Medical Cannabis for the Management of Pain and Quality of Life in Chronic Pain Patients: A Prospective Observational Study

Abstract

Objective

To evaluate the short-term and long-term effects of plant-based medical cannabis in a chronic pain population over the course of one year.

Design

A longitudinal, prospective, 12-month observational study.

Setting

Patients were recruited and treated at a clinic specializing in medical cannabis care from October 2015 to March 2019.

Subjects

A total of 751 chronic pain patients initiating medical cannabis treatment.

Methods

Study participants completed the Brief Pain Inventory and the 12-item Short Form Survey (SF-12), as well as surveys on opioid medication use and adverse events, at baseline and once a month for 12 months.

Results

Medical cannabis treatment was associated with improvements in pain severity and interference (P < 0.001) observed at one month and maintained over the 12-month observation period. Significant improvements were also observed in the SF-12 physical and mental health domains (P < 0.002) starting at three months. Significant decreases in headaches, fatigue, anxiety, and nausea were observed after initiation of treatment (P ≤ 0.002). In patients who reported opioid medication use at baseline, there were significant reductions in oral morphine equivalent doses (P < 0.0001), while correlates of pain were significantly improved by the end of the study observation period.

Conclusions

Taken together, the findings of this study add to the cumulative evidence in support of plant-based medical cannabis as a safe and effective treatment option and potential opioid medication substitute or augmentation therapy for the management of symptoms and quality of life in chronic pain patients.

Medical, Cannabis, Pain, Quality of Life, Opioids

Introduction

Chronic pain is one of the leading causes of disability worldwide, estimated to affect 15–30% of the adult population [1–3]. It poses considerable burden on the quality of life and psychosocial function of patients, as well as significant socioeconomic impacts [4, 5]. In the United States alone, the combined costs associated with health care and lost productivity have been estimated to exceed $500 billion per year [6].

For many chronic pain patients, conventional pharmaceutical medications are ineffective and/or not well tolerated. Opioid medications, in particular, have high abuse potential, questionable long-term efficacy, and have been associated with alarming rates of overdose and death [7–10]. In the United States, it has been estimated that 3.8 million adults misuse prescription opioid medications (POMs), which, due to their high abuse potential, can progress to heroin use [7, 10]. In 2015, POMs were associated with 15,000 deaths; in 2016, that figure more than doubled, with opioid misuse accounting for more than 40,000 deaths [10, 11].

In the midst of an ongoing opioid crisis, there is an urgent need for effective and safer treatment alternatives for chronic pain. With a high margin of safety and postulated regulatory effects of tetrahydrocannabinol (THC) and cannabidiol (CBD) on major endogenous pain circuitry systems, cannabis is emerging as a promising alternative or adjunctive treatment option for the management of chronic pain [12–15].

The analgesic effects of cannabis and cannabinoids have been demonstrated in preclinical as well as clinical studies, although the evidence has been at times inconclusive due to the heterogeneity of study designs, study populations, and study medications. Moreover, many clinical studies have been limited by relatively small samples and short duration and have mostly focused on neuropathic pain [15]. As such, there is a need for large-scale longitudinal studies to evaluate the effects of medical cannabis in chronic pain.

The aim of this study was to evaluate the short-term and long-term effects of plant-based medical cannabis (MC) on outcomes of interest related to pain, quality of life, tolerability, and opioid medication use in a large cohort of chronic pain patients using medical cannabis over the course of one year.

Methods

Study Design

This was a prospective, observational, longitudinal, open-label study design to evaluate the effectiveness of MC in the management of chronic pain symptomatology. Patients were recruited at Apollo Cannabis Clinics in Toronto, Canada, from October 2015 to March 2019. Patients who provided signed informed consent were asked to complete study questionnaires at baseline (before commencing treatment with MC) and then once a month for a total of 12 months (i.e., month 1 follow-up questionnaires, month 2 follow-up questionnaires, etc.). The study was approved by the Advarra Institutional Review Board and was conducted in accordance with research regulations, applicable international standards of Good Clinical Practice, institutional research policies and procedures, and the ethical principles laid out in the Declaration of Helsinki.

Study Population

The study population was comprised of chronic pain patients who were seeking treatment at Apollo Cannabis Clinics and were prescribed MC following an initial clinical evaluation of medical history and health status. Patients were included in the study based on the following inclusion criteria: 1) ≥25 years of age, 2) willing and able to provide informed consent and complete study assessments, 3) chronic pain for three or more months, and 4) newly prescribed MC for the treatment of chronic pain. Patients were excluded if they had used MC in the past three months.

Treatment Regimen

As part of the routine treatment process in the clinic, all patients underwent a clinical evaluation of medical history, health status, and symptomatology by the attending physician and were then advised on a cannabis-based treatment plan, as well as alternative non-cannabinoid-based treatment plans. The prescribing methodology included a personalized treatment plan with specific dosing instructions, route of administration, and a set THC limit, depending on individual medical histories, needs, and tolerability.

Outcomes of Interest

The baseline questionnaire captured demographic information such as age, sex, and medical history, the pain condition for which MC had been prescribed, and the duration of the condition. In addition, the following outcomes of interest were collected at baseline and follow-up:

Brief Pain Inventory

A self-administered rating scale that captures Pain Severity (0=“no pain” to 10=“pain as bad as you can imagine”) and Pain Interference (0=“does not interfere” to 10 =“completely interferes”) [16]. It has high psychometric value, with a Cronbach’s alpha of 0.84–0.94 [17], an intraclass correlation coefficient of 0.84–0.90, and a kappa value of <0.70 [18].

Short Form Health Survey

The Short Form Health Survey is a self-reported, multi-item scale questionnaire that evaluates quality of life. Derived from the SF-36, the 12-item Short Form Survey (SF-12) is a shorter version from which the Physical Composite Summary (PCS) and Mental Composite Summary (MCS) scores are calculated, ranging from 0=“lowest level of health” to 100=“highest level of health” [19, 20]. It has been reported to have a Cronbach’s alpha coefficient of 0.72–0.89 and a test–retest reliability of 0.73–0.86 [21].

