Reduced Working Memory is Associated with Heavier Alcohol Consumption History, Role Impairment and Executive Function Difficulties

Abstract

Both HIV status and heavy alcohol use have been associated with reduced cognitive function, particularly in the domains of working memory and executive function. It is unclear what aspects of working memory and executive function are associated with HIV status and heavy alcohol use and whether performance on these measures are associated with functional impairment. We examined the relationship between HIV, history of heavy alcohol consumption, and HIV/alcohol interaction on speeded tests of frontal inhibitory abilities, a working memory task related to mental manipulation of letters and numbers, cognitive flexibility, and measures of functional impairment. Study participants included 284 individuals (151 HIV +) recruited from two different studies focusing on HIV associated brain dysfunction, one specific to the effects of alcohol, the other specific to the effects of aging. HIV status was not independently associated with working memory and executive function measures. Higher level of alcohol consumption was associated with reduced performance on Letter Number Sequencing. Poorer Letter Number Sequencing performance was associated with role impairment (an inability to do certain kinds of work, housework, or schoolwork) and executive function difficulties. Future studies should examine causal associations and interventions targeting working memory abilities.

Introduction

Substance use is highly prevalent among people living with HIV (PLWH), with estimates ranging from 40 to 70% [1]. Problematic alcohol use among PLWH is nearly double the rate of the general population [2, 3]. HIV and heavy alcohol use have both been associated with deleterious cognitive effects, particularly on working memory and executive function [4-6]. That said, working memory and executive function include a range of subdomains, such as speeded frontal inhibition, mental manipulation of letters and digits, and speeded cognitive flexibility, and it is unclear which components are affected. Additionally, it is unclear whether HIV and alcohol affect cognition additively or synergistically. Another question is what effects of chronic heavy alcohol use persist over time and whether there is cognitive recovery after a period of abstinence. While some cognitive recovery is apparent after several months to a year, in abstinent formerly alcohol dependent individuals, factors such as age may modify recovery trajectory [7-9].

Working Memory and Executive Control

Working memory has been defined as a form of memory that supports the temporary storage and maintenance of internal representations and mediates the controlled manipulation of these representations [10, 11]. Broader definitions of working memory also embrace mental processes employed to reactivate information in storage, such as covert rehearsal [12]. Meanwhile, executive control, often considered as a central component of working memory, allows for the controlled distribution of processing and decision-making and is often performed in the prefrontal cortex [13-15].

Working Memory and Risk Behavior

Several studies have demonstrated a strong relationship between working memory and measures of self-control [16] and impulsivity, particularly delay discounting [16-18]. The ability to adequately store and manipulate information relates to decision making and, in particular, the decision to use substances such as alcohol. Poor working memory function has previously been associated with alcohol misuse, sexual risk behavior, and HIV serostatus [14, 19-21]. Working memory impairment predicts future alcohol problems and risk of relapse [19-21], as well as problematic substance use, including alcohol dependence [22], relapse in smokers [23], and crack cocaine and methamphetamine dependence [14]. Working memory may be a factor in future drinking behavior among alcohol dependent individuals [24, 25]. Alcohol use may also trigger working memory decline, both at the time of intoxication [26, 27] and as a result of heavy drinking over the previous 60 days [28]. In terms of recovery, working memory tends to demonstrate less impairment in those previously alcohol-dependent individuals who have accomplished long-term abstinence as opposed to short-term abstinence. Further, working memory tends to recover more quickly than other domains such as postural stability and motor function [29-31].

Working Memory and Intervention Effectiveness

Previous studies have suggested that higher working memory abilities at baseline are associated with positive outcomes from behavioral interventions [32, 33]. Additionally, working memory training among alcohol-dependent individuals produced effects such as improved quality of life, decreased alcohol craving [25], and reduced alcohol use for greater than one month [24]. A critical component of working memory is the phonological loop, which allows short-term manipulation of auditory verbal information [10]. Retention of semantic information delivered verbally is critical to most behavioral interventions. In addition to working memory’s critical role in processing in-session information, it plays an important role in out-of-session decision making: working memory may mediate between retrieval of within-session memory traces and the out-of-session decision to drink [34].

