Inhibitory Control and Alcohol Use History Ch 2

Chapter 2: INHIBITORY CONTROL & ALCOHOL PREDICT PTSD CHANGE Associations between Baseline Demographic and Clinical Variables with PTSD Symptom Changes In our characterization of baseline variables predicting PTSD changes over time, before examining our measures of interest we first sought to examine demographic and clinical variables predicting PTSD changes. As can be seen in Table 2, correlations between demographic/clinical variables with PTSD symptoms changes (RCISRB) failed to reach significance. We found a very similar pattern when examining demographic, clinical, and daily life functioning differences between groups that showed either Improved, Chronic, or Worsened PTSD symptoms (see Table 2). Table 2 Associations between Baseline Demographic and Clinical Measures and PTSD Changes Over Time Note. Mean ± standard deviation, p-values are from ordinal ANOVAS and χ2 tests comparing the Worsened PTSD, Chronic PTSD, and Improved PTSD groups as well as from correlations (Pearson) between the change in PTSD (RCISRB) and each clinical variable across the full sample (N = 187). q-values indicate significance after .05 false-discovery rate correction. Highlighted cells indicate significantly related to PTSD changes. WTAR = Wechsler Test of Adult Reading, CAPS-IV = Clinician Administered PTSD Scale for DSM-IV, DASS = Depression, Anxiety and Stress Scale, mTBI = mild traumatic brain injury, WHODAS II = World Health. *p < .05. † indicates categorical variables where t-tests or univariate ANOVAs were run with RCI between groups, and Chi-squared tests were run across categorical and PTSD change groups. Association between Baseline Alcohol Use and PTSD Symptom Change We next examined the hypothesized relationships between baseline current and lifetime alcohol use with PTSD symptom changes. For baseline current alcohol use, we first performed correlations between average and maximum drinks on a drinking day. We found that drinks on a drinking day at baseline did not significantly predict PTSD RCISRB change (Average: r = .108, p = .121; Max: r = .073, p = .32). Consistent with this, we also failed to find significant differences between Worsened, Chronic, and Improved groups in baseline average drinks on a drinking day model (χ2 (2) = 2.695, p = .102). However, the maximum drinks on a drinking day model significantly differed between groups (Maximum: χ2 (2) = 4.590, p = .033). We also found that the Improved PTSD group had a numerically lower rate of baseline current alcohol use disorder (0.00%) compared to Worsened (17.86%) or Chronic (20.00%) PTSD groups, but this failed to reach significance (χ2 (2) = 4.932, p = .085). We next examined the relationship between lifetime drinking history and PTSD RCISRB change. We found that higher lifetime alcohol consumption at baseline was significantly associated with worsening PTSD symptoms (r = .24, p = .001, see Table 2). Consistent with this, we found that the Worsened PTSD group had significantly greater lifetime alcohol consumption compared to the Chronic or Improved groups (t(2) = 6.731, p = .010). The difference was driven by the Worsened and Improved groups (Improved M = 7.69, SD = 1.40, Worsened M = 6.69, SD = 1.97, t(147) = .042). Other comparisons between alcohol measures and PTSD change groups were not significant. To better characterize these differences in lifetime drinking, similar to previous studies (e.g., Jacob et al., 2013) we next examined the distinct drinking phases from the LDH starting from the earliest age at which at least one alcoholic drink per month was consumed regularly (see Figure 1 for a comparison on the Worsened, Chronic, and Improved PTSD groups). Using a repeated-measures ANCOVA with average drinks on a drinking day as the dependent measure, lifetime drinking phase as a factor (1, 2, or 3) and PTSD RCISRB as a covariate of interest, we found a significant main effect of phase (F(158) = 4.146, p = .043), where phase 1 drinking was significantly lower than phase 2 and 3 (M = 5.557; 7.682; 7.907 respectively, p < .001). We also found a significant interaction between phase and RCISRB (F(158) = 6.507, p =.012), where there was a greater relationship between RCISRB and average drinks on a drinking day for phase 3 (r = .175, p = .027) than phases 1 or 2 (r = -.033 and -.029; p = .659 and .696). We found a similar pattern of results when examining the Improved, Chronic, and Worsened PTSD groups. Specifically, a repeated-measures ANOVA revealed a significant Group x Phase (phases 1-3 included) interaction