Pharmacological topics of interest for young people: misinformation on TikTok is common.

It is essential that future studies investigate other social media regarding the presentation of pharmacological content. Future studies should include broader sampling strategies and use TikTok Research API. Stricter guidelines for advertisement and quality control of information presented should be established. Considering the findings of this research, we suggest that TikTok use should not be allowed for people younger than 16 years. Children are highly vulnerable to being influenced by social media influencers and advertisements especially concerning junk food, mental health, and addiction behaviour (Pedrouzo and Krynski 2023 ). It is of high importance to establish control mechanisms when registering for use of TikTok, since many children do not use their real birth of date and therefore gain access to unsuitable content (Berg et al. 2023 ). Health care practitioners must be made aware of the consequences of TikTok use and understand pharmacology-related topics and beliefs. Professionals should address this in patient-doctor conversations to achieve better education and understanding for the information discussed. Common misconceptions can possibly be corrected and insecurities arising because of content consumed on social media minimised. Parents must be made aware of the risks of TikTok use and guided to teaching their children safe social media use (Pedrouzo and Krynski 2023 ). TikTok is still rarely accessed by institutional accounts (McCashin and Murphy 2023 ). This is also shown in our research since the majority of the content was created by accounts without expertise. It needs to be investigated in which way professionals can use social media platforms as an opportunity to convey pharmacology-related education and engage with the young population on TikTok.

As shown in Fig.  2 , 11 (9%) of the 121 samples contained a form of advertisement. Only advertisements concerning pharmacological content were considered. 110 (91%) posts did not include any advertisement. The relationship between occupational group and advertisement was investigated. Three (42.86%) posts created by pharmacists and 11 (20.76%) posts created by influencers included advertisements. One (4.76%) video created by the occupational group “unknown/no expertise” also contained advertisement. Because there was only one post created by a scientist that also included advertisement, the results of 1 (100%) are not representative. Videos produced by other occupational groups did not contain advertisement. A recent study has shown similar results about influencers advertising supplements on Instagram (Ricke and Seifert 2024 ). Sixty-three percent of posts analysed contained discount codes for said supplements advertised by influencers (Ricke and Seifert 2024 ). Fig. 2 Advertisement as a pie chart Advertisement as a pie chart Apart from the occupational group, we additionally explored the relationship between advertisements and misinformation present in the video. The results are shown in Fig.  3 . The differences in misinformation among items with and without advertisement were not statistically significant in this study (chi 2 test, p  = 0.3250; Fisher’s exact test p  = 0.3299). Even though results in this category were not significant, the phenomenon of advertisement and misinformation is shown widely across social media platforms and needs to be further investigated (Zenone et al. 2021 ). Fig. 3 Relation between misinformation and advertisements as bar chartgiven as a percentage Relation between misinformation and advertisements as bar chartgiven as a percentage 19 (16%) of 121 posts included one source of reference. Only 7 (6%) contained more than one source of reference. The majority with 95 (78%) did not mention any source of reference (Fig.  4 ). The subcategory containing the highest number of posts with misinformation (33,68%) was no source of reference (Fig.  5 ). Videos with one source of reference contained 31.57% of misinformation among the posts. The subcategory with the least amount of misinformation (14.29%) was several sources of reference. The differences in misinformation were not statistically significant in this study (chi 2 test, p  = 0.5690; Fisher’s exact test, p  = 0.66739. Thus, it is not possible to draw definitive conclusions regarding the influence of the source of reference on misinformation. Fig. 4 Source of reference as a pie chart Fig. 5 Relation of misinformation to sources as a bar chart given in percentage Source of reference as a pie chart Relation of misinformation to sources as a bar chart given in percentage Most posts were created by influencers (53) (Fig.  6 ). Doctors were the second largest group with 25 videos. Twenty-one creators were categorised as unknown/no expertise. Thus, most videos are created by people without any expertise or who have financial interests in creating content. This could result in lower quality of information and large spreading of misinformation among lays. Fig. 6 Occupational groups shown as a pie chart Occupational groups shown as a pie chart Sixty-eight percent of 121 posts did not present any misinformation. Thirty-two percent of videos contained some form of medically related misinformation. English posts included misinformation in 33.3% of videos and German posts contained 25% misinformation. This difference was not statistically significant (chi 2 test, p  = 0.5064; Fisher’s exact test, p  = 0.57959. Posts containing common terms (32.81%) were not significantly more likely to contain misinformation than posts using medical terms (31.58%). Of the videos that used moving visual animation technique, 43.48% contained misinformation. Posts that did not include animation only contained 29.59% misinformation. A recent study about dietary supplements found that two-thirds of posts analysed exceeded the nationally established recommendations for maximum dosage (Ricke and Seifert 2024 ). While supplement overdoses can also cause detrimental effects, false claims about medicinal products as analysed in this study could be more dangerous or even fatal. Examples of misinformation conveyed are shown in Table  4 . Since one-third of our analysed sample contained false claims, it is essential to improve control mechanisms and safety regulations for TikTok users. Most of them are young people who are prone to be influenced by these videos which might be dangerous to their health and education. Influencers make promises of efficacy and trivialize side effects (Ricke and Seifert 2024 ). Young people are known to look for health-related information on social media more frequently over the last few years (Ricke and Seifert 2024 ). Our findings are congruent with the findings of other studies. One study investigated 166 TikTok videos in terms of Covid-19-related misinformation (Baghdadi et al. 2023 ). The study showed that 22% contained moderate level of misinformation and 11% contained high-level misinformation (Baghdadi et al. 2023 ). Additionally, posts containing high-level misinformation seemed to achieve more engagement among viewers (Baghdadi et al. 2023 ). Table 4 Misinformation in content presented on TikTok File number Misinformation Occupational Group Correct information Source Misinformation category Language 2 MDMA, LSD, and psilocybin have the same pharmacological effects Podcast MDMA, LSD, and psilocybin are different drugs and do not have the same effects 1. Nutt et al. ( 2010 ) 2. Hase et al. ( 2022 ) Incorrect information English 5 Benzodiazepines and alcohol cause one of the most dangerous drug interactions Doctor Sedating effects on the CNS may be potentiated but there are many other very dangerous drug/drug interactions that can be fatal 1. Roque Bravo et al. ( 2022 ) 2. Freissmuth et al. ( 2020 ) 3. Gudin et al. ( 2013 ) Incorrect information English 9 CBD oil helps with anxiety. Pets get intoxicated