Examining the use of cannabidiol and delta‐9‐tetrahydrocannabinol‐based medicine among individuals diagnosed with dementia living within residential aged care facilities: Results of a double‐blind randomised crossover trial

Dementia affects individuals older than 65 years. Currently, residential aged care facilities (RACF) use psychotropic medications to manage behavioural and neuropsychiatric symptoms of dementia (BPSD), which are recommended for short‐term use and have substantial side effects, including increased mortality. Cannabinoid‐based medicines (CBM) have some benefits that inhibit BPSD and cause minimal adverse effects (AEs), yet limited research has been considered with this population. The study aimed to determine a tolerable CBM dose (3:2 delta‐9‐tetrahydrocannabinol:cannabidiol), and assessed its effect on BPSD, quality of life (QoL) and perceived pain.


| INTRODUCTION
Dementia is the second leading burden of disease among Australians 65 years and older, and its prevalence is expected to increase exponentially over the next 30 years. 1 Dementia is a progressive reduction in the brain's cognitive and functional ability and manifests as a loss in memory, intellect and rationality. 1The behavioural and psychological symptoms associated with dementia (BPSD) often worsen over time leading to changes in quality of life (QoL) including depression, frustration, clinginess, forgetfulness, wandering, sexual aggression, hoarding, sleep disturbances and 'the sundowner effect' (manifestations of challenging behaviours at the end of the day). 2 The degeneration of dementia often requires the addition of supports within the home or leads to admission into residential aged care facilities (RACF) where 24-h care is available.
Although non-pharmaceutical treatments are preferred, BPSD are commonly managed using pharmacological treatment within the RACF.In particular, the RACF often supports residents living with moderate-to-severe dementia and are often prescribed off-label antipsychotics, sedatives/hypnotics, anxiolytics, acetylcholinesterase inhibitors and antidepressants to reduce troublesome behaviours. 3Many of these medications have substantial side effects 4,5 and are associated with the risk of stroke, falls, injury and increased mortality.Foebel et al. 6 suggest the use of cannabinoid-based medicine (CBM) as a promising alternative.
Studies exploring the adherence and safety of CBM have reported no adverse effects (AEs), the safety and tolerance of the drug and no increased risk of falls. 7Other studies suggest the physiological and psychosocial benefits of CBM help manage chronic pain [8][9][10] and reduce the severity of BPSD such as aggression and agitation. 11nterestingly, some evidence supports the view that CBM use may lead to a reduction in concomitant medications such as opioids, analgesics, antipsychotics, antiepileptics 12 and benzodiazepine, commonly prescribed to improve the QoL in people living with dementia. 13tudies in RACF have found CBM can lead to positive changes in physical measures such as body weight, sleep duration and meal consumption. 11However, the observed changes in BPSD have varied between studies.For example, Shelef et al. 14 reported a reduction in delusions, irritability, sleep and caregiver distress, while Broers et al. 15 observed less rigidity, improved patient care and a decrease in opioid use.The most common reduction in BPSD symptoms across most studies included less agitation 16 and aggression, 17 the most challenging behaviours to manage in those living with dementia.
In part, the variability in CBM outcomes reflects the range of study designs, drug concentrations and duration of treatment resulting in calls for further research to understand the most efficacious formulation, safety profile, drug-to-drug response and CBM use in RACF. 18any studies recommend higher doses (e.g. a range of THC 0-100 mg a day) and longer durations (greater than 3 weeks) and utilising a study design that provides greater rigour to understand the benefits of CBM in this cohort of patients. 18his study seeks to build on these recommendations by (1) determining a tolerable dose of CBM (3:2 delta-9tetrahydrocannabinol:cannabidiol) that does not cause AEs and (2) assess its impact on BPSD, QoL and perceived pain in people living in RACF with dementia.

| METHODS
Ethics approval was received from the University of Notre Dame Australia (UNDA; approval number: 018091F) and registered in the Australian New Zealand Clinic Trials Registry (ACTRN12619000464156).Participants with mild dementia and their families consented to the study.Participants with moderate-tosevere dementia assented and their families consented on their behalf following the guidelines outlined in the researchers to investigate the sensitivity of detecting BPSD changes within the complexity of the disease and concomitant with medications.