when predicting average drinks on a drinking day (F(2,157) = 3.96, p = .004). As can be seen in Figure 1, compared to phase 2, during phase 3, the Improved PTSD group reduced their average drinks on a drinking day, whereas the Worsened and Chronic PTSD groups increased their average drinks. In phase 3, the Worsened PTSD group’s average number of drinks on a drinking day (M = 10.35, SD = 9.91) was nearly double that of the Improved PTSD group (M = 5.24, SD = 3.97, p = .026) and a similar pattern was observed for the max drinks on a drinking day (Worsened PTSD M = 16.52, SD = 12.60; Improved PTSD M = 8.74, SD = 7,54, p = .028) and the rate of binge drinking during this phase (85.71% of the Worsened PTSD; 47.62% of the Improved PTSD group, p = .004). This divergent pattern between the Improved PTSD group and Worsened PTSD group remained for all subsequent phases. The shift during phase 3 in individuals’ early-to-mid 20s is notable, as it coincides with the average age of military enlistment at 21 (Reynolds and Shendruk 2018) and suggests that drinking more versus less during this time has important implications for recovering from post-deployment PTSD. Figure 1 Lifetime Drinking History of Groups with Worsened, Chronic, or Improved PTSD Note. A. Each line represents the average drinks on a drinking day within each phase for the Worsened PTSD (red), Chronic PTSD (green) and Improved PTSD groups (blue). B. The distribution of average drinks on a drinking day for each phase displayed for the Worsened PTSD (red), Chronic PTSD (green) Improved PTSD groups (blue). The error bars reflect the standard deviation around the mean for each group. Within the table, Drinking Phase (%) represents the percent of individuals within each group who reported a new phase of drinking, as not every individual had multiple phases of drinking throughout their lifetime. A phase shift represents a clinically meaningful change in drinking behaviors defined by quantity and/or frequency. Age reflects the age at which the drinking phase started. Days/month represents the frequency of drinking days per month, and phase length reflects the length of time (in months) that those drinking behaviors lasted. Asterisks represent significant group differences between the Improved, Chronic and Worsened PTSD groups. Association between Baseline Cognitive Performance and PTSD Symptom Changes We next examined relationships between baseline cognitive performance and PTSD changes over time, focusing on the hypothesized executive function subdomain of inhibitory control. As can be seen in Table 3, We found that better inhibitory control performance (composite of Color-Word inhibition and Inhibition/Switching) was significantly associated with greater PTSD improvements over time (RCISRB), r = -.183, p = .017, with the Inhibition/Switching task having a numerically stronger relationship with PTSD change (r = .192, p = .012) than the Inhibition task (r = -.13, p = .094). When examining the composite of executive functions excluding inhibitory control measures (Trail Making Test Number/Letter Switching, FAS-Letter Fluency, Category Fluency, Switching, IED, ACT), we did not find a significant relationship with PTSD change (r = -.083, p = .262), suggesting that this association is more specific to inhibitory control rather than other executive functions. We further confirmed the specificity of inhibitory control by performing stepwise linear regression models predicting PTSD RCISRB scores using all of our EF measures. Both backward and forward stepwise regressions identified the model with composite inhibitory control as the only predictor resulting in the best model fit (F(1,166) = 5.768, p = .017). Table 3 Associations between Baseline Cognitive Measures and PTSD Changes Over Time Note. Mean ± standard deviation. p-values are from ordinal ANOVAS comparing the Worsened PTSD, Chronic PTSD and Improved PTSD groups and correlations (Pearson) between change in PTSD (RCISRB) and each cognitive variable. q-values indicate significance after false-discovery rate correction. Highlighted cells indicate significantly related to PTSD changes. CVLT-II= California Verbal Learning Test-Second Edition, D-KEFS=Delis-Kaplan Executive Function System, IED=Intra-Extra Dimensional Set Shift, ACT =Auditory Consonant Trigrams. *p < .05. Corroborating these findings, we found that composite inhibitory control also significantly differed between the Improved, Chronic, and Worsened PTSD groups (Worsened: M = -0.15, SD = 0.85; Chronic: M = 0.02, SD = .82; Improved: M = 