Matching

Allocation bias is the systematic difference in characteristics of
individuals assigned to groups. Such bias may be unconscious or otherwise,
and can occur when allocating patients to treatment groups in trials. For
example, the allocation of younger patients to a new treatment and older
patients to the control arm. This allocation bias may occur if it was
thought that younger patients would have an enhanced response to the new
treatment. The trial results would then appear more favourable for the new
treatment then they should otherwise. Random allocation to treatment
groups is the obvious way to eliminate allocation bias.

In the above case-control study, individuals were not allocated to
the case or control groups. The two groups were selected on the basis of
their disease status: glioma (the cases) or no glioma (the controls). All
cases of glioma were included. Controls were selected at random from
general practices. It is possible that selection bias occurred when
choosing the controls – healthier individuals may have been selected. No
allocation bias could exist because an individual’s group was defined by
their disease status. No allocation to groups occurred.

Competing interests:
None declared