K E Y W O R D S
cannabinoid, crossover trial, dementia, pain, quality of life

Practice Impact
This small Australian study is the first to test an alternative treatment on the management of BPSD for individuals living with moderate-tosevere dementia.This study used a randomised crossover trial allowing each participant to act as their own control.Larger studies are needed to validate these findings.
Western Australian Guardianship and Administration Act (GAA) of 1990.

| Study design
As described elsewhere, 19 the study used a parallel mixed methods design and a phase IIa randomised placebocontrol crossover trial.Participants took part for 18 weeks, which comprised a 2-week eligibility period to medically review participants, collect baseline information and randomly allocate participants into the first 6-week treatment cycle to commence CBM or a placebo.A 2-week washout period occurred prior to and after the second 6-week treatment cycle.The sample size target for this study was 50 participants, which was based on an estimated requirement for 40 cases for a 2 × 2 crossover design based on a two-sided calculation to achieve 81% power (M = 6 7 [SD = 13]) with a significance level of 0.05, and an allowance for a 20% attrition rate.

| Participants and setting
Participants were recruited into the study by approaching the residential aged care clinical and general managers who had relationships with the residents and their next of kin.They provided the research team with a list of those who they thought would be eligible to participate, based on the inclusion criteria of: people living in RACFs, aged 65 years or older, diagnosed with dementia, English speaking, compliant with taking medication and not bedridden.Participants were excluded if they had frontotemporal or Lewy body dementia, epilepsy, anorexia nervosa, comorbid psychiatric conditions or Parkinson's disease.Participants with ongoing or significant congestive heart failure, myocardial infarction, anginal pain, stroke, liver disease and renal disease were also excluded from the study.Participants with a history of heart failure, myocardial infarction, anginal pain, stroke, liver disease, or renal disease were medically reviewed by the independent medical examiner and medical prescriber to assess their eligibility.Participants who were prescribed medications (e.g.primidone, rifampicin, rifabutin, troglitazone, Hypericum perforatum or valproic acid) that interact deleteriously with cannabis metabolism were also excluded.

| Medication
The CBM intervention-oil (medium-chain triglycerides base; CogniCann; MGC Pharmaceuticals Ltd.) was administered in a sealed 10-mL glass spray bottle comprising a mix of THC and CBD (3:2 ratio; 25 mg/ mL THC; 17 mg/mL CBD).Each press of the bottle dispensed 100 μL of oil containing 2.5 mg of THC and 1.7 mg of CBD.The oil was stable at room temperature (below 25°C) for 4 weeks.The placebo was administered in the same 10-mL glass bottle made of a terpene-based oil that contained esters mimicking CBM.Both products were administered by a registered nurse in the RACF with morning and afternoon tea.
The dose was titrated to increase from one spray (2.5 mg THC/1.7 mg CBD) per session to a maximum of 10 (25 mg THC/17 mg CBD) over 2-3 weeks.The presence of any AEs associated with the medication lead to the participant reverting to their previous best-tolerated dose.An upper limit of 50 mg/day of THC was provided to those who did not experience any AEs.Once a participant reached their maximum dose (or a total of 50 mg/THC, 34 mg CBD daily), they continued to receive that dose until the cessation of the 6-week period.The placebo group followed a similar titration process.

