Effectiveness of Mass Media Interventions for HIV Prevention, 1986–2013: A Meta-analysis

"This meta-analysis was conducted to synthesize evaluations of mass media-delivered HIV prevention interventions, assess the effectiveness of interventions in improving condom use and HIV-related knowledge, and identify moderators of effectiveness." The article is from the JAIDS: Journal of Acquired Immune Deficiency Syndromes supplement addressing clinicians and public health scientists in the field of HIV prevention and treatment who might value information on health communication. (Footnotes removed by the editor.)

In this meta-analysis, "[s]tudies were included if they (1) had an HIV/AIDS prevention focus, (2) targeted youth or the general population, (3) evaluated a specific intervention delivered through an audio, visual, or printed mass media channel in a natural setting, (4) quantitatively evaluated the target population using a pre-/post-campaign design, (5) measured condom use and/or HIV/AIDS-related transmission or prevention knowledge, (6) reported sufficient information to calculate effect sizes, and (7) were available in English." Excluded were studies that: focused on a high-risk group or on forced message exposure, sampled only individuals exposed to some campaign components, or used interpersonal media.

As illustrated in Table 1 of the document, interventions were evaluated at pre-/post-campaign using a total of 93 separate sample comparisons drawn from the selected populations. "[M]ost studies were conducted in Africa (50%) or Asia (17%). Interventions were generally conducted nationally (49%) or based in individual communities (43%). Many campaigns pretested message content before campaign launch (39%), and 45% were theory based. Of the 72 interventions, 15% used only 1 channel, and of these, 4 used small media, 3 used radio, 3 used newspapers or magazines, and 1 used television. All other campaigns (83%) used 2 or more channels including signage (72%), radio (70%), television (57%), educational literature (51%), newspapers or magazines (33%), and promotional materials (21%). Overall, 47% of the interventions included an interpersonal component. Most interventions focused on condom promotion (76%), and more than a third used an entertainment-education strategy (39%). Most campaigns had an identifying logo, slogan, or brand (57%), and 42% included condom distribution. The median intervention duration was approximately 8 months, though duration varied widely."

The research report provides details of sample sizes, gender, and other demographic information and then describes measures of efficacy. On condom use: "Greater increases in condom use occurred following interventions conducted in African nations, in countries with lower HDI scores, following longer campaigns, when message content was reportedly matched to the target audience, and when refusal rates were low. (Human Development Index (HDI) is a measure of nation-level social and economic development that combines indices related to life expectancy, educational attainment, and income.) Condom use also increased to the extent that campaigns increased knowledge of transmission (β = 0.56, P = 0.009, k =16) and prevention (β = 0.30, P = 0.03, k = 41).”

On transmission and prevention knowledge, researchers found the largest increases of transmission and prevention knowledge in Asian nations. "Increases in prevention knowledge were larger following longer campaigns. Interventions that reportedly included demonstrations of condom use did not significantly improve prevention knowledge compared with interventions that did not report condom demonstrations. No other coded report characteristics (eg, publication status), campaign characteristics (eg, number of communication channels, message branding, reported theoretical background, condom distribution) or sample characteristics (eg, age, gender, proportion sexually active, pre-campaign rate of condom use) were significant predictors of effect size magnitude for any outcome of interest."

The discussion section reflects upon the lack of use of control groups in the research analysed: "Previous reviews of mass media-delivered interventions to prevent HIV/AIDS have addressed the variability in research design, methodology, and reporting of intervention development, content, and delivery. The most persistent hurdle in determining the efficacy of mass media campaigns is the lack of available comparison groups in the evaluation of these interventions; only 8 of the 72 interventions included in the present meta-analysis provided valid comparison group data. Because of the inability to reliably control access to mass media, it may be both difficult and costly to use true or quasi-experimental designs. As a result, many campaign evaluations compare pre- and post-campaign data or compare individuals who were or were not exposed to the campaign. Both of these evaluation methods have limitations. The former does not allow for direct attribution of changes in knowledge or behavior to the campaign message alone, although it is notable that in the current investigation, campaigns were more successful to the extent that more respondents reported exposure. Additional factors unrelated to a specific campaign, such as the social or political climate, public policy changes, current events, and the presence of multiple campaigns in a targeted area can influence HIV/AIDS-related knowledge and sexual risk behaviors, making it impossible to isolate intervention effects. Evaluations based solely on exposure are also limited in that exposure to mass media may be related to individual characteristics such as income, gender, relationship status, and age....The intervention literature in general would benefit tremendously from the public availability of more quantitative evaluations (ie, using pre-campaign comparison groups or quasi-experimental designs), described in greater detail (eg, specifying media channels used). It would also be beneficial to the literature to have studies include cost information to facilitate cost-effectiveness analyses. Additional information would allow for more nuanced theorizing about what sorts of mass media interventions may work best in different contexts."

The discussion section compares results of meta-analyses of studies that use strategies including interpersonal interaction and finds: "Despite the differences in channels between the current and previous meta-analyses, the effects on behavior change are similar in magnitude - small to medium, depending on moderating factors. Despite their modest size in absolute terms, for statewide or national media interventions the scope of their impact is quite large in absolute terms. Media campaigns are a means of 'going to scale' - taking an intervention to large numbers of people."

The discussion section concludes that mass media campaigns do have power to affect people living in nations with most need for HIV prevention and health promotion interventions.