0.39, SD = 0.47, χ2 (166) = 5.018, p = .026), where the difference was driven by the Improved group, which had significantly greater composite inhibitory control scores than both the Worsened and Chronic groups (t(43) = -2.569, p = .014; t(141) = -1.995, p = .048, respectively). Finally, when examining verbal learning and memory composite from the CVLT-II, we did not observe any associations with PTSD symptom change or differences between Worsened, Chronic, and Improved groups (all p’s > .50). Comparing Lifetime Drinking History and Inhibitory Control in Predicting PTSD Symptom Change Our results showed that both greater lifetime alcohol consumption and worse inhibitory control at baseline were related to worsening PTSD symptoms over a two-year period. Previous studies have observed significant associations between inhibitory control performance and drinking behaviors (Carbia et al. 2018), though in the current study these measures were not significantly correlated (lifetime alcohol consumption and composite inhibitory control, r = .066, p = .396). A regression model including the composite inhibitory control and lifetime alcohol consumption as predictors of PTSD symptom change (RCISRB) was significant (F(2,164) = 7.672, p = <.001, adjusted R2 = 0.074) and both composite inhibitory control and lifetime alcohol consumption significantly predicted unique variance in PTSD symptom changes (betas = -.173 and .225; p = .022 and .003, respectively). Both variables remained significant predictors when controlling for additional covariates (estimated premorbid IQ, age, and education), which did not account for any additional variance in PTSD symptom changes (see Table S3 in the Supplementary Materials). Can Treatment Engagement Explain Individual Differences in PTSD Change? There were no differences between Worsened, Chronic, and Improved groups in treatment participation (see Methods for treatment participation description), including the three broad treatment types (i.e., psychotherapy only, medication management only, and combined treatment; see Supplementary Materials), during the interval between baseline and follow-up (χ2 (2) = 1.42, p = .491). Specifically, 71.43% of the Worsened PTSD group, 84.00% of the Chronic PTSD group, and 61.91% of the Improved PTSD participated in some form of treatment (e.g., medication management and/or psychotherapy). All analyses were repeated after removing those participants with missing treatment frequency data, and group differences yielded identical results (see Supplementary Results and Supplementary Table S8). Thus, additional factors, such as inhibitory control and lifetime drinking history, likely contribute to PTSD symptom trajectories above and beyond the effects of engaging in treatment. Associations between PTSD Changes with Changes in Inhibitory Control and Alcohol Use Though a full examination of relationships between changes in clinical and cognitive variables with PTSD symptom changes is beyond the scope of this study, these analyses can be found in the supplementary materials. Relevant to the current findings, we did find that increases in the average and max drinks consumed on drinking days between the two timepoints were not significantly associated with increased PTSD symptoms over time (r = .102, p = .165; r = .057, p = .443, respectively). All groups decreased their max drinks on a drinking day between timepoints, and for average drinks on a drinking day the Worsened group alone numerically increased, albeit non-significantly (baseline: M = 8.18 SD = 5.32; follow-up: M = 8.50 SD = 6.38; q = .975, see supplementary materials). Further, we found that PTSD symptom changes were not associated with changes in the composite inhibitory control score (r = -.131 p = .095). In general, inhibitory control scores improved over time for all three groups, though this wasn’t significant in any single group after FDR correction except in the Chronic PTSD group (baseline M = .03, SD = .80; follow-up M = .35, SD = .88; q < .001). Discussion This study utilized a well-characterized sample of 187 post-9/11 Veterans with a diagnosis of baseline PTSD who were assessed approximately two years later and examined predictors of continuous PTSD symptom changes as well as differences between subgroups with significantly Improved, Worsened, or Chronic PTSD. Our results demonstrate, both when using a continuous measure of PTSD symptom change and when examining the PTSD change subgroups, that better baseline inhibitory control abilities and lower lifetime drinking history were associated