Ca. Donnelly et al., ANALYSIS OF THE BOVINE SPONGIFORM ENCEPHALOPATHY MATERNAL COHORT STUDY - EVIDENCE FOR DIRECT MATERNAL TRANSMISSION, Applied Statistics, 46(3), 1997, pp. 321-344
In an initial exploratory analysis of the bovine spongiform encephalop
athy (BSE) maternal cohort study data we demonstrate several confoundi
ng effects in the study design. Given these effects, we assess a varie
ty of statistical models to determine the relative contributions of di
rect maternal transmission of the aetiological agent of BSE and of gen
etic susceptibility to the observed maternally enhanced risk of BSE in
the offspring of affected dams. To control for the substantial betwee
n-herd variation in the risk of exposure to the BSE agent it is essent
ial that analyses take into account the matched pair structure of the
data. Maternal exposure is estimated to be most important in animals b
orn within 150 days of disease onset in their dams. The analysis of a
full survival likelihood model indicates that the hypothesis of matern
al transmission with no genetic variation in susceptibility fits the d
ata significantly better than the hypothesis of genetically variable s
usceptibility with no maternal transmission. However, models incorpora
ting both maternal transmission and genetically variable susceptibilit
y fit the data significantly better than pure maternal transmission mo
dels. Although genetic susceptibility cannot be excluded as the cause
of the cohort study results in the absence of detailed genotyping, the
analysis of these study data suggest that low level maternal transmis
sion of BSE is. at least in part, responsible for the significantly en
hanced risk of BSE in the offspring of affected dams. Similar results
indicating significant maternal transmission in the later stages of th
e dam incubation period are obtained from the independent analysis of
data on the dam-offspring relationships among confirmed BSE cases.