Motivational effectiveness of prosocial public health messaging to reduce respiratory infection risk: A network meta-analysis

doi:10.21203/rs.3.rs-4888738/v1

Download PDF

Article

Motivational effectiveness of prosocial public health messaging to reduce respiratory infection risk: A network meta-analysis

https://doi.org/10.21203/rs.3.rs-4888738/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Clear communication is vital for adopting public health interventions that promote protective behaviours against respiratory infections. This systematic review and network meta-analysis assessed the effectiveness of these interventions using behavioural science frameworks, including MINDSPACE contextual influencers and behaviour change techniques (BCTs), to identify key components and mechanisms of action (MoAs). The interventions primarily focused on social distancing, mask-wearing, handwashing, diverse-behavioural intentions, and actual behaviours. The network meta-analysis revealed that prosocial messages, especially those involving loved ones, significantly reduced the risk of respiratory infections (d=0.09; 95% CrI=0.06-0.14; CINeMA: Low). Interventions typically included three contextual influencers (salience, affect, ego) and five BCTs (Information about health consequences; Salience of consequences; Information about social and environmental consequences; Demonstration of the behaviour; Avoidance/reducing exposure to cues for the behaviour). Behaviour intention was the most common MoA. Although further research is needed, this review provides insights into designing effective public health messages for respiratory infection control.

Scientific community and society/Social sciences/Interdisciplinary studies

Scientific community and society/Social sciences/Communication

The COVID 19 pandemic made plain that in the face of a novel viral infection causing severe respiratory disease, among other containment measures, the rapid adoption of population protective and physical social distancing behaviours was required.^1,2 As the waves of infection unfolded and vaccinations became available, it remained the case that behavioural interventions focused on personal protective behaviours (such as hand and respiratory hygiene, mask wearing and social distancing) were still important to contain transmission.^3-5 Respiratory infection is of course not just the consequence of COVID, and into the future containment strategies against influenza in particular, will continue to be important.

Clear communication is essential for the development of public health advice messages that promote effectively behavioural respiratory infection control. Understanding what influences the uptake of non-pharmaceutical interventions may help to inform the development of future effective public health advice messages.^1,6 For instance, simply advising people to adopt these behaviours has been found to be ineffective.⁷ Consequently, in the context of both pandemics and seasonal epidemics, it becomes imperative to employ effective strategies for conveying vital health messages in a succinct and meaningful manner. These strategies should facilitate the adoption of protective behaviours by citizens while mitigating community ambivalence and panic.^8,9

A promising key principle, which is based on social identity¹⁰, social influence¹¹ and moral behaviour¹², promotes care for others rather than individual self-interest.¹³ These ‘protect each other’ messages highlight the benefits of protective behaviours for the wider social group and its most vulnerable members, including loved ones, with evidence of benefits in the COVID-19¹⁴ and other health contexts.¹⁵ Prosocial behaviour is defined as behaviour that benefits others, whether or not it involves an overall cost to self, and includes a variety of important social behaviours such as helping, sharing, and cooperation.^16,17 The term “others” refers to specific individuals or groups of people, and in particular loved ones, vulnerable members (with weakened immune systems such as elderly and chronically ill), health care professionals, co-workers, keyworkers, members of the public, and to some extent society as a whole.¹⁸

Studies have shown that people do consider the social impact of their behaviour during a pandemic, as many participants avoid putting others at risk for their personal benefit.¹⁹ The existing literature illustrates that prosocial public health messages that highlight behaviours related to societal and communal benefits (e.g., “protect each other”), rather than focusing on behaviours that only benefit self (e.g., “protect yourself”), can be a potential mechanism to communicate public health recommendations related to infectious diseases.^20-23

Although the studies mentioned earlier provide some evidence for the effectiveness of prosocial messages in promoting adherence to personal protective behaviours, we still lack robust evidence about the magnitude of this effect.²⁴ As individuals are exposed to conflicting information and misinformation, this can decrease individuals’ perceptions of the severity of the disease and the necessity of adopting preventive behaviours, that in turn may lead to rejection of public health messaging. Therefore, there is a need to examine the extent to which prosocial messages to adhere to personal protective behaviours are effective across different contexts, for better future pandemic and epidemic preparedness. We conducted a systematic review and meta-analysis to identify whether the prosocial messages have the potential to optimise the effect on population behaviour in relation to reducing transmission of respiratory infections. More specifically, we ask the following questions: (1) Are messages focusing on protecting others effective in changing a defined list of behavioural (intention) outcomes compared with other messages/controls?, (2) What actual behaviours or behaviour intentions (e.g., social distancing, hand washing, using hygiene products etc.) do messages about protecting others have positive effects on? (3) What populations do messages about protecting others have positive effects on?

Study selection and characteristics

The search strategy yielded 6,108 studies. After screening the titles and abstracts, 49 studies were selected for full-text screening. A total of 24 full-texts met the inclusion criteria (20 were published^16,24-42 while 4 were pre-prints).^14,43-45 Eight of them included more than one randomised controlled trial study (RCT), yielding 36 RCTs in total. No studies with other design (such as quasi-experimental studies) fulfilled the inclusion criteria. The majority of RCTs (32/36) targeted COVID-19, while 4 RCTs targeted multiple respiratory infections, including (but not limited to) influenza. The characteristics of included RCTs are described in more detail in Table S1 and S2. Sixteen (21 RCTs)^{14,24,28,29, Study 1,31-33,34, Study 1,35-38,40,42,44,45} of the 24 full-texts (36 RCTs) provided sufficient information to be included in the final quantitative analysis (Fig. 1).

Fig. 1: PRISMA Flow Diagram⁴⁶ |* First 100 references from each database and search engine have been kept sorted by relevance

Effectiveness of prosocial messages

Twenty-six RCTs from high-income countries (USA, Denmark, UK, Japan, Germany, Turkey) reported positive effects regarding the included messages focusing on “protect-others” principle^{16,24,26, Study 1 & 2,27-29,32,34, Study 1 & 2,35-39,41,42,44}, one RCT (from USA) reported negative effects²⁵ and nine RCTs (from USA, Italy, France and multiple countries: Spain, Chile, Colombia) reported no difference^{14,26, Study 3,30,31,33,34, Study 3,40,43,45}. None were in low or middle-income countries. Seventeen of the RCTs reporting positive effects regarding the included messages focusing on the "protect-others" principle were conducted in the USA, five in Europe, and four in Asia (see Supplementary file_1 pp.56-57). The main outcomes that positively affected by the messages focusing on protecting others were social distancing intentions (e.g., motivation to adhere to physical distancing, persuasiveness to self-isolate, avoid social gathering, stay home and keep a physical distance with others), mask wearing intentions, hand washing intentions, diverse-behavioural intentions (intentions that could not be categorized into specific groups, such as contact-avoidance intentions, protective behaviour willingness) and actual behaviours (see Supplementary file_1 pp.55-56).

Behavioural Tools and Mechanisms of Action (MoA) ontology evidence

Six of the nine MINDSPACE contextual influencers were identified across 102 intervention arms. On average, each intervention arm adopted 2.52 MINDSPACE contextual influencers. The four most common MINDSPACE contextual influencers were “Messenger” (n=33), “Salience” (n=96), “Affect” (n=42), “Ego” (n=43) (Table S3). The most often applied contextual influencers across the intervention groups focused on protecting others were “Salience” (n=53) and “Ego” (n=32). “Affect” was also present, however, less frequently (n=20).