by licking Voltaren® off of their owner’s skin. Ibuprofen causes inflammation Podcast CBD oil effect on anxiety is not proven. There are no studies concerning pet intoxication with Voltaren®. There is only one paper discussing the possibility of Ibuprofen worsening arthritis inflammation. Ibuprofen in general is known to have anti-inflammatory effects 1. Hazekamp ( 2018 ) 2. Spinella et al. ( 2021 ) 3. Legare et al. ( 2022 ) 4. Soleymanpour et al. ( 2021 ) 5. Dunayer ( 2004 ) 6. Nissen et al. ( 2016 ) 7. Luitjens ( 2022 ) 8. Rogers and Aronoff ( 2016 ) Lack of evidence English 10, 11, 14 CBD oil helps with anxiety Influencer CBD oil effect on anxiety is not proven 1. Hazekamp ( 2018 ) 2. Spinella et al. ( 2021 ) 3. Legare et al. ( 2022 ) 4. Soleymanpour et al. ( 2021 ) Lack of evidence English 18 Cannabis use does not influence lung cancer risk Influencer, scientist Studies have shown that there is a possible relation between lung cancer and cannabis use. The effects still need to be investigated 1. Jeffers et al. ( 2024 ) 2. Connor et al. ( 2021 ) 3. Hoch et al. ( 2019 ) 4. Baumeister et al. ( 2021 ) 5. Tashkin ( 2014 ) Incorrect information English 21 CBD oil does not cause positive drug test results. There are no side effects. It helps with anxiety and muscle pain Influencer The mentioned products contain various levels of THC doses. Studies have shown that through THC contamination, CBD oil use can lead to positive drug test results. CBD can have side effects and its effect on anxiety is not proven. There are no studies showing that it helps with muscle pain 1. Arkell et al. ( 2020 ) 2. Johnson et al. ( 2022 ) 3. Legare et al. ( 2022 ) 4. Spinella et al. ( 2021 ) Lack of evidence English 23 Cannabis eases Parkinson’s symptoms like tremor Unknown/no expertise There are no studies proving this claim. Case numbers are very little so no assumption can be made 1. Legare et al. ( 2022 ) 2. Hoch et al. ( 2019 ) 3. Figura et al. ( 2022 ) 4. Singh et al. ( 2023 ) Lack of evidence English 26 Pictures show brain atrophy in cannabis users. Cannabis limits blood flow to the hippocampus and impairs memory Influencer There is no study showing pictures confirming this claim. The website mentioned is not scientifically verified. Studies show influence on hippocampus and memory but only in terms of acute consumption 1. Filbey et al. ( 2018 ) 2. Kloft et al. ( 2020 ) 3. Kutlu and Gould ( 2016 ) Incorrect information English 27 Cannabis increases risk for psychosis four times Doctor, podcast There are studies confirming a higher risk for psychosis but we could not find a study describing the mentioned numbers 1. Bara et al. ( 2021 ) 2. Connor et al. ( 2021 ) 3. Hoch et al. ( 2019 ) 4. Hill et al. ( 2017 ) 5. Hamilton ( 2017 ) 6. Bossong and Niesink ( 2010 ) Incorrect information English 35 MDMA does not harm the brain and has only been made illegal because of its concurrence to the alcohol industry Unknown/no expertise MDMA use does harm the brain. There is no evidence that the illegalisation of MDMA was related to the alcohol industry https://www.nytimes.com/1985/06/01/us/us-will-ban-ecstasy-a-hallucinogenic-drug.html 1. Kalant ( 2001 ) 2. Mustafa and Mohamad ( 2019 ) 3. Pentney (s2001) Incorrect information, lack of evidence English 37 MDMA reduces dopamine. When used therapeutically, it does not cause “hangover symptoms” News channel Recent studies show that MDMA increases dopamine. Studies investigating treating PTSD with MDMA describe hangover 1. Feduccia and Mithoefer ( 2018 ) 2. Freissmuth et al. ( 2020 ) 3. Kalant ( 2001 ) 4. Mitchell et al. ( 2023 ) 5. Parrott ( 2002 ) Incorrect information English 39 A study showed impaired memory in MDMA users. They forgot 25% more items on the list than non-users Unknown/no expertise There is no study confirming this 1. Freissmuth et al. ( 2020 ) 2. Kalant ( 2001 ) 3. Mustafa and Mohamad ( 2019 ) 4. Pantoni et al. ( 2022 ) 5. Seifert ( 2018 ) Incorrect information German 40 175 people die every day from fentanyl poisoning. Two people went into cardiac arrest because of performing CPR on fentanyl users and therefore inhaling the drug Influencer There is no official source confirming these numbers. Also, there is no official source describing this phenomenon of overdosing while doing CPR. Toxicologists have shown that it is not possible to inhale this much Fentanyl to reach an overdose this quickly 1. Ciccarone ( 2021 ) 2. Del Pozo et al. ( 2022 ) 3. Nelson and Schwaner ( 2009 ) 4. Patocka et al. ( 2024 ) 5. Rauf et al. ( 2021 ) 6. Volkow and Blanco ( 2021 ) Incorrect information English 43 There is no safe dose of fentanyl. It is impossible to know if a drug is laced with fentanyl Influencer In a therapeutic setting like anaesthesia, there is a safe dose of fentanyl. There are test kits to test drugs on fentanyl content 1. Freissmuth et al. ( 2020 ) 2. Norman et al. ( 2023 ) Incorrect information English 45 The video includes a diagram showing US deaths by drug overdose Other expertise The shown numbers could not be verified; the NIDA shows different numbers https://nida.nih.gov/research-topics/trends-statistics/overdose-death-rates Incorrect information English 51 Number of national fentanyl drug overdose deaths in the USA Influencer The shown numbers could not be verified; the CDC shows different numbers CDC ( 2021 ) Incorrect information English 52 The deathliest and most addicting drug is sugar. The second are tobacco and alcohol. Fentanyl is not even close Doctor Fentanyl is the number one cause of death for people aged 18 to 45 in the USA. Also, drugs like Heroin have much higher addiction tendencies and mortality 1. Nutt et al. ( 2010 ) 2. Patocka et al. ( 2024 ) Incorrect information English 62 Size and weight affect the duration of LSD effects Pharmacist There are no studies showing that size has any influence at all. Weight also seems to have little to no effect 1. Dolder et al. ( 2017 ) 2. Freissmuth et al. ( 2020 ) 3. Holze et al. ( 2021 ) 4. Jastrzębski et al. ( 2022 ) 5. Kuc et al. ( 2022 ) 6. Seifert ( 2018 ) Lack of evidence English 67 LSD does not cause addiction Influencer LSD does not cause physical addiction but can cause psychological addiction 1. Freissmuth et al. ( 2020 ) 2. Holze et al. ( 2021 ) 3. Klock et al. ( 1975 ) 4. Nichols ( 2016 ) 5. Schmid and Liechti ( 2018 ) Incorrect information, generalization English 68 Video shows an illustration of a birth control pill stopping the sperm from entering the egg Influencer This is not correct, hormonal birth control prevents ovulation Freissmuth et al. ( 2020 ) Incorrect information English 75 Hormonal birth control causes you to gain less muscle mass. Taking hormonal birth long-term cycle is unhealthy. It can cause galactorrhoea. Progesterone implants can cause PCOS Influencer Hormonal birth control influencing muscle gain is not proven. There are medicinal advantages to a long-term cycle use of contraceptives. There is no evidence supporting that there is a correlation between contraceptives and galactorrhoea. Birth control does not cause PCOS and is even a possible therapeutic option 1. Atluri et al. ( 2018 ) 2. Freissmuth et al. ( 2020 ) 3. Lopez et al. ( 2016 ) 4. Nolan et al. ( 2024 ) 5. Sheehan ( 2004 ) Lack of evidence, incorrect information English 79 Hormonal birth control causes Estrogen dominance. It influences the vaginal flora and the gut microbiome. It increases risk for ovarian cancer. Lactose consumption can trigger Hashimoto’s thyroiditis Unknown/no expertise There are no studies describing an Estrogen dominance. The findings for changes in the vaginal flora are incongruent and not proven. There is no association to the gut microbiome. Hormonal birth control has protective effects concerning ovarian cancer. There are no studies suggesting that Hashimoto thyroiditis is triggered by Lactose; there is a possibility that the absorption of L-thyroxine can be decreased 1. Aminzadeh et al. ( 2016 ) 2. Ampatzis ( 2022 ) 