Safety Measures

Study participants were asked to rate the frequency (“never,” “rarely,” “sometimes,” “frequently”) at which they experienced the following undesired symptoms at baseline and follow-up: fast heartbeat, dizziness, dry mouth, disorientation, increased appetite, drowsiness, feeling faint, fatigue, headache, impaired memory, confusion, feeling “high,” suspiciousness, nervousness, feelings of anxiety, paranoia, imbalance, throat irritation, hallucinations, impairment in motor skills, coughing, nausea, and vomiting.

Opioid Medications

Patients taking opioid medications were asked to provide the name and daily dose of opioid medications they were taking. In order to allow for comparisons across patients taking different types of opioid medications, reported daily opioid doses were converted into Oral Morphine Equivalents (OMEs), as described by Nielsen et al. [22].

Time Point Definition

Given that this was an observational study with no monetary incentives for completion of questionnaires, fewer participants completed their questionnaires at each subsequent monthly follow-up. To reduce the use of variables with missing data, analyses were conducted only on key time points—namely, baseline, month 1, month 3, month 6, and month 12 follow-up data. It should also be noted that many study participants did not complete the monthly follow-up questionnaires in a timely manner. As such, some patients completed their month 1 follow-up questionnaire at month 1 post–treatment initiation, while others may have completed it at month 3 post–treatment initiation. In order to make these data comparable regardless of when the questionnaires were completed, collected data were categorized into specific time points based on how many days had passed since the completion of the baseline report. As such, follow-up time points were defined as follows:

• Baseline (B): day 0, before initiating MC treatment

• Month 1 (M1): 20 to 54 days passed since completion of baseline report

• Month 3 (M3): 85 to 114 days passed since completion of baseline report

• Month 6 (M6): 175 to 204 days passed since completion of baseline report

• Month 12 (M12): 355 to 385 days passed since completion of baseline report

Statistical Analysis

All statistical analyses were conducted using IBM SPSS Statistics for Mac (version 25.0; IBM Corp, Armonk, NY, USA) and XLSTAT Basic for Mac (Addinsoft, Long Island City, NY, USA). For the purposes of this manuscript, only the data collected at B, M1, M3, M6, and M12 were included in the analyses. Normally distributed continuous variables are presented as means with SDs, and non–normally distributed continuous variables are presented as medians, except where noted. Categorical variables are presented as counts and/or as valid percentages of the total. Continuous variables were checked for outliers by inspection of boxplots and for normality using the Shapiro-Wilk test. Changes over time in normally distributed continuous variables were analyzed with one-way repeated-measures analyses of variance (ANOVAs) and, where applicable, Bonferroni post hoc tests; changes over time in non–normally distributed continuous variables were analyzed with the Skillings-Mack test, a nonparametric test that is equivalent to the Friedman test, for use when there are missing data [23].

Exploratory analyses were conducted to evaluate differences between patients who completed all five time point visits (i.e., B, M1, M3, M6, and M12) and patients who completed only four time point visits (B, M1, M3, M6), patients who completed three time point visits (B, M1, M3), and those who completed only two time point visits (B and M1). Categorical variables are presented as valid percentages of the total, and associations between categorical variables, such as gender distribution across time point groups, were analyzed with chi-square tests.

Measures of safety, which comprised categorical response options, were converted into dichotomous variables, where the answer options “never” and “rarely” were combined into the category “never to rarely,” and the answer options “sometimes” to “frequently” were combined into the category “sometimes to frequently.” These variables were tested for the first and second assumptions of sample size adequacy and analyzed with Cochran’s Q test. Where applicable, McNemar’s post hoc tests with Bonferroni corrections were conducted on B, M1, M6, and M12, with statistical significance set at P < 0.0083 for six pairwise comparisons [24].

Results

Study Demographics

During the study period, a total of 1,245 patients provided informed consent and completed questionnaires. However, 494 (40%) completed only the baseline report but no other follow-up reports and, thus, were excluded from the analyses. Only patients who completed the baseline reports and at least one of the follow-up reports (i.e., month 1 and/or month 3 and/or month 6 and/or month 12) were included in the study sample, which resulted in a final study population of 751 (60%) at baseline.

The age of the patient population ranged from 25 to 88 years and was normally distributed, with a mean age of 49.6 ± 14.3 years. The sex distribution was fairly balanced, with 57% female patients.

Study participants reported a heterogeneous array of pain conditions, with the five most commonly reported conditions being back pain (including those with sciatica/neuropathic pain), osteoarthritis, chronic headaches, fibromyalgia, and degenerative disc disease (Table 1). Other less commonly reported pain conditions included pelvic pain, diabetic neuropathy, spinal cord injury, peripheral neuropathy, myofascial pain, and endometriosis, among others. The duration of these pain conditions ranged from three months to >10 years. Moreover, 47% of patients reported having one pain condition, while the rest reported suffering from two or more pain conditions (Table 1).

Table 1

Characteristics of pain conditions afflicting the study population at baseline

Primary pain condition, No. (% of intake total)	Back pain	351 (46.7)
	Osteoarthritis	214 (28.5)
	Chronic headaches	160 (21.3)
	Fibromyalgia	132 (17.6)
	Degenerative disc disease	123 (16.4)
	Sciatica	84 (11.2)
	Rheumatoid arthritis	81 (10.8)
	Herniated disc	73 (9.7)
	Musculoskeletal pain	72 (9.6)
	Spinal stenosis	56 (7.5)
No. of concurrent pain conditions, No. (%)	1	353 (47)
	2–5	364 (48.5)
	6–10	31 (4.1)
	>10	3 (0.4)
Duration of pain conditions, No. (%)	<6 mo	74 (9.9)
	6–11 mo	26 (3.5)
	1–5 y	206 (27.4)
	5–10 y	173 (23)
	Over 10 y	272 (36.2)

Prior Experience and Medical Cannabis Use at Follow-up

At baseline—before initiating MC treatment—60% of the study sample consisted of patients who had no previous experience with cannabis or had only used when younger (Table 2). Patients were prescribed medical cannabis with dosing instructions, route of administration, and set THC limits based on patient-specific needs and tolerability. The reported mean daily doses (in grams of dry flower per day or oil equivalent) were normally distributed at M1, M3, and M6 (P > 0.4), but not at M12 (P = 0.004), and did not differ significantly across time points (F(3,15)=0.798, P = 0.514) (Table 2). Data on exact THC and CBD doses were not collected in the study. However, the MC consumed by the study population typically ranged from 7% to 29% THC and/or CBD.

