Cognitive Intra-individual Variability in the Laboratory Is Associated With Greater Executive Dysfunction in the Daily Lives of Older Adults With HIV

Abstract

Background:

Executive dysfunction, which is common among persons with HIV (PWH), can have an adverse impact on health behaviors and quality of life. Intra-individual variability (IIV) is a measure of within-person variability across cognitive tests that is higher in PWH and is thought to reflect cognitive dyscontrol.

Objective:

To assess whether cognitive IIV in the laboratory is associated with self-reported executive dysfunction in daily life among older PWH.

Method:

Participants included 71 PWH aged ≥50 years who completed six subtests from the Cogstate battery and two subscales from the Frontal Systems Behavior Scale (FrSBe; self-report version). Cognitive IIV was calculated from the Cogstate as the coefficient of variation derived from age-adjusted normative T scores.

Results:

Cognitive IIV as measured by the Cogstate showed a significant, positive, medium-sized association with current FrSBe ratings of executive dysfunction but not disinhibition.

Conclusion:

Higher cognitive IIV in the laboratory as measured by the Cogstate may be related to the expression of HIV-associated symptoms of executive dysfunction in daily life for older PWH.

Mild-to-moderate impairment in executive functions (EFs) is a common feature of HIV (Walker and Brown, 2018). Prevalence estimates suggest that 30–50% of persons with HIV (PWH) experience executive dysfunction (Heaton et al., 2010). Indeed, HIV adversely affects the prefrontostriatal systems that provide neural support for many EF processes (Du Plessis et al., 2014).

Executive dysfunction is a risk factor for HIV transmission (Weinborn et al., 2013); it can emerge in the acute and early stages of HIV infection (Kamat et al., 2015) and may have increased in frequency in the modern treatment era of combination antiretroviral therapy (Heaton et al., 2011). HIV-associated executive dysfunction is heterogenous, meaning that HIV can affect different aspects of EFs, including the subdomains of planning (Cattie et al., 2012), decision-making (Iudicello et al., 2013), cognitive flexibility (Walker and Brown, 2018), novel problem-solving (Heaton et al., 2004), and working memory (Martin et al., 2018), across samples and individuals.

The frequency and extent of HIV-associated executive dysfunction is evident among older adults (Jiang et al., 2016), individuals with concurrent substance dependence (Walker and Brown, 2018), and individuals with lower historical CD4 counts (Heaton et al., 2011). Impairments in EF can make certain activities of daily living (ADLs) challenging for PWH (Casaletto et al., 2017); in fact, EF is one of the most robust cognitive predictors of medication adherence (Hinkin et al., 2002), automobile driving (Marcotte et al., 2004), internet navigation skills (Woods et al., 2017), and aspects of quality of life (Tate et al., 2003) in PWH.

Assessing EF

EF is a challenging construct to define, quantify, and assess in ecologically relevant ways across the life span (Cirino et al., 2018). In the clinic, typical measures of EF assess constructs such as problem-solving, cognitive flexibility, prepotent response inhibition, and verbal fluency (Rabin et al., 2016) using stimuli and paradigms that do not parallel common daily activities that place demands on EFs (Chaytor and Schmitter-Edgecombe, 2003). More naturalistic assessments of EF tend to rely heavily on self- and informant-report approaches that tap into the symptoms of executive dysfunction in daily life (e.g., Frontal Systems Behavior Scale [FrSBe]; Grace and Malloy, 2001) and thus may not reflect one’s ability or capacity per se.

Clinical measures of EF show reasonably robust veridicality as predictors of everyday functioning among PWH (e.g., Casaletto et al., 2017); however, there is a reliable disconnect between standard, performance-based clinical test scores of EF and report-based symptom measures of EF in PWH. This discordance is not unique to PWH nor to EF (e.g., Rourke et al., 1999). As such, there is a need to expand the armamentarium of EF constructs that are measured in the laboratory in order to enhance our ability to draw inferences from clinical EF test performance to the expression of EF symptoms in daily life.