Twenty-eight of the 93 BCTs were identified across 102 intervention arms. On average, each intervention arm adopted 4.03 behaviour change techniques. The seven most common BCTs identified were “Instruction on how to perform a behaviour” (n=96), “Information about health consequences” (n=84), “Salience of consequences” (n=69), “Information about social and environmental consequences” (n=28), “Credible source” (n=34), “Prompts/cues” (n=18) and “Avoidance/reducing exposure to cues for the behaviour” (n=26) (Table S4). Twenty-five of the 28 BCTs identified across the intervention arms focused on protecting others, with the following as the most frequently applied: “Instruction on how to perform a behaviour” (n=54), “Information about health consequences” (n=51), “Salience of consequences” (n=44), “Information about social and environmental consequences” (n=18), “Credible source” (n=14) and “Avoidance/reducing exposure to cues for the behaviour” (n=19).

MoA Ontology applied to 21 of the 24 full-text papers, covering 30 MoA subcategories. On average, each study adopted 3.48 MoAs, with “Behavioural intention” being the most common, followed by “Belief about one's social environment” and “Mental disposition” (Table S5; Table S6).

Cross-country Analysis

Due to the lack of data and the inability to perform a meta-analysis to assess the impact of each MINDSPACE contextual influencer and BCT across the USA and European countries, we calculated the Effective Ratio (ER) to strengthen our narrative analysis. In RCTs conducted in the USA, we evaluate the impact of five MINDSPACE contextual influencers (Fig. 2) and nine BCTs (Fig. 3). The most common MoAs were “Behavioural intention” and “Belief about message” followed by “Belief about consequences of behaviour” and “Willingness to comply”. In RCTs conducted in European countries, we evaluate the impact of four MINDSPACE contextual influencers (Fig. 2) and four BCTs (Fig. 3). The most common MoA was “Behavioural intention”, followed by “Motivation”, “Emotion process”, “Evaluative belief about behaviour” and “Belief about control over behaviour”. In RCTs conducted in Japan and Turkey, ER values could not be estimated for either MINDSPACE contextual influencers or BCTs, as there were no ineffective results.

Figure 2: ERs of MINDSPACE contextual influencers; ER=effective ratio|USA: The highest ER was observed for interventions focusing on “Norms” (ER=6), followed by “Affect” (ER=3.33), “Salience” (ER=2.27), “Ego” (ER=2) and “Messenger” (ER=0.57). For four MINDSPACE contextual influencers, ER values could not be estimated due to the absence of ineffective results. European countries: The highest ER was observed for interventions focusing on “Salience” (ER=7.5), followed by “Affect” (ER=6), “Ego” (ER=6) and “Norms” (ER=1). For five MINDSPACE contextual influencers, ER values could not be estimated due to the absence of ineffective results.

Figure 3: ERs of behaviour change techniques (within letters); ER=effective ratio.|USA: The highest ER in interventions pertained to “Avoidance/reducing exposure to cues for the behaviour” (ER=8), followed by “Salience of consequences” (ER=4.5), “Information about health consequences” (ER=4.25), “Instruction on how to perform a behaviour” (ER=3), “Information about others’ approval” (ER=2), “Information about social and environmental consequences” (ER=1.5), “Credible source” (ER=1.25), “Social support (practical)” (ER=1) and “Prompts/cues” (ER=0.29). For six BCTs, ER values could not be estimated due to the absence of ineffective results. European countries: The highest ER in interventions pertained to “Instruction on how to perform a behaviour” (ER=9), followed by “Salience of consequences” (ER=4.5), “Information about health consequences” (ER=4) and “Prompts/cues” (ER=1). For 19 BCTs, ER values could not be estimated due to the absence of ineffective results.

Populations’ characteristics affected by prosocial messages

Seventeen studies either lacked analysis on the impact of messages about protecting others on demographics or found no significant differences in demographic characteristics. However, six studies showed significant predictors of personal protective behavioural intentions and actual behaviours regarding respiratory infections. Such predictors were the age, gender, employment status, political orientation, race, region, education and health condition. The studies indicated that women, older people, those having less secure employment, more religious people, liberals or left leaning and those who are in worse health conditions intend to respect protective behaviours.

In particular, Browning, et al. ⁴⁴ implied that older age, non-male gender identity, and less secure employment predicted higher intentions for COVID-19 preventative procedures. Conversely, identifying as White or White/Indigenous predicted fewer intentions for these procedures. Capraro and Barcelo ²⁸ found that females, left-leaning individuals, and older individuals are more inclined to wear face coverings. However, the gender difference diminishes where wearing face coverings is mandatory, indicating a stronger impact on men. The study also proposes that messages emphasizing “your community” may be more effective for right-leaning individuals. According to Everett, et al. ¹⁴, older people and more religious people reported stronger behavioural intentions, while people who self-identified as White, male and conservative reported weaker behavioural intentions and felt less personally responsible for preventing the disease spread. Similarly, Hacquin, et al. ⁴⁵ found that women and older people intend to respect protective behaviours and wash their hands more than men and younger people. In addition, education had a negative effect on people’s behavioural intentions regarding protective behaviours and handwashing. Pink, et al. ²⁶ suggested that after seeing any message outlining key behaviours, women tend to increase their intentions to comply more than men. Additionally, the messages were more convincing to individuals leaning more liberal. Falco and Zaccagni ²⁷ found that individuals in poorer health conditions and at higher infection risks are the most affected by the messages focusing on protecting others (families). Conversely, those in better health, facing lower infection risks, and frequently leaving their homes are less impacted by such messages.

Risk of bias

Agreement between the two independent raters in coding the risk of bias criteria was high (91,7%). Overall, we noted a high level of some concerns, which was the result of insufficient reporting of an appropriate analysis used to estimate the effect of assignment to the intervention. An intention-to-treat (ITT) analysis that includes all randomized participants was lacking, resulting in some concerns in all studies. Further, some of studies did not publish a protocol or register the studies on trial registries, making it difficult to assess reporting bias, and this led to downgrading of the evidence quality for the large majority of the included intervention types. The studies were assessed as low risk of bias for the domains “bias arising from the randomization process”, “bias due to missing outcome data” and “bias in measurement of the outcome”. Only one study assessed as high risk of bias due to missing outcome data (Fig. 4; Fig. 5; Table S7).

Fig. 4: Risk of bias assessment. Traffic-light plot of the domain-level judgements.

Fig. 5: Risk of bias assessment. Summary plot of the domain-level judgements.

Network Meta-analysis

Twenty-one RCTs with behavioural intention outcomes (social distancing, mask wearing, handwashing and diverse-behavioural intentions) were considered for quantitative analyses. None of the studies with actual behaviour outcomes met the eligibility criteria for inclusion in the meta-analysis. Nineteen RCTs (n=30 intervention arms (prosocial messages focused on public and loved ones; self-focused messages); n=19 comparator groups) were included in the class-effects model including multiple behavioural intention outcomes except from handwashing intentions outcome, due to inconsistency between direct and indirect evidence. There was a small increase in personal protective behavioural intentions for each intervention group (prosocial messages and self-focused messages) compared to the control group (no message, baseline message or self-protection message), based on effect estimates of standardised mean differences (SMDs) with 95% CrI. The marginally largest effect was observed for prosocial messages focused on loved ones compared to the control message (d=0.09, 95% CrI 0.06 to 0.14, CINeMA: Low). Additionally, the 95% Crl for the comparison between prosocial messages focused on loved ones versus control did not cross 0, indicating that a positive association between prosocial messages focused on loved ones compared to the control group exists in the population of interest. There was relatively low heterogeneity (SD= 0.07, 95% CrI 0.04 to 0.11), therefore there was limited variability between studies in the analysis. The mean value of the total residual deviance (69.52) was similar to the number of data points (65) in the analysis, which indicates a reasonable fit (Table 1; Table S14).

