This review adopted the concept map method based on the recommendation of Novak and Gowin  for both analysing the data and presenting the results. All three researchers (AGh, MP, MDC) were involved in creating the concept maps. The process of concept mapping are; (1) Identify a research question addressing a problem, issue or gap in knowledge, (2) understand the concepts, (3) create the concept map by placing the key concepts at the top of the hierarchy and select and arrange the defined concepts under the key concepts, (4) categorise the concepts in different domains, (5) a specific examples of events or objectives that clarify the meaning of the concept.
The first constructed concept map (Fig. 2) adopts observation and intervention as the two main applications of social media in health promotion research . It illustrates that Facebook was used more for intervention [29, 34, 36, 39] than observation [31, 33] purposes and Twitter was used more for observation [12, 35, 37, 38] than intervention [29, 30, 39] purposes. Sina Weibo  and Instagram were used for observation  and YouTube was used for intervention purposes .
Application of Facebook
Existing research recognised Facebook as a social media platform used to change health behaviour [9, 39] with the potential to deliver strategies in successful health promotion interventions  to promote healthy lifestyles . One study evaluated Facebook posts beyond reach, shares, and likes to identify important components that influence health behaviour . Analysing the content of Facebook accounts of health insurance companies that promote health through education showed that many providers could not actively interact with a wider audience on Facebook . One study investigated the impact of social media to reduce smoking habits and promote healthy behaviour among Indigenous people in Australia by sharing tobacco control content on Facebook . One study explored the use of Facebook by Australian public health organisations to understand what features and strategies are associated with wider and bigger user engagement .
Application of twitter
Twitter is applied as a supplementary tool to enhance social engagement by spreading information and strengthening social networks in various online health promotion programs as a potential source of public and searchable data relating to health behaviour . One study tested the feasibility of designing, implementing, and evaluating social media interventions using Twitter data for prevention purposes . Some studies [35, 37] investigated Twitter data to conclude how the majority of tweets can promote specific preventive behaviours. One study  explored the association between the content of tweets related to national mental health campaign and the real number of suicides to compare the suicide counts during the event to a control timeframe and the associated behaviour change to diminish suicide rates to promote help-seeking and resilience.
Application of health promotion theory frameworks
It is important to observe the shifts in the stages of behaviour change of TTM to evaluate the sustained improvement of health behaviour outcomes that can be prescribed to individuals who are willing to quit an unhealthy habit and modify their behaviours . Social media interactions provide rich data sources to understand the processes of stages of behaviour change and can help to discover the patterns of social behaviour changes . Public health awareness requires the incorporation of some theories of behavioural change into social media health interventions .
The Transtheoretical Model (TTM) has been used in one study to conceptualise the process of behaviour change . One study  identified the highest engagement in pre-action of stages of TTM behaviour change (Precontemplation, Contemplation, and Preparation) in a content intervention on Facebook.
Figure 3 illustrates that the majority of studies [5, 10, 12, 32, 34, 40] adopted social cognitive theory (SCT) as the basis to analyse the social media content and evaluate the relationship between the outcome and self-efficacy postings that influenced online audience engagement. Social cognitive theory (SCT) was used with the assumption that people adopt new behaviour better when they observe the others’ behaviour and demonstrate the desired health behaviour for social media users to learn through observation instead of other health messages like written texts or still pictures .
The adopted methodological approaches in the studies
Existing research discovered that evaluation frameworks for social media interventions can help to measure behaviour change and found exposure, reach, and low- to medium-level of users’ engagement as the most important factors to be measured in the social media campaigns in health promotion .
The results of concept map in Fig. 4 shows that the majority of the articles (65%) adopted quantitative approaches [5, 10, 28,29,30,31,32,33, 36] to evaluate the impact of social media in health behaviour change and 5 studies (35%) used a mixed method [3, 26, 34, 35, 37] for the study design to evaluate the outcome.
The study design used in the original papers
While social media are becoming preferred methods of health promotion for showing their effectiveness in reaching public audiences , the area of need is detected in research to develop models that go beyond calculating reach, shares, and likes and evaluate the outcome of social media campaigns for health promotion [4, 31]. Most research addresses specific interventions and approaches, which vary widely in functionality, and usability. Health promotion evaluations typically seek to understand how effective these approaches are as tools for health promotion and public health communication, education, and behaviour outcomes related to the intervention .
