Medline ® Abstract for Reference 37
of 'Pathophysiology and prediction of chemotherapy-induced nausea and vomiting'
Antiemetic study design: desirable objectives, stratifications and analyses.
Br J Cancer Suppl. 1992;19:S30.
Once the optimal dose and safety of an antiemetic has been established a randomised double blind parallel subjects design is recommended for phase III studies. Randomisation distributes the unknown prognostic variables so that their effects can be allowed for in tests for statistical significance. Stratification can equally distribute the known prognostic factors e.g. prior exposure to chemotherapy, strength of the emetic stimulus, age, sex and prior alcohol consumption. A cross-over design is often proposed because less patients would be required to achieve the same power as a parallel subjects study. The major problem with this design is in being able to estimate and allow for carry over effects or treatment period interactions which can interact with each other and the direct treatment effect. The study must be large enough to detect a statistically significant difference of clinical importance. Interim analyses should be preplanned and early termination should require a difference between the arms with a more significant P value than 0.05. Simple evaluation of efficacy could include quantitation of objective parameters and use of simple ordinal scales to record more subjective phenomena. In a parallel subjects design patients must assess their overall tolerance of therapy which balances the antiemetic's efficacy and toxicity.
Royal Adelaide Hospital, North Terrace, Adelaide, South Australia.