A Retrospective Medical Record Review of Adults with Non-Cancer Diagnoses Prescribed Medicinal Cannabis

Research describing patients using medicinal cannabis and its effectiveness is lacking. We aimed to describe adults with non-cancer diagnoses who are prescribed medicinal cannabis via a retrospective medical record review and assess its effectiveness and safety. From 157 Australian records, most were female (63.7%; mean age 63.0 years). Most patients had neurological (58.0%) or musculoskeletal (24.8%) conditions. Medicinal cannabis was perceived beneficial by 53.5% of patients. Mixed-effects modelling and post hoc multiple comparisons analysis showed significant changes overtime for pain, bowel problems, fatigue, difficulty sleeping, mood, quality of life (all p < 0.0001), breathing problems (p = 0.0035), and appetite (p = 0.0465) Symptom Assessment Scale scores. For the conditions, neuropathic pain/peripheral neuropathy had the highest rate of perceived benefit (66.6%), followed by Parkinson’s disease (60.9%), multiple sclerosis (60.0%), migraine (43.8%), chronic pain syndrome (42.1%), and spondylosis (40.0%). For the indications, medicinal cannabis had the greatest perceived effect on sleep (80.0%), followed by pain (51.5%), and muscle spasm (50%). Oral oil preparations of balanced delta-9-tetrahydrocannabinol/cannabidiol (average post-titration dose of 16.9 mg and 34.8 mg per day, respectively) were mainly prescribed. Somnolence was the most frequently reported side effect (21%). This study supports medicinal cannabis’ potential to safely treat non-cancer chronic conditions and indications.


Introduction
Cannabis (Cannabaceae) has been used medicinally since 400 AD for its analgesic, appetite enhancement, and myorelaxant properties [1][2][3]. Emerging evidence suggests that people with chronic conditions may benefit from using medicinal cannabis for treating chronic pain, multiple sclerosis-related spasticity, epilepsy, Parkinson's disease, insomnia, and anxiety [4]. However, evidence supporting medicinal cannabis for these conditions/symptoms is generally anecdotal and inconclusive [5]. Limitations include short study durations, small sample sizes, differing formulations/modes of administration, lack of follow-up [4], and stigma surrounding its addiction and abuse [6].
In Australia, impressive public support instigated legislative changes in 2016 allowing medical doctors to prescribe cannabis-based medicines for chronically ill patients subject to the approval from the Therapeutic Goods Administration [7] which is the medicine and therapeutic regulatory agency of the Australian Government. Unregistered cannabisbased medicines can be prescribed through the TGA Special Access Scheme, in particular, Category B (SAS Cat-B) [8], and this requires clinical justification for patient use, reasons why registered medications have not been appropriate, and the expected benefits versus CBD: Cannabidiol. ∆-9-THC: ∆-9-tetrahydrocannabinol. a As reported in the medical notes; b Covers around 12,368.2 km 2 and 37.6% (n = 59) of patients travelled 20 km or more to access the clinic; c Not in a relationship, including divorced, widowed, separated; d In a relationship, including married, de facto, same sex; e For each condition class, the conditions with ≥10 patients are shown (neuropathic pain and peripheral neuropathy have been combined in the analysis); f Percentages for 'Condition class' and 'Indications' are given as individual frequencies out of the entire study population, as some patients had more than one condition/indication. Percentages will add to over 100; g Percentages for 'CBM type' and 'Cannabinoid ratios' are given as individual frequencies out of the entire study population, as some patients used more than one type and/or more than one ratio percentages will add to more than 100; h Ratios (∆-9-THC:CBD) included 0:25, 0:100.

Spondylosis
In patients with spondylosis, improved trends were observed for 'fatigue', 'difficulty sleeping', 'mood', and 'QoL' scores to 6 months; and 'pain' and 'bowel problems' to 9 months (Figure 3a). Mixed-effects modelling showed significant changes over time in all symptom scores but no significant improvements in QoL (Table 3). Post hoc analysis found significant decreases compared to baseline for: 'mood' and 'fatigue' to 6 months; and 'pain' and 'difficulty sleeping' to 9 months (Table 3).

