Introduction

An estimated 2,001,140 new cancer cases are expected in 2024 []. Shared decision-making (SDM) describes a process between the clinician and patient to facilitate preference-sensitive choices []. Decision aids, which can support the SDM process, are evidence-based tools designed to provide patients with information, clarify their preferences, and prepare them to make a choice [,]. In this study, we explore the potential of social media as an avenue for engagement with decision-making tools.

SDM has been shown to be important for cancer decision-making, with multiple randomized controlled trials demonstrating that decision aids improve patient knowledge and the quality of decisions [-]. Unfortunately, decision aid use has been limited, and their dissemination has been largely confined to clinical settings. A 2010 study revealed that only 24% of clinicians working with patients with cancer used decision aids [-]. The focus on clinical settings as the singular forum for decision aid deployment is predicated on clinician buy-in and may restrict the use of decision aids to a select cohort of the population [-].

Social media offers a promising means for disseminating decision aids without relying on health care access. It may also provide a more extended and personalized modality for disseminating information []. With 81% of Americans using social media, a number that continues to grow, social media platforms present an underused opportunity to disseminate highly accessible decision-making tools []. Social media can help to overcome challenges associated with traditional clinical encounters (ie, time, workflow, etc) and can enhance the patient-clinician relationship by promoting empowerment, reducing communication barriers, and increasing knowledge about conditions and treatment options [,].

Numerous studies have highlighted the positive impact of web-based decision aids for women, particularly in the context of breast cancer, the leading cause of cancer among females []. Despite the potential benefits of social media for decision aid dissemination, it remains uncertain whether females will use cancer-related decision aids available through social media or other online channels. To address this gap, we examined factors influencing engagement with decision aids on social media and explored health information-seeking behaviors across various platforms. This study aimed to evaluate the feasibility, acceptability, trust, and engagement with social media as a tool to deliver online decision aids to women for cancer treatment. By focusing specifically on women, we aimed to address the unique health and decision-making needs of this population and provide insight for future research on breast cancer-related decision aids.

MethodsSurvey Design

A cross-sectional survey was designed to assess the use and preferences for social media–advertised decision-making tools for cancer care. The survey involved several key areas of inquiry (). Briefly, these areas included “Health-Related Information Behaviors” (5 questions) to assess participants’ behaviors in seeking and using health information; “Sources of Health Information” (3 questions), exploring individuals’ preferences and trust levels in various health information sources; “Social Media Use” (18 questions), examining patterns and motivations behind social media interactions, particularly concerning health information; and “Demographic Data” (10 questions), covering a wide range of personal and socioeconomic factors. Within the “Social Media Use” section, two items related to the main outcomes of the study were embedded, created by the study team, which asked participants to imagine themselves or a loved one deciding about cancer treatment and then assessing their likelihood of viewing cancer treatment information or clicking on a decision aid posted on social media. Other survey questions were adapted from items from the Health Information National Trends Survey []. Items assessing reasons to use social media included categories identified in the literature through the uses and gratifications theory and the social media engagement model [-].

The response formats varied according to the specific inquiry, including multiple-choice options, checkboxes for applicable answers, and Likert scales for assessing attitudes and opinions. Although all 39 questions could be answered, branching logic was used to tailor the survey based on participants’ responses (eg, only those reporting the use of specific social media platforms were asked follow-up questions about their motivations for use). The survey was designed to be completed within 5 to 10 minutes and included 2 attention-check questions. One single question, with the associated branching logic, was shown on the screen at a time. Participants were unable to skip questions (except for the questions asking the frequency of use and reason for use of social media platforms) and were notified, “Please answer this question” if attempted to skip. Participants were able to go back and change their answers if desired. The order of survey items, and answers, was fixed and not randomized, as the survey design prioritized logical flow and ease of navigation for participants. This study is reported in a manner that is consistent with the specified Checklist for Reporting Results of Internet E-Surveys (CHERRIES) guidelines () [].

