Table 4 Quotes from Eligible Studies Supporting Themes

Barnes et al. 2020 [45]

N/A

N/A

“…sampling biases are inherent as Facebook users are not fully representative of the entire Australian population, and potential participants may have been missed if they were not Facebook users, and/or did not have access to the Internet. For the current study, although target numbers were met for breastfeeding participants, and nearly met for pregnant participants, it is not possible to generalise the results to the wider Australian pregnant or breastfeeding population. […] Some limitations to the study also arise from the privacy standards of Facebook itself. It is not possible to report anything about non- responders when using Facebook due to the privacy restrictions that prevent Facebook from giving detailed demographic information about users who were exposed to promoted posts. […] Due to other Facebook privacy restrictions, shares to, or within, closed or secret (private) Facebook groups, or as private messages from outside the research team were not able to be viewed by the first author. […] Finally, it is not possible to determine the exact numbers of participants who found out about the survey on Facebook and viewed the initial survey information page but did not then click through to the actual survey.”

Kawchuk et al. 2020 [48]

N/A

“Twitter misinformation regarding a SMT/immunity link increased dramatically during the onset of the COVID crisis.”

“Although Twitter provides a window into conversations within a social media community, it is limited in that it does not represent all persons in the world. Presently, Twitter ranks 13th in total monthly users; Facebook has 2.45 billion active monthly users compared to Twitter’s 340 million.”

Merten et al. 2020 [49]

Positive: “The majority of pins positively portrayed CBD 91.6% (n1⁄4207) whereas 8.4% (n1⁄419) were balanced, and no pins negatively portrayed CBD.”

“The majority (91.6%) of pins positively portrayed CBD with many claiming a physical or mental benefit including anxiety, depression, pain, and inflammation relief. Most pins did not (98.2%) address potential side effects or recommend dosage…This study revealed widespread acceptance of the use of CBD products with minimal information from reliable public health sources represented.”

“Similar to Twitter and Facebook, Pinterest pins are searchable in search engines such as Google and Yahoo! unless the user has adjusted their privacy settings to add a secret board […] with Pinterest, it is unknown the extent to which people act upon items they pin, there is no demographic information, and there is no precise way to analyze time range.”

Allem et al. 2019 [50]

Positive: “Among health and medical, cannabis was suggested to help with cancer, plantar fasciitis, Crohn’s disease, sleep, pain, anxiety, depression, trauma, and post- traumatic stress disorder, among others.”

N/A

“…findings may not extend to other social media platforms. The posts in this study were collected from an 8-month period and may not extend to other time periods. Data collection relied on Twitter’s Streaming Application Programming Interface, which prevented collection of posts from private accounts. Findings may not generalize to all Twitter users or to the US population. Not all tweets were covered by the established categories, and topics of conversation were not segmented by geographic location, preventing this study from determining the impact of different state cannabis policies on the public’s experience with cannabis.”

Cano-Orón 2019 [52]

Negative: “The majority (79.1%) of the tweets posted using the campaign hashtags were against CTs. Whereas tweets defending the use of CTs represented 3.3% of the sample and neutral ones 11.8%.”

N/A

N/A

Hu et al. 2019 [54]

Negative: “In monitoring system, there were 610 AEs reports associated with CSE, in which 537 (88.03%) were suspected ADRs (10.49% certain). […] In the literature, 5568 AEs were identified, of which 86 (1.54%) were classified as ADRs (1.54% certain). 271 AEs were identified from 108 RCTs (n = 4682). ADR rate in RCTs was 0.021%. Baidu post bar (351 themes) and Sina micro-blogs (309 posts) published a total of 660 posts. We found no useful safety information from Baidu post bar, while we identified from Sina micro-blogs 15 AEs (unassessible/unclassifiable) in which none could be judged as certain or likely ADRs due to vague descriptions.”

N/A

“Our data was only for one Chinese patent medicine (CSE), thus the generalizability of our results is limited.”

