LAUSR

This study examined the effectiveness of anti-smoking messages in positively shifting perceptions of risks related to smoking among adolescents in urban Ghana, both directly through exposure to messaging and indirectly through catalyzing discussions among peers and adults. We used data from two waves of a population-based survey of 3775 adolescent girls and 3279 adolescent boys aged 13-16 years in the urban areas of Accra, Teshie, Kumasi and Sunyani in Ghana. Using an interviewer-directed questionnaire, information was collected on sociodemographic characteristics, tobacco knowledge, exposure to messages about tobacco, frequency of conversations with peers and adults about the health risks associated with smoking, as well as beliefs and attitudes towards smoking. The outcome measure is smoking risk perception in relation to shisha and cigarettes. Using causal mediation analysis, we examine the direct and indirect effects of social media relative to mass media as they work to spur discussions among peers, as well as between peers and adults, about smoking risks and about aligning risk perceptions with objective risk. We find that media exposure - both social and mass media - can impact risk perceptions. We find in particular an 8.6 percentage point increase in shisha smoking risk perceptions due to social media campaign exposure (95% CI: 3.71-13.50) and a 2.3 percentage point increase in cigarette smoking risk perception due to mass media campaign exposure (95% CI: 0.26-4.27). We further find that the indirect effect of conversations with peers mediated 16.3% of the total effect of mass media campaign exposure on cigarette smoking risk perception and 4.8% of the total effect of social media campaigns on shisha smoking risk perception. Social media campaigns increased shisha smoking risk perception among girls by 11.6 percentage points (95% CI: 5.59-17.61), of which 6.8% was mediated by conversations with peers, while conversations with peers did not significantly mediate campaign effect among boys. Conversations with adults were never found to significantly mediate smoking prevention campaigns effects in this sample of Ghanaian adolescents. Behavior change communication programs that rely solely on traditional mass media can miss important opportunities for shifting smoking risk perceptions and sharing information on the harms of smoking. Evaluations that ignore the indirect effects of conversations among peers mediating campaign effects may fail to identify important channels that can be targeted and triggered by mass and social media tobacco control campaigns.