The role of behavioral and neurocognitive functioning in substance use among youth with perinatally acquired HIV infection and perinatal HIV exposure without infection

Abstract

This study examined associations of self-regulatory behavior and cognitive functioning with substance use (SU) to inform interventions for youth with perinatal HIV infection (YPHIV) or exposure but uninfected (YPHEU). Youth aged 7–15 years (YPHIV, n=390; YPHEU, n=211) were followed longitudinally with cognitive testing and behavioral questionnaires including self-report of alcohol, marijuana, tobacco, and other SU. Cox proportional hazards analyses were used to examine correlates of initiating each substance for those without prior use at baseline and generalized estimating equation analyses were used to address associations of cognitive/behavioral measurements with SU prevalence for the entire sample. Lower self-reported self-regulation skills, but higher cognitive functioning abilities, were associated with initiation and prevalent use of alcohol and marijuana regardless of HIV status. Our findings suggest SU screening tools and self-regulation interventions developed for general adolescent populations should be implemented for those with PHIV, who may be at heightened risk for SU-related health consequences.

INTRODUCTION

Youth with perinatally acquired HIV (YPHIV) are now surviving into adulthood in greater numbers than two decades ago. Long-term health, quality of life, co-morbid diseases and disorders, and prevention of secondary HIV transmission have become necessary healthcare priorities for this population. These long-term outcomes may be complicated by typical adolescent risk behaviors including experimentation with substance use (SU). For YPHIV, SU has the potential to impact health through inadequate medication adherence(1), poor viral control (2, 3), lower engagement in care(4), and increased risk for secondary HIV transmission through sexual risk behaviors and genital viral shedding (5–7). Although there is a clear need, few evidence-based interventions focused on prevention and treatment of SU co-morbidities among YPHIV exist. To inform evidence-based programming, a better understanding of risk factors and correlates of SU among YPHIV is critical.

Studies of YPHIV (8–12) suggest that they initiate SU at or slightly below comparable rates as in the general population (5, 13). However, a study of young adults with PHIV showed relatively high rates (28%) of SU disorders (14). Alperen and colleagues (13) found that 60% of YPHIV (mean age 14 years) reported having used alcohol, 27% marijuana, and 25% tobacco at some time. Significant predictors of alcohol or marijuana use included emotional or conduct problems, supporting the importance of mental health contributors to SU for YPHIV. Interestingly, higher academic achievement was associated with greater alcohol drinking risk among YPHIV participating in this study.

Studies have shown that children and youth with perinatally acquired HIV (PHIV) are more likely to have lower cognitive functioning, particularly if they have experienced immunosuppression effects (15, 16). Both YPHIV and perinatally HIV-exposed and uninfected youth (YPHEU) are at increased risk for academic(17) and executive functioning problems(18–21). Among youth in general, aspects of executive functioning such as decision-making and inhibition (22, 23) as well as neurobehavioral characteristics related to self-regulation (e.g., impulsivity and sensation-seeking (24, 25)), are associated with increased risk for SU. Associations of SU with impairments in global intellectual functioning are less well understood, with some studies indicating that impairment is not associated with risk once other contributing factors have been considered (26, 27) and others showing either increased or, more commonly, decreased risk (28).

Although impairment of self-regulation and executive functioning, and possibly global cognition impairment secondary to HIV, may influence SU behavior, we are unaware of studies examining cognitive and neurobehavioral predictors of SU risk among YPHIV. We utilized longitudinal data from the Pediatric HIV/AIDS Cohort Study Adolescent Master Protocol (PHACS AMP) to examine cognitive, executive functioning and self-regulation predictors and correlates of SU in this population. After examining SU characteristics and severity, we compared prospective associations with SU initiation among youth with PHIV and PHEU, age 7–15 years at baseline, as they progressed into or through adolescence. The primary analyses were aimed at testing the hypothesis that lower self-regulation, as indicated by poorer executive functioning and higher sensation-seeking, would place YPHIV at higher risk for initiating SU. Additionally, we examined whether global intellectual functioning alters risk for SU initiation and tested whether these relationships differ for youth with PHIV and PHEU. Last, we conducted exploratory analyses of cognitive and behavioral correlates of prevalent use among the subset of youth who had already initiated SU prior to their study entry.

