In general blinding in trials is

♦ desirable for 'soft' endpoints, such as pain, anxiety, depression (subjective),

♦ often unnecessary for 'hard' endpoints, such as survival (objective),

♦ useless if treatments are known to have very different side effects which would make objective assessment impossible,

♦ dangerous if other treatments the patient may receive could react badly with one of the blinded treatments.

Example 2: The US study of tamoxifen for breast cancer prevention [31] was a doubleblind placebo-controlled trial which randomized patients at 'high risk' of breast cancer to receive five years of tamoxifen or placebo. The main outcome measure of the trial was breast cancer incidence with secondary outcomes being the incidence of fractures and cardiac events - which tamoxifen would be expected to have a beneficial effect on - and also endometrial cancer and thrombotic events - which tamoxifen would be expected to have an adverse effect on. Was blinding necessary here? In its favour, it avoids the possibility of patients being investigated for breast cancer more closely or frequently in one arm. If, for example, control patients were seen more frequently, any breast cancers which occurred would be detected earlier than they would in the active treatment group, leading to an apparent benefit to tamoxifen even if the total number of breast cancers was the same in the two groups. Does the same argument apply to the secondary outcome measures? Given the known potentially adverse effects of tamoxifen, a patient being treated outside the trial might receive anti-thrombotic treatment, which might prevent thrombosis, or undergo additional screening for endometrial cancer which may identify cases at an early stage. However, within the blinded trial it would be impractical to apply such prophylaxis to all patients and it could be argued that this therefore puts patients at a greater risk than would be the case if they received tamoxifen outside the trial.

Blinding may also for instance affect compliance - a woman who is unaware of whether she is receiving active treatment or placebo may be less or more assiduous than one who knows she is taking an active treatment and again this may introduce a difference between the effect of the treatment on a woman within the trial compared with one who receives the treatment outside of the trial. In general, use of blinding can take the administration and monitoring of treatments away from the 'real life' setting and the importance of this needs to be taken into account in deciding whether blinding is necessary or desirable.

0 0

Post a comment