Introduction

Top Abstract Introduction Methods Results Discussion References

The recent clamour for wider access to cannabis or cannabinoids as analgesics in chronic painful conditions has some logic. Humans have cannabinoid receptors in the central and peripheral nervous system,¹ although the functions of these receptors and the endogenous ligands may yet be unclear. In animal testing cannabinoids reduce the hyperalgesia and allodynia associated with formalin, capsaicin, carrageenan, nerve injury, and visceral persistent pain.² The hope then is that exogenous cannabis or cannabinoid may work as analgesics in pain syndromes that are poorly managed. The spasms of multiple sclerosis and resistant neuropathic pain are two obvious targets.

The background to this debate about legitimising cannabis (also called marijuana)from the plant Cannabis sativafor analgesic use is that the drug has been used both therapeutically and recreationally for thousands of years.³ In Britain doctors were able to prescribe cannabis as recently as 1971,⁴ and in a 1994 survey 74% of UK doctors wanted cannabis to be available on prescription, as it had been until 1971.⁵ The debate has included both the natural chemicals that act on cannabinoid receptors and the synthetic cannabinoids. The synthetic nabilone is the only legally available cannabinoid preparation in the United Kingdom and is licensed solely for use in nausea and vomiting induced by chemotherapy. Delta-9-tetrahydrocannabinol (THC) is the most potent cannabinoid, and although it is available in the United States, it is not licensed for use in the United Kingdom.

The evidence used in the public debate about the analgesic efficacy of cannabinoids in humans has been gathered in a less than systematic manner and has often been taken from low quality study designs, such as anecdotal reports, questionnaires, or case series.⁴ The purpose of this systematic review was to find all of the randomised controlled trials of therapeutic use of cannabis in the management of human pain and then to obtain the best estimates of the efficacy of cannabis compared with either conventional analgesics or placebo. We also sought evidence of adverse effects (safety).

Cannabis is used recreationally because of the euphoria that it produces. The adverse psychological effects (including psychomotor and cognitive impairment; anxiety and panic attacks; and acute psychosis and paranoia) may limit therapeutic use.⁶ Other adverse physical effects include dry mouth, blurred vision, palpitations, tachycardia, and postural hypotension.³

Decisions about therapeutic cannabinoids, either about medical availability or about future research, should be based on the best available evidence of efficacy, safety, and tolerability. This systematic review was designed to provide that evidence for cannabinoids used as analgesics.

	Methods

Top Abstract Introduction Methods Results Discussion References

Searching
Two authors (MRT and DC) searched independently, using different search strategies in Medline (for 1966-99), Embase (1974-99), the Oxford Pain Database (1950-94),⁷ and the Cochrane Library (1999, issue 3). The most recent search was done in October 1999. The search included different combinations of the following MeSH headings and "free text" terms: marijuana, marihuana, mariuana, cannabis, cannabinoids, THC, delta-9-tetrahydrocannabinol, nabilone, pain, analgesia, and random*, and different combinations of these terms. Additional reports were identified from the reference lists of retrieved reports and review articles. The search included data in any language. Pharmaceutical manufacturers and authors were not contacted. Only full publications in peer reviewed journals were considered for inclusion in the review. Unpublished data were not sought. Data from review articles, case reports, abstracts, and letters were not included.

Selection and validity assessment
Randomised controlled trials of cannabis and its active constituents (namely, cannabinoids) in human pain were sought systematically. Studies of experimental pain were excluded. Relevant papers had to report on comparisons of cannabis or cannabinoids (experimental intervention, given by any route of administration) with any analgesic or placebo (control intervention).

Data extraction
Data extraction was done by one author (FAC) and cross checked by at least two other reviewers. Discrepancies were resolved by consensus.

Study characteristics
The following information was extracted from each report: the type, dose, and route of administration of cannabinoids; the controls; the types of pain; the sample size; the study design and duration; outcome measures for pain intensity; pain relief; the use of supplementary analgesia; patients' preferences; and adverse effects.

Statistical analysis
Quantitative meta-analysis with pooling of data from the eligible randomised controlled trials was proposed.