Overall, then, it is critical to better characterize the effects of alcohol on working memory, particularly in populations targeted by behavioral interventions, such as people living with HIV.

Aims and Hypotheses

The present study aimed to investigate the effects of alcohol consumption during period of heaviest use, HIV status and the interaction of the two, on working memory and executive function. Furthermore, to understand the functional significance of these relationships, we examined the significantly associated measures of working memory and executive function relative to functional impairment. We hypothesized that individuals with higher levels of alcohol consumption during their period of heaviest use would demonstrate significantly poorer working memory and executive function abilities as measured by Letter Number Sequencing, Trails B, and Stroop tasks. We also hypothesized that the HIV-positive individuals with a heavier drinking history would demonstrate poorer cognitive performance relative to all other groups. Regarding functional significance, we hypothesized that individuals with poorer performance on a task of working memory and executive function measures would report higher levels of executive function difficulties, along with role impairment (unable to do certain kinds of work, housework, or schoolwork because of health) and social functioning impairment.

Methods

Participants

Data for this study were obtained for secondary analyses from two primary studies conducted via the Alcohol and HIV-Associated Brain Dysfunction P01 grant (ARCH—AA019072) and the Age Effects on HIV-Associated Brain Dysfunction R01 grant (HIV and Aging—MH074368). Two hundred eighty-four individuals from the ARCH and HIV and Aging studies were included in the cognitive dataset. Demographic and substance use information is displayed in Table 1.

Table 1.

Study demographics, HIV status and history of alcohol consumption (N = 284)

Variable	N or Mean (% or SD)
Demographics
Male	177 (62.1)
Age (years)	44.52 (10.66)
Race
Caucasian	173 (61)
African American	71 (25)
Other	40 (14)
Education (years)	12.99 (2.68)
HIV +	151 (40)
KMSK Alcohol Score (mean)	10.72 (1.88)
Alcohol Dependent (> 10)	165 (57.9)

HIV human immunodeficiency virus, KMSK Kreek-McHugh-Kellogg-Schluger Scale

Inclusion Criteria of Both Studies

All participants were physically mobile English speakers between 30 and 70 years of age. Additionally, all HIV + participants had been diagnosed with HIV between 2 and 10 years prior to enrollment and had been on antiretroviral therapy for at least 2 years at enrollment.

Exclusion Criteria of Both Studies

All participants were subjected to breathalyzer tests, and detectable BAC above 0.05 led to cancellation of visit. Participants were also excluded for history of (1) significant pre-existing neurological disease, including Alzheimer’s disease, stroke, seizure disorder, head injury greater than mild traumatic brain injury, and opportunistic CNS infections (e.g., toxoplasmosis, progressive MLS, or neoplasm); (2) chronic psychosis (e.g., schizophrenia) diagnosed according to DSM-IV criteria using the Composite International Diagnostic Interview for the DSM-IV (CIDI) [35]; (3) end-stage disease (defined as life expectancy less than 12 months), to optimize likelihood of study completion; (4) current pregnancy; (5) ascites, encephalopathy, esophageal variceal bleeding, hepatorenal syndrome or other severe liver disease. Exclusion criteria for seropositive and seronegative participants were identical.

KMSK Measure

The Kreek-McHugh-Schluger-Kellogg Scale (KMSK) [36] was used to classify alcohol dependence during period of heaviest use. Scores on this measure range from 0 to 13. Previous receiver operating characteristics (ROC) analyses indicate that a sum score of 11, derived from individual quantity, frequency, and duration subscales, is the optimal cut-off for sensitivity and specificity to alcohol dependence [36] and for power to detect an effect [37]. Item 2 “When was this?” was used to calculate number of years since the end of an individual’s period of heaviest use by subtracting the individual’s age from their age at the end of their period of heaviest use. KMSK scales for cocaine and opiate use were used to control for other substance use.