Population Targeting and Survey Distribution

The survey was designed and hosted on the Qualtrics platform (Qualtrics, Provo) and distributed in February 2023 via Prime Panels, a component of Cloud Research []. Prime Panels uses a novel data collection method by aggregation of diverse opt-in market research panels into a comprehensive sampling platform, facilitating the recruitment of participants from existing commercial panel pools. This method supports demographic quotas and specific eligibility criteria, enhancing data representativeness, especially among hard-to-reach populations. Eligible participants were invited to participate through targeted email and dashboard invitations sent by the market research panels within the Prime Panels network, based on the demographic criteria specified for the study []. Participants were required to complete the survey in a single sitting, and no reminders were sent to those who did not finish it during that session. Due to the wide distribution on several platforms, response rates and the total number of invitations sent were not calculated by Prime Panels.

We aimed to gather a representative sample of United States females aged 35-75 years for this study, as this age range represents the peak period for breast cancer diagnosis. The study exclusively enrolled female participants to direct focus toward future research efforts related to breast cancer in women. Using 2022 US Census data, the population of females aged 35-75 years were inputted into the Qualtrics sample size calculator with a 99% CI and a 5% margin of error to determine the required sample size []. Based on these calculations, the survey targeted approximately 660 female participants, with an additional 15% included to account for potential exclusions due to poor response quality, bringing the total recruitment goal to 750 participants. Study specific eligibility criteria incorporated into the Prime Panels query included female participants aged 35 years or older with US IP addresses. Demographic quotas based on United States Census Bureau parameters were set as: 16% Hispanic, Latino, or Spanish origin; 78% White; 13% Black; 5% Asian; 2% American Indian or Alaskan Native; and 2% other races. Age quotas aimed for an equal distribution between the 35-55 years and 56-75 years age ranges to reflect the demographics of breast cancer survivors. To ensure survey security, Qualtrics options for “Bot Detection,” “RelevantID,” and “Prevent Indexing” were enabled.

Statistical Analyses

Data analysis was performed using Microsoft Excel and SPSS Statistics (version 28.0; IBM Corp). Graphs were constructed via R Statistical Software (version 4.1.2; R Core Team 2021). Descriptive statistics characterized the demographic characteristics, health-seeking behaviors, and social media engagement of the study population. Social media engagement was determined based on respondents’ selections of platforms they actively used, followed by survey questions that assessed the frequency of engagement with each platform. These questions categorized usage frequency into 4 levels: multiple times a day, once a day, at least 3 times a week, or less than 3 times a week. Numerical values ranging from 1 to 4 were assigned to these categories, with 1 indicating the least frequent usage and 4 the most frequent. An “Overall Social Media Engagement Score” was computed by aggregating these values across all platforms used by a respondent. Participants were classified into 4 groups based on their “Overall Social Media Engagement Score” to approximate quartiles for analysis. These groups were defined as follows: “Low Engagement” (scores of >0 and ≤3), “Moderate Engagement” (scores of >3 and ≤4), “Moderate-High Engagement” (scores of >4 and ≤8), and “High Engagement” (scores of >8 and ≤23).

To ensure data integrity, surveys were excluded based on the following criteria: completion times shorter than 3 minutes or longer than 20 minutes, flags from Qualtrics indicating duplicate responses, an Amazon Mechanical Turk fraud score above 50, or incorrect responses to two embedded control multiple-choice questions.

Analysis of Trustworthiness of Social Media

To evaluate associations between demographic factors and the perceived trustworthiness of social media as a reliable source of health information, chi-square, and Fisher’s exact tests were performed. To ensure that there were enough observations in each category for the statistical analysis to be reliable, the response categories “Trustworthy” and “Very Trustworthy” were merged into a single “Trustworthy” category, while “Untrustworthy” and “Very Untrustworthy” were combined into “Untrustworthy.”

Analysis of Social Media Engagement

In terms of the two questions assessing the likelihood of engaging with cancer treatment-related information seen on social media, responses were condensed from a 7-tier scale to 3 categories: “Unlikely” (1-3), “Neutral” (4), and “Likely” (5-7) to ensure a more balanced distribution of responses, as some of the original categories had very few observations. Spearman rank correlation coefficients were calculated to quantify the strength and direction of the association between the “Overall Social Media Engagement Score” and the tiered scores representing the likelihood of interacting with cancer-related information. For visual interpretation, the mean likelihood of respondents interacting with cancer-related information was calculated for each unique “Overall Social Media Engagement Score.” For all analyses, statistical significance was set at a P value of less than .05, using 2-tailed testing.

