Using Facebook to promote the uptake of colorectal cancer screening

Background The use of social media presents a unique opportunity for cancer screening programs to motivate individuals to get screened. However, we need a better understanding of what types of social media messages for colorectal cancer (CRC) screening are preferred. The objective of this study was to develop social media messages promoting CRC screening uptake to identify messages preferred by the target audience. Methods We conducted a qualitative descriptive study and collected data through focus groups with Facebook users of screen-eligible age. Participants were presented with social media messages and asked to provide feedback. Messages were informed by the Health Belief Model, current evidence regarding screening communication and health communication and social media best practices. Focus groups were audio-recorded and transcribed and analysis was completed by two independent coders. If messages generated sufficient discussion, we developed a recommendation regarding the use of the message in a future social media campaign. Recommendations included: strongly consider using this message, consider using this message, proceed with caution, and do not use this message. General considerations about social media campaigns were also noted. Results A total of 45 individuals participated in six focus groups. We developed recommendations for 7 out of the 18 messages tested; 1 was classified as strongly consider using this message, 4 as consider using this message and 2 as proceed with caution. The data suggest that participants preferred social media messages that were believed to be credible, educational, and with a positive or reassuring tone. Preferred messages tended to increase awareness about CRC risk and screening and prompted participants to ask questions, and to want to learn more about what they could do to lower their risk. Messages that were viewed as humorous, strange or offensive or that had a negative or excessively fearful tone were less well received by study participants. Conclusions Facebook users prefer social media messages for CRC that have a positive or reassuring tone, are educational, and that have a credible ad sponsor. Campaign planners should proceed with caution when considering messages that use humor or a fearful tone to avoid undermining their campaign objectives. Supplementary Information The online version contains supplementary material available at 10.1186/s12889-022-12732-w.

Ruco et al. BMC Public Health (2022) 22:323 been shown to improve cancer outcomes through early detection and prevention of the disease [2,3]. Organized screening programs are able to define a target population appropriate for screening, send invitations, reminders, and recalls, and track outcomes [4]. However, in many jurisdictions, including those with and without an organized screening program, participation in CRC screening remains suboptimal [5]. For example, an overview of screening programs across the world reported that participation rates for cancer screening with stool-based testing range from 16 to 77% in the first round [5]. This suggests, that a large proportion of eligible individuals are not getting screened. Different and innovative approaches are needed to maximize screening participation in order to reduce the public health burden of the disease.
While traditional mass media campaigns can be used to reach a large proportion of the target population through popular media outlets (e.g. television, radio and print media), these are generally costly and the prominence of such outlets is also declining [6]. Social media is increasingly being used for health promotion and behaviour change interventions, [7][8][9] and affords users the ability to create, discuss and share content in online communities or networks. Individuals use these platforms to access information and connect with others through frequent interactions such as posts or stories that may include tweets, photos, or videos depending on the platform. Social media is now being utilized as a major communication tool across all sectors including healthcare [10]. The bidirectional flow of information, the potential for anonymity in interactions, and the ability to reach and engage individuals from across the globe with relatively low cost makes social media attractive to many users [10].
Evidence has also started to emerge regarding the specific use of social media for cancer screening, prevention and management [7,[10][11][12][13][14][15][16][17]. Literature on the use of theory-based social media interventions for promoting cancer screening is limited, with a limited number of studies focused specifically on CRC [17]. A better understanding of what types of social media messages for CRC screening participation specifically are preferred is needed. Facebook is the most popular social media platform among the population that is eligible for CRC screening (50-74 years). It is estimated that approximately 76% of those aged 55+ use Facebook in comparison with other platforms like Instagram (28%), LinkedIn (40%), Twitter (27%), Pinterest (30%), or Snapchat (6%), [18,19] making Facebook the most appropriate platform for our study.
The objective of this study was to develop social media messages promoting CRC screening uptake on Facebook with users of screen-eligible age (50-74) and to identify messages preferred by the target audience.