Table 2

Cannabis use before and after commencing

Cannabis experience at baseline, No. (% of intake total)	No experience	264 (35.2)
	Used when younger	187 (24.9)
	Light to moderate user	134 (17.8)
	Frequent user	130 (17.3)
	Not reported	36 (4.8)
Daily doses of MC, mean ± SD, g/d	Month 1	1.33 ± 0.41
	Month 3	1.25 ± 0.52
	Month 6	1.33 ± 0.61
	Month 12	1.50 ± 0.55

MC = medical cannabis.

Correlates of Pain

The Brief Pain Inventory (BPI) was used to calculate Pain Interference (BPI-I) and Pain Severity (BPI-S), and changes over time were evaluated with one-way repeated-measures ANOVAs. There were no outliers in BPI-I scores, and the data were normally distributed (P > 0.1). BPI-S scores were also normally distributed across time points (P > 0.2), but there was one outlier at M1 and one outlier at M3. Both of these outliers were within the upper inner fence (i.e., mild outliers), and upon examination of each outlier, it was determined that there were no data collection, entry, or calculation errors associated with these scores. As such, both BPI-S outliers were included in the analyses.

The assumption of sphericity was met for both BPI-I and BPI-S scores, as assessed by Mauchly’s test of sphericity (χ²(9)=14.34, P = 0.112, and χ²(9)=6.30, P = 0.711, respectively). Treatment with MC was found to be associated with significant changes in BPI-I scores (F(4, 84)=8.99, P < 0.0005, partial η² =0.30) and in BPI-S scores (F(4, 84)=9.98, P < 0.0005, partial η² =0.32). Specifically, significant score reductions were observed from B to M1, B to M3, B to M6, and B to M12 (Table 3).

Table 3

Measures of pain interference and pain severity as per the Brief Pain Inventory

	BPI-Interference		BPI-Severity
	Mean ± SD	95% CI, P Value	Mean ± SD	95% CI, P Value
Baseline (N=706)	6.23 ± 1.63		5.58 ± 1.53
Month 1 (N=584)	4.55 ± 2.39*	0.48 to 2.88, 0.003	4.27 ± 1.90*	0.47 to 2.14, 0.001
Month 3 (N=230)	4.08 ± 2.97*	0.34 to 3.96, 0.013	3.89 ± 2.17*	0.55 to 2.83, 0.001
Month 6 (N=105)	4.21 ± 2.64*	0.45 to 3.58, 0.006	3.99 ± 2.18*	0.30 to 2.87, 0.009
Month 12 (N=43)	3.54 ± 2.84*	0.92 to 4.46, 0.001	3.49 ± 2.17*	0.90 to 3.27, <0.001

Treatment with medical cannabis was found to be associated with significant changes in Brief Pain Inventory measures of pain interference (F(4, 84)=8.99, P < 0.0005, partial η2 =0.30) and pain severity (F(4, 84)=9.98, P < 0.0005, partial η2 =0.32).

BPI = Brief Pain Inventory; CI = confidence interval.

*

Denotes statistical significance in relation to baseline.

The scores of the individual domains of the BPI-I subscale (i.e., General Activity, Mood, Walking Ability, Normal Work, Relationships with Other People, Sleep, and Enjoyment of Life) were also examined. Shapiro Wilk’s tests indicated that BPI-I subdomain scores were not normally distributed across all the time points, and thus they were analyzed with Skillings-Mack tests. All the subscores of the BPI-I domain were found to have a computed P value <0.01, indicating statistically significant improvements over the time. A detailed breakdown of the BPI-I subscores at each time point of interest is displayed in the Supplementary Data.

Moreover, given that a proportion of patients reported cannabis use before initiation of MC treatment, two-way repeated-measures ANOVAs were also conducted to evaluate the effects of time and prior cannabis experience/frequency of use on BPI scores. As expected based on the results presented in Table 3, there were statistically significant changes in BPI-I (F(4,72)=9.90, P < 0.0001) and BPI-S scores (F(4,72)=9.94, P < 0.0001) over the course of the treatment observation period. However, there were no statistically significant interactions between time and prior cannabis experience/frequency of use for either BPI-I (F(12,72)=0.79, P = 0.655) or BPI-S (F(12,72)=0.78, P = 0.668) (Supplementary Data).

Quality of Life

The PCS and MCS scores derived from the SF-12 were used as a measure of quality of life. Inspection of scores revealed one upper inner fence and one upper outer fence PCS outliers at M1, two upper inner fence PCS outliers at M3, and one lower inner fence MCS outlier at M6. Each outlier was inspected and determined not to be associated with data collection, entry, or calculation errors, and as such, all five outliers were included in the analyses. Shapiro Wilk’s tests indicated that PCS and MCS data were not normally distributed across all time points of interest. As such, score changes over time were evaluated with Skillings-Mack tests. For both PCS and MCS scores, the computed P values were <0.05, indicating statistically significant improvements over time (Table 4).

Table 4

Physical and mental health–related measures of quality of life as per the Short Form Health Survey

		Mean ± SD	Median	Sum of Ranks	Mean of Ranks	Test Statistics
Physical Composite Summary score	Baseline (N=509)	31.21 ± 8.09	30.18	911	2	Q (observed value) = 18 Q (critical value) = 9 DF = 4 P < 0.05
	Month 1 (N=435)	32.99 ± 9.90	31.10	992	2
	Month 3 (N= 148)	34.05 ± 9.42	31.80	993	7
	Month 6 (N=69)	34.44 ± 10.79	31.71	986	14
	Month 12 (N=31)	33.12 ± 11.12	31.26	977	32
Mental Composite Summary score	Baseline (N=509)	42.83 ± 11.53	41.59	915	2	Q (observed value) = 17 Q (critical value) = 9 DF = 4 P < 0.05
	Month 1 (N=435)	46.55 ± 11.39	48.12	1,006	2
	Month 3 (N=148)	47.26 ± 11.23	48.40	989	7
	Month 6 (N=69)	45.36 ± 11.74	45.92	973	14
	Month 12 (N=31)	51.05 ± 9.42	50.09	976	31

Treatment with medical cannabis was found to be associated with significant changes in the Physical and Mental Health domains of the SF-12 over the course of the 12-month observation period.

