In order to mitigate the potential for nicotine withdrawal effects on cognition, smokers were allowed to smoke ad libitum prior to the assessment and were allowed to take cigarette smoking breaks as requested. Raw scores for neurocognitive measures, except the Luria-Nebraska Item 99 ratio, were converted to age-adjusted or age- and educationadjusted standardized scores via the accompanying normative data. Scaled scores and t-scores for all individual neurocognitive tests were transformed to z-scores to ease readability and interpretation of results using auniversal scaled score for neurocognitive measures. Scaled scores were subtracted by 10 and divided by 3 , while tscores were subtracted by 50 and divided by 10 . Neurocognitive domain scores are the arithmetic average of z-scores for all associated constituent measures. The cognitive efficiency domain consisted of all tests that were timed, or in which the time to complete the task influence the score achieved. For the Luria-Nebraska Item 99 measure, the number correct was divided by time required to complete the task. This ratio was used due to the low ceiling for the number of correct responses , resulting in a non-Gaussian distribution. Finally, the arithmetic average of z-scores for all individual neurocognitive measures was calculated to form a global neurocognition score for each participant. Participants completed the Barratt Impulsivity Scale-11 , a self-report impulsivity questionnaire. The BIS-11 consists of 30 items rated on a scale of “1” to “4” and provides total scores for non-planning, attentional, motor, and total impulsivity. Participants also completed the Balloon Analogue Risk Task , a computerized risk-taking task in which participants pump up balloons to earn increasing monetary reward,microgreens shelving with the potential for loss if a balloon overinflates and explodes. The BART yields a score for the adjusted number of pumps , with higher scores indicating a higher propensity for risk-taking.

Participants also completed the Iowa Gambling Task , a task of decision-making in which participants choose cards from four decks with the goal of winning as much money as possible. The IGT yields a raw Net Total score for each participant based on his or her selections. Raw scores were converted to the demographically-corrected T scores, with higher T scores indicating better decision-making skills.All statistical analyses were performed with SPSS version 22 . Generalized linear models were used in all analyses, employing maximum likelihood parameter estimation, and followed up by pairwise group comparisons; a chi-square statistic and corresponding p-value are generated for each parameter estimate. Three statistical models were tested: primary cross-sectional models compared PSU to AUD at one month of abstinence and included fixed predictors of group ; secondary cross-sectional models investigated potential smoking effects in PSU and AUD at one month of abstinence and included fixed predictors of group , smoking status and the interaction term of group-by-smoking status; and longitudinal models explored change in neurocognition within PSU between approximately 29 days and 128 days of abstinence ; predictors included smoking status , time , and the time-by-smoking status interaction term. Patient characteristics of PSU and AUD at baseline were compared using univariate analysis of covariance for continuous variables and Fisher’s exact test for categorical variables. Polysubstance users and AUD differed in education, gender, AMNART, hepatitis C frequency, and proportion of individuals on prescribed psychoactive medication; these variables were entered as covariates in our generalized linear models comparing AUD and PSU at baseline. Potential covariates and interaction terms were trimmed from the final model when not predictive of the outcome variable. The proportion of study participants reporting a family history of alcohol problems was not significantly different between PSU and AUD . We accounted for the multiplicity of measures by correcting alpha levels via a modified Bonferroni procedure .

This approach considers the mean correlation between variables and the number of tests in the adjustment of alpha levels. All alpha levels were adjusted for both traditional neurocognitive assessment and BIS-11 and their average inter-correlation coefficients in primary and secondary models and in tertiary models . The corresponding adjusted alpha levels for primary and secondary models were p ≤ 0.013 for neurocognitive domains and p ≤ 0.027 for self-reported impulsivity. The corresponding adjusted alpha levels for tertiary models, which included PSU only, were p ≤ 0.011 for neurocognitive domains and p ≤ 0.017 for BIS-11. Alpha levels for risk-taking and decision-making were not adjusted as these are individual tasks measuring separate domains of executive function. Effect sizes for mean differences between groups were calculated with Cohen’s d . We correlated cognitive functioning, risk-taking, decisionmaking and self-reported impulsivity measures to alcohol use in PSU and AUD, and to cocaine, and marijuana use in PSU only at baseline. Since these were exploratory correlations, we chose a less restrictive alpha level of 0.05. As shown in Table 3, and after co-varying for significant differences in AMNART, PSU performed significantly worse than AUD on auditory-verbal memory [x2 = 12.16, p < 0.001, ES = 0.72], and PSU exhibited strong trends to worse performance than AUD on intelligence [x2 = 4.08, p = 0.043, ES = 1.05] and auditory-verbal learning [x2 = 4.62, p = 0.032, ES = 0.54]. For all other domains except fine motor skills, PSU showed numerically lower scores than AUD with effect sizes up to 0.76 but no statistically significant group differences after covariate correction . When smoking status was included as a factor in the cross-sectional group analyses of neurocognitive domains, neither significant group-by-smoking interactions nor main effects of smoking were observed. In addition, gender was not a significant predictor of neurocognitive performance at one month of abstinence, except for fine motor skills which were worse in female than male substance users. Removing the two women from our PSU analyses did not significantly change any of our results. Polysubstance users exhibited trends to worse decision-making than AUD [x2 = 3.64, p = 0.056, ES = 0.33]; the groups were not significantly different on risk-taking .