Lacasse et al. 2019 [55]

Positive: “The most common words associated with #yoga included #fitness, #gym, #workout, and #fit, suggesting that the online depiction of #yoga is based around the physical benefits (e.g., being thin/fit from asana) as opposed to the more in‐depth limbs of yoga, such as meditation or dhyana that leads one to experience yogic enlightenment. According to the CPM, both the author/sender and the audience/receiver may be perpetuating the message of physical benefits based on the most commonly used words (as opposed to spiritual enlightenment), thus supporting the CMC theory of using Instagram as a valuable source of communication that may be changing one’s beliefs and attitudes toward the practice of yoga.”

Positive: “Moreover, popular/emerging themes/text around #yoga suggested that good feelings and appearance were the largest content categories. […] Further, the current data support previous work that practicing was associated with the societal values of reducing stress and feeling positive as well as increasing beauty and muscularity.”

N/A

N/A

Lognos et al. 2019 [56]

N/A

“The study underlines the power of digital social networks to share—disseminate—recommend practices across borders of which health professionals may have little awareness. Some patients become precursors, beta testers, of solutions never proven or whose manufacturing quality remains to be verified. The study raises important questions about the reliability of CAM information available to patients and regulatory authorities’ responsibility for labeling, approval, and surveillance. The results sensitize health professionals and authorities to the power of forums and discussion groups to make known beneficial but also potentially dangerous solutions that currently escape the purview of regulatory and monitoring systems.”

“Given the confidentiality required for the use of the social network data studied and the ethical framework of this study, it was impossible to know the medical characteristics (eg, type and severity of cancer, number of recurrences, treatment period, comorbidities, condition health, and risk behaviors) or personal (eg, age), social (eg, social status), and geographical (eg, France vs Francophonie) information on people who wrote a post. Moreover, it was impossible to know if posts were repeated several times by the same person, including on different social networks. Finally, the rules of confidentiality of the networks do not make it possible to affirm with certainty that all published posts emanate from patients with cancer. For example, companies can use these tools by creating virtual patients to promote their nonpharmacological products. Relatives of a sick person can also register to search for information. Impostors could also be spreading false medical information.”

Majmundar et al. 2019 [57]

Positive: “Themes highlighting User experience (28.90%) and Product appearance (21.80%) were the most predominant followed by posts classified as promotions (10.08%) and those highlighting flavours (1.01%) […] majority of the posts were posted by average Instagram users (24.89%) and vape vendors (20.72%) followed by KandyPens’ official account (17.96%), vaping enthusiasts/advocates (10.75%) and influencers (0.45%).” The authors define “vaping enthusiasts/advocates” as when “the account name mentions vaping or cannabis-related terms but does not sell products” and define “influencers” as when “the account promotes KandyPens by explicitly stating it using specific hashtags such as ‘#ad’ ‘#sponsored’.”

N/A

N/A

Rizvi et al. 2019 [59]

Positive & Negative: “The top three groups with the most number of respective assigned categories and topics, which can be regarded as the information most sought by consumers, are: “use and adverse effects”, “product-related”, and “healthy life style” […] Extracted information pertaining to any symptom or sign could either be an indication or an adverse event of a DS, (e.g., diarrhea, abdominal pain, palpitations, headaches); therefore, uses and adverse effects were combined as one group, “use and adverse effects”. We found a higher number of topics and the associated number of questions concerning: gastrointestinal system (specifically diarrhea and constipation); psychiatric (mainly anxiety and depression); and skin and subcutaneous tissues (primarily acne and UV protection). We also had a “mixed group”, having keywords corresponding to more than one system. For “product-related groups”, we merged categories like dose, dose from, preparation because of their co-occurrence under one topic (e.g., Topic #43). Under the “healthy life style” group, the topics were mostly around eating healthy and weight control/exercise.”

N/A

“The purpose of this research study is to understand the information needs of DS consumers by analyzing questions coming directly from consumers and in their own language. The goal is achieved by using Correlation Explanation (CorEx)—a topic modeling algorithm on the title and body of each question under the Q&A section of the Yahoo! Answers database in order to unveil the “topics” around DS information needs. We generated a list of coherent topics that more accurately represent the areas of DS-related information and associated DS ingredients that consumers are most interested in. We will also evaluate the accuracy of the CorEx method in correctly identifying the topics from social media. In the future, the knowledge gained from this study could be used as a guide for developing more meaningful DS resources for consumers that are better aligned with their information needs.”