| Outcome measures
The study research nurse completed a series of questionnaires on seven occasions (first treatment cycle [Day 0, Day 24 and Day 42], second treatment cycle [Day 56, Day 80 and Day 97] and the last day of final 2-week washout period [Day 112]).These took 20 min to complete and consisted of: • The Neuropsychiatric Inventory Questionnaire-Nursing Homes (NPI-NH) 20 measures 12 neuropsychiatric symptoms (delusions, hallucinations, agitation, depression, anxiety, euphoria, apathy/indifference, disinhibition, irritability, aberrant motor behaviour, nighttime disturbances and appetite).The frequency and severity of each symptom is rated (4-point and 3-point Likert scales) to derive a total score.The 12 individual neuropsychiatric symptoms can be examined to look at the presence of these behaviours.• The Cohen Mansfield Agitation Inventory (CMAI) 21 assesses agitation across three domains: aggressive and physically non-aggressive behaviour, and verbally agitated behaviour.The CMAI comprises 29 items and uses a 7-point Likert scale (Never [1], Less than once a week [2], Once or twice a week [3], Several times a week [4], Once or twice a day [5], Several times a day [6], Several times an hour [7]).A total is calculated, with a maximum score of 203, where a higher score indicates a higher frequency of behavioural disturbances.
• The Quality-of-life Alzheimer's Disease (QOL-AD) 2 measures QOL aspects and consists of 13-items, using a 4-point Likert scale (poor [1], fair [2], good [3] and excellent [4]).The QOL-AD includes both self and proxy report and measures four domains (physical, mental health, social and function) for people with a wide range of dementia severity.With a total maximum score of 52, a higher score indicates a higher QOL.• The Abbey Pain Assessment Scale (APAS) comprise of six items, assessing vocalisation, facial expression, body language and behavioural, physiological and physical change. 22The APAS uses a 4-point Likert scale (Absent [0], Mild [1], Moderate [2], Severe [3]) with a maximum total score of 18.The pain severity is indicated as mild (3-7), moderate (8-13), or severe (14+).

| Impact evaluation
The qualitative phase consisted of two one-on-one interviews (pharmacists and clinical nurse manager), two focus groups (clinical nursing staff and family members), emails (six participants) and 16 presurvey (n = 14)/postsurvey (n = 2) surveys to understand perceptions of CBM.Six questions relating to positive and negative observations, knowledge and changes in perceptions of CBM were asked during the interviews and focus groups.

| Data analysis
The analysis of this study followed the CONSORT 2010 check list for crossover designs. 23These data were screened using IBM SPSS Statistics (version 28) and missing value analysis was applied to missing values in the QoL-AD.Mean and standard deviations (SD) were computed for continuous data (age [years], body weight [kg], NPI-NH, CMAI, QoL-AD, APAS total scores) and frequency distributions (n, %) for categorical (sex, medications) data.
Baseline comparisons between gender and dementia severity groups were assessed via chi-squared or Fisher's exact tests and t-tests or Mann-Whitney U tests for scored data.
Generalised mixed effect models in Stata (Version 17.0; StataCorp) were used to examine the outcomes longitudinally.Mixed effects logit models were used to model binary (indicator) outcomes (NPI-NH) and mixed effects summarised linear models were used to model scale outcomes (CMAI, QoL-AD, APAS).The models were adjusted for treatment differences (placebo/CBM), age, number of medications, total cumulative dose, sex (male/female) and summarised on order (CBM 1st/CBM 2nd).The results were summarised as estimated marginal mean differences and 95% confidence intervals.
QSR NVivo (version 12) was used to manage and store the qualitative data.These data were transcribed verbatim and thematically analysed using (1) an open coding process and (2) line-by-line coding to identify key themes.A triangulated approach, combining all data, was undertaken to avoid bias.Two qualitative experts (AT and CB) discussed the key categories to reach a level of agreement.

| RESULTS
A total of 21 residents, (23% male, 77% female) with an average age of 85 years participated in the study.Men were taller and the women were taking a larger number of medications.No differences were observed between the sexes at baseline, although there was a trend where men had raised levels of non-physical aggression and verbal agitation (Table 1).Two participants withdrew before completing both treatment cycles.One woman was hospitalised for an unrelated incarcerated incisional hernia, and the family of one male withdrew consent at the start of the second treatment cycle due to observing significant behavioural changes.Review of the data revealed this participant was receiving CBM at this time.The data collected from the male participant were included in the analysis but not the female participant, as she withdrew before commencement of a treatment cycle.