	Results

Top Abstract Introduction Methods Results Discussion References

Trial flow
The results of the searches are presented in the figure. The presentation follows the suggested format provided in the QUORUM statement.⁹ Of the 11 excluded trials, three did not use randomised treatment comparisons,^10-12 four did not use subjective pain outcomes,^13-16 two had studied experimentally induced pain, ^{17 18} and one was published as a letter¹⁹ and one as an abstract.²⁰

View larger version (37K): [in this window] [in a new window]   Results of search of Medline, Embase, Oxford Pain Database, Cochrane Library, references on identified papers; and hand searches for work on cannabinoids and pain

Study characteristics
Details from nine randomised controlled trials published in seven reports published between 1975 and 1997 were analysed (table).^21-27 The nine randomised controlled trials comprised a total of 222 adult patients. Five studies that were described in four reports comprised 128 patients with cancer pain.^21-24 Two studies comprised two patients with chronic non-malignant pain (one patient per trial), ^{25 26} and two trials (conducted as a two phase study) comprised six patients with postoperative pain.²⁷ Follow up was six to seven hours in seven trials and six weeks and five months respectively in the two trials on chronic non-malignant pain. ^{25 26} The number of patients in treatment groups ranged from 1 to 37. All studies used a crossover design except the study of postoperative pain.²⁷ The two studies in chronic pain used an "n of 1 within patient crossover" design. ^{25 26} Seven studies described in five reports were single dose evaluations of the analgesic effectiveness of cannabinoids. ^{21-24 27} The median quality score of the trials was 3 (range 3-4) (possible score 0-5). All studies included a placebo group. An adequate method of blindingfor example, tablets of identical shape, colour, and tastewas used in all trials where treatments were given orally. There was no explicit description of the method of blinding in the phase 1 and 2 trials comparing intramuscular levonantradol with placebo in postoperative pain.²⁷

Cancer pain
In the five trials on cancer pain 128 patients were studied (table). In one study oral benzopyranoperidine (a THC congener) 2-4 mg was not as effective as codeine sulphate 60-120 mg and no more effective than placebo in 37 patients.²¹ Oral THC 5-20 mg was found to have an analgesic effect when compared with placebo in 10 patients with pain related to advanced cancer.²² In this study a dose-response relation was shown for analgesia but also for adverse effects. In a further study by the same group oral THC 10 mg was found to be about equipotent to codeine 60 mg, and THC 20 mg was about equipotent to codeine 120 mg.²³ The higher dose was associated with unacceptable adverse effects. In one trial a synthetic nitrogen analogue of THC given orally was superior to placebo and equivalent to about 50 mg of codeine phosphate.²⁴ In a second study in the same report this nitrogen analogue was found to be superior to placebo and to 50 mg of secobarbital.²⁴ In both of these trials the nitrogen analogue of THC was felt to be not clinically useful because of the frequency of adverse effects.

Chronic non-malignant pain
Two patients were studied in two "n of 1 within patient crossover" trials for six weeks and five months respectively (table). In an experienced cannabis user with familial Mediterranean fever, THC was found to be no better than placebo in terms of visual analogue scores for pain intensity.²⁵ Level of morphine use for breakthrough pain was significantly lower, however, while the patient was taking THC than while taking placebo (170 mg v 410 mg per three weeks). In a patient with neuropathic pain and spasticity secondary to a spinal cord ependymoma, THC 5 mg and codeine 50 mg were equianalgesic, and both were superior to placebo.²⁶ Only THC, however, had a beneficial effect on spasticity.

Postoperative pain
Thirty six patients were studied in two trials (conducted as a two phase study) (table).²⁷ Levonantradol was more effective than placebo when given intramuscularly to patients with postoperative pain.²⁷ Adverse effects with levonantradol were common, although considered mild.

Cannabinoids and adverse effects
Adverse effects were reported in all studies. Two patients withdrew from studies owing to adverse effects of THC.²³ THC showed a dose-response relation for adverse effectsfor example, mental clouding, ataxia, dizziness, numbness, disorientation, disconnected thought, slurred speech, muscle twitching, impaired memory, dry mouth, and blurred visionand at 20 mg was highly sedating in 100% of patients, thus prohibiting its use.²³ THC 10 mg was better tolerated, but the frequency of these adverse effects was still higher than with codeine 60 mg or 120 mg.²³ Reductions in arterial blood pressure occurred compared with placebo, but no more than with codeine. Changes in heart rate were not significant. THC 5 mg was well tolerated in neuropathic pain and did not cause an altered state of consciousness.²⁶ Levonantradol caused adverse effects in most patients, but none withdrew.²⁷ The nitrogen analogue of THC did not affect heart rate but caused drowsiness in 40% of patients and was therefore deemed not clinically useful.²⁴ Benzopyranoperidine caused a similar degree of sedation to codeine but was ineffective as an analgesic.²¹

	Discussion

Top Abstract Introduction Methods Results Discussion References

We found nine randomised trials evaluating the analgesic efficacy and safety of cannabinoids. These trials, of either acute or chronic pain, suggest that little useful analgesia can be expected from single dose cannabis in nociceptive pain.