with greater PTSD symptom improvement over a two-year period. For alcohol consumption, we found that Veterans’ drinking history over the past 10-15 years was more associated with change in PTSD symptoms than current alcohol consumption. Together, these findings support trait/pre-existing factor models of inhibitory control and drinking behavior in PTSD, such that stronger inhibitory control abilities at baseline and reduced drinking over a broader time period are protective against experiencing more chronic or worsening PTSD symptoms. These findings provide insights into the factors affecting the progression and chronicity of PTSD and have important treatment implications. In Veterans with a baseline diagnosis of PTSD, we found that improvements in PTSD symptoms over a two-year period were significantly associated with better baseline inhibitory control performance, though PTSD improvements were not significantly associated with other executive functions or cognitive measures. This specific effect was also found when examining the separate PTSD change subgroups (see Figure 2), with the Improved PTSD group showing significantly better baseline inhibitory control performance than the Chronic and Worsened PTSD groups. These findings suggest that inhibitory control deficits play an important role in the maintenance of PTSD symptoms and that individuals with PTSD and above-average inhibitory control abilities may experience better long-term PTSD outcomes compared to those with poorer inhibitory control abilities. This is consistent with studies demonstrating that poorer pre-trauma inhibitory control is a risk factor for developing PTSD after trauma exposure (Samuelson et al. 2020), that inhibitory control is specifically related to greater PTSD symptoms beyond general executive functions in a trauma-exposed population (DeGutis et al. 2015), and that inhibitory control deficits play an important role in maintaining PTSD symptoms (Aupperle et al. 2012). The current study extends these studies by demonstrating that, even in those who meet diagnostic criteria for PTSD, above-average inhibitory control abilities can serve as a protective factor against chronic or worsening PTSD symptoms. These results are more consistent with a model of PTSD in which inhibitory control represents a pre-existing risk factor or protective trait rather than strictly a consequence of PTSD (e.g., PTSD causing increased distractibility and reduced response inhibition). Though we found that better baseline composite inhibitory control performance predicted PTSD symptom improvements, PTSD symptom changes between time points were not significantly associated with changes in composite inhibitory control performance over time (r = .021, p = .793). This emphasizes the importance of more stable, inhibitory control abilities to PTSD chronicity rather than inhibitory control tracking PTSD severity. This is consistent with results from a recent PTSD study of trauma-focused therapy. In particular, Crocker et al. (2018) found that, in OEF/OIF Veterans with PTSD, worse baseline executive functioning (including Inhibition, Inhibition/switching, and working memory) predicted poorer cognitive processing therapy (CPT) outcomes. Further, the addition of compensatory cognitive training targeting executive functions to CPT (SMART-CPT) allowed those with worse inhibitory control to benefit more from therapy (Crocker et al. 2018). These findings, along with the current results and neuroimaging findings during inhibitory control tasks predicting treatment responses (Falconer et al. 2013, Bryant et al. 2021), support a model in which pre-existing inhibitory control abilities affect the maintenance of PTSD symptoms as well as treatment success (Aupperle et al. 2012). Contrasting this model, Haaland et al. (2016) found that, in female Veterans with PTSD, only baseline learning/memory performance predicted PTSD trauma-focused treatment success whereas inhibition and inhibition/switching improvements were associated with PTSD improvements after treatment. Additional studies would be useful to further characterize whether inhibitory control plays more of a prominent role as a pre-existing factor/predictor of PTSD chronicity and treatment success versus a correlate of PTSD symptomatology, as well as determine if this relationship is moderated by gender or type of trauma exposure. In addition to showing the importance of inhibitory control to PTSD chronicity, the current results also found that alcohol use significantly predicted changes in PTSD symptoms. Importantly, lifetime