Spondylosis
In patients with spondylosis, improved trends were observed for 'fatigue', 'difficulty sleeping', 'mood', and 'QoL' scores to 6 months; and 'pain' and 'bowel problems' to 9 months (Figure 3a). Mixed-effects modelling showed significant changes over time in all symptom scores but no significant improvements in QoL (Table 3). Post hoc analysis found significant decreases compared to baseline for: 'mood' and 'fatigue' to 6 months; and 'pain' and 'difficulty sleeping' to 9 months (Table 3). Analysis was possible to 12 months. Bolded, italicized, and highlighted means are statistically significant compared to baseline at p < 0.05 level as analyzed by Dunnett's test. Abbreviations: CI, Confidence interval; DFn, degree of freedom for the numerator of the F ratio; DFd, degree of freedom for the denominator of the F ratio; Diff., difference; M, Mean; QoL, Quality of life.
In patients with spondylosis, improved trends were observed for 'fatigue', 'difficulty sleeping', 'mood', and 'QoL' scores to 6 months; and 'pain' and 'bowel problems' to 9 months (Figure 3a). Mixed-effects modelling showed significant changes over time in all symptom scores but no significant improvements in QoL (Table 3). Post hoc analysis found significant decreases compared to baseline for: 'mood' and 'fatigue' to 6 months; and 'pain' and 'difficulty sleeping' to 9 months (Table 3).

Parkinson's Disease
In patients with Parkinson's disease, improved trends were observed for scores to 3 months ('QoL' to 6 months), with general stabilization over time ( Figure 3b). Mixed-effects modelling showed significant changes over time for 'pain', 'fatigue', and 'difficulty sleeping' scores (Table 3). Post hoc analysis revealed significant decreases compared to baseline for: 'pain' and 'mood' at 3 months; and 'fatigue' and 'difficulty sleeping' to 12 months (Table 3).   Analysis was possible to 9 a,e and 12 b-d months. Bolded, italicized, and highlighted means are statistically significant compared to baseline at p < 0.05 level as analyzed by Dunnett's test. f Analysis was possible to 3 months. Bolded, italicized, and highlighted medians are statistically significant compared to baseline at p < 0.05 level as analyzed by Wilcoxon signed-rank test. Abbreviations: CI, Confidence interval; DFn, degree of freedom for the numerator of the F ratio; DFd, degree of freedom for the denominator of the F ratio; Diff., difference.

Chronic Pain Syndrome
In patients with chronic pain syndrome, improved trends were observed for the scores to 3 months which generally continued to 9 months (Figure 3c). Mixed-effects modelling showed significant changes over time in scores for 'pain', 'fatigue', 'difficulty sleeping', and 'mood' (Table 3). Post hoc analysis found significant decreases compared to baseline for: 'mood' and 'difficulty sleeping' at 3 months, 'fatigue' at 9 months, 'pain' to 12 months, with improved 'QoL' scores at 3 months (Table 3).

Neuropathic Pain and Peripheral Neuropathy
In patients with neuropathic pain/peripheral neuropathy, improved trends were observed for 'fatigue' and 'difficulty sleeping' scores to 3 months; 'pain' and 'QoL' to 6 months; 'bowel problems' to 9 months, and mood to 12 months (Figure 3d). Mixed-effects modelling analysis showed significant changes over time in 'pain', 'fatigue', and 'difficulty sleeping' scores (Table 3). Post hoc analysis showed significant decreases compared to baseline for: 'bowel problems' at 3 months, 'difficulty sleeping' to 6 months, and 'pain' to 12 months, with improved 'QoL' scores at 3 months (Table 3).

Migraine
In patients with migraine, improved trends were observed for 'mood' to 3 months; 'fatigue', 'bowel problems', 'difficulty sleeping', and 'QoL' scores to 6 months, and 'pain' to 18 months (Figure 3e). Mixed-effects modelling showed significant changes over time for 'fatigue' and 'difficulty sleeping' scores (Table 3). Post hoc analysis showed significant decreases compared to baseline for 'fatigue' scores to 6 months (Table 3).

Multiple Sclerosis
There are downward trends for 'pain' (initially to 3 months) and 'bowel problems' (initially to 6 months) scores to 12 months, and 'fatigue', 'difficulty sleeping', and 'mood' (all initially to 3 months) to 15 months, with improvement in 'QoL' to 6 months in patients with multiple sclerosis (Figure 3f). A Wilcoxon signed-rank test showed that the 'fatigue', 'difficulty sleeping', and 'mood' scores were significantly decreased at 3 months compared to baseline (Table 3).