[Table 1 here]

Subgroup Network Meta-analyses by behavioural intention outcomes

There was no inconsistency identified between direct and indirect evidence for social distancing intentions (design-by-treatment interaction model: χ²(6)=3.704, p=0.717), mask wearing intentions (design-by-treatment interaction model: χ²(4)=4.585, p=0.333), diverse-behavioural interventions (design-by-treatment interaction model: χ²(3)=0.140, p=0.987). However, there was evidence of inconsistency for handwashing intentions (design-by-treatment interaction model: χ²(3)=9.115, p=0.028). Local testing of evidence loops identified a statistically significant difference between direct and indirect evidence for the comparison between prosocial public messages and prosocial loved ones messages (d=0.618, 95% CrI 0.04 to 1.19; p=0.03).

Effects on social distancing intentions

Thirteen RCTs were included in the random-effects NMA. There was a small increase in social distancing intentions for each intervention group compared to the control group, with the marginally largest effect to be observed for prosocial messages focused on loved ones (d=0.10, 95% CrI 0.04 to 0.16, CINeMA: Moderate) (Table S8; Figure S1; Table S15).

Effects on mask wearing intentions

Four RCTs were included in the random-effects NMA. There was a small increase in mask wearing intentions for each intervention group compared to the control group, with the largest effect to be observed for prosocial messages focused on public (d=0.16, 95% CrI 0.04 to 0.30, CINeMA: Moderate) (Table S9; Figure S2; Table S16).

Effects on handwashing intentions

Seven RCTs were included in the random-effects NMA. There was a small increase in handwashing intentions for each intervention group compared to the control group, with the largest effect to be observed for prosocial messages focused on loved ones (d=0.20, 95% CrI 0.15 to 0.52, CINeMA: Low) (Table S10; Figure S3; Table S17).

Effects on diverse-behavioural intentions

Nine RCTs were included in the random-effects NMA. There was a small increase in diverse-behavioural intentions for each intervention group compared to the control group, with the largest effect to be observed for prosocial messages focused on loved ones (d=0.17, 95% CrI 0.04 to 0.31, CINeMA: Low) (Table S11; Figure S4; Table S18).

Component network meta-analyses (CNMA) with Mindspace contextual influencers

Handwashing intention outcomes were excluded due to the inconsistency identified. The CNMA model had a reasonable fit (total residual deviance=57.96 from 59 data points) and low heterogeneity (SD=0.05, 95% CrI 0.01 to 0.10). Although limited evidence, a small increase in personal protective behavioural intentions for interventions that incorporated the “salience” (d=0.06, 95% CrI -0.04 to 0.16), “affect” (d=0.06, 95% CrI -0.04 to 0.15) and “ego” (d=0.05, 95% CrI -0.05 to 0.14) compared to interventions that did not include these contextual influencers was found (Table 2). To reduce potential heterogeneity, separate CNMA for self-focused messages was conducted (Table S12).

[Table 2 here]

Component network meta-analyses with BCTs

Handwashing intention outcomes were excluded due to the inconsistency identified. The CNMA model had a reasonable fit (total residual deviance=56.53 from 59 data points)) and low heterogeneity (SD= 0.03, 95% CrI 0.01 to 0.07). A small increase in personal protective behavioural intentions for interventions that incorporated the “information about health consequences” (d=0.12; 95% Crl 0.03 to 0.28) and the “avoidance/reducing exposure to cues for the behaviour” (d=0.19, 95% CrI 0.06, 0.32) compared to interventions that did not include these BCTs was found (Table 3). Separate CNMA for self-focused messages was conducted (Table S13).

[Table 3 here]

Our systematic review supports the premise that public health messages focused on “protect-others” may help to improve respiratory infection control and may be more effective than self-interested messages. We identified 24 full-text articles, comprising 36 RCTs, conducted across 11 countries. Interventions were typically delivered online. An average of three-arm study designs with passive comparators were typically used. The overall quality of included studies was moderate, with only one study rated as ‘high risk of bias’. In particular, 26 of the 36 RCTs found a significant effect of protect-others messages on outcomes such as social distancing intention, mask wearing intention, hand washing intention, diverse-behavioural intentions and actual behaviours. The NMA results support our findings that prosocial messages, especially those referring to loved ones, had a small positive effect on personal protective behavioural intentions. Numerous correlational studies indicate that intentions are associated with behaviour. In particular, current evidence suggests that intentions get translated into action approximately one-half of the time.^47-50 For example, Sheeran ⁵¹ conducted a meta-analysis that encompassed a total of 422 studies from ten previous meta-analyses. The outcome of this analysis revealed a substantial sample-weighted average correlation between intentions measured at one time-point and measures of behaviour taken at a subsequent time-point (r+=0.53).

The systematic review also suggested some demographic characteristics to be predictors of protective behavioural intentions regarding respiratory infections. Women, older people, those having less secure employment, more religious people, liberals or politically left leaning and those who are in worse health conditions intend to enact protective behaviours. Moreover, the studies included in our analysis were conducted in high-income countries. Therefore, further investigation is warranted into the effectiveness of prosocial public health messaging interventions in middle- and low-income countries.

Our findings hold significant implications for public policymakers and communication experts, emphasizing the critical importance of determining how to communicate effectively complex, uncertain, and potentially conflicting health information. Enhancing our understanding of how public health messages can be designed stands to help improve respiratory infection control. The evidence suggests that framing public health messages about social responsibility -by raising awareness about others, especially loved ones (friends and family)- is crucial. Clear communication about risks entails heightening the public's awareness of the consequences for others' health. Tailoring prosocial public health messages requires careful consideration of individuals' perceived susceptibility and differences in message preferences. Framing these messages to focus on how to avoid or reduce exposure to specific cues has proven effective in reducing risk behaviour. This knowledge can be applied to future interventions targeting similar behaviours or addressing different aspects of behaviour.