Analysis of Likelihood to View Cancer-Related Health Information or Click on Decision Aid

Nonparametric tests, specifically the Kruskal-Wallis and Mann-Whitney U tests, were used to evaluate the relationships between demographic characteristics, trust in social media, and the propensity to use decision aids or view cancer-related health information on these platforms. Variables that were found to be significantly related to the use of decision aids or viewing health information at P≤.10 were then checked for multicollinearity via variance inflation factor (VIF) values <5 before inclusion in an ordinal regression model.

Ethical Considerations

This study was reviewed and received approval from the institutional review board at Ohio State University as exempt (protocol 2022E0836). Informed consent was obtained from all participants involved in the study. Participant data were collected anonymously, with no identifying information retained in the dataset. The original informed consent included provisions for the use of deidentified data for research purposes, as reviewed and approved by the institutional review board. Data were stored in a secure, password-protected database accessible only to study investigators. All participants were compensated by Prime Panels in the amount agreed to by the platform through which they entered the survey, which is unknown to study personnel.

Results

A total of 757 responses were initially recorded at the completion of distribution. Of these, 607 met inclusion criteria with a Qualtrics “Response Quality” of 99.0%. Participants completed the survey in a mean SD time of 5.5 (SD 2.6) minutes.

Respondent Demographics

All participants were female, aged 35-75 years (). Of the 607 respondents, most were non-Hispanic (556/607, 91.6%) and White (480/607, 79.1%). The most common education level was some college or an associate degree (201/607, 33.1%). The most common income range was US $20,000 to US $35,000 (119/607, 18.9%), with over half (327/607, 53.9%) earning less than US $50,000 annually.

Table 1. Demographic characteristics of respondents.CharacteristicsRespondents (N=607), n (%)Ethnicity
Hispanic48 (7.9)
Non-Hispanic556 (91.6)
Unknown or prefer not to answer3 (0.5)Race
White480 (79.1)
Black72 (11.9)
Asian27 (4.4)
Native American or Alaskan Native16 (2.6)
Native Hawaiian or Pacific Islander0 (0.0)
Other or prefer not to answer11 (1.8)Age (years)
35-3982 (13.5)
40-49141 (23.2)
50-59149 (24.5)
60-69165 (27.2)
70-7970 (11.5)Highest level of education achieved
High school diploma, GEDa, or less175 (28.8)
Technical training or certificate46 (7.6)
Some college or associate degree201 (33.1)
College degree111 (18.3)
Graduate or professional degree74 (12.2)Annual household income (US $)
<20,000109 (18.0)
20,000-35,000115 (18.9)
35,000-50,000103 (17.0)
50,000-75,000109 (18.0)
75,000-100,00073 (12.0)
>100,00082 (13.5)
Unknown or prefer not to answer16 (2.6)Insurance type
Private229 (37.7)
Government301 (49.6)
Uninsured43 (7.1)
Other8 (1.3)
More than 1 insurance policy26 (4.3)Relationship status
Single129 (21.3)
Married268 (44.2)
Separated or divorced150 (24.7)
Widowed57 (9.4)
Other or unknown or prefer not to answer3 (0.5)Employment status
Full-time207 (34.1)
Part-time71 (11.7)
Not working for pay or unemployed121 (19.9)
Retired200 (32.9)
Other or unknown or prefer not to say8 (1.3)Country of birth
United States555 (91.4)
Outside of the United States52 (8.6)Years of US residency
Less than 15 years20 (3.3)
More than 15 years587 (96.7)

aGED: graduate educational diploma.

Health-Seeking Behavior

In total, 551 out of 607 participants (90.8%) had sought health or medical information from various sources at some point (). Out of 551 respondents, the internet was the most common first source of health information (n=441/551, 80.0%), while 75 or 13.6% consulted a doctor or health care provider.