| Administered daily dose
Eighteen of the 21 (86%) participants reached the maximum daily dose (50 mg THC/17 mg CBD/day) when receiving the CBM.The overall average CBM dose was 30.1 mg THC /20.5 mg CBD per day.The maximum cumulative dose across the study was 1700 mg of CBM; however, due to the drug being titrated for each patient, the average cumulative dose was 1263.3 mg of CBM (75% of the maximum).

| Adverse effects
The incidence of the 16 AEs monitored are shown in Table 2. Fifty-four per cent of all AEs occurred during the titration phase.The number of AEs was significantly higher with CBM (57%, χ 2 = 17.5, p < 0.001).However, no statistical differences were observed between the placebo and CBM treatments at the individual AE level.
The titration period for each participant ranged from 2 to 4 weeks, informed by the onset of an AE.Of the 21 participants, 10 (47%) participants did not experience an AE and reached the maximum dose, and two did not experience any AEs but did not reach the maximum dose.Only nine (43%) reported an AE.The majority of AEs occurred on Days 5 and 8 (Week 1-2) during the titration phase (33%) although some AEs occurred on Day 1 (Week 1), after commencing the medication, or as late as Day 17 (week 3).Of the nine participants who reported an AE, 88% had two or less effects, while 11% reported four or more.From the nine participants reporting an AE, four (44%) increased their dose of CBM, while five (55%) reached the maximum dose.

CBM and placebo treatment cycles
The adjusted mean and SD by treatment (at the end of titration, treatment and the washout phases) for all the outcome measures (NPI-NH, CMAI, QoL-AD and Pain) and subscales of the CMAI (aggressive, non-aggressive and verbal agitated) are shown in Figure 1.No significant differences were seen for any of the measures or were observed between-group/between group-time interactions when participants were receiving the placebo or CBM.

| Comparison of individual characteristics of BPSD in treatment groups
The 12 NPI-NH domains (10 behavioural areas and two types of neurovegetative changes) estimated differences between the adjusted means are shown in Figure 2. Agitation was the only behaviour that significantly decreased at the end of the treatment cycle.

| Post hoc analysis
Post hoc estimates based on our data suggest the study needed 50 participants (accounting for 10% attrition rate) to detect a change for the Total NPI-NH (mean difference of 6 on the Total NPI-NH scale, Power 80%, Alpha 0.05, Beta 0.2).To examine individual differences on the 12 domains of the NPI-NH, a total of 150 participants would be needed (mean difference of 3.0, Power 80%, Alpha 0.05, Beta 0.2).

| Qualitative findings
The qualitative data were collected at the end of the second treatment cycle when the family, aged care staff and researchers were still blinded to the medication allocations.Participants were asked to describe any differences they noticed between treatment cycles.

| Perceptions of CBM on dementia before the trial
The presurveys indicated that families and staff wanted to observe changes in: 'anxiety', 'depression', 'aggression (verbal and physical)', 'pain', 'agitation', 'paranoia', 'appetite' and 'infection'.Many felt positive towards CBM use to manage BPSD.One nurse was unsure about CBM as she had not studied this medication before.

| Observations during the trial
The key categories that emerged were improvements or lack of improvements.the view that sleep and relaxation were the most common improvements, whereas no improvements were attributed to the concomitant of medications and complexities associated with dementia that manifest as pain, cognitive decline and agitation.