METHODS

Participants

Participants were enrolled in PHACS AMP, a prospective cohort study of the long-term effects of perinatal HIV infection and its treatments on biomedical and neurobehavioral outcomes (see https://phacsstudy.org/). Participants were enrolled at one of 15 urban AMP sites throughout the United States, including Puerto Rico, between March 2007 and October 2009. Eligibility criteria included perinatal HIV infection or exposure, age 7 to <16 years, and previous medical care with accessible HIV treatment history. Youth included in these analyses had at least one complete SU questionnaire; data points spanned the age range 7–22 years, with self-reported SU assessed beginning at age 10.

Procedure

Institutional Review Boards at the Harvard T. H. Chan School of Public Health and each PHACS site approved the AMP study. Informed consent and age-appropriate assent were obtained for all youth participants and their primary caregivers according to local IRB guidelines. All measures with the exception of the Behavior Rating Inventory of Executive Functions (BRIEF) were available in Spanish and administered by bilingual personnel to study participants whose primary language was Spanish.

The administration of measures was staggered across study visits to minimize participant burden. Biannual visits included physical exams, medical chart reviews of health and medication status, and structured demographic interviews. Neurodevelopmental evaluations were administered at the time points indicated for each measure below. Quality assurance measures included examiner ratings of effort and mental status, examiner training in detection of signs of intoxication in participants, and examination of all test results by supervising psychologists.

Measures

Primary Outcome

Substance use was assessed using audio computer-assisted self-interview (ACASI), administered in a private location via laptop computer. Youth were informed that data would be transmitted directly to the data management center and unavailable to site personnel or caregivers in order to minimize social desirability of responses and facilitate disclosure of SU. Previous studies have indicated good concordance between self-report data using ACASI and drug toxicology (29). Questions included whether a participant had ever used the following substances: tobacco, alcohol, marijuana, inhalants, over the counter or prescription medications in order to get high, or any other non-prescription drugs (amphetamines/methamphetamine, cocaine/crack cocaine, sedatives, hallucinogens, heroin, or club drugs). Endorsement of any prior use prompted follow-up questions regarding the frequency of use over the prior 3 months, age of first use, and age of first regular use. Participants were asked the amount they typically used for tobacco, alcohol, and marijuana. The interview was administered at data collection time points 6 months, 2.5, 4, 5, 6, 7 and 8 years after entry, for those time points at which the participant was at least 10 years of age.

Cognitive and Behavioral Measures

The Wechsler Intelligence Scale for Children-Fourth Edition (WISC-IV)(30) and Wechsler Adult Intelligence Scale-Fourth Edition (WAIS-IV)(31) are widely used measures of general intellectual ability, providing a Full-Scale IQ and four Index scores: Verbal Comprehension Index, Perceptual Reasoning Index, Working Memory Index, and Processing Speed Index. The Full-Scale IQ and Index scores are age-normed standard scores, computed using published normative data with means of 100 and standard deviation of 15. The WISC-IV was given for children age 6–16 years, and the WAIS-IV for youth ages 17 years and older, at study entry and 3 years, with Processing Speed and Working Memory subtests also administered at 5 and 7 years.

