91Ö±²¥

These include simple methods which are commonly used in medical literature and may be subject to selection bias if patients switching are not typical of the population as a whole.

Methods which attempt to adjust the estimated treatment effect, either through adjustment to the hazard ratio or accelerated failure time models, are also considered.

A simulation study was conducted to assess the performance of each method in a number of different scenarios.

Results

16 different scenarios were identified which differed by the proportion of patients switching, underlying prognosis of switchers and size of true treatment effect. 1000 datasets were simulated for each of these and all methods applied.

Selection bias was observed in naïve methods when the difference in survival between switchers and non-switchers was large.

A number of methods, particularly the AFT method of Branson and Whitehead were found to give less biased estimates of the true treatment effect in these situations.

Conclusions

Simple methods are often not appropriate to deal with treatment switching. Alternative methods such as the Branson and Whitehead method to adjust for switching should be considered.

DSU report

Assessing methods for dealing with treatment switching in randomised clinical trials (PDF, 1.7MB) (July 2010)

Related publication

J Morden, P Lambert, N Latimer, K Abrams, A Wailoo (2011) BMC Medical Research Methodology; Vol 11:4