DF = degrees of freedom; MCS = Mental Composite Summary; PCS = Physical Composite Summary; SF-12 = Short Form Health Survey.

Adverse Events

Study patients were asked to rate the frequency at which they experienced undesired symptoms commonly associated with cannabis use (Table 5, Figure 1).

Figure 1

Frequencies of select adverse effects at baseline and over the course of the study observation period. A) There were no statistically significant changes in the frequencies at which study subjects experienced adverse events commonly associated with cannabis use, such as dry mouth, increased appetite, drowsiness, or impaired memory. B) There were statistically significant reductions in the frequencies at which study subjects experienced fatigue, headaches, feelings of anxiety, and nausea. A full breakdown of adverse event frequencies is displayed in Table 5. *Statistical significance in relation to baseline.

Table 5

Frequency of adverse events at baseline and over the course of the 12-month study observation period

		Baseline (N = 751), %	Month 1 (N = 628), %	Month 3 (N = 248), %	Month 6 (N = 112), %	Month 12 (N = 46), %
Fast heartbeat	Never to rarely	75.2	80.9	83.5	85.7	89.1
	Sometimes to frequently	24.8	19.1	16.5	14.3	10.9
Dizziness	Never to rarely	69.1	73.9	74.2	82.1	82.6
	Sometimes to frequently	30.9	26.1	25.8	17.9	17.4
Dry mouth	Never to rarely	53.4	45.2	47.2	45.5	45.7
	Sometimes to frequently	46.6	54.8	52.8	54.5	54.3
Disorientation	Never to rarely	88.8	90.4	91.5	96.4	93.5
	Sometimes to frequently	11.2	9.6	8.5	3.6	6.5
Increased appetite	Never to rarely	64.3	54.5	54.4	60.7	43.5
	Sometimes to frequently	35.7	45.5	45.6	39.3	56.5
Drowsiness	Never to rarely	50.1	50.6	48	50.9	65.2
	Sometimes to frequently	49.9	49.4	52	49.1	34.8
Feeling faint	Never to rarely	84.2	89.6	89.5	93.8	95.7
	Sometimes to frequently	15.8	10.4	10.5	6.3	4.3
Fatigue	Never to rarely	24	43.3	44.8	50	52.2
	Sometimes to frequently	76	56.7*	55.3	50*	47.8*
Headache	Never to rarely	40.7	60.4	56.5	62.5	58.7
	Sometimes to frequently	59.3	39.6*	43.5	37.5	41.3
Impaired memory	Never to rarely	65	76.4	76.2	77.7	82.6
	Sometimes to frequently	35	23.6	23.8	22.3	17.4
Confusion	Never to rarely	82	90	90.3	92.9	91.3
	Sometimes to frequently	18	10	9.7	7.1	8.7
Feeling high	Never to rarely	82.8	69.4	66.9	68.8	65.2
	Sometimes to frequently	17.2	30.6	33.1	31.3	34.8
Suspiciousness	Never to rarely	91.6	95.2	97.6	98.2	97.8
	Sometimes to frequently	8.4	4.8	2.4	1.8	2.2
Nervousness	Never to rarely	66	82.5	80.6	83.9	93.5
	Sometimes to frequently	34	17.4	19.4	16.1	6.5
Feelings of anxiety	Never to rarely	62.3	78.7	81	83.9	93.5
	Sometimes to frequently	37.7	21.3*	19	16.1*	6.5*
Paranoia	Never to rarely	94.8	96.5	96.8	97.3	100
	Sometimes to frequently	5.2	3.5	3.2	2.7	0
Imbalance	Never to rarely	75.8	84.2	87.5	85.7	84.8
	Sometimes to frequently	24.2	15.8	12.5	14.3	15.2
Throat irritation	Never to rarely	83.8	83	80.6	72.3	76.1
	Sometimes to frequently	16.2	17	19.4	27.7	23.9
Hallucinations	Never to rarely	98.8	99.2	98	98.2	97.8
	Sometimes to frequently	1.2	0.8	2	1.8	2.2
Impaired motor skills	Never to rarely	87	92.7	92.3	87.5	91.3
	Sometimes to frequently	13	7.3	7.7	12.5	8.7
Coughing	Never to rarely	71.4	75.8	79	75	76.1
	Sometimes to frequently	28.6	24.2	21	25	23.9
Nausea	Never to rarely	71.2	83.3	85.1	82.1	89.1
	Sometimes to frequently	28.8	16.7*	14.9	17.9*	10.9
Vomiting	Never to rarely	92.5	97	96.8	95.5	95.7
	Sometimes to frequently	7.5	3	3.2	4.5	4.3

There were statistically significant reductions in the frequencies of fatigue (Q = 16.12, P = 0.002), headaches (Q = 10.51, P = 0.034), feelings of anxiety (Q = 15.85, P = 0.003), and nausea (Q = 17.04, P = 0.001). Bolded lines refer to AEs that demonstrated significant changes over time.

*

Denotes statistical significance in relation to baseline.

Table 6

Opioid medication doses, measures of pain, and quality of life in a subset of patients using opioid medications at baseline

		Mean ± SD	Sum of Ranks	Mean of Ranks	Test Statistics
OME*	Baseline (N =82)	26.23 ± 48.13	164	2	Q (observed) = 26 Q (critical) = 9 DF = 4 P<0.0001
	Month 1 (N =67)	12.69 ± 44.70	119	2
	Month 3 (N =26)	3.28 ± 8.62	128	5
	Month 6 (N =9)	3.00 ± 6.50	131	15
	Month 12 (N =4)	1.38 ± 0.97	134	34
BPI-S*	Baseline (N =76)	5.77 ± 1.77	150	2	Q (observed) = 15 Q (critical) = 9 DF = 4 P = 0.01
	Month 1 (N =65)	4.64 ± 2.23	117	2
	Month 3 (N =23)	4.58 ± 2.11	123	4
	Month 6 (N =9)	5.19 ± 2.21	130	14
	Month 12 (N =4)	4.06 ± 3.16	128	32
BPI-I*	Baseline (N =76)	6.40 ± 2.39	152	2	Q (observed) = 21 Q (critical) = 9 DF = 4 P = 0.0003
	Month 1 (N =65)	5.00 ± 2.87	124	2
	Month 3 (N =23)	4.41 ± 2.81	115	5
	Month 6 (N =9)	5.36 ± 2.95	131	15
	Month 12 (N =4)	3.86 ± 3.26	129	26
SF-12 PCS	Baseline (N =61)	30.24 ± 8.85	104	2	Q (observed) = 5 Q (critical) = 9 DF = 4 P = 0.29
	Month 1 (N =49)	31.54 ± 11.22	106	2
	Month 3 (N =17)	34.31 ± 8.77	114	7
	Month 6 (N =7)	34.18 ± 10.37	113	16
	Month 12 (N =3)	32.40 ± 8.65	112	37
SF-12 MCS	Baseline (N =61)	42.10 ± 11.46	106	2	Q (observed) = 3 Q (critical) = 9 DF = 4 P=0.48
	Month 1 (N =49)	47.21 ± 12.75	116	2
	Month 3 (N =17)	46.14 ± 13.84	112	7
	Month 6 (N =7)	41.62 ± 12.76	105	15
	Month 12 (N =3)	52.68 ± 7.85	110	37