“This information provides essential knowledge on the use of DS for various specific reasons and needs further exploration. […] We analyzed only questions belonging to alternative medicine sub-category under “health” section and might have missed dietary supplement occurrences under other sub-categories, e.g., mental health conditions, general health care. […] There are inherent limitations to topic modeling e.g., topics were generated based on the statistical word distribution within the questions and thus topics with incoherent topic keywords were also generated.”

Yin et al. 2018 [61]

N/A

“In total, the 100 videos were viewed more than 36.80 million times. Among them, 52 were consumer videos; 16 were professional videos; and 32 were news videos… Health professionals could engage more with YouTube by providing clear and authentic information about a popular alternative therapy.”

“The research team focused the study on the 100 most widely viewed, English-language, YouTube videos on cupping, and the results may not be generalizable to the less popular videos and videos in other languages. The study would be strengthened if videos in other languages were included, particularly in places where cupping therapy is highly popular. The cross-sectional design of the study did not capture the changing dynamics of the meta-data of the videos (eg, how the number of views changes with time).”

Dai and Hao 2017 [62]

Positive: “The marijuana related tweets were predominated by supporting opinions and the 213 supporting tweets outnumbered the tweets that were against or neutral about marijuana use by a 214 ratio of 8.6 to 1. The tweets that support marijuana use had a higher number of tweets per user 215 (2.6 vs. 1.8, P < 0.001) and a higher number of followers per user (19,495 vs. 10,467, P < 0.001) 216 than those against or neutral about marijuana use.”

“One of the challenges in working with social media data is the amount of “noise” or “chatter” misinformation included in the data. In our study over 10% of marijuana-related tweets were from top 10 users, suggesting that some of these tweets might be sent through power users or twitter bots (Benevenuto et al., 2010), not reflecting actual attitudes of the public.”

“…our study is a cross-sectional in examining PTSD related tweets over 258 days. We did not assess the trends over time and were unable to establish causal inferences. […] Our research is based on a sample of tweets extracted with selected search keywords related to PTSD and marijuana use. […] Given the sample biases inherent in the Twitter data, the findings from this study should be interpreted with caution and they may be more reflective of perceptions for a certain population. […] Fourth, one of the challenges in working with social media data is the amount of “noise” or “chatter” misinformation included in the data. In our study over 10% of marijuana-related tweets were from top 10 users, suggesting that some of these tweets might be sent through power users or twitter bots, not reflecting actual attitudes of the public.”

Marcon et al. 2017 [63]

Negative: “The “against chiro” group labelled chiropractic and chiropractors with five different negative terms: bullshit/bs; snake oil salesman; witchcraft (voodoo, etc.); a con/scam (taking advantage of gullible people); quacks (quack, quackery, quackropractic); and hacks. This group also used four more elaborate arguments that included: (1) equating chiropractic with the placebo effect; (2) delegitimizing the practice by stating that it is not supported by science, evidence, or evidence-based science; (3) stating explicitly that chiropractic does not cure anybody, thus requiring continual treatment, and (4) suggesting that chiropractic can be risky or dangerous. In addition, commenters in this group (1) provided links to studies/articles to support arguments and (2) expressed nuance, stating there are “good chiropractors” or that chiropractic can be effective in some situations, but that numerous medical benefits should not be attributed to the treatment.”

Positive: “In the “for chiro” group, argumentative characteristics fall broadly into two overarching categories: (1) expounding the benefits of chiropractic and (2) raising critical issues in medical care which chiropractic avoids. Regarding benefits, commenters talk about chiropractic being natural and using natural processes; getting to roots of a problem (not treating merely symptoms but underlying is- sues); helping with migraines; being safe; having a long history (“over 4,000 years”), and being science-based. In addition, personal anecdotes are used to make claims of chiropractic efficacy.”

