The different frequencies of HLA-B*5701 found in patients from different African countries strongly suggest greater utility of HLA-B*5701 screening in some African groups compared with others. Although ethnic-based

pharmacogenetic screening in UK clinics with diverse populations is unlikely to be practical and also likely to be ethically unacceptable, our figures add to the need for caution when considering screening in diverse African settings. Despite an overall sample size of 1502, only two sub-divisions had a sufficiently large sample to allow meaningful interpretation of the prevalence rate [White/Eurasian and Niger-Congo with prevalence rates of 7.95% (CI 5.88%–10.02%) and 0.52% (CI 0.18%–1.52%), respectively]. On the basis of previous GDC 0199 work, we sub-divided our African patients according to linguistic index that language groups might reflect genetic structure [9]. However, more recent data suggest that this

may not necessarily be true in all settings and particularly not African ones [14]. In contrast, a recent, well-publicized study of 121 African populations demonstrated genetic clustering across the Niger-Congo R428 (Niger-Kordofanian) linguistic populations [15]. Our data, where both Ugandan and Zimbabwean populations were classified as Niger-Congo (Bantu) but had very different HLA-B*5701 prevalences, demonstrate the difficulties in using such classifications to distinguish populations either in genetic studies of specific, non-neutral alleles. Our study was not able to distinguish northern (Nilotic) Ugandans from Bantu Ugandans, so it is possible that the different rates among

Zimbabweans and Ugandans were because of cross-classification. However, a significant majority of Ugandans in the United Kingdom are thought to be Bantu and HLA-B*5701 frequency among Nilotics has been reported to be lower than among Bantu [4]. The 244 unclassifiable subjects underline the difficulties in basing decisions on these self-reported measures of ethnicity. Only one of the three HLA-B*5701 positive subjects in the Niger-Congo sub-division self-reported as genetically homogenous reflecting the potential of genetic admixture to complicate analysis. Future studies may consider the use of ancestry information bio-markers to define population groups more accurately. The single false-positive result in a local laboratory reinforces the need for robust laboratory quality assurance. It is reassuring that Sequence specific primer (SSP) methodology failed safe by identifying the patient as HLA-B*5701 positive; by doing so it ensured patient safety was maintained. In the present study, HLA-B*5701 prevalence in the UK was similar to previously reported rates in White HIV-infected subjects but considerably lower than those reported in Black HIV-infected subjects, probably as a result of the large proportion of Black subjects that were of African origin. In addition, 99.