In an article published in the July 1994 issue of the Archives, Silverman et al1 reported on the risks of first-degree relatives of patients with Alzheimer's disease for primary progressive dementia (PPD). They used the family history method for the evaluation of relatives and survival analysis for statistical calculations. The most remarkable finding was a reduction of the hazard rates in the 85- to 89-year age interval in the patient sample and, consequently, the reduction of the familial contribution to the risk for PPD compared with relatives of controls in this age range.
The study had two important shortcomings: first, relatives had not been interviewed personally; second, different informants provided family history information (ie, one or two relatives in the patient group and in the first control group and the index proband in the second control group). When using family history information, it might be that an age-dependent variation