Opioid medication doses were converted to oral morphine equivalents in mg/d to allow for comparisons across different opioid medication types.

BPI-I = Brief Pain Inventory–Pain Interference; BPI-S = Brief Pain Inventory–Pain Severity; SF-12 PCS = Short Form Health Survey–Physical Composite Summary Score; SF-12 MCS = Short Form Health Survey–Mental Composite Summary Score.

At M1, which was the first data collection time point after initiating MC treatment, the most commonly reported adverse events (e.g., reported as “sometimes to frequently” in at least 40% of the study sample) were dry mouth (reported by 54.8% of the study sample), increased appetite (reported by 45.5%), and drowsiness (reported by 49.4%). However, the frequencies at which these symptoms were reported at M1 were not statistically different from baseline.

On the other hand, there were statistically significant changes in the frequencies at which patients experienced fatigue (Q = 16.12, P = 0.002), headaches (Q = 10.51, P = 0.034), feelings of anxiety (Q = 15.85, P = 0.003), and nausea (Q = 17.04, P = 0.001). Post hoc analyses revealed that, compared with baseline, the frequencies of these symptoms decreased at follow-up. Specifically, the proportions of patients reporting fatigue decreased from B to M1, B to M6, and B to M12 (all at P ≤ 0.002); headaches decreased from B to M1 (P < 0.0005); feelings of anxiety decreased from B to M1, B to M6, and B to M12 (all at P < 0.0005); and nausea decreased from B to M1 and B to M6 (both at P ≤ 0.007).

Exploratory Analyses

Opioid Medication Use

Of the 751 study participants included in the analysis, a subset of patients reported taking opioid medications at baseline. This subset consisted of 82 patients, with an age range of 26 to 76 years and a mean age of 50.7 ± 13.4 years. Similar to the full study population, gender distribution was fairly balanced, with 63% females (N = 51). As noted in Time Point Definition, due to the observational nature of this study, each subsequent time point had fewer report completions. For this subset of patients, there were 82 report completions at baseline, 67 report completions at M1, 26 at M3, nine at M6, and four at M12.

Reported daily doses of opioid medications were converted to OMEs as described in Nielsen et al. (2016). Descriptions and tests statistics are displayed in Table 6. These data were not normally distributed and were thus analyzed with the Skillings-Mack test [23], which revealed statistically significant reductions on OME doses over the course of the treatment observation period.

Measures of pain and quality of life were also analyzed with the Skillings-Mack test due to non-normal distribution. These analyses revealed significant improvements in pain severity and pain interference, but not on the SF-12 measures of physical and mental health (Table 6).

Time Point Group Comparisons

Given the high rate of patients lost to follow-up at each subsequent visit, we sought to explore potential differences between patients who completed all five time point visits (i.e., B, M1, M3, M6, and M12; hereinafter referred to as the 5-TMP group) and patients who completed four time points (i.e., B, M1, M3, and M6; or 4-TMP), three time points (i.e., B, M1, and M3; or 3-TMP), or two time points (B and M1 only; or 2-TMP). There were no significant differences in baseline measures between patients who completed all five time point visits and patients who did not (Supplementary Data).

In addition, we evaluated the pattern of treatment response in patients in the 5-TMP, 4-TMP, 3-TMP, and 2-TMP groups by conducting individual repeated-measures ANOVAs on pain scores in each of these patient groups. These analyses were specifically conducted because repeated-measures ANOVAs are tests that exclude cases listwise; that is, only subjects who complete all the time points of interest are included in the analysis. For instance, a patient who completed only the B, M6, and M12 reports would be automatically excluded from an analysis in the 5-TMP group, because M1 and M3 reports were not completed. The BPI results presented in Correlates of Pain are based on analyses of patients who had completed all five time point reports, which included 22 patients only. As this is a small sample size, which considerably limits the generalizability of results, we sought to examine whether the pattern of response observed in the 5-TMP group was comparable to the pattern of response observed in the 4-TMP group (which included 64 patients), 3-TMP group (N = 184), and the 2-TMP group (N = 580). These analyses revealed similar patterns of response across time point groups, up to the last time point included in the analysis. That is, the 5-TMP group had significant changes in BPI scores from B to M1, B to M3, B to M6, and B to M12; the 4-TMP group had significant changes from B to M1, B to M3, and B to M6; the 3-TMP group had significant changes from B to M1 and B to M3; and the 2-TMP group had significant changes from B to M1 (Supplementary Data).

Discussion

Pain Management and Quality of Life

In this prospective, longitudinal, observational study of chronic pain patients, treatment with medical cannabis was associated with significant improvements in measures of pain and health-related quality of life.

Significant improvements in pain severity and pain interference were observed as early as one month after initiating treatment and were maintained over the course of the 12-month observation period during which patients who remained in the study were followed. The mean BPI-S score was 5.58 ± 1.53 at baseline, indicating moderate to severe pain, and 3.49 ± 2.17 at M12, indicating mild to moderate pain [18, 25, 26]. The 2.09-point reduction in pain severity over the course of the treatment observation period is indicative of a clinically significant improvement [27].

