Contamination bias occurs when the members of one group in a trial receive the treatment or are exposed to the intervention that is meant for the other group. The result is a minimization of any real difference that exists between the groups.
For example, in one major study comparing ultrasound to CT for the diagnosis of nephrolithiasis, no difference was noted between the two diagnostic strategies. (Smith-Bindman 2014) However, 40% of the patients in the point of care ultrasound group also had a CT performed. Therefore, instead of comparing CT to point of care ultrasound, the study really compared using CT for everyone to more selective CT use.
Contamination bias is a type of intervention bias.
Smith-Bindman R, Aubin C, Bailitz J, et al. Ultrasonography versus computed tomography for suspected nephrolithiasis. The New England journal of medicine. 2014; 371(12):1100-10. [pubmed]