3. Coelingh Bennink et al. ( 2017 ) 4. Freissmuth et al. ( 2020 ) 5. Kamani et al. ( 2022 ) 6. Krog et al. ( 2022 ) 7. Lee and Syed ( 2022 ) 8. Seitz ( 2023 ) 9. Sung et al. ( 2022 ) 10. Usai-Satta et al. ( 2022 ) 11. Westhoff et al. ( 2013 ) Lack of evidence, incorrect information English 81 Women should not be on hormonal birth control. Your periods are not supposed to just stop. The copper IUD causes the whole body to be inflamed all the time Podcast, influencer There are advantages for hormonal birth control and long-term cycle use, for example endometriosis. The copper IUD only has local effects and does not cause inflammation in the whole body 4. Seitz ( 2023 ) Incorrect information, generalization English 83 Non-hormonal birth control for men is going to have no side effects. Several side effects like trouble breathing and chest pain Influencer, unknown/no expertise Non-hormonal contraception can also have side effects. Many of the mentioned side effects are exaggerated for entertainment purposes 1. Balbach et al. ( 2023 ) 2. Freissmuth et al. ( 2020 ) 3 Seitz ( 2023 ) Generalization, incorrect information English 84 Women on birth control make different mate choices and find different body scents appealing than they normally would. The scent selection takes place because of differences in immune systems. Women on birth control often choose actively against their natural instincts. That is why divorce rates are so high nowadays Podcast, influencer The mentioned study took place in 1995. Recent studies contradict these findings. There is no proven effect on choice based on scent and immune system. Any connection to divorce rates is not proven 1. Cobey et al. ( 2016 ) 2. Havlíček et al. ( 2020 ) 3. Wedekind (1995) Lack of evidence English 85 Hormonal birth control can cause prolactinoma. Symptoms are hot flushes Influencer There is no evidence connecting birth control to prolactinomas. Hot flushes are also not described in literature 1. https://www.niddk.nih.gov/health-information/endocrine-diseases/prolactinoma 2. Vroonen et al. ( 2019 ) 3. Wildemberg et al. ( 2021 ) Lack of evidence English 86 The leading death in women is heart disease. Birth control increases the risk and it is possible that heart disease numbers increase because of birth control Podcast, influencer Heart disease is also known to increase in the male population. There are no studies showing a correlation to birth control but rather a change in lifestyle 1. Naghavi et al. ( 2024 ) 2. Seitz ( 2023 ) Lack of evidence English 89 Hormonal birth control makes the patient’s life worse. Women who smoke should never be given hormonal birth control Doctor, podcast, influencer This is an incorrect generalization. Smoking in general is not an absolute contraindication but a relative one. It depends on other health factors and cigarettes per day Seitz ( 2023 ) Generalization English 92 Side effects of hormonal birth control include depression which women often realize after they stop taking it because they got used to the side effects Influencer Hormonal birth control can cause depression but there is no evidence that side effects become apparent after discontinuation 1. Gemzell‐Danielsson et al. ( 2022 ) 2. Mengelkoch et al. ( 2024 ) 3. Segarra et al. ( 2023 ) Lack of evidence German 94 Hormonal birth control is a modern poisoning of women. Envitox USA classified it as totally cancerous. Synthetic estrogens are one of the groups with the highest destruction potency. 25–60% of users show differences in brain structure. 100% of users show 6% reduced hypothalamus volume. One-third of users have a disease of the thyroid Unknown/no expertise Envitox did not classify hormonal birth control as cancerous. It also does not poison women. Some studies show influence on brain structure but without any of the claimed numbers. There is an association with thyroid diseases, but again, there are no studies describing these numbers 1. Qiu et al. ( 2021 ) 2. Seitz ( 2023 ) 3. Song et al. ( 2023 ) Lack of evidence, incorrect information, generalization German 95 Hormonal birth control causes PCOS and prolactinoma Influencer Hormonal birth control does not cause prolactinoma or PCOS 1. Sheehan ( 2004 ) 2. Siddiqui et al. ( 2022 ) 3. Vroonen et al. ( 2019 ) 4. Wildemberg et al. ( 2021 ) Incorrect information English 97 No women should ever take emergency contraception. It increases risk for ectopic pregnancies. The rise of ectopic pregnancies in Nairobi is due to emergency contraception Doctor This generalization is medically incorrect. Emergency contraception increases ectopic pregnancy risk only when while usage of it a pregnancy develops. Emergency contraception itself does not increase the risk. There are no studies showing a connection between Nairobi ectopic pregnancies and emergency contraception 1. Mutiso and Mugambi ( 2023 ) 2. Restaino et al. ( 2022 ) 3. Seitz ( 2023 ) 4. Shulman ( 2010 ) 5. Zhang et al. ( 2015 ) 6. Zhang et al. ( 2023 ) Lack of evidence, generalization English 102 Emergency contraception increases breast cancer risks by 44% Unknown/no expertise There are no studies supporting this claim Fitzpatrick et al. ( 2023 ) Lack of evidence, incorrect information English 107 When a person exercises intensively, testosterone levels drop a lot sometimes to the level of a castrated man Unknown/no expertise Testosterone can decrease while exercising but not to the level of a castrated man 1. Anderson et al. 2016 ) 2. Bond et al. ( 2022 ) 3. Zouhal et al. ( 2022 ) Incorrect information English 113 Everything gets better with testosterone. Testosterone levels have dropped 50% in the last 20 years Podcast Testosterone therapy also has side effects. There is no study supporting these numbers 1. Bond et al. ( 2022 ) 2. Lokeshwar et al. ( 2021 ) 3. Snyder et al. ( 2018 ) Incorrect information, generalization English 116 Side effects of Viagra® are fatal hypotension. It can cause Priapism which can cause one to lose the penis Influencer Fatal hypotension is mostly described in combination with glycerol trinitrate. Erection is still stimulus-dependent, and therefore, in normal dosages, sildenafil does not cause priapism 1. Freissmuth et al. ( 2020 ) 2. Pușcașu et al. ( 2023 ) 3. Seifert ( 2018 ) Incorrect information German 120 10 mg of Viagra® resulted in priapism and effects as late as 5 h later. It caused numbness in the arms Podcast 10 mg is below the standard dose. The first effects are seen after 30 min and last up to four hours. It does not cause priapism. Numbness is also not a known side effect 1. Fink et al. ( 2002 ) 2. Freissmuth et al. ( 2020 ) 3. Rasmussen et al. ( 2007 ) Incorrect information English Misinformation in content presented on TikTok 1. Nutt et al. ( 2010 ) 2. Hase et al. ( 2022 ) 1. Roque Bravo et al. ( 2022 ) 2. Freissmuth et al. ( 2020 ) 3. Gudin et al. ( 2013 ) 1. Hazekamp ( 2018 ) 2. Spinella et al. ( 2021 ) 3. Legare et al. ( 2022 ) 4. Soleymanpour et al. ( 2021 ) 5. Dunayer ( 2004 ) 6. Nissen et al. ( 2016 ) 7. Luitjens ( 2022 ) 8. Rogers and Aronoff ( 2016 ) 1. Hazekamp ( 2018 ) 2. Spinella et al. ( 2021 ) 3. Legare et al. ( 2022 ) 4. Soleymanpour et al. ( 2021 ) 1. Jeffers et al. ( 2024 ) 2. Connor et al. ( 2021 ) 3. Hoch et al. ( 2019 ) 4. Baumeister et al. ( 2021 ) 5. Tashkin ( 2014 ) 1. Arkell et al. ( 2020 ) 2. Johnson et al. ( 2022 ) 3. Legare et al. ( 2022 ) 4. Spinella et al. ( 2021 ) 1. Legare et al. ( 2022 ) 2. Hoch et al. ( 2019 ) 3. Figura et al. ( 2022 ) 4. Singh et al. ( 2023 ) 1. Filbey et al. ( 2018 ) 2. Kloft et al. ( 2020 ) 3. Kutlu and Gould ( 2016 ) 1. Bara et al. ( 2021 ) 2. Connor et al. ( 2021 ) 3. Hoch et al. ( 2019 ) 4. Hill et al. ( 2017 ) 5. Hamilton ( 2017 ) 6. Bossong and Niesink ( 2010 ) https://www.nytimes.com/1985/06/01/us/us-will-ban-ecstasy-a-hallucinogenic-drug.html 1. Kalant ( 2001 ) 2. Mustafa and Mohamad ( 2019 ) 3. Pentney (s2001) 1. Feduccia and Mithoefer ( 2018 ) 2. Freissmuth et al. ( 2020 ) 3. Kalant ( 2001 ) 