The results confirmed our hypotheses that PTSD would be associated with decreased performance on measures of verbal learning, working memory, attention, and processing speed. Alcohol abuse history was associated with decreased performance on a measure of immediate visual memory. There were no interaction effects of PTSD and alcohol. After covarying for several potential confounds—including education, an estimate of intellectual functioning, depression, current level of alcohol use, and drug history—differences between individuals with and without PTSD remained significant. Given these results, we can more conclusively determine that lower scores on tests of verbal learning, working memory, attention, and processing speed among the participants who were PTSD+ in this sample are not due to the confounding effects of alcohol or depression.
Our findings suggest that memory deficits are not global but restricted to specific tasks. On memory measures, veterans with PTSD showed deficiencies relative to veterans without PTSD in the initial registration and learning of word lists but not on retention. Although we found differences on CVLT performance, we did not see differences between participants who were PTSD+ or PTSD– on paragraph recall, suggesting that verbal memory may be more compromised when the task does not involve a narrative context. In the group with PTSD, working memory was impaired when the task involved auditory material (i.e., Digit Span, Letter Number Sequencing) but not visual material (Spatial Span). Similarly, participants with PTSD did not exhibit deficits on immediate or delayed visual memory. Consistent with previous research (Vasterling et al., 1998
; Yehuda et al., 1995
), these findings of weaknesses in tasks involving attention, working memory, and new learning are suggestive of frontal-limbic abnormalities.
These results also add to the alcoholism literature, suggesting effects of alcohol abuse history on short-term visual memory performance. Because visual memory but not verbal memory was compromised in our participants with significant alcohol abuse histories, our findings offer partial support for the “right hemisphere hypothesis” of alcoholism, suggesting that the right hemisphere may be particularly vulnerable to the effects of alcoholism. These results are consistent with other studies showing impaired visual memory in recently detoxified alcoholics (Donat, 1986
; Leber et al., 1981
). However, in our sample there was not a relationship between frequency of recent alcohol use and visual memory performance, suggesting that impairments seen are due to a longer term history of alcohol abuse or dependence or to genetic or early developmental risk factors for alcohol abuse. These findings highlight the continued need for alcohol abuse researchers to examine visual–spatial and visual memory performance in alcoholics well past the acute withdrawal period. They also suggest the need for familial studies of gene markers and neurocognitive testing to determine if visual–spatial and visual memory deficits are heritable risk factors for alcohol abuse.
Because the groups that were PTSD+ had significantly lower scores on a measure of attention, it was possible that observed differences in verbal memory actually reflected impaired attention. After covarying for the effects of passive auditory attention, the differences on CVLT measures, Letter Number Sequencing, and Digit Symbol remained significant. These results suggest that difficulties in both verbal learning and memory as well as set shifting and mental flexibility are not merely an artifact of reduced passive auditory attention. However, it may also be that the measure of attention used in this study is not sufficiently sensitive to detect alterations in attention that might account for effects on memory. Future studies might continue to study this hypothesis, using a measure of attention such as the Continuous Performance Test (Conners, 1994
Several limitations should be noted when interpreting the results of this study. First, it is difficult to adequately identify individuals with significant alcohol abuse histories. By using the categorical DSM–IV
diagnosis to identify our participants with a history of alcohol abuse, we may be including individuals in our groups with ETOH– who are binge drinkers yet do not meet the diagnostic criteria of abuse or dependence. In addition, many of our participants with a history of alcohol abuse were presently in remission, with some having been sober for nearly 5 years. As alcohol effects have been most consistently identified in current or newly abstinent drinkers (e.g., Beatty, Tivis, Stott, Nixon, & Parsons, 2000
; Grant, 1987
), it is certainly possible that the variation in recency of alcohol abuse or dependence limited our ability to detect a significant effect for alcohol. We attempted to account for this limitation by covarying for current alcohol use in secondary analyses of measures that detected differences in individuals who are PTSD+ and PTSD–; all differences previously seen remained significant. However, our study had limited power to detect effects of recent drinking given the small number of recent drinkers. It is important to note that post hoc analyses revealed the group PTSD+/ETOH+ did not differ from the group PTSD–/ETOH+ on number of drinks in the past 3 months, underscoring our success in controlling for recent heavy alcohol use, arguably the most important PTSD/alcohol confound in neurocognitive testing.
Ideally, neuropsychological studies comparing group differences should match participants in each group on key demographic variables, such as age and education, which will likely have an impact on the dependent variable. When groups are not matched, as was the case in the current study with education level, a less satisfactory approach applies ANCOVA, with differences in demographic variables controlled for (Adams, Brown, & Grant, 1985
). It can be particularly difficult to match participants in PTSD and alcoholism research on education level, but researchers should continue to strive to do so.
In addition, our neuropsychological battery did not include systematic measures of executive function. Past studies have included measures of executive functioning (e.g., Stein et al., 2002
), and future researchers might continue to test these skills by using a design similar to ours. Finally, it is possible that exposure to toxins such as Agent Orange may worsen veterans’ neuropsychological functioning, a possibility not explored in the present study. Future studies might explore in more detail the deployment and toxic exposure histories of enlisted military personnel. Recognizing these limitations, the present findings add to the literature documenting neuropsychological differences in individuals with and without PTSD. Verbal and working memory difficulties were more prominent than visual impairments in veterans with PTSD.
Most important to note is that these differences cannot be accounted for by alcohol abuse history, depression, or attentional disturbances.