drinking history over the last 10-15 years (starting around the age of typical military enlistment) was a stronger predictor of PTSD change than current alcohol use. In particular, the Improved group had less lifetime alcohol consumption than the Chronic and Worsened groups, suggesting that lower lifetime alcohol consumption may be a protective factor against more long-term PTSD. We observed a significant differentiation in drinking habits that occurred in the early-to-mid 20s and was maintained throughout subsequent drinking phases, with the Improved PTSD group decreasing their alcohol consumption during this period and with Worsened/Chronic PTSD groups increasing alcohol consumption. To our knowledge, this is the first study to demonstrate a relationship between alcohol-related behaviors across lifetime drinking phases and PTSD symptom changes. Though the greater association of previous versus current drinking behavior to PTSD changes is notable, it could be possible that previous drinking behavior is simply a more reliable/valid measure than current drinking because participants may be more willing to endorse ‘past’ alcohol problems and may minimize or under-report current drinking (Neumann et al, 2009). Our results demonstrating the importance of drinking history to PTSD changes along with finding that alcohol use change over two years show little-to-no correlations with PTSD changes provide support for the self-medication rather than the mutual maintenance model of alcohol use in PTSD. Both models predict that PTSD is associated with higher alcohol use (Veterans in the current study had ~2-4x the alcohol consumption of the general US population; Chan et al. 2009), but only the mutual maintenance model predicts that increases in alcohol use would significantly worsen PTSD symptoms (Simpson et al. 2014). Inconsistent with mutual maintenance, we found that increases in drinking behavior between timepoints was not significantly correlated with worsened PTSD symptoms (average drinks: r = .102, p = .165; max drinks: r = .057, p = .443). This suggests that increases in alcohol consumption did not appreciably lead to worse PTSD symptoms and that worsened PTSD symptoms did not lead to greater alcohol consumption, which is in line with previous longitudinal studies across two years (Nickerson et al. 2014; Langdon et al. 2016; Kaysen et al 2006, 2011). Though these results are promising, using more fine-grained temporal assessments of alcohol consumption and PTSD symptoms (e.g., ecological momentary assessments) over days/weeks rather than months/years would be important to confirm these findings (Khantzian, 1997; Tripp 2020). Taken together, these results suggest that PTSD symptom chronicity is not significantly related to baseline alcohol consumption (except when considering maximum drinks), nor is it related to changes in alcohol consumption between time points. Rather, the results support the importance of lifetime drinking quantity on PTSD symptom changes. One possibility is that lifetime drinking history may better capture more trait-based or pre-existing differences in behavioral impulsivity (Gröpper et al. 2016; Nigg et al. 2006) than current alcohol consumption. For example, in phase 3 the Improved Group, compared to the Chronic and Worsened groups, may have had more self-control (or exerted more self-control) that allowed them to dampen their drinking behaviors and this was maintained through their subsequent drinking phases. Behavioral impulsivity is a multifaceted personality disposition related to deficient emotional or behavioral self-control in everyday life decisions (Duckworth and Kern, 2011; Whiteside and Lynam, 2001), whereas inhibitory control represents the ability to override impulses to accomplish a goal-directed task (Sadeh et al. 2015). Behavioral impulsivity is consistently related to higher alcohol use (e.g., in Veterans with PTSD symptoms, see Mahoney et al. 2020) and several aspects of behavioral impulsivity could also play a role in maintaining PTSD symptoms, such as difficulty with emotional regulation and a propensity towards risky behavior (a PTSD diagnostic item in the DSM-5). Additionally, PTSD symptoms such as hyperarousal could further contribute to aspects of behavioral impulsivity, especially through negative urgency, or the tendency to act impulsively under negative affect (Roley et al. 2017). Future work would be useful to examine whether specific aspects of behavioral impulsivity are related to the chronicity of PTSD symptoms. The results of the current