4. Mitchell et al. ( 2023 ) 5. Parrott ( 2002 ) 1. Freissmuth et al. ( 2020 ) 2. Kalant ( 2001 ) 3. Mustafa and Mohamad ( 2019 ) 4. Pantoni et al. ( 2022 ) 5. Seifert ( 2018 ) 1. Ciccarone ( 2021 ) 2. Del Pozo et al. ( 2022 ) 3. Nelson and Schwaner ( 2009 ) 4. Patocka et al. ( 2024 ) 5. Rauf et al. ( 2021 ) 6. Volkow and Blanco ( 2021 ) 1. Freissmuth et al. ( 2020 ) 2. Norman et al. ( 2023 ) 1. Nutt et al. ( 2010 ) 2. Patocka et al. ( 2024 ) 1. Dolder et al. ( 2017 ) 2. Freissmuth et al. ( 2020 ) 3. Holze et al. ( 2021 ) 4. Jastrzębski et al. ( 2022 ) 5. Kuc et al. ( 2022 ) 6. Seifert ( 2018 ) 1. Freissmuth et al. ( 2020 ) 2. Holze et al. ( 2021 ) 3. Klock et al. ( 1975 ) 4. Nichols ( 2016 ) 5. Schmid and Liechti ( 2018 ) 1. Atluri et al. ( 2018 ) 2. Freissmuth et al. ( 2020 ) 3. Lopez et al. ( 2016 ) 4. Nolan et al. ( 2024 ) 5. Sheehan ( 2004 ) 1. Aminzadeh et al. ( 2016 ) 2. Ampatzis ( 2022 ) 3. Coelingh Bennink et al. ( 2017 ) 4. Freissmuth et al. ( 2020 ) 5. Kamani et al. ( 2022 ) 6. Krog et al. ( 2022 ) 7. Lee and Syed ( 2022 ) 8. Seitz ( 2023 ) 9. Sung et al. ( 2022 ) 10. Usai-Satta et al. ( 2022 ) 11. Westhoff et al. ( 2013 ) 1. Balbach et al. ( 2023 ) 2. Freissmuth et al. ( 2020 ) 3 Seitz ( 2023 ) 1. Cobey et al. ( 2016 ) 2. Havlíček et al. ( 2020 ) 3. Wedekind (1995) 1. https://www.niddk.nih.gov/health-information/endocrine-diseases/prolactinoma 2. Vroonen et al. ( 2019 ) 3. Wildemberg et al. ( 2021 ) 1. Naghavi et al. ( 2024 ) 2. Seitz ( 2023 ) 1. Gemzell‐Danielsson et al. ( 2022 ) 2. Mengelkoch et al. ( 2024 ) 3. Segarra et al. ( 2023 ) 1. Qiu et al. ( 2021 ) 2. Seitz ( 2023 ) 3. Song et al. ( 2023 ) 1. Sheehan ( 2004 ) 2. Siddiqui et al. ( 2022 ) 3. Vroonen et al. ( 2019 ) 4. Wildemberg et al. ( 2021 ) 1. Mutiso and Mugambi ( 2023 ) 2. Restaino et al. ( 2022 ) 3. Seitz ( 2023 ) 4. Shulman ( 2010 ) 5. Zhang et al. ( 2015 ) 6. Zhang et al. ( 2023 ) 1. Anderson et al. 2016 ) 2. Bond et al. ( 2022 ) 3. Zouhal et al. ( 2022 ) 1. Bond et al. ( 2022 ) 2. Lokeshwar et al. ( 2021 ) 3. Snyder et al. ( 2018 ) 1. Freissmuth et al. ( 2020 ) 2. Pușcașu et al. ( 2023 ) 3. Seifert ( 2018 ) 1. Fink et al. ( 2002 ) 2. Freissmuth et al. ( 2020 ) 3. Rasmussen et al. ( 2007 ) As shown in Table  4 , many claims were made against scientific evidence. This is an immediate threat to the user’s health. For example, one influencer linked her newly diagnosed prolactinoma to her previous progesterone implant and advised users to do their own research and not trust doctors’ advice. This is misleading since there is no scientific evidence describing progesterone implants as a possible cause for prolactinomas (Wildemberg et al. 2021 ). This particular video got 10.7 million views on TikTok. Influencers gain the trust of their often young followers and make promises linked to certain achievements for products advertised (Ricke and Seifert 2024 ). A relationship characterised by dependence is the result (Ricke and Seifert 2024 ). Children liking an influencer make it more likely that they will believe the information presented and imitate behaviour (Coates et al. 2019 ). Therefore, posts like the one mentioned are even more dangerous for this group, especially when they get this much attention and interaction. People could be afraid of using birth control because they fear developing prolactinoma which may lead to more unwanted pregnancies. CBD is also a frequently discussed ingredient on social media. The market for these products has increased rapidly (Soleymanpour et al. 2021 ). Oils and tinctures are exceptionally popular (Soleymanpour et al. 2021 ). The four main therapeutic claims made are effects on anxiety, stress, pain, and sleep disorders (Soleymanpour et al. 2021 ). But CBD has no proven effect on these symptoms (Spinella et al. 2021 ). There is no objective evidence supporting claims on social media, yet many posts of our sample promised immediate anxiety relief. They also advised users to take more than the recommended dosage if they do not feel the desired effects. Some posts also suggested that CBD oil has no side effects at all. Studies investigating CBD use are still very limited (Soleymanpour et al. 2021 ). Therefore, consequences of prolonged use are still unknown (Hazekamp 2018 ). Some studies describe side effects such as diarrhoea and direness (Iffland and Grotenhermen 2017 ). Aside from this, recent research has shown that especially unregulated CBD products are labelled as “THC-free” but are contaminated with levels of THC up to 0.656 mg/mL (Johnson et al. 2022 ). This unintentional consumption of THC can also cause various side effects or even addiction (Johnson et al. 2022 ). Another example for dangerous misinformation occurred in a post discussing LSD. The creator made the generalised statement that LSD does not cause addiction. While it is correct that LSD does not seem to cause physical addiction, it has severe psychologically stimulating effects which may cause psychological addiction to the drug (Passie et al. 2008 ). Because participants in studies had already developed tolerance to the effects after a few doses, it can be assumed that someone dependent on these effects from a psychiatric perspective would need more LSD to reach the desired outcome (Passie et al. 2008 ). This shows that not only false claims but also generalisations can have dangerous consequences. Because of this post, users may view LSD as a “safe drug” and not fear adverse effects. The active ingredients mentioned were as follows: drugs nonspecific, antidepressants, ketamine, ibuprofen, naloxone, glycerol trinitrate, alcohol, benzodiazepines, psilocybin, methamphetamine, heroin, ulipristal acetate, cocaine, levonorgestrel, sildenafil, testosterone/derivatives, LSD, opioids, MDMA, fentanyl, cannabis, estrogen/gestagen. The ingredient discussed most frequently was estrogen/gestagen with 29 mentions (Fig.  7 ). Second was cannabis (26) and third was fentanyl (21). The ingredient class mentioned most often was illicit drugs with 66 mentions. Second were sex hormones (45) and third MOR-agonists (12). Fig. 7 Active ingredients/drugs as a bar chart Active ingredients/drugs as a bar chart The two occupational groups “pharmacist” and “podcast” did not use any source of reference in this research. The group most likely to use several sources was “news channel” (37.5%). Since only one scientist was part of this study, the results concerning scientists are not reliable. The group that used one source of reference most often was the group “other expertise”. With 80% using no source of reference, “doctor” yielded similar results like “influencers” and people with “unknown/no expertise”. All results are shown in Fig.  8 . This finding was statistically significant with a chi 2 test p value of 0.003. Fig. 8 Relation of occupational group and source of reference as a bar chart Relation of occupational group and source of reference as a bar chart The relationship between presence of misinformation and occupational group was investigated. Every occupational group contained posts with misinformation. The occupational group containing the highest number of videos with misinformation was “influencers” with 21 posts (Fig.  9 ). “Podcasts” contained 9 and “unknown/no expertise” contained 8 videos including misinformation. This finding was statistically significant in our analysis with a chi 2 test p value of 0.0191 and Fisher’s exact test p value of 0.0180. This shows that misinformation is more common in occupational groups without expertise. Qualifications and conflicts of interest are mostly unknown, so it is difficult for a user to estimate the quality of information (Zenone et al. 2021 ). Fig. 9 Relation of misinformation and occupational group as a bar chart Relation of misinformation and occupational group as a bar chart Videos created by officials or people with credible expertise are more likely to be viewed than others (Baghdadi et al. 2023 ). Hashtag searches for “medicine” and “doctor” were viewed 1.4 billion and 6.7 billion times (Comp et al. 2021 ). Thus, lay-friendly communication of official sources on TikTok can be an important way to oppose misinformation (Baghdadi et al. 2023 ). Official sources could also convey important information in cooperation via already existing popular accounts on TikTok (Comp et al. 2021 ). The short duration and compressed content on TikTok can achieve active learning within users (Comp et al. 2021 ). The use of hashtags results in a thread of posts about the same topic, which presents an opportunity for repetition learning (Comp et al. 2021 ). Since some ingredients like ibuprofen and antidepressants were only mentioned once, we focused on ingredients that were mentioned more than five times. The lowest relative number of misinformation was found in posts containing content about opioids (16.67%) (Fig.  10 ). Videos on ulipristal acetate (20%) and testosterone/derivatives (22.23%) contained slightly more misinformation. Fentanyl- (23.81%), levonorgestrel- (28.57%), and sildenafil- (28.57%) associated posts were more common to display of incorrect information. The occurrence of misinformation in posts about LSD (30%) and MDMA (30.77%) was even higher. Videos concerning cannabis (38.46%) were the third most common in this category. Second was estrogen/gestagen with a relative amount of misinformation of 41.38%. The relative number of misinformation was highest in videos concerning cocaine (50%). These findings were not statistically significant in this research (chi 2 test, p  = 0.8136). More studies are needed to confirm or negate these results. Table 5  provides a comprehensive summary of all statistical comparisons performed in this study. Fig. 10 Relation of misinformation and active ingredient/drug as a bar chart Table 5. Results of the statistical analysis regarding the association between misinformation and other analysed factors. The table shows calculated p values. Statistically significant p values ( p  < 0.05) are highlighted Parameters Chi 2 test— p value Fisher’s exact test— p value Advertisement and misinformation 0.3250 0.3299 Source and misinformation 0.5690 0.6673 Occupational group and misinformation 0.0191 0.0180 Ingredient and misinformation 0.8136 Sample too large Relation of misinformation and active ingredient/drug as a bar chart Results of the statistical analysis regarding the association between misinformation and other analysed factors. The table shows calculated p values. Statistically significant p values ( p  < 0.05) are highlighted To establish an understanding of which ingredients/drugs are most popular on TikTok, we analysed the number of views for each ingredient (Fig.  11 ). The most popular ingredient was estrogen/gestagen with 4.339.307 views. Second was cocaine (3.823.300) and third opioids with 2.335.256 views. This is concerning because misinformation spreads easily on TikTok and can dominate search results on a specific topic (Baghdadi et al. 2023 ). Since TikTok is mostly used by young people of lower income, it is possible that misinformation on highly popular topics like birth control influence health-related behaviours (Baghdadi et al. 2023 ). Although posts with high-level misinformation present are less likely to be viewed, they are still watched by millions (Baghdadi et al. 2023 ). A study found that one in four posts containing misinformation were commented to express the intention to change behaviour (Baghdadi et al. 2023 ). Fig. 11 Relation of number of views and active ingredient/drug as a bar chart Relation of number of views and active ingredient/drug as a bar chart We found a very strong positive correlation between the number of views and the number of likes ( r  = 0.8984, p  < 0.0001). A linear regression showed that 80.7% of the variance in the number of likes could be explained by the number of views ( R 2  = 0.8070). There was a moderately strong correlation between the number of views and the number of comments ( r  = 0.6732, p  < 0.0001), with R 2  = 0.4532 indicating a lower explained variance. Results are shown in Figs. 12 and 13 . Fig. 12 Linear regression concerning the relationship between views and likes as YX-diagram. The red dots represent highly liked posts and are labeled by their file number Fig. 13 Linear regression concerning the relationship between views and comments as XY-diagram. The red dots represent highly commented posts and are labeled by their file number Linear regression concerning the relationship between views and likes as YX-diagram. The red dots represent highly liked posts and are labeled by their file number Linear regression concerning the relationship between views and comments as XY-diagram. The red dots represent highly commented posts and are labeled by their file number Our results suggest a different user behaviour considering likes and comments. This phenomenon has been investigated in previous studies as well. A user in a study in 2025 stated that even when a user likes controversial posts, they usually do not comment on them. The reason given was that comments are more likely to get you “attacked”. Another used expressed that they feel a “like” does not influence anything different in comparison to a comment (Dvir-Gvirsman 2025 ). There is no statistically significant correlation between the number of likes or the number of followers and comments (Saeidi and Baradari 2023 ). There also does not seem to be a correlation between the quality of a video and engagement (Benajiba et al. 2023 ). Another study found that health-related posts were more likely to be shared with other users but there was no correlation to a high amount of likes or comments (Tenenboim 2022 ). We analysed the highly liked and commented posts shown in Figs. 12 and 13 as red dots. All results are shown below in Tables 6 and 7 , respectively. Posts in both categories showed 67% misinformation. All posts in both categories were created by influencers. The length of the highly liked posts and highly commented posts varied substantially. Table 6 Parameters of extraordinary highly liked posts File number Occupational group Video length in sec Ingredients Source Misinformation 20 Influencer 21 Estrogen/gestagen Several No 40 Influencer 171 Fentanyl, opioids, cocaine None Yes 83 Influencer, no expertise 100 Cannabis None Yes Table 7 Parameters of extraordinary highly commented posts File number Occupational group Video length in sec Ingredients Source Misinformation 20 Influencer 21 Estrogen/gestagen Several No 26 Influencer 25 Cannabis One Yes 40 Influencer 171 Fentanyl, opioids, cocaine None Yes Parameters of extraordinary highly liked posts Parameters of extraordinary highly commented posts In our sample, 8 out of 10 most viewed posts were created by people without expertise with a range from 6,200,000 to 17,000,000 views (Table  2 ). Also, 8 out of 10 creators were categorised as “influencer” among other occupational groups. Four of these videos contained misinformation. All posts were created in English. This finding is supported by other studies. Personal TikTok accounts observably use more digital effects and features for creating their content and as a result achieve high user engagement (McCashin and Murphy 2023 ). This was lacking in public health accounts leading to less attention (McCashin and Murphy 2023 ). Official public health accounts should consider investing more time in using the full range of creational possibilities on TikTok to design their content more appealing to users. Other studies suggest that unless professionals develop a deeper understanding of the app and its characteristics, they may miss important education opportunities especially concerning young users (McCashin and Murphy 2023 ). Results of other studies investigating health-related topics are shown in Table  8 . Table 8 Findings of other studies about TikTok’s influence on health topics Source Topic/focus Method Key