Cognitive Symptoms and HIV

Such efforts are ecologically relevant because cognitive symptoms in daily life are an important component of overall everyday functioning in the diagnosis of HIV-associated neurocognitive disorders (e.g., Matchanova et al., 2020; Woods et al., 2004). Indeed, the widely used Frascati criteria for HIV-associated neurocognitive disorders (Antinori et al., 2007) uses self- and informant-reported cognitive symptoms as one of its major indicators of everyday functioning. For example, mild functional decline is considered a diagnostic possibility when a “patient reports that he or she is experiencing increased difficulty with ≥2 aspects of cognition in daily life” (Antinori et al., 2007, p 1797).

There is growing recognition in the broader cognitive aging literature that subjective cognitive decline (i.e., self- or informant-reported cognitive symptoms in daily life in the setting of normal cognition in the laboratory or clinic) may be an early manifestation of an incipient neurocognitive disorder (Jessen et al., 2014). PWH are 3–5 times more likely than uninfected persons to be diagnosed with subjective cognitive decline (Sheppard et al., 2019; Thompson et al., 2023), which is associated with mild difficulties in ADLs (Sheppard et al., 2019).

Cognitive intra-individual variability (IIV), which is a marker of cognitive dyscontrol, may be relevant to the experience of cognitive symptoms in daily life (Hultsch et al., 2002). Cognitive IIV is a measure of within-person variability in test performance either across cognitive domains or across time (e.g., Stuss et al., 2003). Increased cognitive IIV is evident in a range of populations with neurocognitive deficits, including PWH (Vance et al., 2022). PWH tend to exhibit higher cognitive IIV across a battery of neuropsychological tests (Morgan et al., 2011), which is associated with increased risk of cognitive decline and mortality (Anderson et al., 2018). Imaging studies suggest that greater cognitive IIV is associated with decreased connectivity in the superior longitudinal fasciculus (Jones et al., 2018), as well as decreased gray and white matter volumes in PWH (Hines et al., 2016).

HIV and Cognitive IIV

Among PWH, older adults (Morgan et al., 2011) and persons with substance use disorders (Arce Rentería et al., 2020) are at particular risk of higher cognitive IIV. Higher cognitive IIV across cognitive tests in PWH has been reliably associated with ecologically relevant abilities such as prospective memory (Mustafa et al., 2023) and daily tasks with significant EF demands. For instance, Morgan, Woods, et al. (2012) reported that higher levels of cognitive IIV conferred a 1.6 higher risk of ADL dependence in PWH. Additionally, Thaler et al. (2015) reported that PWH with higher cognitive IIV are at higher risk of suboptimal combination antiretroviral therapy adherence, which can lead to poorer cognitive and health outcomes (Ettenhofer et al., 2010).

Current Study

Although higher cognitive IIV has been associated with ecologically relevant EF abilities such as prospective memory and medication management, the extant literature has not examined whether cognitive IIV is associated with EF symptoms in daily life. We decided to build on existing work by examining the association between cognitive IIV in the laboratory and self-reported executive dysfunction in daily life in a sample of older PWH. We hypothesized that higher cognitive IIV would be associated with higher levels of current EF symptoms in daily life.

METHOD

Participants

Participants in this study were 71 adults with confirmed HIV, aged 50 to 75 years (M = 57.5, SD = 6.4), who were enrolled in an experimental memory study at the HIV Neurobehavioral Research Program at UC San Diego (see Woods et al., 2020; 2021). Participants were included in our study if they had completed the Cogstate battery (www.cogstate.com), had reported being age ≥50 years, and had confirmed HIV infection via MedMira. Age 50 is commonly used as a cut point for aging research in HIV because PWH above this age show higher rates of neural injury, cognitive impairment, and comorbidities that affect cognitive function and brain structure (Mitra et al., 2022).