Content characteristics of prosocial messages associated with effectiveness

In terms of those characteristics that make the prosocial messages effective, the most frequently applied contextual influencers focused on protecting others and the MINDSPACE contextual influencers of “Salience” and “Ego”, with “Affect” also present, although less frequently. The NMA results support our findings, as there was a small increase in personal protective behavioural intentions for interventions that incorporated the “Salience”, “Affect” and “Ego” contextual influencers, compared to interventions that did not include these contextual influencers. Regarding “Salience”, people were more likely to register stimuli that are novel (e.g., messages in flashing lights), accessible, simple (e.g., a snappy slogan), and relevant (e.g., easier to grab attention at moments when people enter a new situation such as a disease outbreak).⁵² Using Instagram stories to share public health messages has been found to engage diverse audiences and make them stick in the memory.⁵³ Sharing messages that increase the salience of risk can affect positively social distancing and mask wearing.²⁴

We tend to behave in a way that supports the impression of a positive and consistent self-image.⁵² In disease outbreaks, individuals can exhibit community-mindedness and sense of duty to protect others by practicing social distancing.⁵⁴ Messages that highlight protection of others (e.g., ‘‘others’’ are more vulnerable to get sick) can motivate behaviour change.^28,35 Moreover, emotional responses to words, images, and events can be very rapid and automatic. Thus, moods, rather than deliberate decisions, can influence judgments.⁵² Social media posts that make positive emotional appeals or provide factual information attract higher levels of user engagement when sharing public health messages.⁵⁵ Inducing empathy for those most vulnerable to the virus encouraged self-isolation and social distancing.³⁸

Results also identified the following six BCTs as potentially important predictors of our outcomes: “Instruction on how to perform a behaviour”, “Information about health consequences”, “Salience of consequences”, “Information about social and environmental consequences”, “Credible source” and “Avoidance/reducing exposure to cues for the behaviour”. The NMA results partially support our findings, as there was a small increase in personal protective behavioural intentions for interventions that incorporated the “Information about health consequences” and the “Avoidance/reducing exposure to cues for the behaviour”, compared to interventions that did not include these BCTs. Theories suggest that associative (such as antecedents), reflective motivational (such as natural consequences and comparison of outcomes) and self-regulatory processes (such as shaping knowledge) BCTs are relevant to successful behaviour change.⁵⁶ In particular, motivation BCTs lead towards a behavioural goal or an intention to enact, as they increase people’s conscious desire to do the target behaviour, by making beliefs or feelings about the behaviour more favourable (or less favourable for unwanted behaviours). Self-regulation system process also involves conscious efforts to modulate thoughts, emotions, and behaviours in order to achieve goals within a changing environment. Thus, self-regulation BCTs enable the motivated individual to self-regulate towards the goal or implement the intention or behaviour. Associative processes BCTs, which do not require conscious motivation or thinking about self-regulation at the time of action, can trigger behaviour compatible with the individual’s motivation and goals or elicit behaviour independent of motivation and goals.⁵⁷

Investigating the MoAs through which interventions have their effects on behaviour is key for optimising their outcomes. The most frequently applied MoA through which the BCTs affected population behaviour was “behaviour intention”. Identifying MoAs is crucial to understanding how the interventions change population behaviour. We can apply this knowledge to similar future interventions, or to those targeting different behaviours.⁵⁸

Strengths and limitations

Our electronic search included four databases of published papers, several databases of “grey” literature. Our hand search involved using search engines such as Google scholar, relevant journals and backward and forward citation searching. We re-ran the same search after a year to include new studies of interest. Data extraction was performed by a single reviewer and verified by a second reviewer. All abstracts, regardless of the language of the original studies, were initially screened for eligibility, as all were written in English. However, none of the studies in languages other than English met the inclusion criteria during the screening process. Only six studies were designed to study behaviours as their primary outcomes.^{16,25,27,39,41,42} The majority were designed to measure participants’ intentions to follow preventive procedures, but we were not able to assess whether these intentions were consistent with participant's behaviours (behaviour – intention gap^39,49,59,60). According to Falco and Zaccagni ²⁷, changes in intentions do not translate into sizeable changes in actions. These gaps can limit the effectiveness of messaging campaigns focused on protecting others. Future studies specifically designed to examine the effectiveness of interventions on behaviours are needed.

As the majority of studies were one-off interventions and only two studies were identified that conducted longer-term follow-ups, we could not assess whether prosocial public health messaging interventions were effective beyond immediate post-intervention effects. Although the systematic review showed that some demographic characteristics predicted protective behavioural intentions regarding respiratory infections, future research is needed to examine the role of protect-others principle in behavioural responses among more diverse samples. As might be expected from the diverse studies included, there were inconsistencies in outcome elements and heterogeneity. In particular, the studies used inconsistent methods of aggregation, inconsistent specific measurements and metrics, and/ or reported data at inconsistent time-points.

The majority of studies were methodologically and statistically heterogeneous. The fact that not all eligible studies were able to be included in the meta-analyses due to inadequate information about results (poorly reported), also limits conclusions. Due to the small number of studies available for meta-analysis, we were unable to group studies by, for example, population type, type of outcome assessment, etc., which would have further improved our understanding of intervention effectiveness. A narrative (descriptive) synthesis was conducted for each of the personal protective behavioural intentions and actual behaviours. Narrative synthesis goes beyond the act of simply describing the main features of included studies and enables summaries of knowledge related to a specific review question ⁶¹.

The GRADE and CINeMA approaches were used to downgrade the evidence when the data were at risk for within-study bias, publication bias, indirectness, imprecision, heterogeneity, and incoherence. Their usage enhanced the robustness of the methodology and strengthened the results of NMAs by providing a systematic and transparent assessment of the certainty of evidence.

This review used an empirically developed MINDSPACE checklist and a taxonomy of intervention techniques to code contextual influencers and BCTs, respectively, present in prosocial public health messaging interventions targeting personal protective behavioural intentions and actual behaviours. The MoAs Ontology was also used to identify the MoAs through which the BCTs can affect behaviour. Two coders for contextual influencers and BCTs and three coders for MoAs independently coded all studies in this review, which is a significant strength. There remains the possibility that a number of MINDSPACE contextual influencers, BCTs and MoAs were not captured due to poor intervention descriptions. The analyses of the intervention content were based on information presented in the published papers and/or available in supplementary materials, although attempts were made to contact study authors for access to detailed intervention descriptions. Furthermore, as most interventions targeting multiple behaviours, did not clearly distinguish contextual influencers and BCTs that targeted one (but not the other) behaviour, it was not possible to consider the extent of MINDSPACE and BCT overlap in the present review.

This systematic review was registered with PROSPERO (CRD42020198874).⁶² A detailed protocol of the review has been published.¹⁸ We used established PRISMA (preferred reporting items for systematic reviews and meta-analyses) guidelines statement for Network Meta-Analyses⁶³ (see Supplementary file_2) and also followed quantitative meta-analysis reporting standards by the APA⁶⁴ and procedures outlined by the Cochrane Handbook for Systematic Reviews.⁶⁵

Search strategy and selection criteria

A search strategy (Table S19) following PICOS was adapted (population, intervention, comparison, outcome and study design), including Medical Subject Headings (MeSH) terms and relevant key words and combining search terms using Boolean operators.^66,67 To identify controlled vocabulary terms for the databases, we first retrieved articles from each database that met the inclusion criteria for the review. We noted common text words and subject terms applied by the indexers and used them for a comprehensive search. To ensure we captured as many relevant records as possible, including older ones, we combined subject terms from the controlled vocabulary with a wide range of free-text terms⁶⁸. Original studies were eligible for inclusion if they were RCTs, non-randomised studies with concurrent controls, or controlled before-and-after studies; reported changes in any behaviour relevant to reducing transmission of respiratory infections; used communication via mass media, social media or print media (such as leaflets and posters) or health professional advice via consultation where messages about protecting others are included¹⁷; and were published in English (Table S20). Studies such as animal experiments, abstracts, case reports, reviews, and systematic reviews were excluded.

A comprehensive literature search of published and unpublished studies in electronic databases was conducted. The following electronic databases were searched from inception to January, 2021 and updated in March 2022: Medline, EMBASE.com, PsycINFO, and Scopus. For unpublished studies we conducted searches in databases of "grey" literature such as PsycEXTRA, Social Science Research Network (SSRN), OSF PREPRINTS database that includes BioHackrXiv, Cogprints, MediArXiv, SocArXiv, PsyArXiv and RePEc. We also conducted supplementary searches in Google Scholar, hand searched relevant journals, and backward and forward citation searching of included studies and relevant reviews.