Although the analgesic effects of MC have been demonstrated in short-term studies of neuropathic pain [28–32], as well as in capsaicin-induced pain and hyperalgesia in healthy volunteers [33], fewer studies have been published on the long-term effects, particularly in non-neuropathic chronic pain. A PubMed search conducted at the time of the study, as well as inspection of reference lists in review articles and meta-analyses, revealed four other published long-term studies on the effects of plant-based cannabis in chronic pain patients. Bellneier et al. [34] conducted a three-month retrospective study of patients using THC/CBD oil capsules. Haroutounian et al. [35] conducted a six-month prospective open-label study of 206 patients using individually optimized doses of MC, and Ware et al. [36] conducted a one-year prospective cohort study of 215 patients using optimized doses of MC. In line with our findings, these studies found significant improvements in correlates of pain. On the other hand, Campbell et al. [15] conducted a large (N ≥ 1,000) four-year prospective cohort study and found no changes in measures of pain associated with cannabis. It should be noted, however, that participants in the study by Campbell et al. [15] had obtained illicitly produced and distributed cannabis, as data collection for this study occurred during 2012–2014, before the decriminalization of medicinal cannabis in Australia in 2016. This is important to note because the cannabinoid and terpene content in illicitly produced cannabis is often unknown to the consumer, and the product itself may contain contaminants, such as pesticides, mold, elevated bacteria levels, and heavy metals that could have potential harmful health effects [37–41]. As such, it is possible that the lack of medical benefits observed in Campbell et al. [15] may be related to the use of unregulated, nonmedical cannabis, rather than a lack of benefits of medical cannabis per se. Moreover, it is important that MC for therapeutic purposes be used under the guidance and supervision of a physician [42]. Although cannabis has a high safety index [42, 43], there are contraindications and precautions that must be taken into consideration for vulnerable populations, such as in those with a history of substance abuse, cardiovascular disease, respiratory disease, and/or anxiety disorders, among other conditions, as well as in patients taking medications that could lead to dangerous drug–drug interactions [44]. Additionally, preclinical studies have found that CBD, as well as co-administered CBD/THC, has biphasic effects, meaning that there is an optimal dose below and above which these cannabinoids are not as effective [45, 46]. More research needs to be conducted to better elucidate the complex mechanisms, as well as the pharmacodynamic and pharmacokinetic properties of cannabinoids, particularly in human populations. Nevertheless, taken together, the precautionary measures that need to be taken into consideration in vulnerable populations and the evidence on biphasic effects of cannabinoids indicate that MC for the purpose of pain management should be used under the care and supervision of a physician, with regular follow-ups to optimize the dose and to monitor issues of tolerability and adverse events.

Notably, our study included subjects who had been cannabis users before initiating MC treatment (light, moderate, and frequent users). As the data collection for the present study took place before the legalization of recreational cannabis in Canada in October of 2018, it is presumed that at least some of the study subjects who reported use before initiating MC treatment would have obtained cannabis through illicit channels, for which there was no information about the type of cannabis obtained, THC and CBD content, or the presence of contaminants. Interestingly, we found that the benefits of MC are beyond those of unregulated and illicitly obtained cannabis. This is evidenced by the findings presented in the Supplementary Data, which indicate significant improvements in mean BPI-S and BPI-I scores compared with baseline, but no significant interactions between the effects of treatment and prior cannabis experience/frequency of use on BPI-S and BPI-I scores (F(12,72)=0.78, P = 0.668, and F(12,72)=0.79, P = 0.655, respectively). This suggests that treatment with MC is associated with improvements in pain regardless of previous experience and/or use of nonmedical cannabis. There are a number of factors that could have contributed to this finding. First—and most importantly—study patients followed a personalized MC treatment plan under the guidance of a physician, with regular checkups to achieve an optimal therapeutic dose while minimizing adverse events due to issues of tolerability. Moreover, by virtue of following such a treatment plan, the psychological factor of guilt and stigma associated with obtaining illicit cannabis is presumed to have been absent or reduced, which in turn would have the effect of reducing anxieties about the use of cannabis, thus further contributing to overall clinical improvement. Finally, illicitly obtained cannabis does not undergo the same testing standards as MC, and inconsistencies in THC/CBD content, as well as cases of the contamination with mold, pesticides, elevated levels of potentially harmful bacteria exceeding Health Canada limits, and heavy metals such as lead—all of which can negatively affect overall health, especially in a population with compromised health—have been reported [37–41]. Taken together, these factors further highlight the importance of following an MC treatment plan under the guidance and supervision of a physician.

Pain alleviation is an important outcome of interest when considering the effectiveness of an analgesic treatment. However, chronic pain is not a condition that is limited to the experience of pain itself. Chronic daily pain also impacts mental and physical health, and consequently negatively affects quality of life [47–49]. As such, quality of life is also an important consideration when evaluating the efficacy of pain treatments [47–49]. In our study, the observed improvements in pain appeared to be associated with positive effects on patients’ health-related quality of life, as evidenced by the observed improvements in measures of physical and mental health in the SF-12. In line with what would be expected in chronic pain patients, PCS and MCS scores were low at baseline, with median scores of 30.18 and 41.59, respectively. Significant improvements were observed over time in both PCS and MCS scores. Given that the analyses of these data were conducted on an unbalanced, incomplete block design, it was not possible to conduct post hoc analyses to further investigate exactly at what time points such changes occurred. However, observation of the ranked mean values would suggest that improvements occurred at M3, with further improvements observed at M6 and M12 on both PCS and MCS scores.

Our findings are consistent with Ware et al. [36], Haroutounian et al. [35], and Bellnier et al. [34], who also conducted long-term prospective open-label studies and found improvements in measures of quality of life after initiation of MC treatment. On the other hand, a recent systematic review and analysis on cannabis-based medicines (including pharmaceutical cannabinoids) for chronic neuropathic pain, as well as a meta-analysis of controlled and observational studies of pharmaceutical cannabinoids, found that these compounds were not superior to placebo in effecting improvements in quality of life [50, 51]. It is difficult to draw comparisons between the findings of Mücke et al. [50], Stockings et al. [51], and our own study findings, as there were some key differences, including study populations, study designs, and outcome measures, among others. It should be noted that the studies by Mücke et al. [50] and Stockings et al. [51] included randomized controlled trials that were shorter in duration, with treatment periods ranging from five days to 14 weeks maximum. Given that in our study (as well as in the studies described above), improvements in MCS and PCS scores were only observed starting at the three-month mark, it is possible that the lack of significant improvements reported by Mücke et al. [50] and Stockings et al. [51] may be partly related to the shorter observation periods of the studies included in their analyses.

