The use of retained patient specimens for haematology quality control.

Patient blood specimens constitute ideal quality control material in many respects. Although stability is a problem, patient specimens are sufficiently stable to allow their use in the control of short-term systematic error. The principal challenges involve the design of a system which combines excellent performance characteristics (probability of error detection and probability of false rejection) with a minimum of extra work. In the past, guidelines have been presented for an optimized quality control program using retained patient specimens in haematology. These guidelines call for the use of three retained specimens initially analysed only once, with subsequent analyses judged 'out of control' if they deviate from the initial result by a prescribed multiple of the long-term standard deviation. In practice, this system may result in a relatively high probability of false rejection of data (Pfr) due to inadequately established control ranges. Also, the maintenance of three retained patient specimens may be an excessive burden on small laboratories. We present data on the optimization of a quality control program using one retained specimen that is appropriate for use by smaller laboratories. The control rules and the number of initial analyses are adjusted to yield the highest possible probability of error detection (Ped while maintaining a low Pfr and a low additional workload. In addition, we present data concerning the control of satellite instruments by sharing patient specimens between the satellite instrument and a 'reference' instrument.