All titles and abstracts retrieved by electronic searching were downloaded to the reference manager database EndNote and uploaded to Covidence, one of Cochrane’s recommended tools.⁶⁹ Duplicates were removed and double-screening was done on a proportion of the retrieved citations until the appropriate agreement (>95%) was achieved (450 studies; kappa=0.969; 95% Cl 0.909-1.000). The remaining studies were screened by one reviewer (AG). The abstracts were included if they met the inclusion criteria. The same procedure was applied to determine the eligibility of studies on the basis of a review of the full texts. Differences in judgment were resolved through discussion and inclusion of a third researcher doing the rating, when required. The selection process was recorded and the PRISMA flow diagram was completed.⁷⁰

Data extraction

Data were extracted from the included studies into a predefined spreadsheet, which covered the following areas: the study design; the communication message; characteristics of the recipient(s) of the communication (protect-others-message); characteristics of the “others” who will be protected due to the message; the manner in which the appeal to protect others is made; intervention features such as the primary outcome(s) and the results. As primary outcomes we considered any actual behaviour or behavioural intention relevant to reducing transmission of respiratory infections. Intentions are assumed to capture the motivational factors that influence behaviour. The stronger the intention to perform the behaviour, the more likely the behaviour will be performed.^49,71 Where more than one reported primary outcome is provided, we included all, reporting their measurements, metrics, methods of aggregations and time-points (when applicable).

Quality assessment

Two researchers (AG and VA) independently assessed the risk of bias of the included studies using the Cochrane Collaboration Risk of Bias Tool (RoB 2).⁷² The percentage of studies with high, low, or some concerns against each criterion was established by consensus between the two assessors. No studies were excluded because of risk of bias.

Data analysis and synthesis

We carried out a systematic review to qualitatively extract the key MINDSPACE conceptual influencers and behaviour change techniques used in prosocial messaging interventions to adhere to personal protective behaviours and quantitatively assess their effectiveness in reducing transmission of respiratory infections.

A narrative synthesis was conducted for each of the personal protective behavioural intentions and actual behaviours. Reviewers’ own descriptions of results were grouped into “positive” effects (e.g., a significant difference between intervention and control groups, favouring the intervention group, was detected), “no difference” (e.g., a significant difference between groups was not detected), “negative” effects (e.g., a significant difference between groups, favouring the control group, was detected). A narratively synthesised section with the attributes of the target population (characteristics of the individuals, groups, sub-populations or populations) that affected by messages about protecting others, was also included.

We content analysed these messages to identify what drives the change in behaviour. Two of the most popular and widely used behavioural tools were chosen due to their relevance and applicability to health policy, the MINDSPACE checklist^73-75 (stands for: Messenger, Incentives, Norms, Defaults, Salience, Priming, Affect, Commitments, and Ego; Table S21), and the Behaviour Change Techniques Taxonomy (BCTT; Table S22).^76-80 The MINDSPACE checklist, developed by Dolan, et al. ⁷³, operates on the premise that many behaviours are primarily driven by automatic, often unconscious psychological processes and can be influenced by the decision-making context.⁷⁵ The BCTT includes 93 techniques aimed at fostering behaviour change by targeting specific psychological processes, including beliefs, attitudes, goals, plans, emotions, habits, and more. There has not been academic investigation of the degree of conceptual (or theoretical) overlap between MINDSPACE and the BCTT (see Vlaev and Dolan ⁸¹ for initial analysis of how those nudges and techniques work through specific, and distinct, neuropsychological processes). Therefore, the possibility for unique insights to exist in each behavioural tool motivated us using both tools in our analysis. To address the challenges of inconsistent or ambiguous terminology in behaviour change intervention outcome measures, we utilized the “Mechanisms of Action” (MoA) Ontology.^82,83 This tool acts as a classification system that labels and defines MoAs and their interrelationships. MoAs describe the processes by which interventions influence target behaviours, such as beliefs, intentions, and behavioural opportunities (Table S23).

We also calculated the effective ratio (ER) to assess the effect of each MINDSPACE contextual influencer and BCT for the USA and European countries. This measure indicates their potential contribution to the effectiveness of the intervention.⁸⁴ The intervention outcomes were classified as effective (indicating statistical significance in intervention(s) and control comparisons) and ineffective (outcome was no statistically significant). The ER is calculated as the total number of effective results involving the MINDSPACE contextual influencer or BCT divided by the total number of ineffective results involving the contextual influencer or the technique. The ER could not be calculated if only one study was available.

A network meta-analysis was conducted as most appropriate for comparing three or more interventions simultaneously in a single analysis by combining both direct and indirect evidence across a network of studies, calculating the standardized mean difference (SMD) with 95% credible intervals (CrIs: the Bayesian alternative to 95% confidence intervals) using WinBUGS, a statistical software for Bayesian analysis. All analyses included 30,000 iterations as burn-in, once convergence was confirmed these iterations were discarded and the model was run for a further 70,000 iterations.

It produces estimates of the relative effects between any pair of interventions in the network, and usually yields more precise estimates than a single direct or indirect estimate.⁸⁵ The simultaneous comparison of all interventions of interest in the same analysis enables the estimation of their relative ranking for a given outcome.⁸⁶ In particular, a class-effects model⁸⁷ was conducted, including all the outcomes reported in RCTs (which met the eligibility criteria for inclusion in the meta-analysis: adequate result information, consistency in outcome elements (e.g., consistent methods of aggregation, consistent specific measurements and metrics)) whilst accounting for dependencies where more than one outcome per RCT are included. In addition, random effects NMAs were calculated using WinBUGS for each outcome separately. A key NMA assumption is the consistency between direct and indirect evidence. Global tests for inconsistency were conducted for each outcome using a design x treatment interaction test.⁸⁸ When a global test identified potential inconsistency, tests for loop inconsistency were conducted for each evidence loop. Network geometry was explored and illustrated using network diagrams. Component network meta-analyses (CNMAs) were conducted to assess the effectiveness of intervention components^89,90 categorised according to the BCTT and Mindspace components. Class effects were also used to account for multiple outcomes reported in RCTs.

For each outcome, including personal protective behavioural intentions and subgroup outcomes like social distancing intentions, hand washing intentions, mask wearing intentions, and diverse-behavioural intentions, we evaluated the confidence in the body of evidence derived from NMA using the Confidence in Network Meta-Analysis (CINeMA) application^91,92, which is broadly based on the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach. The GRADE methodology considers four levels of evidence: high quality, moderate quality, low quality, and very low quality. High quality indicates that the true effect closely aligns with the NMA estimates, while moderate quality suggests a likelihood that the actual effect is similar to the NMA estimates but could differ substantially. Low quality implies a possibility that the true effect differs significantly from the NMA estimates, and very low quality indicates a high likelihood of substantial differences. We determined the certainty of the evidence using the online CINeMA software (https://cinema.ispm.ch), which assesses criteria such as within-study bias, reporting bias, indirectness, imprecision, heterogeneity, and incoherence. If any concerns, whether minor or major, are identified, the certainty of the evidence is downgraded. Although CINeMA is a valuable tool for assessing the certainty of evidence in NMAs, its direct applicability to CNMAs may be limited due to differences in analytical approach and scope of assessment. Thus, the CINeMA has not been used for our CNMAs.