Opioid Medication Use

Given the cumulative evidence on the potential benefits of MC as a substitute for opioid medications in the management of pain, exploratory analyses were conducted in a subset of patients who reported using opioid medications at baseline. Examination of OMEs over the course of the study indicated statistically significant reductions in daily opioid medication doses, with reductions observed at M3, and further reductions at M6 and M12. Interestingly, this subset of patients also exhibited significant improvements in pain interference and pain severity. It should be noted, however, that these findings are the results of exploratory analyses in a small subset of patients, and the generalizability of results remains limited due to the high rate of patients lost to follow-up at each subsequent visit.

The mechanisms of action by which cannabinoids confer analgesic effects involve a multitude of regulatory physiological processes. The endocannabinoid system appears to work both independently and synergistically with other molecular targets within major endogenous pain circuitry systems [52]. Cannabinoids are postulated to act at least in part through an opioid receptor mechanism, and preclinical studies have consistently shown synergistic analgesic effects between THC and opioids [53]. Moreover, cannabis has been shown to increase dopamine levels in the nucleus accumbens [54–56], and in humans, it has been found to be associated with reductions in opioid withdrawal symptoms [57, 58], which is postulated to facilitate the process of reducing, and in some cases ceasing, the use of opioid medications. Our findings, albeit exploratory in nature and limited to a small subset of patients, are in line with other large-scale studies that also reported reductions in the use of opioid medications after commencing MC treatment [14, 35, 59–61], suggesting that MC may be an effective and safe treatment option and substitute for opioid medications in the management of pain. This is especially relevant in light of the current opioid crisis.

Safety Profile

Whole-plant cannabis is known to have a high safety index—a one-year prospective cohort study reported that MC is associated with mild to moderate nonserious adverse events, and on its own is not associated with serious adverse events [36]. Accordingly, there have been no reported cases of deaths directly related to cannabis overdose, and the most commonly reported adverse events include dry mouth, increased appetite, and “feeling high” [10, 13, 62, 63]. Our findings were in line with these reports—there were no serious adverse events, and the most commonly reported adverse events after initiating MC were dry mouth, increased appetite, and drowsiness. However, the reported frequencies of these symptoms were not significantly higher than baseline.

Interestingly, in our study, the reported frequencies of fatigue, headaches, feelings of anxiety, and nausea decreased compared with baseline. This finding supports the possibility that MC may be helpful in the management of these symptoms. Of course, such reductions in negative symptomatology may also be attributed to reduction or discontinuation of other medications, switching from recreational/illicitly obtained cannabis to MC, relief of symptoms associated with pain, and/or improvements in physical and mental well-being. Taken together, our findings on safety profile further highlight the importance of following an MC treatment program under the guidance and supervision of a physician to maximize medical benefits while minimizing adverse events.

Generalizability of Results

As indicated in Time Point Definition, and perhaps due to the observational nature of this study, each subsequent follow-up time point had fewer report completions, which may have contributed to biased results and limited the generalizability of results. To investigate potential differences between patients who completed all five time point visits and those who did not, additional analyses were conducted (Supplementary Data). These revealed no significant differences in baseline demographic measures such as age, gender, and type and duration of pain between patients who completed all five time point reports and those who did not. In addition, we evaluated the pattern of treatment response as measured by the BPI in the 5-TMP, 4-TMP, 3-TMP, and 2-TMP groups. Significant improvements were observed from baseline to month 1 in all of these patient groups, and such improvements were maintained up to the last completed time point visit in each group. In other words, significant improvements in BPI were maintained up to month 12 in the 5-TMP group, up to month 6 in the 4-TMP group, and up to month 3 in the 3-TMP group. These results serve as robust evidence that MC effects immediate pain alleviation, as significant improvements were observed after one month of MC treatment in all these patient groups, including the 2-TMP group, which had 580 patients in the analysis, thus suggesting generalizability of results.

On the other hand, it is difficult to determine whether the long-term maintenance of such improvements is also generalizable, as there were fewer report completions at each subsequent visit after month 1 due to patients being lost to follow-up. Given that the improvements in BPI scores were comparably maintained up to month 3 in the 3-TMP, 4-TMP, and 5-TMP groups, and up to month 6 in the 4-TMP and 5-TMP groups, it may be inferred that improvements in BPI scores were similarly maintained in patients who did not complete all the time point reports up to the 12-month follow-up. On the other hand, it is also possible that in patients who did not complete the M3, M6, and/or M12 reports, the improvements in pain observed after one month of treatment may have reverted back to baseline pain levels and then worsened further thereafter. Indeed, there have been reports of pain exacerbation associated with cannabis use [33, 64]. However, such reported negative outcomes appear to be related to cannabis dose and frequency of use, rather than the use of medical cannabis per se, and thus may not be applicable to the present study, where patients followed a treatment regimen under the guidance of a physician. For instance, the study by Wallace et al. [33] reported significant increases in pain at higher doses, which the authors attributed to the existence of a therapeutic window or optimal dose for cannabinoids to effect analgesia, a point that is discussed in Pain Management and Quality of Life, and which further highlights the importance of following a treatment regimen under the guidance of a physician. More recently, Boehnke et al. [64] reported that higher frequency of use is associated with higher pain severity and pain interference than light and moderate frequency of use. However, it must be noted that the differences in pain scores, albeit statistically significant, were rather small (mean BPI-I score of 4.2 in light-frequency users compared with 4.7 in high-frequency users; and mean BPI-S score of 5.2 in light-frequency users compared with 5.6 in high-frequency users). Moreover, the higher pain scores observed in high-frequency compared with moderate- or light-frequency users cannot be concluded to stem from cannabis use, as the study by Boehnke et al. [64] was a cross-sectional study without baseline pain scores to serve as comparison at pre- and postinitiation of MC treatment.