The Children’s Color Trails Test (Color Trails)(32) is a standardized paper-and-pencil assessment of executive functioning processes including alternating and sustained visual attention, sequencing, psychomotor speed, cognitive flexibility, planning and inhibition-disinhibition. The first condition (Color Trails 1) requires the individual to connect randomly placed numbered circles in order as quickly as possible; the second condition (Color Trails 2) requires similarly connecting numbered circles but while alternating between two circle colors. Age-norm-referenced scores include standard scores for each trial completion time (mean =100, standard deviation=15; higher scores indicate faster performance), and percentile ranges for the Interference Index ([Color Trails 2 Time raw score–Color Trails 1 Time raw score]/Color Trails 1 Time raw score), a measure of cognitive flexibility and interference susceptibility (higher percentiles indicate less interference). The Color Trails test was administered at study time points 1, 3, 5, and 7 years for youth < 17 years.

The Behavior Rating Inventory of Executive Functioning, Parent-Report Form (BRIEF-PR)(33)and the BRIEF-Self-Report (BRIEF-SR)(34) are norm-referenced rating inventories that assess executive functioning in the performance of everyday tasks, across multiple domains. Included are: Behavioral Regulation Index (combines Inhibit, Shift, and Emotional Control scales); Metacognition Index (combines Initiate, Working Memory, Plan/Organize, Organization of Materials, and Monitor scales); and General Executive Composite, computed by combining scores from the Behavioral Regulation Index and Metacognitive Index. Age-norm-referenced T-scores (i.e., M=50, SD=10) are computed, with higher T-scores indicating more difficulties and T-score ≥65 indicating clinical significance. Caregivers completed the BRIEF-PR for youth up to age 18 years. Per BRIEF-SR guidelines, children and adolescents age 11 and older completed the BRIEF-SR. The questionnaire was administered as an interview to minimize literacy concerns, at study time points 1, 3, 5, and 7 years.

The Behavior Assessment System for Children, 2^nd Edition (BASC-2)(35) measures perception of youth emotional and behavioral health. The Sensation Seeking scale from the Self-Report of Personality was used to measure youths’ perception of their own propensity to take risks and seek out excitement; the scale T-score with mean of 50 and standard deviation of 10, computed using published age norms, was used in analyses. As with the BRIEF, the BASC-2 was administered as an interview, and higher scores indicate greater reported problems. It was administered at study entry and at follow-up time points 2, 2.5, 4, 6 and 8 years.

Demographic and Caregiver Information

Information collected via structured interview with the primary caregiver included age, sex, race/ethnicity, and primary language of child; household income; and the caregiver’s relationship to the child, education, psychiatric diagnoses, and SU and mental health treatment. Presence of any current psychiatric syndrome was determined using the Client Diagnostic Questionnaire (36, 37), administered by a psychologist in either English or Spanish. The Client Diagnostic Questionnaire is a validated structured screening interview used to identify current symptoms of psychiatric disorder(s), including depression, anxiety, post-traumatic stress disorder, and psychotic disorder: not otherwise specified, as well as alcoholic and nonalcoholic substance abuse/dependence, tailored to populations affected by HIV.

HIV-related Health Information for YPHIV

Chart abstraction provided current (study entry) and peak HIV-1 RNA viral load (VL; copies/mL); current and nadir CD4+ T-lymphocyte count and percent; CDC classification of HIV disease(38); diagnosis of encephalopathy; and current antiretroviral treatment.

Statistical Methods

Substance use was examined in three sets of analyses. Because some participants reported SU prior to entry, they were excluded from survival analyses examining predictors of use initiation for that substance. Separate prevalence analyses were performed to examine associations of SU prior to entry with the first measured cognitive or behavioral measurement and study entry demographic characteristics. Separate sets of analyses were conducted for alcohol, marijuana, tobacco, and “other substance” (all other drugs combined due to low prevalence) use.

Substance use characteristics:

Substance use severity was described by summarizing participants’ highest frequency use of each substance across all time points for the entire cohort.

Prevalence analysis:

This analysis modeled the odds of past SU at the time of the first cognitive or behavioral measurement. Generalized estimating equation models were used with a logistic link, binomial model, site level clustering and an exchangeable working correlation matrix to estimate adjusted odds ratios (ORs).

