Enthusiasm for medical cannabis is rapidly outstripping the evidence to support its use. Randomized controlled trials are needed to provide guidance to patients, clinicians, and policy makers. However, in the meantime, these results offer useful preliminary guidance for the use for medical cannabis, and can suggest directions for future research. There are four results in particular that may be clinically useful and which should prompt future research.

First, these results underscore the importance of the THC:CBD ratio as an important attribute that patients and clinicians should consider in treatment decisions. This is particularly true for neuropathic pain, insomnia, and depressive symptoms, for which this ratio seems to explain considerable variance in perceived effectiveness. For neuropathic pain in particular, an increase in THC:CBD ratio over the full range (0–100%) more than doubled perceptions of effectiveness.

Second, some symptoms may not be as responsive to increases in the THC:CBD ratio. For PTSD-related flashbacks and anorexia, although adjusted proportions showed a trend toward an increase, this trend was not significant. Indeed, the evidence to support the use of cannabis for PTSD-related symptoms is still very sparse,¹¹ although several clinical trials are underway.

And for anxiety symptoms, increases above a ratio of 1:1 were associated with reduced effectiveness. This inverted U-shaped curve seen for anxiety symptoms is not surprising. One well-recognized side effect of THC is anxiety symptoms, sometimes accompanied by paranoia. So it is reasonable that even if THC offers some symptom relief, higher ratios may have the opposite effect.

Two other results are more surprising and warrant further exploration. First, the finding that increasing THC:CBD ratios are associated with greater effectiveness for neuropathic pain is unexpected. There is increasing evidence that CBD offers benefits in the treatment of neuropathic pain, and some suggestion that CBD-only formulations may be effective.^9,12–17 And yet these results indicate a strong effect of increasing THC:CBD ratio, in a linear manner. It is also noteworthy that the overall improvement rate noted here for neuropathic pain (47%) is higher than that reported in a recent meta-analysis (20%).¹⁸ Further research is needed to explore the observed differences between these findings in a real-world population and the previous literature.

Second, these results did not indicate that a higher THC:CBD ratio is associated with greater effectiveness in the treatment of anorexia. This is also unexpected, because much of the evidence for the use of cannabis for this symptom comes from studies of either THC-dominant cannabis strains or synthetic THC (dronabinol).^19–21 Those studies are generally consistent in demonstrating modest effectiveness. Therefore, the results reported here suggest that CBD may also have a role to play in the treatment of anorexia.

Although this study offers some surprises and can make valuable suggestions for future research, there are three limitations that should be considered in interpreting the results reported here. First, this study relied entirely on patients' reports of their own symptoms. We cannot establish, for instance, that these patients actually suffered from clinically diagnosed depression or neuropathic pain. Although this is a significant limitation, it is a limitation that would be expected to introduce noise into potential associations of interest. That is, errors in diagnostic classification would tend to reduce the ability to detect differences in the association between THC:CBD ratio and perceived effectiveness across symptoms. The net effect of diagnostic misclassification would be to produce underestimates of relationships. Therefore, although this study may have failed to detect real associations (e.g., for anorexia or depressive symptoms), one can still be confident that the associations that were identified are genuine.

Second, in this study we had access to very little additional data about patients. For instance, we are unable to determine the etiology for a patient's report of neuropathic pain or anorexia. Even the presence of diagnosis data cannot be used to ascribe etiology with any certainty. Nor could we use additional data such as body mass, functional status, or other medications that may be associated with responsiveness to cannabinoids. These and other limitations could be remedied by future registry studies that link to electronic health records.

Finally, it is important to note that this study may overestimate THC's value because of the placebo effect. Because THC has psychoactive effects, it is easy for patients to be cued by the sensation of euphoria to expect benefits. This limitation is not specific to this sort of observational study, of course. In fact, it is a broad concern about all clinical studies of THC-containing cannabis, which makes true blinding impossible.²²

Although these limitations are important, this observational study provides important and novel evidence that this approach can offer important insights into use patterns and perceptions of effectiveness. By leveraging large number of patients, using cannabis in their daily routines, large-scale registry studies can provide valuable real-world data. Although such studies cannot replace randomized controlled trials, they can—and should—be an essential part of the growing science of medical cannabis.