Methods
We developed social media messages promoting CRC screening and evaluated them using a qualitative descriptive study. The study was performed in accordance with relevant guidelines and regulations and informed consent was obtained from all participants. The study was approved by the Research Ethics Board at St. Michael's Hospital, Unity Health Toronto (REB# 19-084). We used the Standards for Reporting Qualitative Research (SRQR) checklist to report on our findings [20].

Social media message specifications
Facebook ads can include an image, primary text, a headline, a description and a call-to-action ( Fig. 1) [21]. The primary text appears at the top of the post, and should be 125 characters in length or less as per Facebook recommendations [21]. The headline usually appears right below the image and is recommended it not be longer than 40 characters. The description (30 characters) appears below the headline and can include a website link where users are redirected to if they click on the ad. The call-to-action button invites users to engage with the post. For example, a common call-to-action for Facebook posts includes 'Learn More' where users are invited to find out more information about the post.

Development of social media messages
We developed 18 messages (Additional File 1) informed by the Health Belief Model (Fig. 2), [22,23] current evidence regarding screening communication, [24,25] messages used in other jurisdictions and campaigns, expert consultations (staff at Ontario Health (Cancer Care Ontario) who have experience developing screening communications) and health communication best practices [23,24]. The Health Belief Model was selected because a systematic review on individual-level factors for behaviour change pertaining to CRC screening found the most supportive evidence for this behaviour change theory [22]. Messages were written in plain language using available health plain language thesauri [26,27]. An iterative process with several rounds of feedback was carried out to refine messages. Additionally, a marketing firm executive and copy editor with experience in social media campaigns in the healthcare industry reviewed the messages and revised them according to best practices. We also ensured we had variability among the messages regarding tone, framing, vividness, emotion and source of the message. Each message was classified according to the construct in the Health Belief Model.

Participants and recruitment
Participants included screen-eligible age individuals (50-74 years) who were current Facebook users (had an active user account), resided in Ontario and could communicate in English. An external research firm was contracted to conduct recruitment of participants using random digit dialing (inclusive of landlines and cell phone numbers) between February and April 2020. For focus groups conducted virtually, participants were also required to have access to appropriate technology to be able to join an online discussion. We aimed to recruit participants for six focus group as previous work empirically assessing saturation within focus groups found that five focus groups was enough for saturation [28].

Data collection
We collected data through focus groups held in person in the Greater Toronto Area or virtually (due to the COVID-19 pandemic) with residents in Southwestern, Eastern, and Northern Ontario to ensure geographic variation in our sample. Participant socio-demographic information was collected using a questionnaire. A facilitator led all focus groups using an interview guide with prompts to generate discussion and obtain feedback on the messages. Messages were shown in the same format as they would appear in a Facebook post with an image placeholder ( Fig. 1). Due to the time limit, it was not feasible to test all messages in every focus group. Each focus group lasted 90 min and participants received a $110 honorarium. The research team included both experienced and early career researchers in the field of cancer screening and prevention, behavioral science, and consumer and producer perspectives on social media. The research team had no prior relationship with focus group participants.

Data analysis
Focus group discussions were audio-recorded, transcribed, and reviewed for quality assurance prior to analysis. NVivo 12 software (QSR International) was used for data management. The focus group transcripts were analyzed by two independent coders (AR, DL). The coders met multiple times to review the data; they resolved any discrepancies through discussion. Participant comments were identified as relevant if they directly addressed any of the messages being tested. Relevant participant comments were classified as positive or negative (magnitude coding) depending on the sentiment expressed, and were summarized using short phrases that mirrored the language used by the participant (in vivo coding) [29]. The coders then reviewed the data for all messages and selected messages for further analysis where sufficient discussion in focus groups and a range of opinions was received in order to support a confident recommendation regarding the use of the message.
Selected messages were assessed to determine the extent to which they might support the uptake of timely and appropriate stool testing for CRC screening. For this assessment, we considered the participant comments against a list of outcomes that are known to precede the adoption of a health behaviour [30]. For example, ability to recall the message, beliefs about its relevance, and reported knowledge gains from the message can contribute to individual behaviour change [30]. We also considered the severity of the negative comments to distinguish messages that might be only ineffective from those that might be harmful or those with a mixed reaction. The balance between positive and negative comments was considered to offer a recommendation regarding the use of each of the selected messages in a future social media campaign. Initial recommendations were made by the two coders independently and the results were discussed with the broader study team to finalize the recommendations. Recommendations included: strongly consider using this message, consider using this message, proceed with caution, and do not use this message. Comments that provided general insight into how to design a social media campaign for CRC screening were also identified and summarized. Participant socio-demographic characteristics were reported in descriptive summary form.