Pain
In patients with pain, improved trends were observed for all symptom scores to 3 months ('QoL' to 6 months) with general stabilization over time (Figure 4a). Mixed-effects modelling showed significant changes over time for all symptoms scores (Table 4). Post hoc analysis found significant decreases compared to baseline for: 'bowel problems' and 'mood' to 9 months; and 'pain', 'fatigue', and 'difficulty sleeping' to 12 months, with improved 'QoL' scores to 12 months (Table 4).

Muscle Spasm
In patients with muscle spasm, initial improved trends were observed for all symptom scores to 3 months; 'pain', 'fatigue', and 'bowel problems' to 12 months; and 'difficulty sleeping', 'QoL', and 'mood' to 15 months (Figure 4b). Mixed-effects modelling showed significant changes over time for 'pain' and 'fatigue' scores (Table 4). Post hoc analysis showed significant decreases compared to baseline for: 'pain' at 3 months, 'fatigue' at 3, 9, and 12 months, and 'difficulty sleeping' to 6 months (Table 4).

Sleep
In patients with sleep issues, improved trends were observed for 'fatigue', 'bowel problems', and 'mood' scores to 3 months; 'difficulty sleeping' to 6 months; and 'pain' and 'QoL' scores (initially to 3 months) to 9 months (Figure 4c). Wilcoxon signed-rank test showed 'fatigue' and 'difficulty sleeping' scores were significantly lower at 3 months compared to baseline (Table 4).

Muscle Spasm
In patients with muscle spasm, initial improved trends were observed for all symptom scores to 3 months; 'pain', 'fatigue', and 'bowel problems' to 12 months; and 'difficulty sleeping', 'QoL', and 'mood' to 15 months (Figure 4b). Mixed-effects modelling showed significant changes over time for 'pain' and 'fatigue' scores (Table 4). Post hoc analysis showed significant decreases compared to baseline for: 'pain' at 3 months, 'fatigue' at 3, 9, and 12 months, and 'difficulty sleeping' to 6 months (Table 4).

Sleep
In patients with sleep issues, improved trends were observed for 'fatigue', 'bowel problems', and 'mood' scores to 3 months; 'difficulty sleeping' to 6 months; and 'pain' and 'QoL' scores (initially to 3 months) to 9 months (Figure 4c). Wilcoxon signed-rank test showed 'fatigue' and 'difficulty sleeping' scores were significantly lower at 3 months compared to baseline (Table 4).
Around 64.8% of patient treatments were ongoing at the time of data collection, whilst 22.3% had ceased treatment. The treatment status was unknown for 22.9% of patients (n = 157). Thirty-nine reasons were reported for treatment cessation, with the main being inefficacy (51.3%), followed by side effects (25.6%), and relief from other medications/treatment (12.8%) (Figure 5b). One patient died from disease complications unre- . Inefficacy = no effect on condition/indication, Loss of efficacy = initial effect wore off over time.
Around 64.8% of patient treatments were ongoing at the time of data collection, whilst 22.3% had ceased treatment. The treatment status was unknown for 22.9% of patients (n = 157). Thirty-nine reasons were reported for treatment cessation, with the main being inefficacy (51.3%), followed by side effects (25.6%), and relief from other medications/treatment (12.8%) (Figure 5b). One patient died from disease complications unrelated to medicinal cannabis treatment.

