrative/statistical evidence rather than narrative/anecdotal information, the likelihood of getting a mammogram did not differ significantly between those who received gain-framed and a loss-framed messages (p=0.06).
Based on this single study, we graded the overall strength of evidence as insufficient because of the small sample sizes in each intervention group (n=29), the use of a convenience sample (reflected in the risk of bias assessment of moderate), and the imprecision of the results that were evaluated only immediately after exposure to the intervention (precluding any conclusions).
Framing (Gain, Loss) Versus Targeting (Yes/No)
One trial used a 2 × 2 factorial design to examine the effect of message framing (gain vs. loss) and targeting on the basis of ethnicity of the women receiving the information; the two options were targeting Latinas only or taking a multicultural orientation (women of various ethnic backgrounds).74 The purpose of the messages was to motivate breast cancer screening in low-income women who are medically underserved. The investigators hypothesized that targeting would enhance attention to the message, especially for the loss frame. They showed women older than 40 one of four videos with four different message strategies: (1) gain frame and non-targeted/multicultural, (2) loss frame and non-targeted/multicultural, (3) gain frame and targeted toward Latinas, and (4) loss frame and targeted toward Latinas. Participants self-reported information about mammography use 6 and 12 months after exposure to the videos. The investigators contacted study participants by either telephone or mail (stamped, preaddressed envelope).
On average, 41 percent of participants reported having a mammogram within the past 12 months 6 months after exposure to the intervention. Those who received the loss frame and non-targeted (multicultural) video reported the highest percentage of mammograms (Group 2; 50%), followed by those exposed to the gain frame and Latina targeted video (Group 3; 41%). Among women receiving the gain frame and non-targeted (multicultural) video and those receiving shown the loss frame and Latina targeted video (Groups 1 and 4), 37 percent of participants had a mammogram within the past 12 months 6 months after exposure to the intervention. Using hierarchical logistic regression, controlling for the past year’s screening usage, framing and targeting had a significant interactive effect on the probability of getting a mammogram. “Within 6 months after participation, loss framed videos persuaded more participants to obtain mammograms than the gain framed videos, but only among those viewing the non-targeted multicultural context. Unexpectedly, the loss framed, non-targeted/multicultural message was more persuasive in terms of mammography use than the other three kinds of messages.” No psychological mediators (e.g., perceptions of risk, attitudes) were systematically influenced by the framing and targeting interaction74, p.260
On average, 57 percent of participants reported having a mammogram within the past 12 months 12 months after exposure to the intervention. The pattern of differences among study groups was similar at 6 and 12 months. Within 12 months after participation, and controlling for the past year’s screening usage, the framing/targeting interaction in the logistic regression model was not statistically significant.
One RCT also examined the effects of gain versus loss message frames when they are targeted to audience segments based on cultural differences, specifically individualistic or collectivistic orientation.79 Individualism includes a tendency to focus on the self,81 whereas collectivism incorporates the self as part of a larger group.82 The investigators sought to determine if there was an interactive effect between framing and targeting the messages in this way. The messages focused on preventing influenza by getting immunized. Messages were delivered as part of a brochure with the inside of the brochure manipulated to create four different versions: (1) a loss frame with an individualistic appeal (self-loss message: Skipping a Flu Shot May Put You at Risk); (2) a loss frame with a collectivistic appeal (other-loss message: Skipping a Flu Shot May Put Many at Risk); (3) a gain frame with an individualistic appeal (self-gain message: Getting a Flu Shot May Benefit You); and (4) a gain frame with a collectivistic appeal (other-gain message: Getting a Flu Shot May Benefit Many). Several aspects of the brochure (the headline, a quote from a doctor, the primary content, and the call to action) reflected these nuances. Other aspects remained constant.
The investigators found a significant interaction between message framing and this type of cultural appeal. Those who received the loss framed messages oriented toward benefitting others, were more likely to intend to get a flu shot relative to those with a loss-framed message oriented toward the self. This finding held for both study populations—Hong Kong, Chinese (6.04 versus 4.51) and Americans (6.49 versus 4.39) based on a 10-point Likert scale containing statements such as “I intend to behave in ways that are consistent with the message.”79 The investigators also conducted a mediation analyses and found that for Hong Kong Chinese, perceived severity of influenza increased behavioral screening intentions. In the American sample, “gain-self and loss-other appeals promoted behavioral intention through changing people’s cognitive perceptions of the issues and attitudes toward the behavior” (p. 143).
The two trials of varying sizes, had moderate risk of bias, precise and consistent estimates, but the results in one trial held only at 6 months and not at 12 months, and the targeting was done on different factors. Therefore, we graded the SOE as insufficient.
Targeting Versus Tailoring
One randomized trial evaluated the effectiveness of two types of letters to encourage low-income women served by county public health clinics to adhere to cancer screening recommendations.75 One group of women (Group 2, denoted PF) received a form letter targeted to women age 40 and older; another group of women (Group 3, denoted PT) received a personally tailored letter (based on information from their medical record about their personal risk of cancer) (e.g., “Mrs. Smith, you may be at risk of breast cancer because…); and a third group (Group 1) served as a control. Both interventions sought to prompt women to get screened for breast and cervical cancer.