Results
A total of 45 individuals participated in six focus groups (Table 1). Briefly, just over half of the participants were female (56%, n = 25) and most participants were between the ages of 55-64 (58%, n = 26). Twenty-one participants (47%) had previously had CRC screening with colonoscopy being the most common type of screening test completed. Over 70% of participants reported using Facebook at least 2-3 times/week (n = 32) and 98% (n = 44) reported using Facebook at least once a week. We developed recommendations for 7 out of the 18 messages tested in focus groups (Table 2). Briefly, only one message (#4) addressing perceived barriers to getting screened received a recommendation to strongly consider using the message. Participant comments suggested that the message may be effective because it is credible, educational, comforting, and positive. Participants suggested the message addressed barriers to screening by clarifying that it is possible to get screened without undergoing a rectal test or a colonoscopy. No evidence that the message may be ineffective or harmful was found.
Four messages (#1, #2, #3, #11) received a recommendation to consider using the message. These messages addressed perceived susceptibility, perceived benefits and social norms. These messages were seen as credible, positive or reassuring, simple and easy to understand; they also increased awareness about CRC risk and screening and prompted participants to ask questions or want to learn more about what they could do to lower their risk. Although these messages also attracted negative comments, these generally suggested the messages may be ineffective in some cases but not harmful.
Three messages (#1, #2 and #7) were designed to increase the perception of CRC risk in the audience. These messages elicited a range of reactions, including that the messages were scary enough to prompt people to take steps to protect their health, not scary enough, or too scary. One of the messages (#7) received a recommendation to proceed with caution because of the negative tone of the message and the fear triggered by it.
One message (#12) was designed to use humour to engage the audience. While some participants said the message caught their eye and appreciated the playful tone, others felt the message was offensive and unprofessional, leading to a recommendation of proceed with caution.
Focus group participants also discussed general considerations that may be useful when designing social media campaigns for cancer screening. Some participants distinguished between social and traditional mass media and said social media messages need to 'stand out. ' Specifically, participants talked about the sheer amount of content and information that is competing for their attention on these platforms and that messages can easily get lost in this sea of information. Some participants look for 'lighter content' when using Facebook or other social media platforms and therefore they may actively avoid messages that sound like bad news. While some participants felt that Facebook was not a trusted source of health information, an important feature of the tested messages that conveyed credibility included the use of a Canadian domain (.ca). This suggested to the participants that the information was from a trusted source and not 'just another ad trying to sell you something. '