Table 2. Characteristics of health-seeking behaviors.QuestionRespondents (N=607), n (%)Have you ever looked for information about health or medical topics from any source?551 (90.8)The most recent time you looked for information about health or medical topics, where did you go first?551 (100)Doctor or health care provider75 (13.6)Internet441 (80)Brochure or pamphlet, etc.10 (1.8)Friend or coworker3 (0.5)Family11 (2)Cancer organization2 (0.4)Newspapers1 (0.2)Books5 (0.9)Library2 (0.4)Telephone information number1 (0.2)The most recent time you looked for information about health or medical topics, who was it for?551 (100)Self408 (74)Someone else72 (13.1)Both oneself and someone else71 (12.9)Which of the following sources have you used in the last month as a source of news or information about health topics?a592 (100)Blogs or personal websites72 (12.2)Center for disease control and prevention133 (22.5)World Health Organization63 (10.6)Government46 (7.8)Community or faith leaders19 (3.2)Online news256 (43.2)Email48 (8.1)Family and friends204 (34.5)Health professionals282 (47.6)Radio22 (3.7)Podcasts27 (4.6)TV69 (11.7)Social media119 (20.1)Print media46 (7.8)Video sharing sites53 (9)Have you ever looked for information about cancer from any source?607 (100)Yes397 (65.4)No210 (34.6)In the past 12 months, have you used the internet to look for cancer information for yourself?397 (100)Yes185 (46.6)No212 (53.4)Where do you access your social media accounts?a603 (100)Computer or laptop258 (42.8)iPad or tablet159 (26.4)Smartphone497 (82.4)

aParticipants were able to check all that apply.

Social Media Use and Engagement

In total, 80 out of 607 or 95.6% of respondents used social media. Of these, Facebook was the most popular platform, used by 511 or 84.2%, and was used primarily for social interactions by 338 out of 487 respondents (69.4%). YouTube (Alphabet Inc) and Instagram (Meta) were primarily used for entertainment (189 out of 251 or 75.3% and 110 out of 193 or 57.0%, respectively). 18.5%, or 112 out of 607 respondents, demonstrated a “Low Engagement” pattern regarding social media use.

Trustworthiness of Social Media

The majority of the 607 respondents found social media trustworthy (73/607, 12.0%) or neutral (257/607, 42.3%) for health information. Black or Asian race, younger age, and longer duration of US residency were associated with greater trust in social media. Among Black respondents, 14 out of 72 (19.4%) considered social media trustworthy, compared to 49 out of 480 (10.2%) of White respondents (P=.003). Asian respondents showed even higher trust levels, with 7 out of 27 (25.9%) rating social media as trustworthy. Younger individuals also reported greater trust, with 17 out of 82 (20.7%) of those aged 35-39 years trusting social media compared with 12 out of 70 (17.1%) among those aged 70-79 years (P<.001). In addition, respondents with longer US residency (more than 15 years) showed greater trust, with 272 out of 587 (46.3%) indicating trustworthiness in social media, compared with only 5 out of 20 (25.0%) of those with less than 15 years of residency (P=.004). In total, 277 respondents (45.6%) noted social media to be untrustworthy ().

Table 3. Factors associated with perceived trustworthiness of social media as a source for health information.FactorsTrustworthyNeutralUntrustworthyRespondents, n (%)P valueTotal73 (12)257 (42.3)277 (45.63)607 (100)N/AaEthnicity.14
Non-Hispanic64 (11.5)233 (41.9)259 (46.58)556 (100)

Hispanic9 (18.8)23 (47.9)16 (33.33)48 (100)

Unknown or prefer not to answer0 (0.08)1 (33.3)2 (66.67)3 (100)
Race.003b
White49 (10.2)194 (40.4)237 (49.4)480 (100)

Black14 (19.4)32 (44.4)26 (36.1)72 (100)

Asian7 (25.9)15 (55.6)5 (18.5)27 (100)

Native American or Alaskan Native2 (12.5)7 (43.8)7 (43.8)16 (100)

Native Hawaiian or other Pacific Islander0 (0)0 (0)0 (0)0 (100)

Other or prefer not to answer1 (8.3)9 (75)2 (16.7)12 (100)
Age (years)<.001b
35-3917 (20.7)33 (40.2)32 (39)82 (100)

