Toxicity data will be collected and presented in nearly all trials. For example, in the ICON2 ovarian cancer trial [4], data on the toxic effects of treatment were collected on 875 of the patients randomized. A subset of these data is presented in Table 9.10.

Issues raised in the analysis of toxicity data are the same as those for quality of life data and the reader is referred to Section 9.4.9 for a full discussion of the analysis of such data. Some more general issues are covered here.

There is a strong temptation to perform significance tests for each toxicity. However, it is usually inappropriate to do this because standard advice for significance tests is not usually followed. In particular, as described with reference to Table 9.6, null and alternative hypotheses are not clearly stated, a lack of significant difference between arms is treated as a 'proof' of equivalence and adjustment is not always made for multiple testing. Further, it should be noted that the trial alone is unlikely to be the only source of toxicity data for the treatments being compared. For example, in the above ICON2 trial it would not be informative to perform a significance test of whether the rate of alopecia

Table 9.10 Number of patients with grade 3 or 4 toxic effects reported (using World Health Organization criteria) during treatment in the ICON2 ovarian cancer trial. Reprinted with permission from Elsevier Science (The Lancet, 1998, 352, 1571-6)

Toxic effects

CAP (n = 430)

Carboplatin (n=455)


300 (70%)

20 (4%)


153 (36%)

44 (10%)

Nausea and vomiting

84 (20%)

39 (9%)


26 (6%)

73 (16%)


1 2 (3%)

5 (1 %)

is different in the two treatments, before other issues are addressed. Issues are: how important is the side effect of alopecia to patients and what sort of differences between alopecia are important? Thus, if the relative toxicity of the treatments are likely to be critical in deciding between treatments, then this needs to be stated at the beginning of the trial, and hypotheses should be specified and power calculations should be performed for the most important toxicities. For the remaining toxicities, the information should be presented as in Table 9.10, perhaps allied to confidence intervals for the difference between percentages.

Hair Loss Prevention

Hair Loss Prevention

The best start to preventing hair loss is understanding the basics of hair what it is, how it grows, what system malfunctions can cause it to stop growing. And this ebook will cover the bases for you. Note that the contents here are not presented from a medical practitioner, and that any and all dietary and medical planning should be made under the guidance of your own medical and health practitioners. This content only presents overviews of hair loss prevention research for educational purposes and does not replace medical advice from a professional physician.

Get My Free Ebook

Post a comment