This research explored influences on the prescribing behaviour of physicians regarding medical cannabis. The research approach allowed a qualitative and in‐depth ‘lived experience’ of physicians allowing interpretation and understanding of the physicians' perceptions of (non)prescribing of medical cannabis, going beyond statistical and descriptive information found in previous quantitative research. This is especially important in a changing legislative environment as in‐depth interviews allow unexpected findings to emerge [39]. This research finds that knowledge, trust (physician‐patient relationship), concerns and implications for reputation influence prescribing behaviours. We find that physicians develop distinctive, yet unique, personal construal systems which they apply to a wide variety of aspects in their practice of medicine, including their perceptions and prescribing inclinations regarding medical cannabis.

Consistent with the findings of other studies from other countries, physicians described their legislative and procedural knowledge surrounding prescribing medical cannabis and creating dosage regimes to be very poor [e.g. 3, 27]. As such, physicians rely on reinforced internalised guidelines (i.e. when medicinal cannabis is [not] suitable, dosage) which have been informed mostly by tacit sources of knowledge such as their interactions with colleagues, patients, opinion leaders and pharmaceutical representatives, and their early medical training and experience. Previous research demonstrated that over half of New Zealand GPs surveyed were aware of available pharmaceutical‐grade cannabinoid preparations but many are also confused about whether they needed specialist/Ministry of Health approval or PHARMAC funding and there was disagreement about the levels of evidence for the effectiveness of medical cannabis [3]. Indeed, more than half of oncology physicians have reported they have insufficient understanding of the prescribing process [27]. Our research suggests that due to these beliefs, physicians reserve medical cannabis as a last resort for patients who are non‐responsive to standard treatments [e.g. 13, 24]. This was due in part to concerns such as: (i) lack of clinical evidence [e.g. 17, 18]; (ii) adverse side effects [e.g. 22] including interactions with other drugs and impact on brain structure—particularly the highly neuroplastic brains of children and adolescents that are highly susceptible to structural change induced by THC exposure, which can potentially lead to lower cognitive function and poorer outcomes later in life; and (iii) exploitation potential [e.g. 25, 26]; as well as (iv) high cost of medical cannabis—New Zealand GPs find cost one of the largest impediments encountered to prescribing medical cannabis [3, 27].

The research suggests that beyond prescription knowledge, physicians emphasise the physician–patient relationship, highlighting the key role of trust. This is also seen in a shift in viewing patient treatment from what fits the majority (as determined by large clinical studies) to what may benefit each person on a case‐by case basis (i.e. personalised medicine) [40]. This shift is seen and most often achieved by giving patients a more active role in decision‐making processes alongside health‐care professionals, by clarifying acceptable medical options and allowing the patient to choose their preferred course of care/treatment [41]. As such, trust is a big factor in physician–patient relationships [42] and its subsequent impact on medical cannabis prescription. The research suggests that GPs in our study are less trusting of their patients due to usually being the public's first point of medical contact and seeing a wider variety of individuals; while specialists are more trusting of their patients as they see individuals already filtered through GPs. A lack of trust and concern for drug‐seeking individuals may likely be due to their experiences in general practice which may led to inadvertently stereotyping patients. For example, patients directly asking for medical cannabis may be viewed with scepticism (stereotyping), particularly because medical school training has placed a strong emphasis on recognising drug‐seeking individuals [43]. Past research also indicates that a barrier to prescribing medical cannabis is the perceived potential abuse [e.g. 26]. The traditional position of cannabis as an illegal drug in society is interpreted to have a stronger influence on the need for patient trust than for other prescription drugs, due to the perceived stigma that may arise in providing a prescription.

It is suggested that a physician's unique training, professional experiences and personal history mean they develop construals about the areas of medicine they find most interesting, and the way they engage with patients [28]. The impact that patient relationships may have on a physician's prescribing behaviour was apparent in the analysis of personal consturals regarding patient interactions based on their professional medical experience as well as their personal life experiences and beliefs. As an example, Jacob favoured a very holistic and personalised medical approach that respects the values and decisions of the patient. Therefore, his system of personal construals is likely to be strongly informed by his experiences of practicing medicine in this highly holistic and personalised style, thereby making him more likely to trust his patients. This is in sharp contrast to Diana and Luke's relationship with their patients. While scientific or evidence‐based medicine can only recommend treatments that correspond with their proven level efficacy (most often across large patient groups in controlled clinical trials), holistic medicine may recommend treatments that do not have a high level of proven efficacy because they have been reported to work anecdotally on an individual basis [44]. As such, we can see that physicians' personal construal systems are at times shaped, constrained or in conflict with dominant logics that exist within the institution of medicine, particularly, the logic of standardisation [45, 46]. Indeed, the main reason for not prescribing was indicated as insufficient evidence base by 50–75% of respondents [3, 27]. Despite the rigid professional expectation for physicians to adhere to the established institutional logics that shape the medical practice [36] the findings of this study suggest that physicians allow their personal construals to influence their prescribing behaviour more than previous literature suggests.