40-4920 (14.2)68 (48.2)53 (37.6)141 (100)

50-5913 (8.7)71 (47.7)65 (43.6)149 (100)

60-6911 (6.7)67 (40.6)87 (52.7)165 (100)

70-7912 (17.1)18 (25.7)40 (57.1)70 (100)
Highest level of formal education achieved.40
High school diploma, GEDc, or less24 (13.7)81 (46.3)70 (40)175 (100)

Technical training or certificate5 (10.9)19 (41.3)22 (47.8)46 (100)

Some years of college or associate degree23 (11.4)92 (45.8)86 (42.8)201 (100)

College degree12 (10.8)40 (36)59 (53.2)111 (100)

Graduate or professional degree9 (12.2)25 (33.7)40 (54.1)74 (100)
Annual household income (US $).63
<20,00011 (10.1)53 (48.6)45 (41.3)109 (100)

20,000-35,00017 (14.8)48 (41.7)50 (43.5)115 (100)

35,000-50,00015 (14.6)47 (45.6)41 (39.8)103 (100)

50,000-75,00012 (11)43 (39.5)54 (49.5)109 (100)

75,000-100,00010 (13.7)24 (32.9)39 (53.4)73 (100)

>100,0008 (9.8)35 (42.7)39 (47.6)82 (100)

Unknown or prefer not to answer0 (0)7 (43.8)9 (56.3)16 (100)
Insurance type.16
Private28 (12.2)98 (42.8)103 (45)229 (100)

Government42 (14)123 (40.9)136 (45.2)301 (100)

Uninsured1 (2.3)24 (55.8)18 (41.9)43 (100)

Other insurance or more than 1 policy2 (5.9)12 (35.3)20 (58.8)34 (100)
Relationship status.79
Single15 (11.6)58 (45)56 (43.4)129 (100)

Married33 (12.3)102 (38.1)133 (49.6)268 (100)

Separated or divorced18 (12)70 (46.6)62 (41.3)150 (100)

Widowed7 (12.3)25 (43.9)25 (43.9)57 (100)

Other or unknown or prefer not to answer0 (0)2 (66.7)1 (33.3)3 (100)
Employment status.09
Full-time22 (10.6)96 (46.4)89 (43)207 (100)

Part-time10 (14.1)31 (43.7)30 (42.3)71 (100)

Not working for pay or unemployed17 (14.1)58 (47.9)46 (38)121 (100)

Retired24 (12)68 (34.0)108 (54)200 (100)

Other or unknown or prefer not to say0 (0)1 (50)1 (50)2 (100)
Country of birth.28
United States64 (11.5)233 (41.98)258 (46.49)555 (100)

Outside of the United States9 (17.3)24 (46.15)19 (36.54)52 (100)
Years of US residency.004b
Less than 15 years7 (35)8 (40)5 (25)20 (100)

More than 15 years66 (11.2)249 (42.4)272 (46.3)587 (100)
Platforms used (multiple selections allowed)N/A
Facebook65 (12.7)220 (43.1)226 (44.2)511 (100)

Twitter (rebranded as X)23 (16.9)65 (47.8)48 (35.3)136 (100)

Instagram35 (12.6)125 (45.1)117 (42.2)277 (100)

YouTube53 (13)179 (45.4)162 (41.1)394 (100)

WhatsApp (Meta)17 (21.3)34 (42.5)29 (36.3)80 (100)

TikTok (ByteDance)22 (12.9)85 (49.7)64 (37.4)171 (100)

Other or unknown4 (13.3)10 (33.3)16 (53.3)30 (100)

aN/A: not applicable.

bP<.05.

cGED: graduate educational diploma.

Among social media platforms, the highest proportion of trustworthy users was noted among the 80 WhatsApp users (n=17, 21.3%), followed by 23 out of the 136 (16.9%) Twitter users. Amongst the 511 respondents who used Facebook, the most frequently used platform, 65 or 12.7% reported trust in social media for health information.