Exclusion criteria were a prior diagnosis of non-HIV-related major neurocognitive disorder, severe psychiatric disorder, seizure disorder, opportunistic infection that involved the CNS, head injury with loss of consciousness >30 minutes, or stroke with neurologic sequalae. Individuals were also excluded from our study if they met the diagnostic criteria for substance dependence within the past 30 days or tested positive on a screen for illicit drugs (excluding marijuana) on the day of testing.

Written, informed consent was provided by all of the participants, who received nominal financial compensation for completing a research neuropsychological, psychiatric, and medical evaluation. The institutional review board at UC San Diego reviewed and approved the parent study procedures.

Demographic information and HIV characteristics for participants in this study are provided in Table 1.

TABLE 1.

Demographic, Psychiatric, and Medical Characteristics of the Study Participants

Variable	Sample (N = 71)	Range
Demographic characteristics
Age (years)	57.5 (6.4)	50–75
Education (years)	14.3 (2.5)	8–20
Sex (% women)	21.1	—
Ethnicity
% White	57.7	—
% Black/African American	21.1	—
% Hispanic	19.7	—
% Native American	1.4
Psychiatric characteristics
POMS total mood disturbance	31.1 (34.1)	0–156
Generalized anxiety disorder (%)†	22.5	—
Major depressive disorder (%)†	70.4	—
Substance use disorder (%)†	70.4	—
Alcohol use disorder (%)†	60.6
Cannabis use disorder (%)†	36.6
Stimulant use disorder (%)†	50.7
HIV disease characteristics
Estimated HIV duration (years)	22.0 (8.2)	2.9–38.0
Current CD4	707.7 (317.9)	120–1892
Nadir CD4	187.8 (180.0)	0–725
HIV plasma viral load (% detectable)	2.9	—
AIDS (%)	70.4	—
Prescribed antiretroviral therapy (%)	95.7	—
Number of other medical conditions	2.1 (1.6)	0–6
Pulmonary disease (%)	5.6
Cardiovascular disease (%)	5.6
Type II diabetes (%)	15.5
Liver disease (%)	5.6
Hepatitis C (%)	23.9
Hypertension (%)	54.9
Hyperlipidemia (%)	57.8
Renal disease (%)	4.2
Cancer (% past)	14.1

Values are presented as M ± SD unless otherwise noted.

^†Lifetime diagnoses.

POMS = Profile of Mood States.

Assessments

Cogstate

Six subtests from the Cogstate, a computer-based neuropsychological battery, were administered to all of the participants by trained research assistants who followed manualized instructions (Table 2).

TABLE 2.

Cognitive Characteristics of the Study Participants

Variable	Sample (N = 71)	Range
Cogstate Battery
Mean T score	45.9 (6.3)	21.9–58.7
Intra-individual SD	8.2 (2.3)	3.6-14.8
Coefficient of variation	0.18 (0.07)	0.06–0.48
Global impairment (%)†	39.4	—
FrSBe (mean T score)
Executive dysfunction	53.9 (16.2)	28–109
Disinhibition	54 (14.0)	27–107

^†Impairment is defined by a >0.5 cut point on the global deficit score (Carey et al., 2004).

FrSBe = Frontal Systems Behavior Scale.

Carey CL, Woods SP, Gonzalez R, et al. 2004. Predictive validity of global deficit scores in detecting neuropsychological impairment in HIV infection. J Clin Exp Neuropsychol. 26:307–319. doi:10.1080/13803390490510031

• The Detection subtest measures psychomotor speed; the primary outcome variable is response time. Participants press yes as quickly as possible when a card turns face up.

• The Identification subtest measures visual attention; the primary outcome variable is response time. Participants indicate whether a card is red by pressing yes or no as quickly as possible.

• The One-Back subtest measures focused attention; the primary outcome variable is response time. Participants indicate whether the current card matches the last card shown (one back) by pressing yes or no as quickly as possible.