Survival analysis:

This prospective analysis included youth who had not used a given substance (e.g., alcohol) at the age of the first cognitive or behavioral measurement. It was designed to examine associations with the initiation of use of each substance. The initiation of SU included alcohol other than a few sips, any use of marijuana, at least one whole cigarette, and any other use, each evaluated separately. The Cox proportional hazards models with site specific baseline hazards and a robust variance estimator specified at the caregiver level was used to estimate hazard ratios (HRs) for the time to first SU. The neurocognitive variables of interest (e.g. Color Trails, or Full-Scale IQ) and age were treated as time dependent covariates.

Incidence and prevalence analyses were adjusted for SU at baseline (other than the substance for the modeled outcome, incidence analysis only), caregiver type, sex, Hispanic ethnicity, race, and age of the child using multivariable GEE and Cox models, as appropriate. To minimize multiple comparisons, incident and prevalence analyses were conducted by HIV status when the p-value was less than 0.1 for a Wald test of the interaction term between HIV status and the cognitive or behavioral measurement.

SAS version 9.4 was used for all analyses. Confidence intervals were reported as 95% intervals.

RESULTS

Group Characteristics

Of 678 youth, 601 completed at least one ACASI. Among those who did not complete an ACASI, the most common reason (16 participants) was that the youth was severely cognitively impaired and unable to independently complete the interview. Other reasons included caregiver refusal, caregiver unavailability to grant permission for the SU measure, and lack of time. Youth without an ACASI were younger at study entry than those with a completed ACASI (median of 10 and 11 years, respectively) and less likely to be Hispanic (83.1% and 70.5% non-Hispanic, respectively). A higher percentage of YPHEU than YPHIV completed an ACASI (94% vs 86%). Note that the percent of missing ACASIs that were due to severe cognitive impairment was similar between groups (YPHIV: 58.6%, YPHEU: 56.3%). In addition, there were regional differences, with a higher percentage (49.4%) of those who did not complete an ACASI from the South compared to those who did complete an ACASI (31.9%). Caregiver characteristics were similar. The majority (56%) of youth without an ACASI was missing cognitive and behavioral assessments as well; thus, these characteristics were not compared.

Table 1 presents means or proportions of demographic and other potential confounding variables by group (YPHIV and YPHEU) for participants and caregivers included in the prevalence analysis. The median age of study entry for the overall cohort was 11 years; YPHIV were on average 2 years older. At the last visit, median ages were 17 (YPHIV) and 16 (YPHEU) years. Among YPHIV, 75% were Black, compared to 64% of YPHEU; 24% of YPHIV and 37% of YPHEU were of Hispanic ethnicity. Among YPHIV, 44% were in the care of a biological parent, substantially fewer than YPHEU (80.6%).

Table 1.

Demographic and HIV disease characteristics of participants with (YPHIV) and without (YPHEU) HIV.

			Youth HIV Status (Median (Q1, Q3) or Percent)
	Characteristic		YPHIV (N=390)	YPHEU (N=211)
Youth
	Age at Study Entry	Median (Q1, Q3)	12 (10, 14)	10 (8, 12)
	Age at the Last ACASI Visit	Median (Q1, Q3)	17 (16, 19)	16 (14, 17)
	Sex	Female	53.0%	47.9%
	Ethnicity	Hispanic	24.4%	37.0%
	Race	White	24.0%	34.3%
		Black	74.7%	63.7%
		Other	1.4%	2.0%
	Region	Puerto Rico	6.2%	11.8%
		Northeast/Midwest	47.4%	31.3%
		South	30.8%	34.1%
		West	15.6%	22.7%
	Language use in Home	English	79.2%	71.6%
		Spanish	13.8%	19.4%
		Bilingual	5.1%	5.2%
		Other	1.8%	3.8%
YPHIV only
	Viral Load	>400 copies/ml	36.3%
	CD4 Cell Count at Entry	Median (Q1, Q3) cells/mm3	738 (521, 959)
	Nadir CD4 Cell Count	Median (Q1, Q3) cells/mm3	663 (462, 880)
	CDC Class	Mildly symptomatic	33.9%
		Moderately symptomatic	29.2%
		Severely symptomatic	23.1%
		Not symptomatic	13.7%
Caregiver
	IQ	Median (Q1, Q3)	87 (78, 100)	83 (74, 95)
	Any Psychiatric Syndrome?	Yes	29.3%	41.0%
	Race	White	35.7%	36.9%
		Black	62.9%	60.5%
		Other	1.4%	2.6%
	Positive for Drug Use?	Yes	2.0%	3.6%
	Treated for Alcohol/Drugs?	Yes	8.5%	16.3%
	Ethnicity	Hispanic	24.5%	36.9%
	Income	$≤20K	45.7%	66.0%
		$21–50K	36.1%	52.6%
		$>51K	18.2%	8.4%
	Caregiver Relationship	Biological parent(s)	43.6%	80.6%
		Other biological relative	24.9%	9.5%
		Non-relative	31.5%	10.0%

