|
 |
|
|||||||
|
|||||||||
Correspondence: When an article is eligible for submission of Correspondence, a link to the response form is available within the full-text article. You must be a current subscriber who has activated the online portion of your subscription in order to send a Correspondence. Any reader can read published Correspondence.
Correspondence to:
|
|
Correspondence:Correspondence published:
-
![[Read Correspondence]](/icons/misc/sectionDOWN.gif)
APOE influences on neuropsychological function after mild head injury: Within-person comparisons
- Alexander Collie, PhD, Paul Maruff, PhD, Marina Falleti, BSc(Hons) (6 July 2004)
-
Reply to Collie et al
- Anna Sundström, Lars Nyberg (6 July 2004)
|
|
|||
|
Alexander Collie, PhD, University of Melbourne 51 Leicester Street, Carlton South, Victoria, 3053 Australia, Paul Maruff, PhD, Marina Falleti, BSc(Hons)
Send Correspondence to journal:
acollie{at}cogstate.com Alexander Collie, PhD, et al.
|
Sundstrom et al [1] compared neuropsychological function in 11 ApoE-e4 positive and 23 ApoE-e4 negative individuals with mild head injury (MHI). The authors conducted two analyses on their data. First, they compared neuropsychological performance before and after MHI in both groups separately using paired sample t-tests. This showed significant post-injury decline in the ApoE-e4 positive group but not in the ApoE-e4 negative group.
Second, they calculated the proportion of individuals in each group whose performance declined by one SD on at least two of nine measures. This revealed a greater probability of post-injury decline within ApoE-e4 positive individuals than in ApoE-e4 negative individuals. The authors concluded that “…the risk that MHI will have a negative effect on neuropsychological functions is increased for individuals with the ApoE-e4 allele.” We believe that the conclusion cannot be supported on the basis of the analysis presented. First, the hypothesis that change in cognition was greater in ApoE-e4 carriers than in non-carriers should have been tested using a group by assessment repeated measures ANOVA. As the between groups effect sizes at the post-injury assessment are very small (<.3), we are not optimistic that re-analysis would reveal a significant group by assessment interaction. The second analysis fails to control for false classification of decline (i.e., Type 1 error). It is well known that the probability of false classification increases with the number of measures analyzed, even if the Type I error is controlled for each measure. The one-tailed probability of decline of 1 SD occurring on 2/9 tests is approximately 50% [2]. Sundstrom et al have observed no greater decline in their ApoE e4 allele group than should occur by chance. There are two ways in which the authors can reduce experiment-wise error in this analysis. First, a more strict definition of decline could be set (e.g., change greater than two SD on 2/9 tests). Second, if the authors wish to keep their definition of decline (two tests impaired at one SD) then it is necessary to reduce the number of measures studied [2]. Given the potential implications of conclusions that ApoE e4 allele predisposes an individual to worse outcome after MHI, it is important that appropriately conservative analyses are undertaken in studies addressing this issue. References 1. Sundstrom A, Marklund P, Nillson LG, et al. APOE influences on neuropsychological function after mild head injury. Within-person comparisons. Neurology 2004; 62: 1963-1966. 2. Ingraham LJ, Aitken CB. An empirical approach to determining criteria for abnormality in test batteries with multiple measures. Neuropsychology 1996; 10: 120-124. |
|||
|
|
|||
|
Anna Sundström, Psychology Department of Psychology, Umeå University, S-901 87 Umeå, SWEDEN, Lars Nyberg
Send Correspondence to journal:
anna.sundstrom{at}psy.umu.se Anna Sundström, et al.
|
||||