01.06.2013 Views

Statistical Methods in Medical Research 4ed

Statistical Methods in Medical Research 4ed

Statistical Methods in Medical Research 4ed

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

might, for example, be thought that a drug cannot take effect before at least a<br />

week has elapsed, and that therefore any adverse events occurr<strong>in</strong>g dur<strong>in</strong>g the first<br />

week can be discounted. These events should be omitted only if the underly<strong>in</strong>g<br />

assumption is universally accepted; otherwise, the possibility of bias aga<strong>in</strong> arises.<br />

Similarly, adverse events (such as accidental deaths) believed to be unrelated to<br />

the disease <strong>in</strong> question should be omitted only if their irrelevance is beyond<br />

dispute (and this is rarely so, even for accidental deaths). However, events<br />

especially relevant for the disease under study (such as cardiovascular events <strong>in</strong><br />

a trial for treatments for a cardiovascular disease) should be reported separately<br />

and may well form one of the primary endpo<strong>in</strong>ts for the trial.<br />

The policy of <strong>in</strong>clud<strong>in</strong>g <strong>in</strong> the analysis, where possible, all patients <strong>in</strong> the<br />

groups to which they were randomly assigned, is called the <strong>in</strong>tent(ion)-to-treat<br />

(ITT) or as-randomized approach. It follows the pragmatic approach to trial<br />

design (§18.3), <strong>in</strong> that groups receive treatments based on ideal strategies laid<br />

down <strong>in</strong> the protocol, with the recognition that (as <strong>in</strong> rout<strong>in</strong>e medical practice)<br />

rigidly prescribed regimens will not always be followed. In a modified form of<br />

ITT it is occasionally thought reasonable to omit a small proportion of patients<br />

who opted out of treatment before that treatment had started. In a doublemasked<br />

drug trial, for <strong>in</strong>stance, it may be clear that the same pressures to opt out<br />

apply to all treatment groups. This would not be so, however, <strong>in</strong> a trial to<br />

compare immediate with delayed radiotherapy, s<strong>in</strong>ce the group assigned delayed<br />

treatment would have more opportunity to opt out before the start of radiotherapy.<br />

If a high proportion of patients abandon their prescribed treatment regimen,<br />

perhaps switch<strong>in</strong>g to an alternative regimen under study, the estimates of differences<br />

<strong>in</strong> efficacy between active agents will be biased, probably towards zero. In<br />

a later subsection we discuss whether adjustments can be made to allow for this<br />

<strong>in</strong>complete compliance.<br />

Withdrawals and miss<strong>in</strong>g data<br />

18.6 Protocol departures 607<br />

An ITT analysis may be thwarted because response variables, or other relevant<br />

<strong>in</strong>formation, are miss<strong>in</strong>g for some patients. These lacunae can arise for various<br />

reasons. The patient may have withdrawn from the <strong>in</strong>vestigator's care, either<br />

because of dissatisfaction with the medical care or because of removal to another<br />

district. In a follow-up study, <strong>in</strong>formation may be miss<strong>in</strong>g for some prescribed<br />

visits, because the patient was unwell. Technical equipment may have failed,<br />

lead<strong>in</strong>g to loss of test results, or other adm<strong>in</strong>istrative failures may have<br />

occurred.<br />

In such situations it may be possible to retrieve miss<strong>in</strong>g <strong>in</strong>formation by<br />

assiduous enquiry at the medical centre. In trials for which mortality is the<br />

primary endpo<strong>in</strong>t, it is possible <strong>in</strong> some countries (such as the UK) to determ<strong>in</strong>e

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!