Does PTSD cause dementia?


Post traumatic stress disorder (PTSD) is a common condition in Australian ex-service personnel and veterans.  Depending on the type of deployment the rate of chronic PTSD among veteran’s ranges between 10-20%.  As veterans age they enter epochs associated with increasing risk of dementia.  The Vietnam era veterans are now entering their sixth and seventh decades.  As they age more and more will be exposed to the risk of Alzheimer’s disease, vascular dementia, and combined forms, as well as Lewy body dementia and frontotemporal dementias.  An important question is whether veterans suffering from PTSD are more likely to develop dementia. 


Link between PTSD and dementia


Over last 10 years interest has increased in the possible causal association between PTSD and dementia (primarily Alzheimer’s disease and vascular dementia).  A number of potential explanations have been proposed.


1. PTSD is associated with impaired cognition in domains of attention, working memory, verbal memory, new learning, and executive functions.  PTSD can be considered as a ‘disorder of memory’.  The cognitive changes seen in PTSD reduces cognitive reserve and this can predispose to development of dementia.


2. The cognitive changes seen in PTSD may in fact be very early markers for dementia developing among PTSD sufferers.


3. PTSD and dementia share common risk factors such as traumatic brain injury (TBI), low IQ, limited education, substance abuse, and risk factors for vascular disease.


4. PTSD is regarded as a stress-related condition.  Chronic stress can predispose to dementia in the following ways.


(i) Chronic stress is associated with alterations in hypothalamic-pituitary-adrenal axis function and raised pro-inflammatory cytokines.   Reduced cortisol levels and allosteric down regulation of glucocorticoid system leads to chronic CNS inflammation and increased cognitive decline.


(ii) Chronic stress is associated with damage to hippocampus and reduced hippocampal volumes.  Smaller hippocampal volume is correlated with deficits in short-term memory performance and hipppocampal atrophy is seen in Alzheimer’ disease.


5. PTSD may accelerate the general aging process.


6. Cognitive decline might ‘unmask’ PTSD in older veterans.  The combination of dementia and PTSD may cause difficult behaviour problems in these patients.


So far there has been no evidence of a causal link between PTSD and dementia.  But this has changed with the publication of two new studies.


New evidence


The two studies that have raised the possibility of a causal link between PTSD and dementia are –

Yaffe et al.  PTSD and risk of dementia among US veterans.  Arch Gen Psych 2010; 67: 608-613.

Qureshi et al.  Greater prevalence and incidence of dementia in older veterans with PTSD.  J Am Geriatr Soc 2010; 58: 1627-1633.


Study design


Both investigations are retrospective cohort studies over a seven to 10-year period (1998-2008) using US Department of Veterans Affairs (VA) administrative databases recording clinical contacts in VA clinics and facilities.  Diagnoses of PTSD and dementia are based on ICD-9-CM criteria.  Dementia diagnoses included Alzheimer’s disease, vascular dementia, senile dementia, frontotemporal dementia, Lewy body dementia, and dementia not otherwise specified.  Large numbers of subjects were involved.


Yaffe study


This study used VA National Patient Care Database coding clinical information on patients seen from 2000 to 2007.  This is an incidence study of new cases of dementia over a seven-year period and the association of dementia with PTSD diagnosis at baseline.  Dementia cases at baseline were removed from sample.  Potential confounder and shared risk factors between PTSD and dementia were assessed and controlled in data analyses (using adjusted Cox proportional hazard models giving Hazard Ratios [HR]).  Potential confounders assessed were age, socioeconomic status, sex, educational and income strata, medical co morbidity (hypertension, diabetes, ischemic heart disease, cerebrovascular disease, neuropsychiatric conditions (clinical depression, substance abuse and head injury), and number of inpatient and outpatient visits to VA clinics.


This study involved 181,000 subjects; 53,000 with PTSD, 128,000 without PTSD.  The mean age was 68 years, and 96% were men.  The cumulative incidence rate of dementia over seven years was 10.6% among PTSD patients and 6.6% among non-PTSD patients at baseline.  This was a strongly statistically significant difference (significant = *).  The uncontrolled HR for PTSD versus non-PTSD was 2.31*.  That is, those with PTSD had over twice the risk of developing dementia.  The HR after controlling for confounders was 1.77*.  No difference was found in the HR for the different types of dementia.  A comparison of PTSD with other non-PTSD psychiatric disorders revealed a HR of PTSD for dementia of 1.47*.


Qureshi study


This study used the Veterans Integrated Service Network 16 Data Warehouse database (10 medical centers in south-central USA) coding clinical information on patients seen from 1998 to 2008.  This is a prevalence and incidence study of cases of dementia over a 10-year period and examined the association of dementia with PTSD diagnosis at baseline.  Potential confounder and shared risk factors between PTSD and dementia were assessed and controlled in data analyses (using multivariate logistic regression models giving Odds Ratios [OR]).