Table 1. Effect estimates on personal protective behavioural intentions

Comparison	Effect estimates: SMD	95% CrI	Confidence rating
Prosocial message focused on public	0.08	0.04, 0.12	Low
Prosocial message focused on loved ones	0.09	0.06, 0.14	Low
Self-focused message	0.08	0.03, 0.13	Low
Model fit and heterogeneity	Between-study SD 0.07	95% CrI 0.04, 0.11

Total residual deviance: Mean 69.52 from 65 datapoints; handwashing outcomes were not included due to inconsistency between direct and indirect evidence

Table 2. CNMA effect estimates of MINDSPACE contextual influencers on personal protective behavioural intentions (prosocial messages)

Comparison	Effect estimates: SMD	95% CrI
Messenger	-0.01	-0.08, 0.05
Norms	-0.04	-0.26, 0.12
Defaults	-0.04	-0.24, 0.16
Salience	0.06	-0.04, 0.16
Affect	0.06	-0.04, 0.15
Ego	0.05	-0.05, 0.14
Model fit and heterogeneity	Between-study SD 0.05	95% CrI 0.01, 0.10

Total residual deviance: Mean 57.96 from 59 datapoints

Table 3. CNMA effect estimates of BCTs on personal protective behavioural intentions (prosocial messages)

Comparison	Effect estimates: SMD	95% CrI
4.1. Instruction on how to perform the behaviour	-0.06	-0.21, 0.02
5.1. Information about health consequences	0.12	0.03, 0.28
5.2. Salience of consequences	0.03	-0.03, 0.07
5.3. Information about social and environmental consequences	0.02	-0.07, 0.10
6.1. Demonstration of the behaviour	0.03	-0.10, 0.26
6.2. Social comparison	-0.01	-0.08, 0.07
7.1. Prompts/cues	0.01	-0.08, 0.11
9.1. Credible source	-0.01	-0.06, 0.06
12.3. Avoidance/reducing exposure to cues for the behaviour	0.19	0.06, 0.32
Model fit and heterogeneity	Between-study SD 0.03	95% CrI 0.01, 0.07

Total residual deviance: Mean 56.53 from 59 datapoints

Author Contributions: AG is the corresponding author and the primary author of the study, conceived the study, contributed to the development of the search strategy, developed the inclusion and exclusion criteria and data extraction criteria, was involved in the conceptualization of the research questions and conducted the systematic review (behavioural coding and analyses of the results) and meta-analysis. VA contributed to the development of the search strategy, was involved in the behavioural coding of the papers and in the meta-analyses, and provided written feedback on the manuscript. NM contributed to the Network Meta-analysis and provided written feedback on the manuscript. CB conceived the study, contributed to the development of the selection criteria and data extraction criteria, was involved in the conceptualization of the research questions and provided written feedback on the manuscript. SM conceived the study and provided written feedback on the manuscript. MK contributed to the development of the selection criteria and data extraction criteria and provided written feedback on the manuscript. IV conceived the study, contributed to the development of the selection criteria and data extraction criteria, was involved in the conceptualization of the research question and revised the manuscript critically and contributed to it intellectually. All authors have read and approved the final version of the manuscript.

Acknowledgment: The authors thank Fiona Beyer, for her expertise in developing the search strategy, as well as Paulina Schenk, for her contribution to the identification of mechanisms of action. The authors gratefully acknowledge Louise Tanner and Ryan Kenny for their time and consultation regarding the meta-analysis.

Funding statement:

This study/project is funded by the National Institute for Health Research (NIHR) [Policy Research Unit in Behavioural Science (project reference PR-PRU-1217-20501)]. The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care.

Competing interests’ statement: None declared.

Data availability statement: The data presented in this study are available upon reasonable request from the corresponding author.