| DISCUSSION
Qualitative feedback from families and staff of RACF was positive, although only a reduction in agitation was significant at the end of the CBM treatment cycle.This study did not detect any other psychosocial changes attributable to CBM, although some general trends were seen.While we cannot dismiss this finding, the lack of statistical power in the study may have attributed to this result.Initial sample estimates from published data by van den Else et al. 24 suggest a total of 50 participants were required to detect a change in the total NPI-NH score.The recruitment process took longer than anticipated and was significantly impacted by the coronavirus (COVID-19) pandemic causing delays with medication delivery, lack of access into RACF, changes in nursing routines and restrictions to family visits.These changes resulted in the trial extending for over 2.5 years.The initial study approval was only granted for those with mild dementia due to the GAA being silent on consent to medical research.Fortunately, after the first two participants completed the trial, changes to this Act were approved allowing the next of kin to consent on behalf of someone with a moderate-to-severe cognitive decline.This approval allowed the researchers to expand their participant inclusion.Martin et al. 25 published ethical considerations to help researchers navigate CBM trials.Despite this, only 21 participants enrolled into the study.Post hoc estimates from our data support the initial estimate of 50 participants required to detect changes in the Total NPI-NH.Significant changes across the 12 domains on the NPI-NH would require 150 participants.This study is still one of the largest to date with this population group, with other studies ranging from two 26 to 39 participants. 16he crossover design is considered to be robust and was selected for this study 23 ; however, the use of self-reported questionnaires may have limited the ability to detect small changes related to CBM use.For example, the CMAI asks participants to observe the frequency of behaviours over the last 2 weeks, 21 where the NPI-NH assesses the presence of a behaviour. 20These self-reported questionnaires were selected as they are the most robust and considered the gold standard in RACF. 16,27To the authors' knowledge, no BPSD questionnaires ask whether behaviours have changed due to the level of care or treatment provided. 9erhaps future studies should consider combining the use of questionnaires with new technologies, such as wearable devices that provide real-time feedback when detecting changes in movement to improve the accuracy in observations. 28Due to RACF enforcing lock-out restrictions during COVID-19, the researchers reduced the number of intended visits from twice a week to collect the survey responses.Once restrictions were eased, the researchers resumed their twice-weekly visits.Similarly, limited family visits during the COVID-19 period may have led to uncertainty when reporting changes in behaviour.As such, changes in a participant's behaviour during the course of the day reported by others 16 may have been missed.Furthermore, the complexity in reporting was compounded by the severity of the BPSD(s) present, the fluidity of the symptoms (e.g.sundowning), and the severity of the dementia masking the effects of the medication. 2 This made it difficult to observe whether changes were the result of the medication or the dementia.Future studies could assess the well-being of the families and caregivers of those living with dementia as these compounding factors may impact their primary care and support.
Another consideration from this study was the large number of medications the participants were already taking, masking any effects of CBM.The polypharmacological effect from the different types of medication (e.g.benzodiazepine, antidepressants and antipsychotics) may have impacted the ability for CBM to interact with the endocannabinoid system, 14,18 as some medications used to treat BPSD are metabolised within the liver and interact with chromosome CPY540, 29 which is similar to CBM.Participants of this study were generally taking 10-15 different types of medication.While the ultimate hope is that medical cannabis can be used to replace a number of these, it was decided to request that participants did not change their prescribed medications during the course of the study.This decision was informed by van den Elsen et al. 7 and the hope that it would assist in the interpretation of the results.The 6-week duration provided enough time to establish the safety and efficacy of the drug 12 as no severe AEs were detected.Future studies should consider a longer duration for the medication administration to allow for small adjustments to previously prescribed medications that interact with chromosome CPY540 to occur. 29The study duration was also shorter than the treatment required in the management of BPSD, leading to inconclusive effects of this medication.
A limitation from the original protocol 19 was that the medication administration (9 am and 12 pm) was scheduled to fit in with the medication routines already established in RACF.New trials exploring CBM use should consider a later administration of the second dose to improve sleep hygiene and afternoon energy levels.In addition, the staff had minimal experience with CBM use.The pharmacist demonstrated the administration (e.g.priming the bottle) of the medication on the first day of administration; however, further information sessions may ease concerns with administering CBM.This would help lower reservations regarding CBM use and the proposed psychoactive effects of THC.This study demonstrated CBM with the addition of THC appeared safe and tolerated.The individual effects of CBM on each participant were not explored.For example, the effectiveness of the medication by classifying individuals as a T A B L E 3 Description and supporting quotes for key themes and subthemes summarised from this study.responder or a non-responder would benefit this field of research and supports a crossover design. 23inally, the limited effect seen in this study may be related to the formulation of medication (3:2, THC:CBD).While a consensus exists that the best result is a mixture of THC and CBD, 14 the actual composition of the product remains unclear. 25,29In many instances, a balanced ratio of 1:1 for THC:CBD (30) or a higher ratio of CBD are being used: for example, a study is examining the effects of CBD among those with mild dementia.Therefore, further studies are needed to compare various products and to explore the use of minor cannabinoids such as cannabigerol (CBG) as this compound develops earlier allowing for timely harvest, does not hold psychotropic effects and provides neuroprotection to the brain. 30