• The Two-Back subtest measures working memory; the primary outcome variable is accuracy. Participants indicate whether the current card matches the card shown two cards ago (two back) by pressing yes or no as quickly as possible.

• The One Card Learning subtest measures learning and attention by assessing continuous visual recognition; the primary outcome variable is accuracy. Participants indicate whether the current card was previously presented by pressing yes or no.

• The Continuous Paired Associate Learning subtest measures visual memory; the primary outcome variable is the number of errors. Participants are asked to tap on the screen to indicate the location of an object that they had previously viewed.

Intra-individual Variability

We transformed the raw cognitive subtest scores to age-adjusted normative T scores using the Cogstate manual. An overall mean was calculated for each participant by averaging the six Cogstate T scores. An intra-individual standard deviation (iSD) was calculated by generating the square root of the variance relative to the mean for each participant across all six subtests collectively.

Our primary variable of interest was the coefficient of variation (CoV), which is calculated by dividing each individual participant’s iSD by his or her own mean (Abdi, 2010; Christensen et al., 2005). A higher CoV indicates greater cognitive IIV across cognitive tasks (sample range = 0.06–0.48). In dividing the iSD by the overall mean, the CoV metric controls for global cognition and provides greater confidence that the observed associations between cognitive IIV and FrSBe ratings are not simply an artifact of overall cognitive ability. The correlations between the CoV and the mean and iSD are displayed in Table 3.

Table 3.

Cognitive Correlates of the FrSBe Scale T scores in 71 Older Adults With HIV Disease

Cogstate Variable	Cogstate CoV	FrSBe Executive Dysfunction	FrSBe Disinhibition
CoV	---	0.26*	0.10
iSD	0.93*	0.19	0.15
Mean T score	−0.56*	−0.23	0.07
Detection	−0.43*	−0.19	−0.07
Identification	−0.49*	−0.31*	−0.06
One-back	−0.51*	−0.25*	−0.13
Two-back	−0.18	−0.27*	0.00
One Card Learning	−0.32*	0.10	0.30*
CPAL	−0.19	−0.14	0.02

Data represent Spearman’s rho values.

^*P < 0.05.

CoV = coefficient of variation. CPAL = Continuous Paired Associate Learning. FrSBe = Frontal Systems Behavior Scale. iSD = intra-individual standard deviation.

EFs in Daily Life

The FrSBe is a self-rated questionnaire that assesses the frequency of neurobehavioral symptoms that are related to frontally mediated processes (see Table 2). The factor structure of the FrSBe, as examined in both healthy individuals and those with CNS injury, supports the existence of separate subscales for apathy, executive dysfunction, and disinhibition (Stout et al., 2003). For our study, we used ratings from the Executive Dysfunction (e.g., planning and organization) and Disinhibition (e.g., impulsivity and risky behavior) subscales of the FrSBe.

Many of the items on the FrSBe Executive Dysfunction subscale reflect cognitive symptoms that are relevant to daily life (e.g., “Cannot do two things at once,” “Is (un)able to plan ahead,” “Mixes up a sequence,” and “Is a poor problem solver”). The FrSBe Executive Dysfunction subscale shows reliable, positive associations with ADLs in PWH (e.g., Kamat et al., 2016). Although many models of executive dysfunction include disinhibition as a subcomponent of executive skills, the context and psychometric properties of the FrSBe provide evidence of both convergent and divergent validity for the separation of disinhibition as a separate construct, particularly in HIV (e.g., Kamat et al., 2014; 2016). Therefore, we examined the FrSBE Executive Dysfunction and Disinhibition subscales separately within the context of this study.

Participants rated their current symptom frequency using a scale that ranged from one (almost never) to five (almost always), meaning that higher raw scores indicated greater levels of executive dysfunction and disinhibition. Demographically adjusted T scores were derived from the FrSBe manual’s normative sample. Internal consistency for both subscales was acceptable (Executive Dysfunction α = 0.81; Disinhibition α = 0.84).