Taken together, the results of the analyses discussed herein suggest that the improvements in pain scores observed after one month of treatment are generalizable to a larger chronic pain population, as improvements were seen in a large sample size of 580 patients. Moreover, given that i) such improvements were similarly maintained in the 5-TMP, 4-TMP, and 3-TMP groups and ii) reports of pain exacerbation associated with cannabis appear to be dose-related and the result of the biphasic effects of cannabinoids rather than the use of MC per se, we propose that the maintenance of pain relief observed in this study may also be generalizable to a larger pain population, as long as patients follow a treatment regimen under the supervision of a physician. Nevertheless, the reproducibility of results in the long term must be confirmed in a study with improved strategies to ensure patient retention for data integrity.

Study Limitations

There were a number of limitations associated with this study. This was an open-label study, and there was no control group. Moreover, the heterogeneity in dose and consistency of cannabis regarding THC/CBD ratio and methods of intake used during the study prevented a more in-depth analysis of dose-related effects with respect to THC and CBD content, as well as the synergistic effects of other cannabinoids, flavonoids, and terpenes, which have been postulated to contribute to the analgesic effects of plant-based MC [52]. It should be noted, however, that great variability in strains and dosing effects is an unavoidable aspect of cannabis therapeutics, and as such, there are limitations to the generalizability of the results of studies that utilize predefined doses vs the customization of best evidence-based practices for individual patients [42, 65]. For instance, dose response is strongly influenced by cannabinoid bioavailability, which has a high degree of intra- and interindividual variability stemming from factors that have been excellently summarized in a review by Huestis [66]. These include—but are not limited to—the mode of administration (e.g., bioavailability can vary from 20% to 30% when taken orally, or 10% to 60% via inhalation), behavioral factors (e.g., recent meals, depth of inhalation, duration of breath holding, and temperature of vaporizer can all affect cannabis absorption), and genetic factors (e.g., high intra- and interindividual variability in the activity rates of enzymes involved in cannabinoid metabolism, such as CYP 450 2C9 and 2A), among others [66]. Accordingly, MC doses are individually determined, and the general approach to MC treatment is to “start slow, go slow, and stay low,” increasing the dose as needed and as tolerated [42]. In this regard, medical cannabis is a highly individualized evidence-based medicine, where a high degree of variability in cannabis strain and dosing is to be expected, as the patient serves as the sole unit of observation for efficacy and tolerability in finding the optimal intervention or combination of plant varieties and dosage forms [42].

Another limitation of the study was that, as discussed in Generalizability of Results, each subsequent follow-up time point had fewer report completions, which may have contributed to biased results and limited the generalizability of results. The issue of unbalanced data sets due to patients lost to follow-up is not uncommon, particularly in large-scale, long-term studies [67, 68]. In this study, the issue may have been exacerbated by the lack of financial incentives to complete reports and the fact that prompts to complete follow-ups reports were sent via e-mails that contained links to the questionnaires. While this method may have been beneficial in terms of time- and cost-efficiency, it is also likely to have been disadvantageous in that it lacked the “personal touch” necessary to maintain good rapport and motivation for study participants to complete follow-up study reports [68]. In this regard, the data collection methods and strategies to retain long-term patients were limitations of the study.

Another limitation of the study was the lack of standardized reporting for adverse events associated with MC treatment. As this was an observational study, we tried to keep the time burden of completing questionnaires low. As such, several aspects of adverse event data collection common in clinical trials (e.g., duration, severity, action[s] taken with the intervention, rescue medications, outcome) were not collected in this study, which prevented more in-depth analyses.

Finally, other limitations of the study include selection/volunteer bias, as this was a study of patients at a medical cannabis clinic who had chosen MC to treat their chronic pain and thus was not a full representation of the chronic pain population at large, and recall and expectancy bias, as data in the study were collected by means of subjective self-report questionnaires.

Conclusions

In this large, open-label, prospective, observational study, we found that treatment with MC was associated with significant improvements in measures of pain as well as quality of life, which were sustained over the course of the 12-month observation period in patients who remained in the study for the full duration. Improvements in measures of pain were observed as early as one month after initiating treatment with MC, while improvements in quality of life were only observed starting at three months after initiating treatment. This suggests that while the analgesic effects have an immediate onset of action, the global health benefits of MC necessitate a longer treatment course. We also found that MC was associated with significant improvements in measures of pain regardless of previous recreational use, indicating that the benefits of an MC treatment course extend beyond those of unregulated, recreational cannabis. Such benefits are postulated to stem from the facts that medical-grade cannabis meets higher standards of purity and consistency and that there are no contaminants such as mold, pesticides, or other substances that could negatively affect health. Additionally, following a treatment course that is tailored to the needs and tolerability of the patient, under the guidance of a physician, is likely to contribute to acceptance and treatment adherence by the patient, which in turn may have further contributed to improved pain scores and quality of life. Moreover, the MC treatment course in this study was not associated with increases in the frequency of undesired adverse events, but rather decreased the frequency of headaches, fatigue, feelings of anxiety, and nausea, further supporting the importance of following a treatment course that is optimized to the needs and tolerability of the patient under the supervision of a physician. Finally, in a small subset of patients who reported using opioid medications at baseline, exploratory analyses revealed that opioid medication doses decreased, while measures of pain were significantly improved over the course of the study treatment observation period. Taken together, the results of this study add to the cumulative evidence in support of plant-based MC as a safe and effective treatment option and potential opioid substitute or augmentation therapy for the management of chronic pain symptomatology and quality of life.

Funding sources: This study was funded by Apollo Applied Research.

Disclosure and conflicts of interest: Authors Vahid Salimpour and SunYoung Rosalia Yoon were affiliated with and Bryan Hendin and Genane Loheswaran were employees of Apollo Applied Research during the time the study was conducted and/or the manuscript was written. Apollo Applied Research is an independent subsidiary of Canopy Growth Corp. Author Gordon Ko was affiliated with Apollo Applied Research at the time the study was conducted/the manuscript was written, and he has received CME honoraria or support, consultancy fees, or grants from the following companies during 2016–2019: Aurora (CanniMed, MedReleaf), Canopy (Mettrum, Tweed, Apollo), Canntrust/Tetra BioPharma; Tilray (CBM talks/clinical advisory boards participation). Authors Ramin Safakish and Imrat Sohanpal have no conflicts of interest to declare.

Acknowledgments

We would like to thank Dr. Paula Williams and Bijan Rafat for their intellectual contributions to study design and data collection.

Supplementary Data

Supplementary data are available at Pain Medicine online.