Patients who receive some but not all their allocated treatment

For many cancers, treatment is given over a period of many weeks, months or sometimes years. As a consequence, it is almost inevitable that some patients will not complete all their allocated treatment. This may happen because the patient dies during treatment, or the patient's disease progresses (or improves) rapidly making further treatment inappropriate, or the patient experiences unacceptable toxicity and refuses further treatment, or the doctor considers it not in the patient's best interests to continue with it. Generally, patients who do not complete the allocated treatment are likely to have an inherently poorer prognosis than those who do complete it, and this poor prognosis is not a consequence of their non-completion of the course of treatment. Further, in cancer, the experimental treatment is likely to be more toxic than the control treatment and thus lead to more patients stopping treatment early. Hence, exclusion of such patients from the analysis will again introduce systematic differences between the two groups being compared, and this bias will be introduced into the analysis and results. An extreme example of this is if we are comparing six courses of adjuvant cytotoxic therapy versus no adjuvant cytotoxic therapy. In many trials it is common to find that only approximately 80 per cent of all patients actually receive all six courses of adjuvant therapy, with those perhaps least well stopping therapy early. If we exclude these patients then we are comparing the 80 per cent most 'well' patients in the adjuvant therapy group with 100 per cent of patients in the no adjuvant therapy group, leading to a clearly biased estimate of the difference between the two.

0 0

Post a comment