Our research also indicates that physicians are worried about their reputation if they were to prescribe medical cannabis. Again, this may be interpreted as being due to the unique status of medical cannabis with its association to the illicit drug trade which can lead to a fear of professional judgement if they are seen to be prescribing to patients where other professionals may not see the treatment as warranted. This is similar to patients who report hiding their medical cannabis use to avoid judgement [47, 48]. This worry appears to be linked to the insufficient evidence base for medical cannabis [3]. As there are no perceived clear clinical guidelines for prescribing medical cannabis due to insufficient research surrounding its safety and efficacy, it is perhaps unsurprising that some physicians may instead choose not to prescribe medical cannabis altogether. However, no previous study has identified reputation as a barrier to prescribing medical cannabis. Physicians' personal experiences of over‐work and burnout, or those of their colleagues, may lead them to form construals to determine what their maximum patient capacity and workload is, as well as how to not exceed it. Therefore, physicians may perceive certain actions, like providing new procedures or offering new treatments as high‐risk behaviours that may lead to increased patient numbers, a greater workload, stress and eventually burnout. Although education was seen as a possible way to upskill, the reluctance to self‐educate was strong due to resistance to pre‐conceived notions about medical cannabis. It is recommended that CME on medical cannabis should be well advertised and encouraged in district health boards. It may be that physicians' multiple prescribing behaviours will impact patients who are accessing care and seeking advice (and prescription) about medical cannabis and result in frustration, confusion and illegal possession.

Our research provides several recommendations for policy and education. Given the lack of medical cannabis knowledge, any future guidelines for medical cannabis in New Zealand should be comprehensive yet succinct, be consistent when issued by different medical authorities, have clear dosage or titration regimes and be explicit about any legal intricacies [49, 50]. Previous literature should be consulted to inform future clinical guidelines on the prescription of medicinal cannabis in ways that would make physicians more likely to read and adhere to them. There also needs to be clarity around the prescribing process in the New Zealand context [27]. There were no significant thematic differences between GPs and specialists other than suitable indications for medicinal cannabis, with all psychiatrists conveying that there are no indications (and that there may even be contraindications) for it within their specialisation. This finding suggests that clarity around suitable indications may also be necessary and importantly, as the physicians were unsure about prescription guidelines and dosage (and relying on untrusted manufacturer recommendations), more information must be given about prescription guidelines. Research has shown that the preferred education methods for GPs in New Zealand by a majority is CME sessions, followed by CME online modules and information sheets [3]. Thus, education should preferably be through CME sessions. The Royal New Zealand College of General Practitioners could actively encourage GPs to enrol in courses via their communications platforms to help alleviate unconscious biases and perceived stigma. Lastly, the cost of prescription needs to be clearly communicated (if covered) and PHARMAC must consider the number of medications funded to enable greater access to patients.

The limitations of our study also provide avenues for future research. Our sample is small because of the exploratory and qualitative nature of the study which effects the generalisability of the findings. Future research based on a large‐scale quantitative survey would be of benefit. The generalisability of the findings may also be limited to the sampling method (i.e. organisational database) and future research would benefit from a random sampling approach. Participants were also mostly GPs, with only three specialists being interviewed, meaning the findings will likely not be significant in the context of any specialisations outside general practice.

Future research is needed into medical cannabis to alleviate physician concerns (i.e. side effects, cost) about prescribing and ways of fostering trust in the physician–patient relationship. We recommend that the issue of reputation should be addressed by the cannabis industry to ensure both the social and legal parameters of prescribing medical cannabis are made clear to physicians. In this vein, greater discussion around responsibility should be included in CME. Furthermore, the strategies used to assess drug seeking behaviour should be further investigated in future research. While our research and previous research find that physicians generally will follow clinical guidelines based on institutional logics (i.e. the standardised approach to medicine) [45, 46], we find that physicians often allow their personal construals to determine their perceptions and prescribing behaviour to a considerable extent when they practice medicine. Thus, future research should explore how physicians' multiple prescribing behaviours impact patients who are accessing care (e.g. emotional reductions, accessing the black market).