The concept map (Fig. 5) structured the methods used in the selected research papers based on the functionality of social media to evaluate the outcome of health promotion campaigns for behaviour change. The review identified communication and education as the most influential factors to evaluate the outcome of health promotion programs for behaviour change .
Communication factor to evaluate the sustainability of health promotion outcome
Of the nine studies that adopted communication factor to evaluate the outcome of the health promotion program [5, 12, 29,30,31,32, 35, 37, 38], seven conducted content analysis (text analysis) to measure the engagement rate [29,30,31, 35, 37], exposure/control rate  and self-efficacy  as the parameters to evaluate the outcome. Three studies implemented online surveys [5, 12, 30] and used factor analysis [5, 12] and focus group interviews  to measure self-efficacy , respondents’ attitude  and participants’ attitude and evaluate the outcome.
Content analysis to evaluate communication outcome
A national mental health campaign involved a content analysis of suicide-related posts on Twitter associated with the campaign to determine the association between change in the number of suicides during and immediately following the campaign . Counts during exposure and control windows were compared. The content of tweets related to eating disorder campaigns were analysed according to five categories of promotion health behaviour stages in eating disorders and measured the parameters of potential impact, potential reach, number of contributors, percentage of very active users, and percentage of very influential users to evaluate the outcome . The reactions of Twitter users were measured in an oral health campaign by performing content analysis of the tweets and manually classifying the tweets based on Neiger’s model of social media engagement [35, 45]. One study conducted a content analysis of the China Tobacco Control Media Campaign on Sina Weibo and dummy coded the segmented posts to three persuasive content characteristics for behaviour change strategy . They used regression to test the effects of the independent variables on online audience engagement. The Twitter account of a regional cancer charity who had hosted a prevention skin cancer campaign were evaluated . The volume of tweets related to a list of predefined keyword search terms was compared pre and post campaign to track the messages based on the location of the participants. They measured the most common social media metrics such as impressions, engagements, engagement/impression rate, likes and shares to show when a message spreads by detailing the number of users who see it, who respond to it, or who subsequently share that message with their friends or followers.
Online surveys to evaluate communication outcome
One researcher evaluated the outcome of a cervical cancer prevention campaign by conducting an online survey engaging the social media users who followed the campaign . One study investigated the health behaviour of social media users on 3 social media channels (Facebook, Twitter, and Instagram) and the online discussions concerning diabetes . They surveyed the users of diabetes social media groups through an online questionnaire to assess the impact of health promotion intervention on their lifestyles and online health behaviours . A quasi-experimental study assessed the feasibility of social media intervention by conducting a web-based survey that was promoted on social media platforms . One study developed an integrated conceptual model and assumed social media interventions would increase public protection and prevention through the interaction between public awareness and behaviour changes as mediating factors. They conducted an online survey questionnaire to test the defined hypothesis .
Education factor to evaluate the sustainability of health promotion outcome
From four studies adopting the education factor [10, 33, 34, 36], three studies used content analysis (text and image) to measure the engagement rate [33, 36] and attitude of the interviewed participants . One study conducted online surveys to measure self-efficacy  to evaluate the outcome of health promotion campaigns for behavioural change.
Content analysis to evaluate education outcome
One researcher performed content analysis of text and image data on the Facebook account of public and private health insurance providers related to health topics . One study conducted a Facebook campaign to actively target young adults applying the stage of behaviour change of TTM by measuring the engagement of users in sharing comments on Facebook instead of likes . They used t-tests and negative binomial regression analysis using TTM to examine the relationship between engagement and characteristics of participants. The TTM based post content was dummy coded and the deviation was used to estimate the difference in engagement between each individual post and the overall engagement . One study conducted an educational intervention campaign with 5 self-regulatory techniques. The participants were encouraged to engage in an existing social network to meet their weight loss achievements. The level of engagement with the Facebook post was quantified and a subset of participants were interviewed to evaluate the passive online engagements .
Online survey to evaluate education outcome
An online survey was conducted to explore the moderating role of colour in YouTube animated cartoons on health behaviour of social media users . A hierarchical multiple regression model was used to compute the capacity of YouTube animated cartoons regarding Covid-19 in predicting the health behaviour of social media users. The results show that colour significantly moderate the impact of COVID-19 YouTube animated cartoons on health behaviours such as avoiding handshake, regular hand washing, using hand sanitizers and face masks. They used self-efficacy to evaluate the outcome of sustainable health behaviour among social media users who were exposed to COVID-19 YouTube animated cartoons .