Characteristics of Substance Use

Table 2 shows the highest frequency of alcohol, marijuana, tobacco and other SU reported at any time point for the entire cohort and characteristics of their use. About half (48%) reported ever having had more than a few sips of alcohol, and of those, a majority (79%) reported having used alcohol within the past 3 months (13.9% at least weekly and 1.4% daily) and 56% reported ever having more than 5 drinks on one day. Slightly less than half (43%) of all participants reported having tried marijuana; of those, 78% had used it in the past 3 months (17.5% at least weekly, and 22.2% daily). A third of all participants (34%) had tried cigarettes, with 23% of those reporting daily use. However, the majority (86%) reported smoking 5 or fewer cigarettes per day. Only 3.3% of participants reported ever having used a drug other than marijuana.

Table 2.

Substance use frequency and characteristics (worst ever report)

Characteristic	Category	Percent1	Percent2
Alcohol
Lifetime use (N=6012)	Never Used		52.1%
Frequency of use in the past three months (N 1=288; N=6012)	No use in the past 3 months	21.2%	10.1%
	Once a month or less	39.6%	19.0%
	> once a month, < once a week	24.0%	11.5%
	One or more times a week	13.9%	6.7%
	Every day	1.4%	0.7%
Greater than 5 drinks on one day – lifetime (N 1=277; N=6012)	Never		52.1%
Greater than 5 drinks on one day in the past three months (N 1=277; N=6012)	None in the past 3 months	43.6%	16.5%
	Less than monthly	33.5%	12.6%
	Monthly	11.9%	4.5%
	Weekly	9.3%	3.5%
	Daily or almost daily	1.8%	0.7%
Marijuana
Lifetime use (N=6012)	Never Used		57.2%
Frequency of use in the past three months (N 1=257; N=6012)	No use in the past 3 months	22.2%	9.5%
	Once a month or less	24.1%	10.3%
	> once a month, < once a week	14.0%	6.0%
	One or more times a week	17.5%	7.5%
	Every day	22.2%	9.5%
			0.0%
Number of joints per day (N 1=253; N=6012)	Never Used		57.2%
	1–2	42.3%	17.8%
	3–5	30.8%	3.5%
	>5	26.9%	11.3%
Tobacco
Lifetime use (N=6012)	Never Smoked		65.4%
Frequency of use in the past three months (N 1=101; N=6012)	No use in the past 3 months		17.8%
	One or more puffs	17.8%	3.0%
	1 cigarette	28.7%	4.8%
	2–5 cigarettes	39.6%	6.7%
	6–15 cigarettes	5.9%	1.0%
	16–25 cigarettes	5.9%	1.0%
	More than 1 pack per day	2.0%	0.3%
Illicit drugs other than marijuana
Lifetime use (N 2=601)	Never Used		96.7%
Frequency of use in the past three months (N 1=57; N 2 =601)	No use in the past 3 months	52.6%	0.5%
	Once a month or less	29.8%	1.0%
	> once a month, < once a week	8.8%	0.3%
	One or more times a week	7.0%	0.3%
	Every day	1.8%	0.0%