findings Baghdadi et al. ( 2023 ) COVID-19 misinformation on TikTok Content analysis of TikTok videos with #coronavirus High engagement with COVID-19 misinformation; potential threat to public health behavior Berg et al. ( 2023 ) Social media policies on cannabis promotion Review of platform policies Loosely defined guidelines may promote cannabis marketing and underage exposure Bozzola et al. ( 2022 ) Social media use in children and adolescents Scoping review Identified risks like cyberbullying, addiction, and exposure to harmful content Coates et al. ( 2019 ) Influencer marketing and children’s food intake Randomised trial with children Unhealthy food promotion increased intake; healthy food promotion had no effect Johnson et al. ( 2025 ) Portrayal of prescription stimulants on TikTok Systematic content analysis Predominantly positive depictions may influence medication requests McCashin and Murphy ( 2023 ) TikTok and youth mental health Systematic review and content analysis; descriptive table in Microsoft Excel TikTok is under-researched, considering psychological topics, despite its popularity among youth; adverse effects on mental health are possible Montag et al. ( 2021 ) Psychology of TikTok use Narrative review TikTok’s design may lead to addictive behaviors and affect well-being Olvera et al. ( 2021 ) TikTok tics phenomenon Descriptive and quantitative assessment of TikTok content Surge in tic-like behaviors among adolescents linked to TikTok content Pedrouzo and Krynski ( 2023 ) Children and adolescents on social media Narrative review TikTok’s algorithm is highly addictive to children; excessive use may lead to adverse cognitive effects Powell and Pring ( 2024 ) Social media influencers and health outcomes Systematic review Influencers can significantly impact health behaviors and outcomes Radin and Light ( 2022 ) TikTok in science communication Case study TikTok presents opportunities for teaching and learning science online Ranpariya et al. ( 2022 ) Dermatology influencers on TikTok Content analysis Promotional content often lacks disclosure; potential misinformation; accounts focused on dermatology, lacking formal qualifications, demonstrated the greatest frequency of promotional material Wu et al. ( 2022 ) Cigar promotions on TikTok Systematic content analysis Influencers promote cigars without age restrictions; potential youth exposure Zenone et al. ( 2021 ) TikTok and public health research agenda Commentary Areas for future research on TikTok’s public health impact should include health-risk product marketing, portrayal of illness and medical advice, and (mis)information diffusion Findings of other studies about TikTok’s influence on health topics

For this paper, we analysed videos on the social media platform TikTok. Every video analysed was publicly accessible. The search function on TikTok was used to find the videos and afterwards filtered for the highest number of Likes. The use of popularity filters was chosen intentionally to reflect the highest influence and relevancy for users. This aligns with our pilot study’s focus on public health and adolescence. Hashtags that were used in the search process are as follows: “birth control”, “birth control pill”, “estrogen”, “morning after pill”, “planb”, “Viagra”, “sildenafil”, “testosterone”, “anabolic”, “steroids”, “opioids”, “opioid crisis” “drugs”, “party drugs”, “drug awareness”, “cannabinoid”, “weed”, “cbd”, “cbd-oil”, “thc”, “ecstasy”, “molly”, “lsd”, “acid”, “fentanyl awareness”. Because of community guidelines, it is not possible to search for “cannabis” or “fentanyl”. The final sample contained 121 videos in German and English. All data were documented and analysed in an Excel table. The videos and screenshots of the Number of views were downloaded and archived. The following 13 parameters were analysed: number of likes in absolute numbers, number of Likes in relative numbers, number of comments in absolute numbers, number of comments in relative numbers, number of views in absolute numbers, occupational group of the creator, video length, advertisement, drug/active ingredient mentioned, source reference, presence of incorrect information, animation and language. The language was analysed in terms of the presence of medical terms versus common terms and the language in which the post was created. The initial classification of misinformation was performed by established medical literature. To evaluate the presence of incorrect information, the content was analysed by comparing the claims made with information on PubMed, in medical literature, or other reliable sources like medical journals. Only pharmacological and medical-related claims were analysed. Statistical testing and linear regression were performed with GraphPad Prism ( https://www.graphpad.com/features ). For statistical testing, the Chi 2 test and Fisher’s exact test were applied. Linear regression graphics were also designed in GraphPad Prism. To categorize the profession of each creator, we generated 8 different subcategories. These categories were as follows: doctors, pharmacists, news channels, podcasts, influencers, other expertise, unknown/no expertise, and scientists. To be categorised as an influencer, a creator had to have at least a follower crowd of 100,000 people and/or advertisements. When a creator did not fit the mentioned criteria for an influencer or any of the other professions, he or she was categorised as unknown/no expertise. Some creators were assigned several subcategories, for example a doctor spoke about cannabis in a podcast. The parameter source reference was subcategorised into one source reference, several source references, and no source reference. Figure  1 provides an overview of the methodological procedure. Table 1  provides a summary of important term definitions. Table 2 provides an overview of the general characteristics of the TikTok items analysed. Table 3 provides a code for the entries shown in Table  2 . Fig. 1 Flowchart describing the process of data collection Table 1 Important term definitions Term Definition Citations Occupational group “influencer” A creator with ≥ 100,000 followers and/or product promotion - TikTok influencers are people who have a considerable number of followers and let them participate in their lives. Many have cooperations with brands and promote their products accordingly or post discount codes ( https://www.marketinginstitut.biz/blog/tiktok-influencer/#dim-h2-nr4 ) - Makro-Influencers have over 100,000 followers (Sesar et al. 2022 ) Occupational group “unknown/no expertise” No verified medical, pharmacological, or other scientific qualification Occupational group “other expertise” Counseling centers, non-profit associations/awareness programs, physician assistant The involvement of nonprofit organizations has been observed to be integral to public health promotion initiatives (Vedel et al. 2020 ) Occupational group “news channel” Media formats with journalistic aspirations Misinformation Contradiction to established scientific literature or clinical guidelines “Misinformation is false information. Importantly, it is false information that was not created with the intention of hurting others. Misinformation is often started by someone who genuinely wants to understand a topic and cares about keeping other people safe and well. It is then shared by others who feel the same. Everyone believes they are sharing good information – but unfortunately, they are not. And depending on what is being shared, the misinformation can turn out to be quite harmful.” ( https://www.who.int/news-room/spotlight/let-s-flatten-the-infodemic-curve ) Advertisement Explicit marketing for drugs, supplements, or otherwise pharmacologically related products Social media can be utilised as part of advertising and promotion strategies to communicate incentives such as discounts, special offers, and product-related benefits (Radu et al. 2017 ) Table 2 Overview