One of the great strengths of CSP based concurrent programming languages (such as occam) is the support provided to the programmer in avoiding the creation of erroneous programs. One such support – parallel usage checking – detects program behaviours that may leave a variable in an unpredictable state. Current implementations of this check are safe but can lead to inefficient program implementations. In some cases, a simple program transformation can be used to demonstrate the safety of programs that would otherwise fail existing tests. By presenting a simple (but generic) example, I will show that using such a transformation allows the creation of more efficient programs.