Neuropsychological Battery

The battery consisted of the following tests, which were chosen for their sensitivity to HIV-associated Neurocognitive Disorder (HAND), particularly with regard to working memory and executive function deficits [38]: Stroop Color and Word Test [39]; Trail Making Test, Part B [40]; and Letter-Number Sequencing tests from the Wechsler Adult Intelligence Scale—Third Edition (WAIS-III) [41]. The Stroop Color and Word Test is a speeded naming test involving three trials (word, color and color/word), with the final trial involving inhibiting a previously learned response (word reading) to read the color which the ink is printed in, while ignoring the word, which requires frontally mediated inhibition. Trail Making Test Part B (Trails B) is the second trial of a speeded sequencing task, which taps working memory, requiring the examinee to switch between letters and numbers in ascending order. The Letter Number Sequencing Task requires the examinee to listen to a series of letters and numbers, tapping auditory learning, and manipulate them in working memory in order to recite back first numbers in ascending order, then letters in alphabetical order. Demographically corrected T-scores for the Trail Making Test were calculated using established Heaton norms [42]. Scaled scores were calculated for the Letter Number Sequencing task using WAIS III age corrected norms [41]. These scaled scores for Letter Number Sequencing were then converted to T-scores using a T distribution, in order to be consistent with other outcome measures. Domain composite score for working memory/executive function performance was calculated by averaging the T-scores of all tests in the domain (Letter Number Sequencing, Trail Making Test B, Stroop interference). This approach has previously been used to capture broader domains of cognitive function, rather than narrowly defined capacity to perform any specific test [37, 43].

Functional Impairment

Significant cognitive associations with alcohol, HIV or their interaction were examined relative to a scale of HIV-related quality of life and functional outcomes known as the HIV Medical Outcomes Survey (HIV-MOS) [44], which is a 34 item scale. We examined specific functional items related to basic activities of daily living “Eating, dressing, bathing, or using the toilet,” role impairment “Does your health keep you from working at a job, doing work around the house or going to school?” “Have you been unable to do certain kinds or amounts of work, housework, or schoolwork because of your health?” social functioning impairment “How much of the time, during the past 4 weeks, has your health limited your social activities (like visiting with friends or close relatives)?” and an item of subjective executive function difficulties “How much of the time, during the past 4 weeks, did you have difficulty reasoning and solving problems, for example, making plans, making decisions, learning new things?” Items are measured using a six point Likert scale (1 = all of the time, 2 = most of the time, 3 = a good bit of the time, 4 = some of the time, 5 = a little of the time, 6 = none of the time). The HIV-MOS measure was only administered during the HIV and alcohol study; therefore, data from these analyses come from 186 individuals from the ARCH study.

Statistical Analysis and Sample Size

Statistical analyses were performed using IBM SPSS Statistics Version 24. A hierarchical multiple regression was conducted to examine the relationship between alcohol consumption during period of heaviest use, HIV, and working memory and executive function measures using the following steps: step (1) KMSK sum (2) HIV status (3) KMSK sum x HIV. T-score models were able to preserve power by accounting for age and education while examining other key predictors. A sample size of 284 was deemed adequate given the five independent variables to be included in the analysis. The assumption of singularity was met. As tolerance and VIF statistics were within acceptable limits, the assumption of multicollinearity was deemed to have been met [45]. Additionally, univariate analyses were conducted to examine whether individuals with a KMSK score associated with alcohol dependence (> 10) performed more poorly than individuals who reported a score that was not associated with alcohol dependence. The Heaton normative sample includes individuals from both urban and rural areas who were recruited and tested in a number of U.S. states, including California, Washington, Colorado, Texas, Oklahoma, Wisconsin, Illinois, Michigan, New York, and Virginia, as well as the Canadian province of Manitoba. The original Heaton normative sample included 1212 individuals who completed the Halstead-Reitan Battery.

Results

Two hundred eighty-four individuals were included in the analyses (140 ARCH, 144 Aging; 97 subjects were not included due to lack of KMSK data or having no history of drinking, and one outlier was deleted due to improbable reported alcohol consumption). One hundred fifty-one individuals were HIV-positive (40%). Prior to hierarchical multiple regression, relevant assumptions were tested. One hundred sixty-five individuals met the cut-off criteria for lifetime alcohol dependence (57.9%). One hundred thirty-four individuals (47%) met the cut-off criteria for lifetime cocaine dependence. Thirty-five individuals (12.3%) met the cut-off criteria for lifetime opiate dependence. Ninety-eight individuals (34.4%) met criteria for lifetime dual dependence diagnosis (alcohol dependence + another substance). The average number of years since the end of an individual’s period of heaviest use was 14.6 years (SD = 11.6), with 79.3% of the sample reporting that this period lasted for more than a year.