| Study strengths and limitations
In addition to the challenges faced in this study in obtaining good quantitative data, these trials may suffer from observer bias, meaning some behavioural changes (e.g.sundowning) may not have been captured due to these occurring outside visiting hours.The researchers mitigated this by inviting family members and staff to qualitative sessions, although bias cannot be ruled out.The small sample may have increased the possibility of a false-negative result, so these results need to be considered with caution.Biological markers to collect pharmacodynamic and pharmacokinetic information were not collected in this study.This decision was made due to the age of the participants and the intrusive nature of these types of tests.Further studies should consider the addition of collecting biological markers.Despite these challenges, the study was robustly designed and larger than many studies in this area.

| CONCLUSIONS
This study demonstrated that a maximum of 50 mg THC/34 mg CBD per day was safe and well-tolerated in individuals living with dementia.While the study failed to identify significant differences in BPSD, QOL and pain level, CBM was associated with a significant reduction in agitation among residents.Qualitative analysis demonstrated that many family members and staff felt that the medication assisted some participants in becoming more relaxed and improved sleep quality.Due to recruitment difficulties during COVID-19, these results encourage further studies of this type that involve 50 participants or more.

ACKNO WLE DGE MENTS
The research team would like to thank the participating aged care facilities, residents and their family members for their involvement in the study.In addition, this study would have not been possible without the involvement of the medical prescriber, research nurse and pharmacy.Open access publishing facilitated by The University of Notre Dame Australia, as part of the Wiley -The University of Notre Dame Australia agreement via the Council of Australian University Librarians.

CONFLICT OF INTEREST STATEMENT
This research is financially supported by MGC Pharmaceuticals Ltd., an Australian registered company with global connections that specialise in the manufacture of cannabis-based medicine (CBM) in its Good Manufacturing Practice (GMP)-certified laboratory, but they were not involved in the data collection, analysis or interpretation of the data.The authors declare that, as there may be a perceived conflict of interest of undertaking research funded by MGC Pharmaceuticals Ltd., the manufacturer of the drug used in the trial, a number of precautionary steps have been implemented to minimise these.These included obtaining input on study design and drug doses in the development of the research protocol, and obtaining Therapeutic Goods Administration (TGA) approval for provision of the drug.All MGC Pharmaceutical staff were excluded from the trial itself including the data collection, analysis and interpretation of the data.Joint management meetings only discussed the progress of the study in broad terms to ensure compliance with budgetary issues and appropriate responses to any serious adverse events.The University of Notre Dame Australia (UNDA) has written permission from MGC Pharmaceuticals Ltd., for a worldwide non-exclusive, royalty free license to use the Project Intellectual Property for non-commercial research purposes including consent to publish research findings regardless of the results.The results will be disseminated via brief reports provided to the participating aged care facilities, through manuscript publications and conference presentations.

DATA AVAILABILITY STATEMENT
The 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.ORCID Amanda Timler https://orcid.org/0000-0003-2619-5937

F I G U R E 2
Estimated proportion from generalise linear model with probabilities for difference between the two treatment arms at each indicated time point.
Baseline information for participants by gender.Sex difference comparisons between men and women include chi-squared or Fisher's exact tests for categorical data and t tests or Mann-Whitney U tests for continuous data.Bolded depicts a statistically significant difference between men and women.
T A B L E 1Note: Adverse effects grouped by placebo and cannabinoid-based medicines (CBM) during the study.
T A B L E 2 F I G U R E 1 Scale outcomes by treatment with adjusted mean difference estimates at each indicated time point.

Table 3
presents quotes supporting She just sleeps, even when you just go and visit her, she would just be sleeping… and then she complained about being sleepy all the time!(Focus group, family member of those living with dementia).