Medical and Psychiatric Evaluation

A research nurse conducted a medical evaluation with each study participant, which included a record review to ascertain the duration of HIV infection, nadir CD4 (cells/μL), current medications (e.g., antiretroviral regimen), and common medical comorbidities (e.g., vascular disease). A blood draw was performed to obtain a current CD4 (cells/μL) and HIV RNA plasma viral load. Certified research assistants conducted the Composite International Diagnostic Interview (version 2.1; World Health Organization, 1998) to determine the presence of lifetime and current (i.e., 30 days from evaluation) diagnoses of substance use disorders, major depressive disorder, and generalized anxiety disorder (American Psychiatric Association, 1994). Descriptive data for the relevant psychiatric, medical, and sociodemographic data are presented in Table 1.

Statistical Analysis

We conducted the data analysis in JMP version 16.0 (SAS). The critical alpha was set to 0.05 for all of the analyses. The primary variables of interest were not normally distributed (i.e., Shapiro-Wilk P ≤ .01). As such, we used a nonparametric approach (Spearman’s ρ) for all of the analyses whenever possible.

We generated and interpreted effect sizes (Woods et al., 2023) according to recent recommendations for psychological research (Funder and Ozer, 2019), which consider values ≤ 0.05 to be very small, 0.06–0.10 small, 0.11–0.20 medium, 0.21–0.30 large, and ≥ 0.4 very large. Methodologists generally recommend three possible ways of considering covariates in cross-sectional studies such as ours (Field-Fote, 2019):

• a priori approach, whereby covariates are chosen in advance, as determined by prior theory and evidence;

• standard covariate approach, whereby covariates are included solely on their demonstrated relationship with the dependent variable/criterion; and

• confound approach, whereby only variables that relate to both the independent variable/predictor and the dependent variable/criterion are included.

We used the confound approach for considering the covariates due to the relatively small sample of PWH and the corresponding risk of Type II error in oversaturated models (Field-Fote, 2019). Specifically, any demographic, psychiatric, or medical variables displayed in Table 1 were included in the model if they related significantly (P < 0.05) to both the Cogstate CoV and the FrSBe variables of interest. Sex and current CD4 were significantly related to the Cogstate CoVs (Ps < 0.05), and number of medical conditions, lifetime substance use disorder, and Profile of Mood States (McNair et al., 1971) total mood disturbance were significantly related to the current FrSBe T scores (Ps < 0.05). However, no characteristics were related to both variables of interest, so no covariates were included in the analyses.

RESULTS

Relationship of Cognitive IIV and Executive Dysfunction

The Cogstate CoV was significantly positively associated with FrSBe current self-ratings of executive dysfunction at a medium effect size, ρ = 0.26; P = 0.03 (Table 3). Table 3 shows nonsignificant but broadly medium effect sizes for the relationship of FrSBe executive dysfunction with the Cogstate mean and iSD. The univariate associations with the individual Cogstate subtest T scores are shown for descriptive purposes.

The Cogstate CoV was not significantly associated with FrSBe current self-ratings of disinhibition, and the observed effect was small, ρ = 0.10; P = 0.40. A similar pattern of results was observed for the Cogstate mean and iSD.

Next, the specificity of the main effect of the Cogstate CoV on FrSBe current self-ratings of executive dysfunction was determined. Fisher’s z transformation, modified for use with dependent samples, was used to test the significance of the difference between the correlations of Cogstate CoV with FrSBe current self-ratings of executive dysfunction and current disinhibition. The difference between these two correlations approached significance: z = 1.6, P = 0.05.

DISCUSSION

PWH show evidence of executive dysfunction both in the laboratory and in their daily lives, which increases their risk of ADLs dependence and poorer quality of life. Our study extends prior work by demonstrating an association between performance-based executive dysfunction as measured by cognitive IIV and EF symptoms in daily life as measured by the FrSBe.