Discussion
This retrospective medical record review describes the population characteristics of patients using medicinal cannabis at a clinic in Sydney, Australia and provides data on the effectiveness and safety of medicinal cannabis treatment on patient conditions and indications. In our study, the mean age of patients was higher than in international studies (40-50 years) [14][15][16][17][18][19]. In Australia, patients must have trialed mainstream therapies before their medicinal cannabis application can be approved [4]. Patients were predominantly female, consistent with previous prospective studies of non-cancer patients [14,16,20]. Most patients were from the Greater Sydney Region which agrees with another study which found 82.6% of patients lived in urban areas [21]. This is indicative of medicinal cannabis prescribers and dispensers being established in larger urban areas [22]. More than half of the patients were partnered. This may make it easier for patients to undertake daily errands, including attending medical appointments whilst on medicinal cannabis as driving with cannabis containing ∆-9-THC is illegal in Australia [23].
It difficult to ascertain the actual numbers of patients that had used cannabis before attending the clinic as patients are under no obligation to state previous cannabis use. However, the results here are similar to another Australian study which showed that 14.1% had not used cannabis before using it for medical reasons [24].
In this study, most patients were prescribed oral oil cannabis-based medicines of balanced ∆-9-THC:CBD ratios reflecting their availability for prescription in Australia [4]. This could be due to cannabis flower being very expensive and the ease of administrating oils compared to vaporization [4]. In contrast, international patients predominantly inhale cannabis; however, oral preparations are becoming more prevalent due to greater physician advocation [15,16,21,25].
Literature shows that therapeutic doses of ∆-9-THC (5-20 mg) tend to be smaller than CBD (50-1500 mg) [4] and our study supports this finding. This is expected as CBD is non-psychoactive and may circumvent the side effects of ∆-9-THC [17]. Our study showed that higher doses of ∆-9-THC were used which may indicate that people with non-cancer conditions/indications may be able to tolerate higher doses compared to people with cancer, and this needs to be ascertained in clinical use [19].
Most patients in our study sought treatment for neurological and musculoskeletal conditions, with pain as the main indication. This agrees with data indicating that most approvals in Australia are for chronic pain, and pain ranks as the number one indication in several studies [4,14,16,21]. In previous studies, the frequency of insomnia and psychiatric disorders is higher [16,18,21], with one study reporting 9.7% and 27% of patients, respectively [26]. Our study's lower incidence (1.3% and 0.6%, respectively) can be explained by the prescribing medical doctor referring patients with sleep and mental health disorders to other specialists to investigate/rule out specific causes of these conditions and measure their severity.
The presence of symptoms such as pain, fatigue, sleep difficulties, and poor mood in this study indicates high symptom burden across these chronic diseases [27,28]. For overall effectiveness, just over half of the patients believed medicinal cannabis to be 'beneficial', with mixed-effects modelling of all patients showing statistically significant improvements in all SAS scores except for nausea.
When comparing age groups, 56% of patients under <65 years found medicinal cannabis to be beneficial compared to 51.2% of patients aged ≥65. A study of patients with cancer indicated that younger age was associated with the therapeutic success of medicinal cannabis, which could indicate more tolerance to higher doses, and this may be reflected in the present study [29].
In terms of patient-perceived benefit, the results suggest that medicinal cannabis is more effective at treating neuropathic pain conditions compared to nociceptive pain conditions such as spondylosis (the lowest of all conditions/indications), findings congruent with available evidence on different pain conditions [11,[30][31][32]. A prospective observational study in patients with chronic pain showed significant score reductions in measures of pain severity and interference in daily life over a 12-month study duration compared to baseline (p < 0.001), as well as significant improvements in QoL (p < 0.05) [14]. These results are supported by other studies which demonstrated significantly improved pain scores and QoL over time [15,16]. Within the conditions, SAS pain scores in our study showed significant pain decreases over time for spondylosis, chronic pain syndrome, and neuropathic pain/peripheral neuropathy to at least 9 months. Moreover, for pain as an indication, there were sustained significant decreases in all SAS scores to at least 9 months compared to baseline which shows an overarching effect of medicinal cannabis on other symptoms. Due to the many different pain etiologies, it is difficult to ascertain pain location and/or patient's response and tolerance to pain [33].
Medicinal cannabis was beneficial to around 60% of patients with Parkinson's disease. A systematic review of the effects of cannabis on Parkinson's disease, including RCTs and non-RCTs, reported improvements in sleep and pain, which agree with our findings [34].
For migraine, our study revealed that under half of the patients believed treatment to be beneficial, with significant SAS scores for 'fatigue' and a downward trend for 'difficulty sleeping'. One retrospective chart review analyzing migraine showed that 85.1% of patients reported a decrease in mean migraine headache frequency from 10.4 to 4.6 at follow-up (p < 0.0001) [25]. Similarly, archival data from an app showed migraine ratings reduced by 50% (p < 0.001) [35]. Another retrospective review showed improvements in headache profile, sleep, and mood [36].