Discussion
Our study identified several social media messages that may be effective in encouraging CRC screening among Facebook users of screen-eligible age. Of the 7 messages with an overall recommendation, one was classified as strongly consider using this message, four as consider using this message and two as proceed with caution. Participants preferred social media messages that were believed to be credible, educational, and with a positive or reassuring tone. These messages tended to increase awareness about CRC risk and screening and prompted participants to ask questions and to want to learn more about what they could do to lower their risk. Messages that were viewed as humorous or offensive, or that had a negative or fearful tone were not well received by all participants and may garner mixed reactions. Using nationally specific websites to foster trust/credibility is suggested.
Our results are aligned with previous research exploring what types of messages may be most effective for CRC screening. For example, Kiviniemi et al. [22] found the greatest evidence for the behavioural change constructs of benefits, barriers, and perceived susceptibility, which underpin 5 of the 7 messages that our study is recommending for use in a future social media campaign. There are also similarities between our results and that of Weaver et al. [31] who developed text messages for CRC screening through focus groups with adults aged 50-75 years of age. Participants in their study identified the following characteristics for appealing text messages including having a positive or reassuring tone, and avoiding content that contains bad news or test results [31]. Although text messages may be inherently different from social media messages, some similarities likely remain given the bidirectional flow of information and the fact that social media accounts are also frequently accessed through mobile phones.
Our findings suggest that a fine balance is needed between instilling a sense of urgency through the use of fear-based messaging and instilling too much fear where the audience is put off by a message. Fear appeals including messages that address perceived susceptibility have been shown to be associated with screening outcomes such as intention or participation [22,32]. This was also supported by our data where participants commented that certain messages were not 'scary enough' to prompt action (messages #1, #3). Participants in our study also described some messages as instilling a sense of urgency to act (message #2), while others felt that some messages may be       'very scary' and thus prevent people from taking action (message #7). The use of fear-based messaging requires further consideration by campaign planners given the mixed reactions including whether these could be addressed by audience segmentation. Our data suggest that some people may be turning to Facebook for 'lighter content'; as such, planners may wish to avoid any fear-based messaging. Interestingly, Carcioppolo et al. [33] explored the addition of humour to fear-based messages about colonoscopy screening and tested these online using Amazon's Mechanical Turk platform. The authors found that messages with a mixed appeal (fear and humour combined) may be more effective at increasing screening intentions than fear-based messaging alone among those with high cancer worry [33]. This may suggest that the addition of humour may soften fear-based messaging that may otherwise scare the target audience. The one message that included humour in our study also had a mixed appeal with the perceived barriers construct and reactions to this message led to a recommendation of proceed with caution. Additionally, Carcioppolo et al. [33] identified opportunities for audience segmentation including among those with high cancer worry which may be more receptive to mixed appeal messages. Further exploration regarding which segment of social media users would engage with mixed appeal messaging is needed.
The results of our study must be considered in light of the study's strengths and limitations. Due to the limited time available in each focus group, we were not able to test all messages in all focus groups. This resulted in some messages not garnering sufficient comments to make a recommendation. However, we were still able to make an informed recommendation on 7 of the tested messages. Our study was rigorously conducted as analysis was completed independently by two coders and supported by extensive discussion within the broader study team. Given that participants in our study were only from Ontario, our findings may not be generalizable to other populations. However, our study utilized random digit dialing (inclusive of both landlines and cellphones) for recruitment and to ensure variation among participant demographics. While we did not limit inclusion criteria to those at average risk of disease, the messages in our study were targeted to those at average risk of disease and some consideration to this and the potential for audience segmentation must be given in future work (e.g. age, gender, prior screening history). It is also important to note that our study focused on one social media platform. As such, our findings may not be transferable to other platforms. Despite this, our study fills an important gap in the literature as it identifies potential messages that program planners may use to maximize screening efforts and characteristics of effective social media messages for CRC screening specifically.

Conclusions
Our study provides insights into how behaviour change theory and focus group input can be used to develop social media messages for cancer screening. Our study has identified social media messages that may be preferred by individuals of screen-eligible age. These messages, coupled with appropriate images or design can be used in a randomized controlled trial on Facebook to evaluate effectiveness of social media messaging to increase CRC screening uptake. Further research could also extend this work to explore whether the messages would need to be refined to address the needs of those at increased risk of disease and whether audience segmentation could be used for messages with mixed reactions.  FDN-148470, GSO-157853). The funding agency had no role in the design or conduct of the study, collection, analysis and interpretation of data and in writing of the manuscript.

Availability of data and materials
The dataset used and analysed during this study are available from the corresponding author on reasonable request.

Declarations Ethics approval and consent to participate
The study was performed in accordance with relevant guidelines and regulations and informed consent was obtained from all participants. The study was approved by the Research Ethics Board at St. Michael's Hospital, Unity Health Toronto (REB#19-084).