Holford, D. L., Juanchich, M. & Sirota, M. Ambiguity and unintended inferences about risk messages for COVID-19. J Exp Psychol Appl28, 486-508, doi:10.1037/xap0000416 (2022).
Morstead, T., Zheng, J., Sin, N. L., King, D. B. & DeLongis, A. Adherence to Recommended Preventive Behaviors During the COVID-19 Pandemic: The Role of Empathy and Perceived Health Threat. Ann Behav Med56, 381-392, doi:10.1093/abm/kaab107 (2022).
Kim, D. & Lee, Y. J. Vaccination strategies and transmission of COVID-19: Evidence across advanced countries. J Health Econ82, 102589, doi:10.1016/j.jhealeco.2022.102589 (2022).
Yang, J. et al. Despite vaccination, China needs non-pharmaceutical interventions to prevent widespread outbreaks of COVID-19 in 2021. Nature Human Behaviour5, 1009-1020, doi:10.1038/s41562-021-01155-z (2021).
Perski, O. et al. Interventions to increase personal protective behaviours to limit the spread of respiratory viruses: A rapid evidence review and meta-analysis. Br J Health Psychol27, 215-264, doi:10.1111/bjhp.12542 (2022).
Teasdale, E., Santer, M., Geraghty, A. W., Little, P. & Yardley, L. Public perceptions of non-pharmaceutical interventions for reducing transmission of respiratory infection: systematic review and synthesis of qualitative studies. BMC Public Health14, 589 (2014).
Bish, A. & Michie, S. Demographic and attitudinal determinants of protective behaviours during a pandemic: a review. Br J Health Psychol15, 797-824, doi:10.1348/135910710X485826 (2010).
Contreras, G. W. & MEP, M. Getting ready for the next pandemic COVID-19: Why we need to be more prepared and less scared. Journal of Emergency Management18, 87-89 (2020).
Smith, J. A. & Judd, J. COVID-19: Vulnerability and the power of privilege in a pandemic. Health Promotion Journal of Australia31, 158-160, doi:10.1002/hpja.333 (2020).
Moran, M. B. & Sussman, S. Translating the link between social identity and health behavior into effective health communication strategies: An experimental application using antismoking advertisements. Health Communication29, 1057-1066 (2014).
Haslam, S. A., Reicher, S. D. & Platow, M. J. The new psychology of leadership: Identity, influence and power. (Psychology Press, 2010).
Haidt, J. The righteous mind: Why good people are divided by politics and religion. (Vintage, 2012).
Bonell, C. et al. Harnessing behavioural science in public health campaigns to maintain ‘social distancing’in response to the COVID-19 pandemic: key principles. Epidemiol Community Health74, 617-619 (2020).
Everett, J. A., Colombatto, C., Chituc, V., Brady, W. J. & Crockett, M. The effectiveness of moral messages on public health behavioral intentions during the COVID-19 pandemic. (2020).
Shen, L. Targeting smokers with empathy appeal antismoking public service announcements: a field experiment. Journal of Health Communication20, 573-580 (2015).
Grant, A. M. & Hofmann, D. A. It’s not all about me: Motivating hand hygiene among health care professionals by focusing on patients. Psychological science22, 1494-1499 (2011).
Penner, L. A., Dovidio, J. F., Piliavin, J. A. & Schroeder, D. A. Prosocial behavior: Multilevel perspectives. Annu. Rev. Psychol.56, 365-392 (2005).
Grimani, A. et al. Effect of prosocial public health messages for population behaviour change in relation to respiratory infections: a systematic review protocol. BMJ Open11, e044763, doi:10.1136/bmjopen-2020-044763 (2021).
Campos-Mercade, P., Meier, A. N., Schneider, F. H. & Wengström, E. Prosociality predicts health behaviors during the COVID-19 pandemic. Journal of Public Economics195, 104367, doi:https://doi.org/10.1016/j.jpubeco.2021.104367 (2021).
Balbo, L., Jeannot, F. & Estarague, J. in 31ème Congrès de l’Association Française du Marketing-AFM 2015 (2015).
Curtis, V. et al. How to Set up Government-Led National Hygiene Communication Campaigns to Combat COVID-19: A Strategic Blueprint. BMJ Global Health5, e002780, doi:doi:10.1136/bmjgh-2020-002780 (2020).
Ghio, D. et al. What influences people’s responses to public health messages for managing risks and preventing disease during public health crises? A rapid review of the evidence and recommendations. BMJ open11, e048750, doi:https://doi.org/10.1136/bmjopen-2021-048750 (2020).
Li, M., Taylor, E. G., Atkins, K. E., Chapman, G. B. & Galvani, A. P. Stimulating influenza vaccination via prosocial motives. PloS one11, e0159780 (2016).
Bokemper, S. E., Huber, G. A., James, E. K., Gerber, A. S. & Omer, S. B. Testing persuasive messaging to encourage COVID-19 risk reduction. PLoS One17, e0264782, doi:10.1371/journal.pone.0264782 (2022).
Banker, S. & Park, J. Evaluating prosocial COVID-19 messaging frames: Evidence from a field study on Facebook. Judgment Decision Making15, 1037-1043 (2020).
Pink, S. et al. The effects of short messages encouraging prevention behaviors early in the COVID-19 pandemic. PLoS One18, e0284354, doi:10.1371/journal.pone.0284354 (2023).
Falco, P. & Zaccagni, S. Promoting social distancing in a pandemic: Beyond good intentions. PLoS One16, e0260457, doi:10.1371/journal.pone.0260457 (2021).
Capraro, V. & Barcelo, H. The effect of messaging and gender on intentions to wear a face covering to slow down COVID-19 transmission. Journal of Behavioral Economics for Policy4, 45-55 (2020).
Ceylan, M. & Hayran, C. Message Framing Effects on Individuals' Social Distancing and Helping Behavior During the COVID-19 Pandemic. Front Psychol12, 579164, doi:10.3389/fpsyg.2021.579164 (2021).
Favero, N. & Pedersen, M. J. How to encourage “Togetherness by Keeping Apart” amid COVID-19? The ineffectiveness of prosocial and empathy appeals. Journal of Behavioral Public Administration3, 1-18 (2020).
Frias-Navarro, D., Pascual-Soler, M., Berrios-Riquelme, J., Gomez-Frias, R. & Caamaño-Rocha, L. COVID–19. Effect of Moral Messages to Persuade the Population to Stay at Home in Spain, Chile, and Colombia. The Spanish Journal of Psychology24, e42, doi:10.1017/SJP.2021.39 (2021).
Gillman, A. S., Iles, I. A., Klein, W. M. & Ferrer, R. A. Increasing Receptivity to COVID-19 Public Health Messages with Self-Affirmation and Self vs. Other Framing. Health Communication38, 1942–1953. (2022).
Heffner, J., Vives, M. L. & FeldmanHall, O. Emotional responses to prosocial messages increase willingness to self-isolate during the COVID-19 pandemic. Pers Individ Dif170, 110420, doi:10.1016/j.paid.2020.110420 (2021).
Jordan, J., Yoeli, E. & Rand, D. Don’t get it or don’t spread it: Comparing self-interested versus prosocial motivations for COVID-19 prevention behaviors. Scientific reports11, 1-17 (2021).
Luttrell, A. & Petty, R. E. Evaluations of self-focused versus other-focused arguments for social distancing: An extension of moral matching effects. Social Psychological Personality Science12, 946-954 (2021).
Miller, S., Yardley, L., Little, P. & team, P. Development of an intervention to reduce transmission of respiratory infections and pandemic flu: measuring and predicting hand-washing intentions. Psychol Health Med17, 59-81, doi:10.1080/13548506.2011.564188 (2012).
Miyajima, T. & Murakami, F. Self-interested framed and prosocially framed messaging can equally promote COVID-19 prevention intention: A replication and extension of Jordan et al.’s study (2020) in the Japanese context. Frontiers in psychology12, 1341 (2021).
Pfattheicher, S., Nockur, L., Böhm, R., Sassenrath, C. & Petersen, M. B. The Emotional Path to Action: Empathy Promotes Physical Distancing and Wearing of Face Masks During the COVID-19 Pandemic. Psychological Science31, 1363-1373, doi:10.1177/0956797620964422 (2020).
Sasaki, S., Kurokawa, H. & Ohtake, F. Effective but fragile? Responses to repeated nudge-based messages for preventing the spread of COVID-19 infection. The Japanese Economic Review72, 371-408, doi:10.1007/s42973-021-00076-w (2021).
Utych, S. M. & Fowler, L. Age-based messaging strategies for communication about COVID-19. Journal of Behavioral Public Administration3, 1-14 (2020).
Varma, M. M., Chen, D., Lin, X., Aknin, L. B. & Hu, X. Prosocial behavior promotes positive emotion during the COVID-19 pandemic. Emotion23, 538–553. (2022).
Yardley, L., Miller, S., Schlotz, W. & Little, P. Evaluation of a Web-based intervention to promote hand hygiene: exploratory randomized controlled trial. Journal of medical Internet research13, e1963 (2011).
Barari, S. et al. (MedRxiv, 2020).
Browning, A., Moss, M. E. & Berkman, E. Leveraging Evidence-Based Messaging to Prevent the Spread of COVID-19. (2021).
Hacquin, A.-S., Mercier, H. & Chevallier, C. Improving preventive health behaviors in the COVID-19 crisis: a messaging intervention in a large nationally representative sample. (2020).
Page, M. J. et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. Bmj372, n71, doi:10.1136/bmj.n71 (2021).