Cognitive IIV and EF Symptoms

The association between cognitive IIV and daily EF symptoms was accompanied by a medium effect size (Funder and Ozer, 2019) and was not confounded by any demographic, psychiatric, or medical variables (Field-Fote, 2019). Importantly, the confound approach to determining covariates for the planned analyses limits the conclusions that can be drawn from these data. We cannot determine whether CoV has incremental ecological value in association with daily EF symptoms; rather, we can only state that the observed relationship in this sample is not expressly confounded by the factors listed in Table 1.

Executive Dysfunction

To our knowledge, this is the first study to link cognitive IIV with self-reported executive dysfunction in daily life for PWH. As such, these data align with the recent report from Webber et al. (2022), who observed that cognitive IIV was independently associated with clinician-rated cognitive fluctuations in daily life among healthy older adults, individuals with dementia with Lewy bodies, and individuals with Alzheimer disease. Taken together, these studies suggest that cognitive IIV may play a role in HIV-associated symptoms of daily executive dysfunction. It follows that cognitive IIV can help conceptualize daily EFs, especially when mean-level performance is unrevealing, which can be a problem in conditions like HIV with mild-to-moderate effects on cognition.

It is important to mention that the cognitive symptoms measured herein by the FrSBe are not to be mistaken for ADLs, and that the effect sizes, although moderate, are not sufficiently large to guide current clinical decision-making. Nevertheless, our findings have considerable ecological relevance.

• Cognitive symptoms in daily life are part of the Frascati criteria for HIV-associated neurocognitive disorders (Antinori et al., 2007), are central to the diagnostic criteria for subjective cognitive decline (Jessen et al., 2014; Sheppard et al., 2019), and are reliable predictors of ADLs (e.g., Kamat et al., 2016) and quality of life in PWH (e.g., Woods et al., 2015).

• The effect sizes that are typically observed between cognition and ADLs tend to be very modest across the literature; for example, a meta-analysis showed that cognition explained only 23% of the variance in functional outcomes among individuals with mild cognitive impairment (Mcalister et al., 2016). It can be quite challenging to discover associations between phenomena in the laboratory and their daily life counterpart in complex medical samples (Woods, 2021). Indeed, the conceptual pathway from a laboratory score to a daily functioning outcome is complicated by issues of task relevance, insight, compensatory behaviors, and the conceptual and practical match between the two constructs in those different settings (Casaletto et al., 2017; Chaytor and Schmitter-Edgecombe, 2003). There are a multitude of factors that can contribute to poor everyday functioning, and cognition is only one of those factors; thus, the effect sizes tend to be modest but meaningful.

Disinhibition

Our study revealed no association between cognitive IIV and disinhibition (e.g., risky behaviors and impulsivity) as measured by the FrSBe in older PWH. The observed correlations were nonsignificant and demonstrated a small effect, suggesting that the association between cognitive IIV and executive dysfunction is unique to the facet of EF that is tapped by this specific FrSBe subscale. In particular, the EF-related symptoms represented on the FrSBe Executive Dysfunction subscale largely involve decision-making, working memory, and cognitive flexibility, which are aspects of EF that are known to be impacted by HIV (Walker and Brown, 2018).

It is conceptually plausible that variability in cognitive control could promote lapses in inhibitory function both in the laboratory and in daily life, particularly since it is known that disinhibition is higher in PWH as compared to seronegatives (e.g., Martin et al, 1992). Nevertheless, disinhibition is also evident pre infection and does not appear to increase in the setting of acute, early HIV (Kamat et al., 2016). Moreover, the proposed neural pathways of disinhibition are distinct from the frontal pathways that are closely related to the Executive Dysfunction subscale of the FrSBe (Kamat et al., 2014). As such, future work with other measures of inhibition and in other populations is warranted.