Unexpectedly, women who received the personally tailored intervention (Group 3, PT) were less likely to schedule (13.6 percentage points less) or obtain (20.2 percentage points less) cervical screening within the first 12 months after receiving the letter than women who received targeted form letter (Group 2, PF). Percentages for control group members fell between the targeted and tailored groups. The differences among the three study groups for breast cancer screening were generally comparable to those for cervical cancer screening.
Another RCT examined innovative approaches to changing lifestyle behaviors to reduce dietary fat and to increase fiber among Latinas and their families in two counties in Southern California.72,78 One intervention group (the “tailored print” material option) received 12 weekly newsletters tailored using baseline survey data provided by the participants. The messages were tailored on a variety of factors including the meals prepared at home most often, readiness to change behavior, points of influence in one’s life. The newsletters contained activity inserts and a story (novella) about a woman gaining control of her personal life. This group also received supporting materials (e.g., recipes and magnets to prompt behavior change). All materials were delivered to women in their homes via the U.S. postal service. The other intervention group (the “lay health worker/promotora tailored print material” option) received weekly home visits or telephone calls from promotoras (lay health workers in their community) over a 12-week period plus the 12 newsletters and activity inserts also delivered by mail. Finally, a control group received “off the shelf” materials by mail covering the same modules and content as the intervention groups; these women received no tailoring or personal interaction, but the materials were targeted to Latinas.
Using analysis of covariance, controlling for baseline levels, the investigators found no statistically significant differences in terms of a percentage point decrease in calories consumed from fat over a 12 week period among the three groups. No statistically significant differences emerged in terms of the difference in percentage point decrease in dietary fiber consumption among the three groups. The investigators found no statistically significant changes in dietary outcomes over time when comparing 12 month data to 12 week data (information about calories from fat was not reported). Based on these two trials with moderate risk of bias and varying levels of precision and inconsistent findings, we graded the strength of the evidence as low.
Another RCT compared the effect of targeted versus tailored interventions to increase colorectal cancer screening rates among patients in a large, urban medical practice.80 Study participants were between 50 to 74 years of age. One study group (Group 2, denoted SI for standard intervention) received a version of the intervention targeted “to individuals who were not up to date with screening according to guidelines”80(p. 2084). Another study group (Group 3, denoted TI) received a version of the intervention the standard version of the intervention plus two tailored ‘message pages’. The pages were tailored based on personal barriers to screening identified through baseline data. A third intervention group (Group 4, denoted TIP) received the standard intervention, the tailored message pages, and a telephone reminder about screening from a trained health educator. After initial contact which occurred within 30 days of randomization, participants were contacted two more times—approximately 12 months later, and again approximately 24 months after baseline (if they had not been screened at 12 months).
The investigators used multiple sources of data (an endpoint chart audit, billing data, and self-report) 24 months after exposure to the intervention. While all three intervention groups were more likely than the control group to have been screened within 24 months, no significant differences emerged between the intervention groups based on univariate or multivariate analyses.
Based on these three trials of varying sizes, moderate risk of bias and precise and inconsistent estimates, we graded the strength of the evidence as insufficient. The mixed results could be due to different factors used for targeting and tailoring across the three studies.
Targeting and Tailoring Versus Targeting Only
One large (N=5,500) trial developed and evaluated interventions to promote breast cancer screening in a nationally representative sample of U.S. women veterans.76 The trial included women ages 52 and older who were no longer on active duty. One group (targeted only) received a less personalized, targeted intervention; it included a letter for women to use to discuss mammography with her health care provider, educational booklets, and a pamphlet about mammography screening services available through the Veterans Health Administration (VA). Another group (targeted and tailored) received the same materials, but their tailored letter consisted of individualized messages that addressed each participant’s responses to questions on a baseline survey reflecting her attitude and opinions about screening, including her stage of readiness to be screened. A control group received no intervention. All groups completed the baseline survey.
The more intensive intervention (tailored and targeted) was no more effective than the less intensive intervention (targeted only) in terms of screening rates at 12 months (46.0% vs. 46.9% screened, respectively) and at 24 months after the intervention (24.8% vs. 24.8% screened, respectively). Further, differences between the both intervention groups and the control group at both time periods were not statistically significant based on multivariate analyses (namely, Cox proportional hazard modeling and intention to treat [ITT] analyses). Modified ITT was used to evaluate screening at 24 months because it was conditional on having a mammogram at 12 months. This single trial had low risk of bias and precise estimates given adequate power and thorough analyses, and we graded the strength of the evidence as low.
- Results—Key Question 1: Communication Strategies - Communication and Disseminati...Results—Key Question 1: Communication Strategies - Communication and Dissemination Strategies to Facilitate the Use of Health-Related Evidence
- A Framework To Facilitate the Use of Systematic Reviews and Meta-Analyses in the...A Framework To Facilitate the Use of Systematic Reviews and Meta-Analyses in the Design of Primary Research Studies
- Methods Guide for Effectiveness and Comparative Effectiveness ReviewsMethods Guide for Effectiveness and Comparative Effectiveness Reviews
- Grading the Strength of a Body of Evidence When Comparing Medical Interventions ...Grading the Strength of a Body of Evidence When Comparing Medical Interventions - Methods Guide for Effectiveness and Comparative Effectiveness Reviews
- Future Research Needs for the Comparison of Percutaneous Coronary Interventions ...Future Research Needs for the Comparison of Percutaneous Coronary Interventions with Bypass Graft Surgery in Nonacute Coronary Artery Disease
Your browsing activity is empty.
Activity recording is turned off.
See more...