Cannabinoids for control of chemotherapy induced nausea and vomiting: quantitative systematic review

Tramer et al's assertion that '[cannabinoids] are superior to
conventional antiemetics'(1) is flawed. Such a comparison is limited by
the inadequacy of the methods of measurement of outcome.

In this systematic review the endpoint of primary interest was
defined as 'the complete control of nausea or vomiting in the first 24
hours of chemotherapy'. However, nausea and vomiting may occur for a much
wider time frame: 'delayed' nausea and vomiting (peaking at 48 - 72 hours
post chemotherapy) are well recognised phenomena (2). In a study of 105
patients receiving outpatient chemotherapy post treatment nausea and
vomiting were experienced by 73% and 33% of patients respectively (3). No
patient indicated that the period during treatment was the worst, rather
these symptoms tended to be at their worst some 24 hours or more after
chemotherapy infusion. Tramer et al's study does not tell us how patients
faired for the time period when most problems related to nausea and
vomiting occur.

This review also found high rates of side effects caused by
cannabinoids including sedation, hallucinations and substantial effects on
mood. Under these conditions patients' judgements of their experience can
be severely impaired (2). Nausea can only be measured by patients' self-
report. Tramer et al found that the numbers needed to be treated with
cannabinoids (NNT) for control of nausea is less than the NNT for control
of vomiting (NNT 6 and 8 respectively). This could, in part, be explained
by patients' inability to accurately remember their experience of nausea
and vomiting rather than an actual reduction in the experience in reality.

Tramer et al used a validated score to rate the quality of studies
(4). This scoring system assesses the adequacy of various aspects of
research methodology. An assessment of the choice of outcome measure is
also important. In a meta-analysis of 392 studies of the use of
antiemetics Morrow et al found that whilst effect size may be relatively
independent of the type of outcome measured (for example nausea or
vomiting), the mean effect size did vary according to how the outcome was
quantified (5). The mean effect size measured for frequency was
statiscally different from measures of duration and furthemore measures of
severity were statistically different from measures of duration.

The nature and quality of outcome measures (and hence study
endpoints)can have a substantial effect on the overall meaning and value
of a study, and profound effects on the results of systematic reviews or
meta-analyses. In order to be fully relevant, outcomes must address the
major problems found in the clinical setting.

References

(1) Tramer MR, Carroll D, Campbell FA, Reynolds DJM, Moore RA, McQuay
HJ. Cannabinoids for control of chemotherapy induced nausea and vomiting:
quantitative systematic review. British Medical Journal 2001; 323: 16-21

(2) Olver IN. Antiemetic study design: desirable objectives,
stratifications and analysis. Britsih Journal of Cancer 1992; 66: S30 - 34

(3) Watson M, McCarron J, Law M. Anticipatory nausea and emesis, and
psychological morbidity: assessment and prevalence among outpatients on
mild to moderate chemotherapy regimens. British journal of cancer 1992;
66: 862 - 866

(4) Jadad AR, Moore A, Carrol D, Jenkinson C. Assessing the quality
of reports of randomsied clnical trials: is blinding necessary? Controlled
Clnical Trials 1996; 17: 1-12

(5) Moorow GR. Methodology and assessment in clinical antiemetic
research: a meta-analysis of outcome parameters. British journal of Cancer
1992; 66: S38-S41