N/A

N/A

Bitcon et al. 2016 [64]

N/A

N/A

“Views of bloggers writing in languages other than English were not captured in this search. Secondly, the keywords used for the initial sample were compiled from existing academic literature therefore bloggers that used different terms to describe their blog would have been missed. This bias would have then carried through to the second stage of snowball samples and potentially resulted in missing other online communities and themes. In addition, the research may have missed less widely linked blogs as Google blog search was the only engine used for the initial collection of blogs.”

Marcon et al. 2016 [67]

Negative: “Of all tweets analyzed in Corpora 1 and 2 (n = 1200), a total of 77 tweets (6.42%), 95% CI (2.52%-10.32%) contained skeptical or critical sentiment. Following in-depth analysis, 25 of the 77 tweets contained explicitly skeptical or critical content, representing 2.08% of the more general Twitter discourse, 95% CI (0%-5.98%). In Corpus 3: “spinal manipulation” (n = 67), 25 tweets, 37% of the corpus, contained skeptical or critical sentiments. Following in-depth analysis, 9 of the 25 tweets contained explicitly skeptical or critical content, representing 13% of the Corpus.”

“In the abundance of tweets substantiating and promoting chiropractic and SMT as sound health practices and valuable business endeavors, the debates surrounding the efficacy and risks of SMT on Twitter are almost completely absent. Although there are some critical voices of SMT proving to be influential, issues persist regarding how widely this information is being disseminated.”

N/A

Keim-Malpass et al. 2013 [69]

Positive: “Through CAM use, and yoga in particular, an awakening was described, where the participants were able to reconnect with their bodies in a more meaningful way

Many of the young women were acutely aware of their bodily experiences throughout the cancer trajectory. The young women became “in tune” with minute sensations and changes that they were not aware of prior to the cancer diagnosis. Such descriptions were found in several narratives and were often recounted as a positive encounter that coincided with the CAM experience. […] This awakening and reconnection through CAM use was uniformly characterized as a positive sentiment. For bodies that were somewhat alienated as “not fully their own” during the cancer journey, use of CAM represented an option where young women could learn to trust their bodies again.”

Positive: “… suggests that many found strength within CAM modalities (particularly yoga) to help reshape and focus on the evolving identity post-diagnosis.”

Positive: “For many of the young women, using CAM signified the women taking positive directions to control their health. The descriptions of “the good stuff” embedded within their narratives represented food and nutrition modalities used to make changes in nutritional choices, support the immune system, and help alleviate side effects from treatment.”

Positive: “One of the most powerful elements of CAM was the sense of release that would accompany yoga, acupuncture, meditation, or guided imagery. The release that occurred allowed the women to transcend their physical body and be present in the moment. […] The descriptions and use of CAM by the young women highlights the moments of solace and reconnection to themselves during periods of cancer treatment. The CAM use was often initiated by a high symptom burden and an empowering desire to take a proactive and positive complementary approach to treatment. The described CAM use had both short-term impacts for combating negative disease experiences as well as longer term impacts for enhanced cancer survivorship and positive lifestyle changes through stress reduction, mindfulness, physical activity, and more balanced nutritional choices.”

N/A

“Because qualitative methodology was used, the data lack generalizability beyond the experiences presented. In addition, because the participants were only accessed through online public websites, identity was not captured. Therefore, no disease or treatment-related details could be confirmed by a medical record.”

Walden 2013 [71]

N/A

N/A

“I do not wish to overstate the functional appeal of blogs. As discussed earlier, a search for naturopathic blogs revealed several dozen blogs that appeared at the top of search engine results but were apparently abandoned. Three bloggers who were originally interviewed for this study stopped blogging at their sites in 2010, while another three interviewees had published five or fewer posts through November 2011. Such limited activity in nearly a third of this study’s sample reveals the challenge of regularly blogging. Most interviewees, regardless of the number of entries they write, observed that blogging takes a lot of time. There are pressures to get it right and to “represent all of us” in an accurate and science-based manner. To spread positive information about the discipline, naturopathic physicians appear to start blogs with great enthusiasm yet blogging interest may wane after a certain point.”