A family study of panic disorder: reanalysis using a regressive logistic model that incorporates a sibship environment.

Previous analysis of affection status in parents and siblings of 117 probands with panic disorder by a log-linear model for binary pedigree data found a common concordance across biological first-degree relatives and no spouse association [Hopper JL, Judd FK, Derrick PL, Burrows GD: Genet Epidemiol 4:33-41, 1987]. In this paper the data were reanalyzed using a regressive logistic model that modelled both vertical transmission and a shared sibship environment. Ascertainment correction was made by a) an "ascertainment assumption-free" procedure, following Ewens and Shute [Theor Pop Biol 30:388-412, 1986] and compared with b) complete ascertainment and c) single ascertainment. Under every scheme there was evidence for vertical transmission from parents to offspring. Inclusion of a sibship environment gave an improved fit, suggesting that vertical transmission alone may not be sufficient to explain the familial aggregation observed in these families. The effects of an affected parent, of the postulated environmental factor (present for all siblings if it was present for one sibling), and of the prevalence of the rare environmental factor were estimated and found to be roughly similar under the different schemes. Model predictions of lifetime prevalence were consistent with other population-based studies. Under the same assumption-free method, standard errors approximately doubled and computation time increased compared with the other ascertainment schemes that made specific, although not necessarily correct, assumptions. The regressive logistic model used less computation time and gave greater insight into the pattern of familial aggregation than did the previous modelling.