Effect of laboratory or clerical error on presymptomatic risk calculations for Huntington disease: a simulation study

Linked markers are useful in prenatal diagnosis as well as presymptomatic diagnosis in late age-of-onset diseases such as Huntington disease (HD). It is widely assumed that most laboratory or clerical errors will be detected because of incompatibility of marker haplotypes within the family. However, errors in marker phenotypes that are compatible but wrong may result in a consultand being given an incorrect risk estimate. We have addressed this issue using simulated marker data in pedigrees similar to those seen in our HD testing program. In Family Structure I (an 11-member, 3-generation family), a particular family was more likely to be detected as inconsistent than incorrectly assigned. In a small nuclear family (Family Structure IV), fewer errors would be detected, and more would appear consistent but give incorrect risk estimates (e.g., low risk misclassified as noninformative or high). Given the presence of tight linkage, risk estimates are often calculated based on a small number of relatives. However, these computer simulations demonstrated that increasing the number of relatives typed decreases the probability that the family will remain consistent with an error present, and, therefore, decreases the probability of an incorrect assignment of risk. It is important to decrease the level of such errors by duplicated readings of raw marker data and validation of computer input.