Aim The European Huntington's Disease Network REGISTRY project analyses CAG repeat length measurements centrally which can be compared with the original laboratory result held at the local study site from 15 countries. We aimed to assess the extent of the discrepancies between duplicate results on the same sample obtained between January 2004 and June 2009.

Methods We subtracted the result reported by the central EHDN laboratory, BioRep, from that obtained at the local laboratory. We assessed the impact of the discrepancy at the clinically significant boundaries of 35–36 repeats and 39–40 repeats. We then re-analysed the data using a proposed acceptable measurement error of ±1 repeat for results ≤42 repeats and ±3 repeats for results ≥43 repeats. In both cases, we will present results anonymously for 10 countries which contributed more than 20 samples. We compared duplicate results for 348 samples analysed by BioRep and the Tübingen laboratory.

Results The results between BioRep and the local laboratory were discordant in 672 (51%) of the cases; in 52 (4%) cases, this resulted in a change at one of the clinical boundaries; and in five (0.4%) cases there was a potential misdiagnosis, one of which can be explained by a labelling error. When the data were re-analysed using a proposed acceptable measurement error, the number of discrepancies reduced to 129 (9.7%). Analysis of the upper allele for the 348 duplicate samples analysed at both BioRep and Tübingen gave a discrepancy rate of 6.9% but complete concordance when the proposed acceptable measurement error was used.

Conclusions Clinicians need to be aware of a measurement error when CAG results are reported. We strongly recommend that service laboratories should participate in external quality assessment schemes but in addition, should use known standards regularly to calibrate their results.