Interlaboratory comparison of autoradiographic DNA profiling measurements. 4. Protocol effects

The observed total interlaboratory uncertainty in restriction fragment length polymorphism (RFLP) measurements is sufficiently small to be of little significance given current forensic needs. However, as the number of RFLP data increase, further reduction in the total uncertainty could help minimize the resources required to evaluate potential profile matches. The large number of data available enable quantitative estimation of the within-laboratory imprecision and among-laboratory bias contributions to the total uncertainty. Some small but consistent among-laboratory measurement biases can be attributed to specific procedural or materials differences. The bias direction is often fragment-specific and thus unpredictable for unknown samples. Actions that would minimize currently recognized sources of interlaboratory bias include the following: (1) all laboratories should use the same algorithm for data interpolation, (2) all laboratories should use the same sizing ladders, (3) each laboratory should prepare control DNA and sample DNA in the same manner and with the identical reagents, (4) all laboratories should adopt a uniform policy on ethidium bromide use, and (5) all laboratories should adopt the same control DNA sizing acceptability criteria.