Use of Cancer Information or Decision Aids Through Social Media

Participants who considered social media “Trustworthy” (n=73) were more likely to view cancer information (n=61, 83.6%) or click on a decision aid through social media (n=61, 83.6%) than the 277 respondents who viewed social media as “Untrustworthy” (view: n=133, 48.0%; click: n=125, 45.1%) ( and ). Younger participants, particularly those aged 35-39 years were more likely to view cancer-related information through social media. Only 10 out of 57 (12.2%) in the 35-39 years age group rated their likelihood as “unlikely,” compared with 54 out of 89 (32.7%) aged 60-69 years. Among respondents aged 35-39 years, 54 out of 82 (65.9%) were likely to click on the decision aid, while in the 60-69 years age group, 87 out of 165 (52.7%) indicated they were likely to click.

Table 4. Nonparametric analysis of factors influencing viewing of cancer-related health information on social media: social media trustworthiness and demographic insights.VariableLikelihood of viewing cancer-related health information seen on social mediaRespondents, n (%)P value
UnlikelyNeutralLikely

Total141 (23.23)82 (13.51)384 (63.26)607 (100)N/AaTrustworthiness of social media
<.001b
Untrustworthy99 (35.7)45 (16.2)133 (48)277 (100)

Neutral36 (14)31 (12.1)190 (73.9)257 (100)

Trustworthy6 (8.2)6 (8.2)61 (83.6)73 (100)
Ethnicity
.86
Hispanic130 (23.4)75 (13.5)351 (63.1)556 (100)

Non-Hispanic11 (22.9)6 (12.5)31 (64.6)48 (100)

Unknown or prefer not to answer0 (0)1 (33.3)2 (66.7)3 (100)
Race
.17
White110 (22.9)68 (14.2)302 (62.9)480 (100)

Black23 (31.9)9 (12.5)40 (55.6)72 (100)

Asian4 (14.8)2 (7.4)21 (77.8)27 (100)

Native American or Alaskan Native2 (12.5)3 (18.8)11 (68.8)16 (100)

Native Hawaiian or Pacific Islander0 (0)0 (0)0 (0)0 (0)

Other or prefer not to answer2 (16.7)0 (0)10 (83.3)12 (100)
Age (years)
.008b
35-3910 (12.2)15 (18.3)57 (69.5)82 (100)

40-4927 (19.1)20 (14.2)94 (66.7)141 (100)

50-5931 (20.8)15 (10.1)103 (69.1)149 (100)

60-6954 (32.7)22 (13.3)89 (53.9)165 (100)

70-7919 (27.1)10 (14.3)41 (58.6)70 (100)
Highest level of education attained
.65
High school diploma, GEDc, or less36 (20.6)25 (14.3)114 (65.1)175 (100)

Technical training or certificate11 (23.9)8 (17.4)27 (58.7)46 (100)

Some college or associates degree43 (21.4)27 (13.4)131 (65.2)201 (100)

College degree29 (26.1)13 (11.7)69 (62.2)111 (100.0)

Graduate or professional degree22 (29.7)9 (12.2)43 (58.1)74 (100)
Annual household income (US $).83
<20,00025 (22.9)20 (18.3)64 (58.7)109 (100)

20,000-35,00023 (20)16 (13.9)76 (66.1)115 (100)

35,000-50,00029 (28.2)8 (7.8)66 (64.1)103 (100.0)

50,000-75,00024 (22)18 (16.5)67 (61.5)109 (100)

75,000-100,00020 (27.4)7 (9.6)46 (63)73 (100)

>100,00016 (19.5)10 (12.2)56 (68.3)82 (100)

Unknown or prefer not to answer4 (25)3 (18.8)9 (56.2)16 (100)
Insurance type.39
Private50 (21.8)28 (12.2)151 (65.9)229 (100)

Government74 (24.6)40 (13.3)187 (62.1)301 (100)

Uninsured6 (14)9 (20.9)28 (65.1)43 (100)

Other insurance or more than 1 policy11 (32.4)5 (14.7)18 (52.9)34 (100)
Relationship status.29
Single34 (26.4)19 (14.7)76 (58.9)129 (100)

Married56 (20.9)32 (11.9)180 (67.2)268 (100)

Separated or divorced36 (24)20 (13.3)94 (62.7)150 (100)