Two characteristics were used to group subjects; PTSD or no PTSD at baseline, and Purple Heart recipient (PH) or no PH.  Soldiers receive a Purple Heart if they are physically injured in active service.  To some degree the Purple Heart qualifier might be a proxy for the intensity of combat experienced by the veteran.  This study compared four groups; PTSD+/PH-, PTSD+/PH+, PTSD-/PH+, and PTSD-/PH-.


In all 10,481 subjects were included.


Prevalence and incidence of dementia by group


N                        Prevalence %                        Incidence %


PTSD+/PH-                        3660                        11.1                                    9.5

PTSD+/PH+                        153                        5.9                                    6.8

PTSD-/PH+                        1503                        9.2                                    5.6

PTSD-/PH-                        5165                        4.5                                    4.0


Odds ratio after controlling for confounders


Prevalence OR            Incidence OR


PTSD+/PH- v PTSD-/PH-                        2.3*                                    2.2*

PTSD+/PH+ v PTSD-/PH-                        1.4                                    1.4

PTSD-/PH+ v PTSD-/PH-                        1.2                                    1.2


PTSD+/PH- v PTSD-/PH+                        2.0*                                    1.7*

PTSD+/PH+ v PTSD+/PH-                        0.6                                    0.6

PTSD+/PH+ v PTSD-/PH+                        1.2                                    1.1


The group that included PTSD positive but PH negative veterans had over twice the risk of developing dementia than PTSD negative and PH negative veterans.  This group also had nearly twice the risk of developing dementia than the PTSD negative and PH positive veterans.  This study also used anti-dementia medication use (cholinesterase inhibitors and memantine) as a proxy for dementia diagnosis – no change in pattern of results was observed.




Both studies have methodological limitations or threats to their validity.  First, administrative diagnoses may not be as valid as prospective clinical evaluation.  Second, eligibility and access to VA care may bias prevalence and distribution of predictor (PTSD) and outcome (dementia) and confounder variables (Berkson’s bias).  It is interesting to note that the PTSD prevalence in the two studies was 2-3%, much lower than usually found in surveys of combat veterans.




These studies represent a first step in establishing a causal link between PTSD and dementia.  They provide information about criteria required to substantiate causal links in medicine (the Bradford Hill criteria).


1. Strength of association (effect size etc)


Both studies showed a substantial and significant risk of developing dementia in older veterans who had PTSD at the start of the observation period.  The risk was doubled compared to subjects without PTSD.


2. Consistency of association (most studies give similar results)


The findings of both studies were consistent; both in direction of the association and the strength of the association.


3. Temporal relationship (the risk factor comes before the disorder)


Although the studies were retrospective examinations of existing databases, they used data collected prospectively and were able to assess onset of new cases of dementia and the relation to baseline PTSD.  Both studies showed that PTSD predicts an increased risk of dementia.


5. Specificity (one factor, one condition)


Both studies used multivariate statistical techniques to control for shared and confounding risk factors.  Even after controlling for these variables the risk of PTSD for dementia remained strong suggesting that the presence of PTSD has a unique contribution to risk of dementia.  However, the risk of dementia was similar across all the dementia types.  One study found that PTSD but not non-PTSD psychiatric conditions had an increased risk of dementia.  These results suggest that PTSD is a specific risk factor for dementia as a whole, but not for any particular type of dementia.


Further studies will need to assess whether a biological gradient is present; if the presence of more severe or prolonged PTSD is associated with a higher risk of dementia.




The importance of these findings demand that further studies are undertaken, especially prospective cohort studies.  Further scientific enquiry needs to examine the direct links between PTSD and different types of dementia.


It will be interesting to see if these observations can be replicated in Australian veteran populations.  How could this be achieved?  The Australian Department of Veterans’ Affairs (DVA) has administrative databases on information about accepted service-related disabilities, hospital admission diagnoses, and the use of medications by veterans.  These databases could be used to identify cases of PTSD and dementia and examine the relationship between them over time.  A very preliminary examination of this data shows trends that seem similar to the findings of the USA studies.  More work needs to be done.


Planning dementia services for veterans may need to take into account the higher risk of dementia in veterans suffering from PTSD.  There may be a case for dementia to be considered as a service-related compensable condition.


Finally, opportunities for prevention of dementia may be possible.  First, avoid or limit the development of PTSD in military personnel.  Second, identify and treat PTSD early and effectively in order to reduce the risk of dementia.  And third, early detection of cognitive impairment or dementia in veterans suffering from PTSD may provide a chance to prevent deterioration.



Prof Philip Morris.