Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Dexamethasone and chemotherapy nausea and vomiting

A wider perspective

From time to time along comes a paper that makes you wish that you had written it. That might be because it will make you famous (but who wants reporters nagging them anyway?). It's more likely to be because it was a smashing bit of work that was helpful and well done. One such is an examination of dexamethasone used to control chemotherapy-induced nausea and vomiting [1].

This may not be a topic that affects many as professionals, but for patients it is important. It is also important because palliative care has too few good trials, let alone good systematic review, so exemplars matter.


Start by thinking of the problems that a review might encounter. First is the setting. Different chemotherapy regimens are thought to have different effects on emesis, with some being highly emetogenic and others less so. The patients may be having a first cycle, or have had some cycles before, and may or may not have had previous chemotherapy. They may have a history of chemotherapy-induced emesis. The type of cancer may be the same or different. The age of the patients may differ.

Then there is the treatment and the comparator. Dexamethasone dose may vary, as well as the mode of delivery. The comparator may be placebo, but that would be unfair, so it may be placebo with cover from other antiemetic, or other antiemetic, or combinations.

Then there is the period over which emesis is measured. Is the first 24 hours different from later periods, and over what period is it important to measure any effect of dexamethasone?

Then there is what you measure. Is it vomiting alone, or nausea alone, or retching, or complete control of emesis?

Then there is study design and reporting. Do the trials have properties that may confer bias? Are the studies using valid methods?

Anyone undertaking such a review might be forgiven for feeling a little nauseated at the prospect of sorting this lot out. Is it possible to make sense of it all?


The review sought randomised trials of dexamethasone for controlling chemotherapy-induced nausea and vomiting using a variety of electronic databases, as well as hand searching journals. Comparison was with placebo, no treatment, active agents or combinations of these, and a number of different languages was permitted.

The main outcome was prevention of vomiting because that is more objective than nausea, but prevention of nausea was also considered. Periods were split between the first 24 hours (acute phase) and up to 8 days after chemotherapy (delayed phase).


There were 32 studies included with 42 different comparisons and information on 5,500 patients. Doses of dexamethasone varied between 8 mg and 100 mg. Half used 20 mg for the first 24 hours, and the mean dose over the acute and delayed phase was 56 mg. Information was collected about type of patient, cancer, chemotherapy, previous cycles, prior chemotherapy and history of chemotherapy-induced emesis. Comparison was predominantly with placebo or no treatment supplemented with other antiemetic agents.

The overall results for control of emesis in the acute phase are shown in Figure 1. With dexamethasone 1320/1911 patients (69%) did not vomit (and 31% did vomit) compared with 937/1713 patients (55%) who did not vomit (and 45% who did vomit) with control. The relative benefit was 1.3 (1.2 to 1.4) and NNT 7 (6 to 9).

Figure 1: Acute phase: first 24 hours of chemotherapy

The overall results for control of emesis in the delayed phase are shown in Figure 2. With dexamethasone 887/1461 patients (61%) did not vomit compared with 754/1679 patients (45%) with control. The relative benefit was 1.3 (1.2 to 1.5) and NNT 6 (5 to 8).

Figure 2: Delayed phase: 2-5 days after chemotherapy


What was interesting was that both random effects and fixed effects were used in calculating odds ratios, risk ratios and risk difference, with a number of comparisons and end points. Potential variability in antiemetic efficacy was examined by looking at doses. Meta-regression, a complex statistical approach for most of us, was used to examine potential effects of trial design (blinding), and concealment of treatment allocation, and of different settings like moderately and highly emetogenic treatments. None made much difference. To some extent the consistency of response can be seen in the L'Abbé plots (Figures 1 and 2), both for big studies (large symbols) and small studies (smaller circles).

The sensitivity analysis underpinned the overall conclusion that about six patients need to be treated to prevent one from experiencing emesis in either phase.

The value of the meta-analysis is not just in demonstrating that dexamethasone treatment works, and how well it works, but in the superb methodological workup to give us confidence in that result. An excellent example of a paper that could be used for training in what makes for acceptable standards in preparing and writing systematic reviews.

A wider perspective

On its Internet site, Bandolier also examines a series of other systematic reviews concerned with treatment of acute vomiting after chemotherapy and radiotherapy. High dose metoclopramide and conventional treatment do relatively poorly, with about 50% vomiting. With 5HT-3 receptor antagonists alone about 40% vomit, while with 5HT-3 receptor antagonists plus dexamethasone, only 25% vomit. A nice example of relative efficacy in a difficult area.


  1. JP Ioannidis et al. Contribution of dexamethasone to control of chemotherapy-induced nausea and vomiting: a meta-analysis of randomised evidence. Journal of Clinical Oncology 2000 18: 3409-3422.
previous or next story in this issue