We thank Drs Prasher & Haque for their interest in our paper and are pleased that they agree with our conclusions. It is quite obvious that the difference in findings in the meta-analysis between our study and Prasher et al's study was due to the inclusion of data in our study that were not available at the time of Prasher et al's study. According to our calculation, our meta-analysis has 92% power (95% CI 88-96%) at the 5% level. However, traditional power calculation is not applicable in this case because instead of simply adding allele frequencies among all studies, we have used the computerised version of the Woolf (Reference Woolf1995) method of meta-analysis that takes account of each study individually. Also, because of the varied nature of studies included in the meta-analysis we did not feel it appropriate to calculate specificity and sensitivity in the traditional way.

It was not stated in Prasher et al's (1997) paper which 31 patients (15 with and 16 without dementia) out of 40 patients with Down's syndrome, presented in Wisniewski et al's (1995) study, were included in their meta-analysis. The age of death of patients reported in Wisniewski et al's study ranged widely between 15 and 69 years. They mentioned at the bottom of their table that “The presence of dementia is defined as a deterioration of competence, as judged by the physician following the patient”. No detail about diagnosis was mentioned in the text and no patient over age 30 had an ϵ4 allele. For these reasons we chose not to include this study in our meta-analysis. However, as Prasher & Haque point out inclusion of this study would have made little difference to our findings.

Whereas Prasher & Haque rightly suggest that further research is needed to clarify the role of ApoE ϵ4 in Alzheimer's disease in people with Down's syndrome, we were surprised to see that they have recommended ApoE genotyping as a possible screening test for dementia in this population. This will be totally inappropriate at this stage considering the uncertain relationship between Alzheimer's neuropathology and ApoE genotype in people with Down's syndrome, as we mentioned in the last paragraph of the Discussion in our paper.

We agree with Prasher & Haque that the presence of ϵ4 allele is neither necessary nor sufficient for the development of Alzheimer's disease.