on the analysed TikTok content Number of video files Number of likes Number of comments Occupational group Number of views Advertisement Ingredient Source Misinformation Language 1 33,800 252 1 1,200,000 2 9 2 2 1 2 49,500 892 5 711,100 2 6, 10, 7, 8 3 1 1 3 19,000 167 7 217,500 2 6, 5, 11, 12, 13 3 2 1 4 488,200 3965 1 4,900,000 2 6, 11, 12, 13 3 2 1 5 64,700 1687 1 722,300 2 12, 15, 16, 11 3 1 1 6 32,400 1156 7 397,800 2 7, 10, 6, 17 1 2 1 7 6676 189 7 339,100 2 11 1 2 1 8 13,700 1444 6 421,900 2 11 3 2 1 9 34,300 452 5 782,300 2 6, 18 3 1 1 10 16,100 842 6 645,700 1 6 3 1 1 11 11,600 506 6 525,100 1 6 3 1 1 12 150,800 1461 1, 5 1,000,000 2 6 3 1 1 13 8571 97 6 160,700 1 6 3 2 1 14 18,072 319 6 395,462 1 6 3 2 1 15 34,000 252 1 277,600 2 6 3 1 1 16 106,700 3282 8 1,800,000 2 6 3 2 1 17 19,300 432 6 196,500 2 6 3 2 2 18 355,700 5673 6, 9 3,800,000 1 6 1 1 1 19 100,900 3017 4 1,100,000 2 6 3 2 2 20 1,300,000 36,300 6 6,200,000 2 6 2 2 1 21 32,200 1684 6 2,100,000 1 6 3 1 1 22 15,400 833 1 352,900 2 6 1 2 1 23 207,700 1070 8 2,300,000 2 6 3 1 1 24 92,700 6640 8 1,800,000 2 6 3 2 1 25 78,800 2907 6 1,600,000 2 6 3 2 1 26 362,400 13,000 6 4,100,000 2 6 1 1 1 27 332,400 8347 1, 5 3,000,000 2 6 3 1 1 28 12,000 511 4 141,800 2 6, 15 1 2 2 29 16,000 365 6 162,400 2 8 3 2 2 30 89,400 3331 1 1,100,000 2 8 3 2 1 31 21,900 359 2 570,300 2 8 3 2 1 32 75,500 1036 7 871,900 2 8 3 2 1 33 209,200 4512 4 2,400,000 2 8 2 2 2 34 171,000 5570 6 1,800,000 2 8 1 2 1 35 70,200 1493 8 496,200 2 8 2 1 1 36 200,000 2151 6 2,100,000 2 8 3 2 1 37 4992 64 4 77,500 2 8 3 1 1 38 36,700 477 4 791,000 2 8 2 2 1 39 30,300 671 8 314,200 2 8 1 1 2 40 2,600,000 27,600 6 14,500,000 2 5, 11, 12 3 1 1 41 266,200 2857 7 2,700,000 2 5, 11, 19 3 2 1 42 102,000 1619 6 2,300,000 2 5 3 2 1 43 31,300 1108 6 576,100 2 5 3 1 1 44 213,700 2885 8 1,900,000 2 5 1 2 1 45 58,400 1476 7 1,300,000 2 5, 12, 13, 14, 16, 19 1 1 1 46 11,681 297 6 323,770 2 5, 11, 19 3 2 1 47 382,500 7152 6 3,400,000 2 5, 19 3 2 1 48 128,800 7046 6 1,900,000 2 5, 11 3 2 1 49 94,500 4479 7 1,300,000 2 5, 20 1 2 1 50 2265 73 6 74,000 2 5, 11 3 2 1 51 538,700 12,300 6 6,700,000 2 5 1 1 1 52 24,900 827 1 438,900 2 5, 15 3 1 1 53 83,800 405 6 1,500,000 2 5 3 2 1 54 57,200 958 7 1,300,000 2 5, 13, 12 1 2 1 55 85,100 725 1 1,500,000 2 5 1 2 1 56 17,100 586 4 1,300,000 2 5, 11 3 2 2 57 10,400 74 1, 7 773,900 2 5 3 2 1 58 10,600 323 1, 7 1,000,000 2 5, 6, 16 3 2 1 59 23,700 1558 1 624,500 2 5, 11 3 2 1 60 242,400 6367 6 2,400,000 1 7, 10 3 2 1 61 1,400,000 14,500 4 10,700,000 2 7 2 2 1 62 10,900 170 2 162,500 2 7, 10 3 1 1 63 79,500 1508 7 523,000 2 7 3 2 1 64 109,600 1166 6 646,300 2 7 1 2 1 65 130,900 1341 6 894,700 2 7,8 3 2 2 66 86,800 2313 1 787,200 2 7 3 2 1 67 233,900 2150 6 1,500,000 2 7 3 1 1 68 183,700 692 6 3,700,000 2 1 3 1 1 69 246,200 960 1 4,300,000 2 1 3 2 1 70 129,600 572 6 1,400,000 2 1 3 2 2 71 142,800 1315 8 1,900,000 2 1 3 2 1 72 15,100 238 6 234,300 2 1 1 2 2 73 217,300 5453 1 2,600,000 2 1 3 2 2 74 70,300 688 6 2,100,000 2 1 1 2 1 75 517,300 6173 6 4,900,000 2 1 3 1 1 76 1,500,000 9036 5, 6 17,000,000 2 1 3 2 1 77 364,900 3293 1 2,800,000 2 1 1 2 1 78 15,500 250 2, 6 420,800 1 1 3 2 1 79 68,200 2255 8 990,900 2 1 3 1 1 80 1034 77 4 54,600 2 1 3 2 1 81 206,000 3572 5, 6 k.A 2 1 3 1 1 82 1,400,000 13,400 8 11,500,000 2 1 3 2 1 83 1,200,000 7385 6, 8 4,300,000 2 1 3 1 1 84 598,200 2518 5, 6 5,000,000 2 1 3 1 1 85 796,100 7488 6 10,700,000 2 1 3 1 1 86 227,600 3498 5, 6 2,000,000 2 1 3 1 1 87 486,600 4631 6 5,000,000 2 1 3 2 1 88 250,600 2165 1 1,400,000 2 1 2 2 1 89 588,700 5820 1, 5, 6 8,100,000 2 1 3 1 1 90 1,100,000 7263 5, 6 13,500,000 2 1 3 2 1 91 139,500 1507 8 1,700,000 2 1 3 2 1 92 184,500 2204 6 2,200,000 2 1 3 1 2 93 175,600 1459 6 1,800,000 2 1 3 2 2 94 201,000 8025 8 4,000,000 2 1 3 1 2 95 542,700 9844 6 4,500,000 2 1 3 1 1 96 63,200 335 2 2,000,000 2 2, 21 3 2 1 97 108,700 1547 1 1,100,000 2 2 3 1 1 98 359,900 3158 1 3,400,000 2 1 3 2 1 99 119,000 655 8 1,600,000 2 21 3 2 1 100 134,132 534 8 1,400,000 2 2, 21 3 2 1 101 13,900 162 6, 8 803,200 1 2 3 2 1 102 35,500 393 8 429,200 2 2, 21 1 1 1 103 32,100 93 2, 6 528,500 1 2, 21 3 2 1 104 73,100 629 1 1,900,000 2 2 3 2 1 105 370,900 3712 5, 8 7,700,000 2 3 3 2 1 106 223,700 922 5 3,100,000 2 3 3 2 1 107 190,800 1093 8 2,900,000 2 3 3 1 1 108 60,800 189 6 462,500 2 3 3 2 1 109 26,400 408 6 349,600 2 3 3 2 1 110 61,900 270 8 626,300 2 3 3 2 1 111 36,700 391 8 394,100 2 3 3 2 1 112 39,600 958 8 558,172 2 3 3 2 1 113 20,100 300 5 602,900 2 3 3 1 1 114 77,500 277 6 812,600 2 4 3 2 1 115 44,000 806 2, 6 490,500 2 4, 22 3 2 1 116 34,900 198 6 353,300 2 4 3 1 2 117 29,700 127 1 767,600 2 4 3 2 2 118 27,400 411 1 264,600 2 4 3 2 1 119 41,100 196 1 345,800 2 4 3 2 1 120 476,000 878 5 3,800,000 2 4 3 1 1 121 25,500 97 2, 6 170,300 1 6 3 2 1 Table 3 Number code for categories shown in Table  2 . Occupational group, advertisement, ingredient, source of reference, misinformation, and language. Because of the formality for the category occupational group, the number 3 for nursing is mentioned, although it was not present in this sample Occupational group Advertisement Ingredient Source of reference Misinformation Language 1 = Doctor 2 = Pharmacist 3 = Nursing 4 = News channel 5 = Podcast 6 = Influencer 7 = Other expertise 8 = Unknown/no expertise 9 = Scientist 1 = Advertisement 2 = No advertisement 1 = Estrogen/Gestagen 2 = Levonorgestrel 3 = Testosterone/Derivatives 4 = Sildenafil 5 = Fentanyl 6 = Cannabis 7 = LSD 8 = MDMA 9 = Drugs unspecific 10 = Psilocybin 11 = Opioids 12 = Cocaine 13 = Methamphetamine 14 = Antidepressants 15 = Alcohol 16 = Benzodiazepines 17 = Ketamine 18 = Ibuprofen 19 = Heroin 20 = Naloxone 21 = Ulipristal acetate 22 = Glycerol trinitrate 1 = One source 2 = Several sources 3 = No source 1 = Misinformation 2 = No misinformation 1 = English 2 = German Flowchart describing the process of data collection Important term definitions - TikTok influencers are people who have a considerable number of followers and let them participate in their lives. Many have cooperations with brands and promote their products accordingly or post discount codes ( https://www.marketinginstitut.biz/blog/tiktok-influencer/#dim-h2-nr4 ) - Makro-Influencers have over 100,000 followers (Sesar et al. 2022 ) Overview on the analysed TikTok content Number code for categories shown in Table  2 . Occupational group, advertisement, ingredient, source of reference, misinformation, and language. Because of the formality for the category occupational group, the number 3 for nursing is mentioned, although it was not present in this sample 1 = Doctor 2 = Pharmacist 3 = Nursing 4 = News channel 5 = Podcast 6 = Influencer 7 = Other expertise 8 = Unknown/no expertise 9 = Scientist 1 = Advertisement 2 = No advertisement 1 = Estrogen/Gestagen 2 = Levonorgestrel 3 = Testosterone/Derivatives 4 = Sildenafil 5 = Fentanyl 6 = Cannabis 7 = LSD 8 = MDMA 9 = Drugs unspecific 10 = Psilocybin 11 = Opioids 12 = Cocaine 13 = Methamphetamine 14 = Antidepressants 15 = Alcohol 16 = Benzodiazepines 17 = Ketamine 18 = Ibuprofen 19 = Heroin 20 = Naloxone 21 = Ulipristal acetate 22 = Glycerol trinitrate 1 = One source 2 = Several sources 3 = No source 1 = Misinformation 2 = No misinformation 1 = English 2 = German

