This study shows that over half of the webpages containing information about medical cannabis are from news websites, which indicates the newsworthiness of this topic. It should be mentioned, however, that the ranking made by Google prioritizes government websites, those from non-profit organizations and health portals over news outlets. Websites bearing the HONcode, an independent health information quality certification, were also ranked significantly higher, in agreement with our previous findings on information online about probiotics (Neunez et al., 2020), confirming the observation that Google uses effective criteria to prioritize high-quality information.

Content analysis in terms of disease/indications showed a mismatch between the therapeutic areas mentioned on the Web and those for which there is regulatory approval. The therapeutic area most frequently mentioned on the Web is pain. In the UK, NICE (The National Institute for Health and Care Excellence) guidelines specifically recommend not to prescribe cannabis products for chronic pain unless as part of a clinical trial (NICE, 2019) and pain treatment is not an approved indication for any cannabis product in the US as of February 2020 (FDA, 2020).

The second most frequent indication mentioned on the Web is epilepsy. It should be noted, however, that the only cannabis-based medicine approved for this indication, plant-derived CBD, is licensed exclusively for the treatment of specific epileptic syndromes, namely Dravet and Lennox-Gastaut Syndromes (Friedman & Devinsky, 2015).

Another indication frequently mentioned is multiple sclerosis, despite the limited approval of cannabis-derived products for this disease. So far, an oro-mucosal spray containing plant-derived THC and CBD (nabiximols) is only approved, for instance in the UK, for the treatment of multiple sclerosis-associated spasticity.

A number of webpages mention cancer and chemotherapy. In fact, there is moderate evidence that THC, dronabinol and nabilone may be useful for treating refractory chemotherapy-induced nausea and vomiting, although they failed to show superiority as compared to conventional drugs, particularly prochlorperazine (Smith et al., 2015). In patients with human immunodeficiency virus (HIV)/AIDS, the studies have been of short duration and limited to a small number of patients, preventing any solid conclusion of efficacy (Lutge et al., 2013).

Finally, scarce evidence suggests the CBD might be useful for the treatment of other neurological and psychiatric disorders, such as schizophrenia and anxiety (Black et al., 2019). Although there is little clinical evidence for these indications (Crippa et al., 2018; Rohleder et al., 2016), anxiety is mentioned by a significant number of webpages.

Our research complements a recent study by Kruger et al. (Kruger et al., 2020) that analyzed the information online on medical cannabis. Despite using different search terms in Google, the authors reported, like us, that pain, epilepsy, nausea, and multiple sclerosis are the medical conditions most frequently described online (Kruger et al., 2020). We also found a percentage of webpages reporting potential side effects, 22%, similar to what reported by Kruger (20%). However, our sub-analysis by website typology (reported in Table 1) identified a large variability, with side effects mentioned more frequently by health portals and non-profit organizations and much less frequently by commercial websites and news outlets, thus suggesting a bias associated with commercial interests and newsworthiness. Poor reporting on the risks associated with medical cannabis was also observed in studies on print news outlets in California (Halvorson et al., 2018) and in Canadian news media (Gunning & Illes, 2021).

Therefore, the general picture is that there is a partial mismatch between the indications mentioned on the Web for cannabis-based products and the regulatory approval, particularly for the treatment of pain. Often, an indication is frequently mentioned on the web despite there being few RCTs listed in the Cochrane database, and this is, for instance, the case of nausea, cancer, and anxiety disorders. There is also a mismatch between the frequency in which an indication is mentioned on websites and the level of evidence as defined by the NASEM report.

This breadth of online information might potentially raise the interest of the public in the use of medical cannabis for a range of indications that is broader than that of indications that are actually approved, potentially making it more attractive for the public to use cannabis supplements as self-medication in absence of a medical prescription.

In addition, there might be a misconception regarding the safety of cannabis-based products. As a result, self-initiated use may lead to side effects, drug interactions, use despite contraindications, and non-adherence to medical treatments. Among the reported side effects of cannabis-based products are nausea, vomiting, sedation, and motor impairment (Arnold, 2021; Bar-Lev Schleider et al., 2018). As for the drug interactions, cannabinoids may interfere with the effects of various psychoactive drugs in different ways, especially after long-term use (Gottschling et al., 2020). Contraindications for the use of certain cannabis-based products include brain disorders, particularly those with psychotic features, such as schizophrenia and bipolar disorder (Gottschling et al., 2020). Finally, self-medication may lead to poor adherence to doctor-oriented treatments, interfering with therapeutic effectiveness (Osterberg & Blaschke, 2005).

On the other hand, we found that the large majority of websites (88%) had a neutral stance on the use of cannabis, indicating that the information available online is not particularly polarized. One exception is represented by commercial websites that have a largely positive stance (67%), which can be explained by the fact that commercial websites are often selling cannabis-derived products.

Another aspect of health information online is that of the completeness of the information provided, which was suggested as a criterion for health information quality (Dutta-Bergman, 2004). In two recent studies, we noted that a key aspect of completeness of health information is the mention of potential side effects and regulatory issues (Neunez et al., 2020; Manley & Ghezzi, 2018). In the present study, while regulatory issues were often mentioned, only 22% of websites mentioned the potential side effects, and this was due, in particular, to the paucity of this type of information in news outlets and commercial websites. The websites providing the most complete information were those from health portals, all of them mentioning the potential adverse effects.

The main limitation of the present study is the protocol through which the sample of webpages was collected, as it may vary with time and with the search query. Although we used generic and neutral search terms, the results could be different when searching for cannabis and a specific health condition. This limitation, however, could have been at least partially circumvented by evaluating a large number of webpages. Another limitation is that we only used Google as a search engine. These may not be a representative sample of the infosphere, because Google recently implemented high-quality standards in the page ranking of what they define as “your money your life” pages. According to these guidelines, the algorithm used by Google gives a higher ranking to pages written by people or organizations with medical expertise, authoritativeness and trustworthiness, and where the information provided is aligned with the scientific consensus on the topic (Google, 2020). As we noted elsewhere, other search engines provide more lower-quality results than Google (Ghezzi et al., 2019). On the other hand, as Google has around 90% of the search engine market share, the websites it returns are the ones that the user would likely find. Not only different search engines might provide different results, but a recent study found that one third of the health-related webpages are present on Facebook (Libert, 2015), which is also used for information seeking (Anita & Williams, 2013).

Another cautionary note is the fact that we refer to the Cochrane database and the NASEM conclusions for assessing the level of evidence for the effectiveness of cannabis products in specific indications. Conclusions on the effectiveness of cannabis products in disease are not within the scope of the present study, and the field is developing rapidly with new results from clinical trials being reported. Furthermore, there are different cannabis products being tested and it is often difficult to compare the evidence obtained for different diseases with different products.