Sheeran, P. & Webb, T. L. The intention–behavior gap. Social and Personality Psychology Compass10, 503-518, doi:10.1111/spc3.12265 (2016).
Manski, C. F. The use of intentions data to predict behavior: A best-case analysis. Journal of the American Statistical Association85, 934-940 (1990).
Conner, M. & Norman, P. Understanding the intention-behavior gap: The role of intention strength. J Frontiers in Psychology13, 4249 (2022).
Shmueli, L. Predicting intention to receive COVID-19 vaccine among the general population using the health belief model and the theory of planned behavior model. BMC Public Health21, 804, doi:10.1186/s12889-021-10816-7 (2021).
Sheeran, P. Intention—behavior relations: a conceptual and empirical review. European review of social psychology12, 1-36 (2002).
Smith, H. S. et al. A Review of the MINDSPACE Framework for Nudging Health Promotion During Early Stages of the COVID-19 Pandemic. Popul Health Manag25, 487-500, doi:10.1089/pop.2021.0269 (2022).
Dreibelbis, R., Kroeger, A., Hossain, K., Venkatesh, M. & Ram, P. K. Behavior change without behavior change communication: nudging handwashing among primary school students in Bangladesh. International journal of environmental research public health13, 129 (2016).
Baum, N. M., Jacobson, P. D. & Goold, S. D. “Listen to the people”: public deliberation about social distancing measures in a pandemic. The American Journal of Bioethics9, 4-14 (2009).
Kite, J., Foley, B. C., Grunseit, A. C. & Freeman, B. Please like me: Facebook and public health communication. PloS one11, e0162765 (2016).
Black, N. et al. Behaviour change techniques associated with smoking cessation in intervention and comparator groups of randomized controlled trials: A systematic review and meta‐regression. Addiction115, 2008-2020 (2020).
Dixon, D. & Johnston, M. MAP: A mnemonic for mapping BCTs to three routes to behaviour change. British Journal of Health Psychology25, 1086-1101 (2020).
Hagger, M. S., Moyers, S., McAnally, K. & McKinley, L. E. Known knowns and known unknowns on behavior change interventions and mechanisms of action. Health Psychology Review14, 199-212, doi:10.1080/17437199.2020.1719184 (2020).
Faries, M. D. Why We Don't "Just Do It": Understanding the Intention-Behavior Gap in Lifestyle Medicine. American journal of lifestyle medicine10, 322-329, doi:10.1177/1559827616638017 (2016).
Sniehotta, F. F., Scholz, U. & Schwarzer, R. Bridging the intention–behaviour gap: Planning, self-efficacy, and action control in the adoption and maintenance of physical exercise. J Psychology health20, 143-160 (2005).
Lisy, K. & Porritt, K. Narrative synthesis: considerations and challenges. JBI Evidence Implementation14, 201 (2016).
Grimani, A. et al.The effect of communication strategies for population behaviour change in relation to infectious disease: a systematic review, <https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42020198874> (2020).
Hutton, B. et al. The PRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses of health care interventions: checklist and explanations. Annals of internal medicine162, 777-784, doi:10.7326/m14-2385 (2015).
Appelbaum, M. et al. Journal article reporting standards for quantitative research in psychology: The APA Publications and Communications Board task force report. American Psychologist73, 3-25, doi:10.1037/amp0000191 (2018).
Cumpston, M. S., McKenzie, J. E., Welch, V. A. & Brennan, S. E. Strengthening systematic reviews in public health: guidance in the Cochrane Handbook for Systematic Reviews of Interventions, 2nd edition. J Public Health (Oxf), doi:10.1093/pubmed/fdac036 (2022).
Huang, M., Névéol, A. & Lu, Z. Recommending MeSH terms for annotating biomedical articles. Journal of the American Medical Informatics Association18, 660-667 (2011).
Methley, A. M., Campbell, S., Chew-Graham, C., McNally, R. & Cheraghi-Sohi, S. PICO, PICOS and SPIDER: a comparison study of specificity and sensitivity in three search tools for qualitative systematic reviews. BMC health services research14, 579 (2014).
Higgins, J. P. et al. (John Wiley & Sons, Chichester (UK), 2019).
Covidence, <https://community.cochrane.org/help/tools-and-software/covidence> (
Moher, D. et al. Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Systematic reviews4, 1 (2015).
Bagozzi, R. P. The self-regulation of attitudes, intentions, and behavior. Social Psychology Quarterly55, 178-204, doi:10.2307/2786945 (1992).
Sterne, J. A. et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ366, l4898 (2019).
Dolan, P. et al. Influencing behaviour: The mindspace way. Journal of Economic Psychology33, 264-277 (2012).
Thaler, R. H. & Sunstein, C. R. Nudge: Improving decisions about health, wealth, and happiness. (Penguin, 2009).
Vlaev, I., King, D., Dolan, P. & Darzi, A. The theory and practice of “nudging”: changing health behaviors. Public Administration Review76, 550-561 (2016).
Michie, S., Atkins, L. & West, R. The behaviour change wheel: a guide to designing interventions. (Silverback Publishing, 2014).
Michie, S., Van Stralen, M. M. & West, R. The behaviour change wheel: a new method for characterising and designing behaviour change interventions. Implementation science6, 42 (2011).
Michie, S. et al. Behaviour change techniques: the development and evaluation of a taxonomic method for reporting and describing behaviour change interventions (a suite of five studies involving consensus methods, randomised controlled trials and analysis of qualitative data). Health Technology Assessment19, 1–188 (2015).
Armitage, C. J. et al. Investigating which behaviour change techniques work for whom in which contexts delivered by what means: Proposal for an international collaboratory of Centres for Understanding Behaviour Change (CUBiC). British Journal of Health Psychology26, 1-14 (2021).
Michie, S. et al. The behavior change technique taxonomy (v1) of 93 hierarchically clustered techniques: building an international consensus for the reporting of behavior change interventions. Annals of behavioral medicine46, 81-95 (2013).
Vlaev, I. & Dolan, P. Action change theory: A reinforcement learning perspective on behavior change. Review of General Psychology19, 69-95, doi:10.1037/gpr0000029 (2015).
Michie, S. et al. The Human Behaviour-Change Project: harnessing the power of artificial intelligence and machine learning for evidence synthesis and interpretation. Implement Sci12, 121, doi:10.1186/s13012-017-0641-5 (2017).
Michie, S. et al. Representation of behaviour change interventions and their evaluation: Development of the Upper Level of the Behaviour Change Intervention Ontology. Wellcome open research5, 123 (2020).
Martin, J., Chater, A. & Lorencatto, F. Effective behaviour change techniques in the prevention and management of childhood obesity. International Journal of Obesity37, 1287-1294, doi:10.1038/ijo.2013.107 (2013).
Rouse, B., Chaimani, A. & Li, T. Network meta-analysis: an introduction for clinicians. Internal and emergency medicine12, 103-111, doi:10.1007/s11739-016-1583-7 (2017).
Chaimani, A., Caldwell, D. M., Li, T., Higgins, J. P. & Salanti, G. in Cochrane handbook for systematic reviews of interventions (eds Higgins JPT et al.) Ch. 11, 285-320 (Cochrane, 2019).
Dias, S., Ades, A. E., Welton, N. J., Jansen, J. P. & Sutton, A. J. Network meta-analysis for decision-making. (John Wiley & Sons, 2018).
Jackson, D., Barrett, J. K., Rice, S., White, I. R. & Higgins, J. P. A design-by-treatment interaction model for network meta-analysis with random inconsistency effects. Statistics in medicine33, 3639-3654, doi:10.1002/sim.6188 (2014).
Welton, N. J., Caldwell, D., Adamopoulos, E. & Vedhara, K. Mixed treatment comparison meta-analysis of complex interventions: psychological interventions in coronary heart disease. American Journal of Epidemiology169, 1158-1165 (2009).
Freeman, S. C. et al. Component network meta-analysis identifies the most effective components of psychological preparation for adults undergoing surgery under general anesthesia. Journal of Clinical Epidemiology98, 105-116 (2018).
Nikolakopoulou, A. et al. CINeMA: An approach for assessing confidence in the results of a network meta-analysis. PLoS medicine17, e1003082, doi:10.1371/journal.pmed.1003082 (2020).
Papakonstantinou, T., Nikolakopoulou, A., Higgins, J. P. T., Egger, M. & Salanti, G. CINeMA: Software for semiautomated assessment of the confidence in the results of network meta-analysis. Campbell systematic reviews16, e1080, doi:10.1002/cl2.1080 (2020).

There is NO Competing Interest.

Download PDF

Version 1

posted

You are reading this latest preprint version

Motivational effectiveness of prosocial public health messaging to reduce respiratory infection risk: A network meta-analysis

Status:

Version 1

Abstract

Figures

Main

Results

Discussion

Methods

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1