All the trials had a quality score of 3 or above and therefore are unlikely to be biased. They were predominantly single dose experiments. In eight of the nine trials intramuscular and oral cannabinoids were more effective analgesics than placebo but no more effective than oral codeine 50-120 mg.

Acute pain
In the two postoperative pain trials levonantradol was superior to placebo but no more effective than codeine.²⁷ Such a level of efficacy makes cannabinoids unlikely to be useful, certainly for moderate or severe postoperative pain. Meta-analyses of single dose studies in patients with acute pain found that the number needed to treat for at least 50% pain relief ranged from 2 to 5 compared with placebo for non-steroidal anti-inflammatory drugs and paracetamol. The number needed to treat for codeine 60 mg was much less useful, at 16.²⁸ If cannabinoids can deliver analgesia only equivalent to codeine 60 mg, with a presumed number needed to treat of about 16 for at least 50% pain relief, they are unlikely to have a place in acute pain treatment.

Cancer and non-malignant pain
No large trials examined cannabinoids in cancer pain and chronic non-malignant pain. Only two trials had treatment group sizes of more than 30. ^{21 23} All five trials in cancer pain were single dose, and four found the cannabinoid as effective as codeine, but with dose limiting adverse effects. ^{22 23 24} Benzopyranoperidine, tested in one trial, was ineffective compared with both codeine and placebo.²¹ In chronic non-malignant pain we found two "n of 1 within patient crossover" trials. In a patient with abdominal pain related to familial Mediterranean fever, neither THC nor placebo produced pain relief, but with THC the patient used less additional morphine for breakthrough pain.²⁵ In Maurer et al's n of 1 study of THC 5 mg for neuropathic pain and spasticity, the reduction in pain intensity was similar to that for codeine 50 mg, but only THC reduced spasticity.²⁶ We found no trials evaluating smoked cannabis for pain management, but one trial compared the effect of smoked marijuana with smoked placebo on postural balance in patients with spastic multiple sclerosis.¹⁶ The smoked marijuana was associated with subjective improvement of symptoms and with objectively measured impaired posture and balance in all subjects.

Adverse effects
Adverse effects associated with the cannabinoids were common and sometimes severe in six of the eight trials that showed efficacy. The predominant adverse effect seemed to be depression of the central nervous system. Cardiovascular effects were generally mild and well tolerated. Levonantradol was commonly associated with adverse effects (predominantly drowsiness or sedation, or both), of which over half were considered to be moderate or severe. THC 10-20 mg showed a dose-response relation for adverse effects, with depressant effects on the central nervous system occurring in most patients receiving either dose. In Holdcroft et al's patient²⁵ no adverse effects were attributable to THC 50 mg a day, but the patient was an experienced cannabis user. Maurer et al's patient experienced no altered state of consciousness taking THC 5 mg for neuropathic pain; the cannabinoids might be stimulant at low doses and depressant at higher doses, and perhaps this was the reason for lack of sedation in this patient.²⁶ The nitrogen analogue of THC had a side effect profile similar to codeine.²⁴ This cannabinoid was as sedating as secobarbital, which has no analgesic properties, thus it is unlikely that any sedation caused by cannabinoids contributes to their analgesic effect. Other studies have shown that barbiturates and tranquillisers given with analgesics contribute nothing to pain relief.²⁹

Conclusion
The best that can be achieved with single dose cannabis in nociceptive pain is analgesia equivalent to single dose codeine 60 mg, which rates poorly on relative efficacy compared with non-steroidal anti-inflammatory drugs or simple analgesics. Increasing the cannabinoid dose to increase the analgesia will increase adverse effects. More intriguing perhaps than these relatively negative analgesic results in nociceptive pain are the suggestions of efficacy in spasticity and in neuropathic pain, where the therapeutic need is greater than in postoperative pain.