All results shown above indicate possibly dangerous consequences concerning health-related content on TikTok. The statistically significant association between occupational group and misinformation and source suggests that the professional background of creators may influence the accuracy of the information presented and whether a source of reference is used. With 32% of misinformation present, it is imperative to develop better control mechanisms for TikTok content, not only to prevent misinformation from spreading, but also to regulate advertisements. Guidelines concerning advertising on social media are often vague and are not rigorously followed (Berg et al. 2023 ). It is important to investigate TikTok independently from other social media platforms because the audience and content can be highly different to other social media even on the same profiles (Comp et al. 2021 ). For example, a TikTok profile with 1.1 million followers can simultaneously have less than 1000 followers on Instagram (Comp et al. 2021 ). Aside from that, TikTok creates a sense of community and belonging through its nature of sharing and commenting (Comp et al. 2021 ). The effects need to be observed especially for vulnerable users like children. Other studies also found that health-related information on TikTok often does not meet scientifically established standards (Montag et al. 2021 ). This is especially concerning for young users since it is known that they are more likely to show risky behaviour online including substance abuse (Bozzola et al. 2022 ). With estrogen/gestagen, cocaine, and opioids being the most popular topics, it is clear where interests in mostly young people lie. Misinformation considering topics such as cannabis or other drugs may lead to a false understanding and trivialization of these substances. Especially for addicting drugs, content needs to be supervised and banned if needed. Major alcohol brands gained popularity on social media by advertising their products and evidence shows that viewers of this content normalize the behaviour presented (Bozzola et al. 2022 ). A correlation of social media use and addiction has already been established (Bozzola et al. 2022 ).

Since only 121 posts were analysed, no general conclusions on the presentation of pharmacological content on TikTok can be drawn. Further research is needed to investigate whether the findings of the present study are generally applicable. Since this paper focused specifically on popular posts and the number of likes was filtered, less popular posts with different quality of information may not have been found. Thus, our findings may include selection bias. Additionally, the classification of misinformation was conducted based on today’s level of medical research and consensus. The category misinformation was analysed in binary outcomes. This may oversimplify our findings. In Table  4 , we subcategorised misinformation, but future research should also focus on the nuances of misinformation on a deeper level so that findings are more specific. The reliability of our findings may be reduced because of the initial classification of misinformation by one coder. Future studies should include inter-coder testing. In future studies, more extensive statistical analysis of data sets should be performed. Because of different levels of popularity concerning the drugs/active ingredients, there were varying numbers of contributions to the individual groups. This causes a restricted comparability between the ingredients. Sometimes several drugs/active ingredients were mentioned per video, so the presence of misinformation cannot be clearly assigned to one individual ingredient. Aside from that, smaller groups of active substances were only mentioned in one post. These smaller groups are not representative. The numbers of views and likes were collected over a period of several weeks. Because of varying levels of fluctuations on TikTok, this might cause restricted comparability between the individual videos. One video was deleted while this research took place so the number of views could not be established. This study did not investigate the impact of language and cultural differences on perception and creation of posts. Despite these limitations, our findings suggest a concerning trend of misinformation on pharmacological topics on TikTok.

Social media has become an increasing market in our society and has a large influence on people all around the world (Powell and Pring 2024 ). Since social media content is largely unregulated, it is essential to investigate its effects on public health matters (Powell and Pring 2024 ). Recent studies show that social media is the most important health resource for young people (McCashin and Murphy 2023 ). The key role of influencers in health matters was already observed in the Covid-19 pandemic. Influencers promoting hygiene habits like hand disinfection led to positive changes in overall health behaviour (Powell and Pring 2024 ). While some studies showed these positive effects, others have found negative effects of influencer marketing on social media especially for young people (Powell and Pring 2024 ). Advertising healthy foods had no effect on children’s behaviour and promoting unhealthy food choices was linked to an increased energy intake (Powell and Pring 2024 ). Highly idealised body images created by influencers are proven to have negative effects on body image and self-esteem (Powell and Pring 2024 ). Searching for health information on the Internet has been increasingly observed, especially in younger people. A study showed that 92% of participants had used the Internet at least once in the last 12 months to find health-related information (Chu et al. 2017 ). Ninety-seven percent stated that they choose the Internet as their first medium for research considering health information. One reason for the popularity of the Internet and medical information is that users feel it enables them to be more proactive in medical decision-making. Users can find a wide range of information considering health topics. This is underlined by the fact that all participants in the study expressed their wish for clarity and confirmed health issues as a main reason to use the Internet for their research. The four main advantages stated are convenience, the wide range of accessible information, self-awareness, and the emotional support from the online community. The social aspect of community was particularly important for females. Many participants emphasised that doctors did not give them enough time during consultation to ask questions or to talk about issues. Apart from that, patients where traditional medicine “failed” were prone to search for alternative treatments via the Internet (Chu et al. 2017 ). TikTok is an originally Chinese App existing since 2016 (Pedrouzo and Krynski 2023 ). It has gained a large number of users especially, while the Covid-19 pandemic and is now the most rapidly expanding platform among the youth (McCashin and Murphy 2023 ). Data collected in the USA suggests that 32.5% of TikTok users are 10–19 years old (McCashin and Murphy 2023 ). It is estimated that globally, the majority of TikTok users are pre-teens (McCashin and Murphy 2023 ). TikTok allows use of the app for people aged 13 or older (Montag et al. 2021 ). Sending direct messages is only allowed for people who are 16 or older (Montag et al. 2021 ). TikTok is one of the most addicting social media platforms because of its highly personalised algorithm that collects information about the viewer in a shorter time period than any other social media app (Pedrouzo and Krynski 2023 ). With the assistance of artificial intelligence and a variety of device permissions required, it can show specific content and advertising based on the user’s activity on several apps (Pedrouzo and Krynski 2023 ). The continuous stimulation of the dopaminergic system leads to excessive use especially in children (Pedrouzo and Krynski 2023 ). The focus of studies published concerning this topic has been food intake and body image (Powell and Pring 2024 ). Since this is only a small range of topics, our goal was to extend this research from a pharmacological perspective. The effects of TikTok content about popular illicit drugs and sex-related medication have been investigated and are presented in the following. To our knowledge, our study is the first of its kind to analyse pharmacological content on TikTok. Due to the rising attention and importance of the topic, there is great potential for research in this area, and the present study has the character of a pilot study.