No significant group-by-smoking interactions or main effects for smoking were observed on either IGT or BART. Polysubstance users self-reported significantly higher BIS-11 total and nonplanning impulsivity, a measure of cognitive control, than AUD , and being on a prescribed psychoactive medication significantly predicted higher total and nonplanning impulsivity. With smoking status included in the analyses, no significant group-by-smoking interactions were observed for any of the BIS-11 measures. However, self-reported motor impulsivity showed a trend for a group-by-smoking interaction [x2 = 3.259, p = 0.071], a significant main effect for group [x2 = 2.005, p = 0.006], and a trend for a smoking effect [x2 = 1.499, p = 0.066]. Follow-up pairwise comparisons showed significantly higher motor impulsivity in smoking PSU compared to both smoking and nonsmoking AUD . Between baseline and follow-up, neurocognitive functions in abstinent PSU improved markedly in the following domains: general intelligence, cognitive efficiency, executive function, working memory, and visuospatial skills , and weaker improvements were observed for global cognition and processing speed . Abstinent PSU did not change significantly in the domains of learning and memory or fine motor skills. Preliminary analyses indicate that the lack of significant changes in the domains of visuospatial memory and fine motor skills were related to significant time-by-smoking status interactions ,greenhouse tables where only nonsmokers increased on fine motor skills and only smokers improved on visuospatial memory. The BART scores increased significantly with abstinence , whereas the IGT scores did not change during abstinence. Self-reported total and motor impulsivity decreased significantly with abstinence and the nonplanning score tended to decrease . The following changes were observed when restricting our longitudinal analysis to only those 17 PSU with baseline and follow-up data: general intelligence, executive function, working memory , visuospatial skills , global cognition , and processing speed . The 19 PSU not studied longitudinally differed from our abstinent PSU restudied on lifetime years of cocaine use . PSU not restudied performed significantly worse at baseline than abstinent PSU on cognitive efficiency, processing speed, and visuospatial learning . Furthermore, they did not differ significantly on years of education, AMNART, tobacco use severity, and proportions of smokers or family members with problem drinking, or the proportion of individuals taking a prescribed psychoactive medication.In PSU, more lifetime years drinking correlated with worse performance on domains of cognitive efficiency, executive function, intelligence, processing speed, visuospatial skills, and global cognition . More cocaine consumed per month over lifetime correlated with worse performance on executive function and greater attentional impulsivity .

More marijuana consumed per month over lifetime correlated with worse performance on fine motor skills and tended to correlate with higher BIS-11 motor impulsivity ; in addition, more marijuana use in the year preceding the study correlated with higher nonplanning and total impulsivity. Interestingly, more lifetime years of amphetamine use correlated with better performance on fine motor skills, executive function, visuospatial skills, and global cognition . Similar to the associations found in PSU, more lifetime years drinking in AUD correlated with worse performance on cognitive efficiency, visuospatial skills, and global cognition , and worse performance on visuospatial memory correlated with greater monthly alcohol consumption averaged over the year preceding assessment and over lifetime . In addition, longer duration of alcohol use in AUD was related to worse auditory-verbal learning and memory . Earlier age of onset of heavy drinking in AUD was associated with worse decision-making .Our primary aim was to compare neurocognitive functioning and inhibitory control in onemonth-abstinent PSU and AUD. Polysubstance users at one month of abstinence showed decrements on a wide range of neurocognitive and inhibitory control measures compared to normed measures. The decrements in neurocognition ranged in magnitude from 0.2 to 1.4 standard deviation units below a zscore of zero, with deficits >1 standard deviation below the mean observed for visuospatial memory and visuospatial learning. In comparisons to AUD, PSU performed significantly worse on measures assessing auditory-verbal memory, and tended to perform worse on measures of auditory-verbal learning and general intelligence. Chronic cigarette smoking status did not significantly moderate cross-sectional neurocognitive group differences at baseline. In addition, PSU exhibited worse decision-making and higher self-reported impulsivity than AUD , signaling potentially greater risk of relapse for PSU than AUD . Being on a prescribed psychoactive medication related to higher self-reported impulsivity in PSU. For both PSU and AUD, more lifetime years drinking were associated with worse performance on global cognition, cognitive efficiency, general intelligence, and visuospatial skills. Within PSU only, greater substance use quantities related to worse performance on executive function and fine motor skills, as well as to higher self-reported impulsivity. Neurocognitive deficits in AUD have been described extensively. However, corresponding reports in PSU are rare and very few studies compared PSU to AUD during early abstinence on such a wide range of neurocognitive and inhibitory control measures as administered here . To our knowledge, no previous reports have specifically shown PSU to perform worse than AUD on domains of auditory-verbal learning and general intelligence at one month of abstinence. Our studies confirmed previous findings of worse auditory-verbal memory and inhibitory control in individuals with a comorbid alcohol and stimulant use disorder compared to those with an AUD, and findings of no differences between the groups on measures of cognitive efficiency . Some of the cross-sectional neurocognitive and inhibitory control deficits described in this PSU cohort are associated with previously described morphometric abnormalities in primarily prefrontal brain regions of a subsample of this PSU cohort with neuroimaging data . Our neurocognitive findings also further complement studies in subsamples of this PSU cohort that exhibit prefrontal cortical deficits measured by magnetic resonance spectroscopy and cortical blood flow . Our secondary aim was to explore if PSU demonstrate improvements on neurocognitive functioning and inhibitory control measures between one and four months of abstinence from all substances except tobacco. Polysubstance users showed significant improvements on the majority of cognitive domains assessed here, particularly cognitive efficiency, executive function, working memory, self-reported impulsivity, but an unexpected increase in risk-taking behavior . By contrast, no significant changes were observed for learning and memory domains, which were also worst at baseline, resulting in deficits in visuospatial learning and visuospatial memory at four months of abstinence of more than 0.9 standard deviation units below a z-score of zero. There were also indications for significant time-by-smoking status interactions for visuospatial memory and fine motor skills, however these analyses have to be interpreted with caution and considered very preliminary, considering the small sample sizes of smoking and nonsmoking PSU at followup.