Overall Cognitive Performance

One interpretive issue that deserves consideration is whether the association between IIV and EF symptoms in daily life can be better explained by the underlying influence of global cognitive performance, or perhaps by an anomalous single measure. This important question has been addressed in several ways across the cognitive IIV literature, for example, by including mean-level performance in the statistical model or by extracting the variance associated with mean-level performance.

Given the small sample size and the nonnormal distribution of the variables in our study, we used the CoV to account for overall cognitive performance. The CoV, which is obtained by dividing the iSD by the mean, is widely used in the cognitive IIV literature (e.g., Musso et al., 2015; Sheppard et al., 2018; Webber et al., 2023). Furthermore, the data displayed in Table 3 show that the relationship between the CoV and executive dysfunction as measured by the FrSBe was as strong or stronger than that of the individual or mean Cogstate T scores. Likewise, the effect size of the association between the CoV and the FrSBe Executive Dysfunction subscale was comparable to that of the iSD and the FrSBe Executive Dysfunction subscale. Taken together, these factors increase our confidence that the observed relationship between cognitive IIV and the FrSBe is not solely an artifact of mean-level performance, although these two factors are certainly intertwined.

Study Limitations

Several limitations deserve consideration in interpreting the findings from our study. First, we used the self-report version of the FrSBe, which has shown strong evidence of construct validity in individuals with HIV (e.g., Kamat et al., 2014) but can nevertheless be biased by issues of insight and mental health factors (e.g., mood). As such, collateral and/or clinician reports of cognitive IIV and EF symptoms in daily life (e.g., Mayo Fluctuations Scale; Ferman et al., 2004) might be useful to examine in future investigations.

A second limitation was our use of the relatively brief computerized Cogstate battery to measure cognitive IIV. Although prior studies provide some support for the usefulness of the Cogstate in this regard (e.g., Cho et al., 2023; Mustafa et al., 2023), future work may benefit from including well-validated standard clinical measures of EF such as continuous performance tasks. Third, our findings are restricted to the measurement of cognitive IIV by way of dispersion (i.e., score variability across a battery of tests), which may differ in meaningful ways to the assessment of cognitive IIV as measured by inconsistency across a single task (e.g., Rutter et al., 2020).

Fourth, a major limitation of all work related to cognitive IIV is that normative values for cognitive IIV across a typical battery of neurocognitive tests do not currently exist. As such, it is difficult to identify abnormally high or low levels of cognitive IIV, which would be of tremendous clinical value. Likewise, the Cogstate does not include normative adjustments for educational quality or race/ethnicity, which may affect its diagnostic accuracy among diverse samples of PWH (Norman et al., 2011).

Finally, these findings are likely not specific to older PWH and could plausibly be observed in healthy adults across the life span as well as in groups with CNS dysfunction (e.g., Webber et al., 2022). Although there is evidence that aging with HIV has additive, adverse effects on brain structure and function, including cognitive IIV (Morgan et al., 2011) and daily functioning (Morgan, Iudicello, et al., 2012), questions remain whether the association between cognitive IIV and daily cognitive symptoms would be amplified in older PWH. Because there was no seronegative comparison group in this study, the inclusion of a well-matched control sample will be important for further studies in order determine the uniqueness of the relationship between cognitive IIV and EF in PWH.

CONCLUSION

Findings suggest that cognitive IIV may play a modest but meaningful role in symptoms of executive dysfunction in the daily lives of PWH. Cognitive IIV appears to be a promising candidate for ecologically valid measurement of daily cognitive problems, which are notoriously difficult to measure using laboratory-based neurocognitive metrics (Chaytor et al., 2006).

Glossary

ADL

activity of daily living

CoV

coefficient of variation

EF

executive function

FrSBe

Frontal Systems Behavior Scale

IIV

intra-individual variation

iSD

intra-individual standard deviation

PWH

persons with HIV