Letter Number Sequencing T-Score

Hierarchical multiple regression revealed that at stage one, level of alcohol consumption during period of heaviest use contributed significantly to the regression model, [F (1, 282) = 4.295, p < 0.05)] and accounted for 1.5% of variation in Letter Number Sequencing performance correcting for normative scores. Introducing HIV status explained an additional 0.03% of variance in Letter Number Sequencing performance correcting for normative scores. When the HIV x alcohol interaction was entered, alcohol consumption during period of heaviest use remained significant and uniquely accounted for 2.3% of the variance. Results are displayed in Table 2. History of alcohol dependence was associated with poorer Letter Number Sequencing performance [T(283) = 2.460, p < 0.05]. To examine if this relationship would remain while incorporating other substance use, we added level of cocaine and opiate use during period of heaviest use. Alcohol consumption during period of heaviest use remained significant in the model.

Table 2.

Summary of hierarchical regression analysis for variables predicting letter number sequencing T-score

Variable	β	t	r	r2	Δr2
Step 1			0.122	0.015
KMSK alcohol*	−0.122	−2.072*
Step 2			0.133	0.018	0.003
KMSK alcohol*	−0.122	−2.059*
HIV status	−0.051	−0.866
Step 3			0.164	0.027	0.009
KMSK alcohol*	−0.233	−2.599*
HIV status	−0.048	−0.815
HIV × KMSK alcohol	0.148	1.647

KMSK Kreek-McHugh-Kellogg-Schluger Scale, KMSK Alcohol level of alcohol consumption during period of heaviest use

N 284

^*p < .05

^**p < .01

^***p < .001

Working Memory/Executive T-Score

Hierarchical multiple regression revealed that at stage one, level of alcohol consumption during period of heaviest use contributed significantly to the regression model, [F(1, 284) = 4.943, p < 0.05)] and accounted for 1.7% of variance in the working memory and executive function domain composite score. Introducing HIV status explained an additional 0.04% of variance in this domain. Results are displayed in Table 3. History of alcohol dependence was associated with poorer working memory and executive function [T(283) = 2.265, p < 0.05].

Table 3.

Summary of hierarchical regression analysis for variables predicting working memory/executive T-score

Variable	β	t	r	r2	Δr2
Step 1			0.131	0.017
KMSK alcohol*	−0.131	−2.223*
Step 2			0.146	0.021	0.004
KMSK alcohol*	−0.130	−2.210*
HIV status	−0.065	−1.104
Step 3			0.149	0.022	0.001
KMSK alcohola	−0.165	−1.842a
HIV status	−0.064	−1.085
HIV × KMSK	0.046	0.519
alcohol interaction

KMSK Kreek-McHugh-Kellogg-Schluger Scale; KMSK alcohol level of alcohol consumption during period of heaviest use N = 284

^ap < .07

^*p < .05

^**p < .01

^***p < .001

Trails B T-Score

None of the models or predictors significantly explained variance in Trails B performance corrected for normative scores. History of alcohol dependence was not significantly associated with Trails B performance.

Functional HIV-MOS Items

Higher Letter Number Sequencing T-score was correlated with reduced role impairment [r = 0.189, p < 0.05], and less frequent executive function difficulties (difficulty reasoning, solving problems, making plans, decisions and learning new things) [r = 0.186, p < 0.05]. Similarly, Working Memory/ Executive T-score was correlated with reduced role impairment [r = 0.240, p < 0.05] and less frequent executive function difficulties [r = 0.186, p < 0.05]. Other functional items were not significantly correlated with these cognitive items. Table 4 notes the results of the regression analyses examining LNS and Working Memory/Executive T-score as predictors of reduced role impairment and executive function difficulties. All models were significant. Working Memory/Executive T-score [β = 0.201, p < 0.05] and HIV status [β = −0.272, p < 0.05] significantly predicted role impairment [r² = 0.164, p < 0.05]. Additionally, LNS predicted executive function difficulties, even when including covariates [β = 0.165, r² = 0.076, p < 0.05]. HIV status [β = −0.267, p < 0.05] independently predicted role impairment, and LNS was trending [β = 0.155, p < 0.07].