study have important clinical and treatment implications. First, they suggest that individuals with PTSD and poorer inhibitory control and/or greater lifetime drinking history are at an increased risk for chronic PTSD and should be more closely monitored and potentially targeted with interventions. All groups generally improved in the inhibitory control tasks across timepoints, suggesting that improvements may be a practice effect. However, Worsened and Chronic groups never reached the Improved group’s performance on inhibitory control tasks (see Supplementary Table S3). Our results suggest that interventions specifically targeting inhibitory control may be particularly beneficial and, similar to Crocker et al. (2018), could be readily paired with trauma-focused therapies to reduce PTSD symptoms. These include computer-based inhibitory training programs shown to improve PTSD symptoms such as attentional control training (Badura-Brack et al. 2015; Lazarov et al. 2019), response inhibition training (Echiverri-Cohen et al. 2021), and interference control training (Bomyea et al. 2015). They could also include inhibitory control enhancing pharmacological interventions such as methylphenidate (McAllister et al. 2016) and cognitive skills training targeting inhibitory control (Fortier et al. 2018; Maraver, Bajo, and Gomez-Ariza 2016). The current results also highlight the importance of broadly assessing alcohol use across the lifespan beyond current drinking in individuals with PTSD. Clinicians might benefit from focusing more on the onset of problem drinking, earlier life patterns of alcohol use, and the historical function of alcohol in an individual’s life. Though the current findings are robust and have important implications, there are limitations, and future studies would be useful to build on these results. First, we assessed participants with a PTSD diagnosis, and are unable to determine if inhibitory control differences were present before their diagnosis or trauma-exposure. Samuelson et al. (2020) found that in OEF/OIF Veterans, pre-deployment inhibitory control performance significantly predicted postdeployment PTSD symptom severity, suggesting that the performance differences we observed may have been present before trauma exposure. It could be that PTSD symptoms further exacerbate inhibitory control abilities (see Vasterling and Hall 2018, though our data does not necessarily support this) and this would be important to test in future studies. A second limitation is that we assessed inhibitory control only using two variants of a widely used non-emotional or ‘cool’ inhibitory control task that may have biased our results to find more pre-existing/trait-like effects. Incorporating additional inhibitory control tasks, especially ‘hot’ inhibitory control tasks (i.e., that use trauma-related or emotional distractors such as in the emotional Stroop, Ashley et al. 2013), would be useful to determine if performance on these measures also predict PTSD changes or rather fluctuate more with PTSD symptom severity. Another limitation is that treatment-related information was obtained from clinical notes/reports, and additional details, such as the length of treatment, were limited. Similarly, additional timepoints would allow for a better examination of the trajectory of PTSD symptomology. Finally, given that the sample was composed of mostly male (87%) post-9/11 Veterans, it is unclear if these results apply to females or other types of trauma, necessitating replication in a non-Veteran population. The current findings advance the understanding of PTSD in several ways. They suggest that pre-existing inhibitory control abilities and early adulthood drinking behavior are important predictors of the long-term chronicity of PTSD. They also suggest that the associations between PTSD change and inhibitory control/lifetime drinking are specific, with other cognitive, clinical, and alcohol consumption measures not significantly predicting PTSD symptom changes. These findings complement studies showing the importance of inhibitory control and alcohol consumption to the development of PTSD and provide a framework for future investigations into the mechanisms of impulsivity and inhibitory control in PTSD. Finally, the current findings have clear clinical implications, suggesting that clinicians should closely monitor individuals with these risk factors and use innovative targeted interventions such as inhibitory control training and pharmacotherapies to improve PTSD and functional outcomes in this vulnerable group.