We thank Drs Saxon and McFall for their thoughtful comments on the issue of PTSD, dementia risk, and tobacco use. Our study clearly shows that in a very large population of veterans, those with diagnoses of PTSD are much more likely to develop dementia than those without PTSD. As we noted in the article, this association does not necessarily prove causation, and unmeasured confounding is certainly one of the potential explanations for this association.1 Large administrative databases have the advantages of studying these issues at the population level but often have incomplete reporting of confounders. In our study, history of tobacco use was determined by the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code V15.82 in the Department of Veterans Affairs National Patient Care database.2 As acknowledged in the limitations, ICD-9 codes are insensitive and most likely underestimate smoking and other comorbidities in the general and Veterans Affairs population. However, we did indeed adjust for history of tobacco use in the multivariate hazard models. While smoking is clearly associated with almost all psychiatric disorders, including PTSD, the extent and magnitude of this association with dementia are controversial,3 but even the highest estimates of a smoking-dementia association would not be sufficient to explain the PTSD-dementia association in our study. Based on this and with our statistical adjustment, we feel that our results are credible and merit further studies on the relationship between dementia and PTSD.
Yaffe K, Vittinghoff E, Lindquist K, Barnes D, Covinsky KE, Neylan T, Kluse M, Marmar C. Tobacco Use Data Questioned in Veterans Study—Reply. Arch Gen Psychiatry. 2010;67(12):1324-1325. doi:10.1001/archgenpsychiatry.2010.167