Symptoms such as fatigue, pain, sleep, and mood disorders are also prevalent in multiple sclerosis [37]. In this study, 60.0% of patients with multiple sclerosis believed treatment to be beneficial, with symptoms of 'pain', 'fatigue' 'difficulty sleeping', and 'mood' as possible targets for medicinal cannabis treatment. For the indication muscle spasm, these results were comparable. A recent review supported Sativex's ® (an oral spray of balanced ∆-9-THC and CBD) effectiveness on multiple sclerosis pain and spasticity, with improvements in sleep and QoL [38]. Another prospective study found 70.5% of patients reported a ≥20% improvement in symptoms after one month [39], which agrees with our results. As Sativex was not widely used in our study, it is difficult to compare these findings to our study as predominantly oil formulations were used. However, over 80% of patients in this study used balanced formulations and our results could mirror these findings.
Poor sleep is an important symptom in chronic disease as it coincides with poor symptom control and impacts QoL [40]. Many studies treat sleep as a secondary outcome [41], with a literature review suggesting that cannabis could improve sleep in patients with chronic pain [42]. Although results for patients with sleep as an indication showed the greatest rate of perceived benefit of all indications/conditions analyzed, this data should be treated with caution due to a small sample size. Studies have shown insufficient and conflicting evidence for medicinal cannabis use in sleep disorders and suggest that the cannabinoid type, combination, dose, and route of administration need to be considered when designing RCTs [42,43]. Comparatively, our results showed significant improvement in SAS scores for 'difficulty sleeping' for spondylosis to 9 months and Parkinson's disease to 12 months, with improvements in other conditions to at least 3 months, with a downward trend for migraine. This also occurred with 'pain' and 'fatigue' (and sometimes 'mood') in the listed conditions and indications which shows that tackling one symptom with medicinal cannabis may reciprocate on another.
Medicinal cannabis was generally safe as shown by the follow-up SAS scores to 12 months for many of the conditions/indication. Moreover, dosing was tailored to each patient ("start low, go slow" approach) by the prescribing medical doctor, thus minimizing dose-related toxicities [44]. A scoping review found that minor side effects were reported including somnolence, dizziness, and dry mouth which are congruent to this study's findings [45]. In our study, the most reported reason for treatment cessation was perceived medicinal cannabis inefficacy followed by side effects, with our findings congruent to a prospective clinical trial [20].
The strength of this study is that patient data up to 30 months can be assessed which is longer than prospective and clinical trials [14,15]. Although missing data is unavoidable [46,47], mixed-effects modelling allowed for a valid analysis whilst reducing associated biases [13,47].
One limitation relates to the location of the study (undertaken at one clinic) which can introduce locality bias, and retrospective design of the study which limits the ability to evaluate the causality of medicinal cannabis use and improvements in symptom scores. Confounding factors may have influenced patient SAS scores. For example, a patient with spondylosis may have a higher score for their pain distress due to another developing issue. Whilst the PCOC SAS can accurately and reliably measure symptom distress in a palliative population, it lacks specific measures for many condition-specific indications e.g., mobility for Parkinson's disease [48], number of headaches experienced by patients with migraine, etc. Moreover, SAS scores of bowel problems, mood, and QoL did not significantly improve with medicinal cannabis over time for the conditions/indications examined in this study. This could be due to these symptoms having multifaceted constructs which are not covered in the single SAS score. Thus, using specific measures, more frequently/consistently recorded, may have led to more significant results. Other limitations included differentiating between these differing pain etiologies, interpreting results in smaller samples, unclear dosing regimen, and lack of follow-up. It is not known whether this missing data can accurately reflect greater improvement and/or sustainability over time. Although the cannabis-based medicines were mostly balanced, other formulations and their effectiveness could not be determined. In addition, this study was unable to relate dose and safety, tolerance to previous use, and adverse drug-drug interactions.

Conclusions
This study indicates that medicinal cannabis, in a balanced formulation, may address a variety of non-cancer conditions and indications concurrently and can be safely prescribed by a medical doctor. The results can be used to guide future clinical trials which could investigate the conditions and indications identified here as sensitive to treatment. It is recommended that 'pain' and 'difficulty sleeping' (which may improve fatigue) be investigated further as treating these two symptoms may have reciprocating effects on each other, on mood, and QoL. Future studies should address how cannabinoids interact in the body and with one another in differing ratios and formulations, and include relevant patient-reported outcome measures and longer-term safety studies. It is recommended that patients are assessed individually to determine whether medicinal cannabis is an appropriate treatment option, considering the associated safety risks to potential therapeutic benefit. Understanding how medicinal cannabis can be used in mainstream medicine is crucial as it has the potential to positively impact millions of lives around the world.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/jcm12041483/s1, Table S1: All patient conditions; Table S2 Informed Consent Statement: Patient consent was waived as the research was deemed low risk as there are no therapeutic products being tested; there was no foreseeable risk of inconvenience and discomfort due to the confidential nature of the project and pooling of data; informed consent would be a major undertaking.

Data Availability Statement:
De-identified data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.