Table 4.

Summary of regression models examining working memory, executive function HIV, and alcohol predictors of functional impairment (N = 186)

Variable	β	t	r	r2
Role impairment and LNS			0.383	0.147
LNSa	0.155	1.850a
KMSK	−0.05	−0.413
HIV Status*	−0.267	−3.281*
HIV × KMSK interaction	−0.187	−1.586
Role impairment and WM/ Exec T			0.405	0.164
WM/Exec T*	0.201	2.453*
KMSK	−0.05	−0.432
HIV Status*	−0.272	−3.392*
HIV × KMSK interaction	−0.168	−1.459
Executive difficulties and LNS			0.276	0.076
LNS*	0.165	1.910*
KMSK	−0.017	−0.137
HIV status	−0.119	−1.411
HIV × KMSK interaction	−0.165	−1.350
Executive difficulties and WM/Exec T			0.272	0.074
WM/Exec Ta	0.157	1.835a
KMSK	−0.033	−0.266
HIV status	−0.119	−1.417
HIV × KMSK interaction	−0.147	−1.216

KMSK Kreek-McHugh-Kellogg-Schluger Scale, WM/Exec T Working Memory/Executive T-score, N = 186

^ap < .07

^*p < .05

Discussion

As predicted, higher levels of alcohol use during period of heaviest use was associated with poorer working memory and executive function. Specifically, these deficits included reduced performance on a working memory task involving mental manipulation of letters and numbers (Letter Number Sequencing). Level of alcohol use during period of heaviest use was not associated with ability to inhibit prepotent information and speeded word and color reading (Stroop). Results were also consistent when examining individuals with KMSK scores consistent with a history of alcohol dependence to non-dependent individuals. While alcohol explained only 2.3% of the variance in predicting Letter Number Sequencing performance, previous studies have also found that history of problematic alcohol use is associated with poorer Letter Number Sequencing performance [46, 47], which suggests that this is a consistent finding in the literature. Furthermore, Letter Number Sequencing performance and working memory/executive domain performance were significantly associated with role impairment and executive function difficulties, suggesting some additional functional utility of these cognitive measures. HIV was associated with role impairment related to work responsibilities as well. Given the high rate of alcohol use disorders in people living with HIV, it is important to consider the potential lasting cognitive effects, particularly on working memory. Our findings demonstrated that a history of heavy drinking is associated with poorer working memory and executive function. These findings are also consistent with the finding that brain networks associated with working memory and executive function are negatively impacted by previous alcohol consumption, particularly in individuals living with HIV [48]. Previous studies have suggested that working memory abilities prior to behavioral intervention are associated with positive intervention outcome [32, 33]. Additionally, working memory training among alcohol dependent individuals demonstrated effects such as improved quality of life, decreased craving [25] and reduced alcohol use for greater than one month [24].

Limitations

Several limitations to this work must be considered. Firstly, our cognitive tests did not control for a number of proximal factors, such as fatigue, effort, and diet, which may influence test performance. Furthermore, the KMSK measures alcohol use during period of heaviest use, but does not account for more recent alcohol use, which may influence performance and may, in part, explain why the KMSK explained a relatively small proportion of the variance. A measure of more recent alcohol use was not included across the entire sample. Furthermore, the HIV and Aging participants included were not recruited for their alcohol use and therefore may have diluted some of the effects due to reduced alcohol use in participants from this sample. However, findings from Cohen et al. [4] suggest that consumption during period of heaviest use may be a robust predictor of cognitive performance, above and beyond certain measures of more recent drinking. Additionally, while incorporating level of cocaine and heroin use during period of use did not change the general relationships between alcohol and cognition, it is possible that other forms of substance use played a role in cognitive decrements. Additionally, these findings are cross-sectional and thus we cannot comment on the directionality of the relationships between alcohol consumption and cognition

Data Availability

The authors confirm that the data supporting the findings of this study are available within the article.