¹Among youth who endorsed ever using the substance with available data

²Among all youth with an ACASI

The Supplemental Table shows self-reported characteristics and consequences of SU. Among the 341 participants who acknowledged ever using alcohol (more than a sip) or illicit drugs, approximately 42% reported using substances to relax or feel better and 42% reported using while alone, 30% reported that family or friends told them they should use less, and 23% had gotten into trouble or forgotten things while using. Approximately 13% reported that they were unable to do regular activities at least once in the past due to SU other than alcohol.

Predictors of Incident Substance Use

Results of unadjusted and adjusted Cox proportional hazards regression analyses evaluating predictors of SU initiation are shown in Table 3. Cognitive tests: In adjusted analyses, better (faster) performance on the Color Trails 1 was associated with lower rate of marijuana (HR=0.97, 95% CI = 0.95–0.99) and tobacco use (HR=0.97, 95% CI = 0.94–1.00); better performance on Color Trails 2 was associated with a higher rate of marijuana use (HR=1.02, 95% CI = 1.00–1.04). For the WISC, higher Full-Scale IQ was associated with a higher rate of Other Substance use (HR=1.43, 95% CI = 1.00–2.05); with regards to index scores, higher Working Memory score was associated with alcohol use (HR=1.21, 95% CI = 1.04–1.42), and higher Verbal Comprehension score with use of Other Substances (HR=1.43, 95% CI = 1.02–2.00). Incident cognitive test interactions with HIV status: No interactions with HIV status were observed for the findings reported above or for alcohol or marijuana use in general. However, youth with PHIV and higher Verbal Comprehension scores were less likely, and YPHEU more likely, to use tobacco. Behavioral measures: Higher self-reported Sensation Seeking was strongly associated with a higher rate of alcohol (HR=1.32, 95% CI = 1.11–1.58) and marijuana use (HR=1.29, 95% CI = 1.07–1.55). Youth self-reported BRIEF Metacognition Index scores in the clinically significant range were associated with reduced rate of later alcohol use (HR=0.39, 95% CI = 0.16–0.96); there were no strong associations with the caregiver BRIEF. Incident behavioral measure interactions with HIV status: No interactions with HIV status were observed for the reported findings regarding Sensation Seeking and BRIEF Metacognition Index scores. A tendency for caregiver-reported BRIEF Behavioral Regulation Index to interact with HIV in associations with marijuana use reflected higher odds of marijuana use among YPHIV with caregiver-reported Behavioral Regulation Index scores in the clinically significant range (HR=2.02, 95% CI = 1.05 – 3.88), but the result was inconclusive among YPHEU (HR=0.83, 95% CI = 0.31–2.26).

Table 3.

Association of incident risk by substance with behavioral and cognitive predictors.

			Unadjusted		Adjusted
Substance Use Outcome	Exposure of Interest	Incident Cases in the Adjusted Model	Hazard Ratio (95% CI)	p-value	Hazard Ratio (95 % CI)	p-value
Alcohol	Interference Index (Raw Score)	119	1.05 (0.84, 1.32)	0.67	0.99 (0.78, 1.27)	0.94
	CCTT-1: Time T score	119	0.97 (0.95, 0.99)	<0.001	0.98 (0.96, 1.01)	0.15
	CCTT-2: Time T score	119	1.00 (0.99, 1.02)	0.72	1.00 (0.98, 1.02)	0.79
	Child BRIEF: GEC ≥65	103	0.86 (0.43, 1.73)	0.67	0.54 (0.24, 1.21)	0.13
	Child BRIEF: BRI ≥65	103	1.46 (0.78, 2.74)	0.24	1.09 (0.54, 2.20)	0.80
	Child BRIEF: MI ≥65	103	0.68 (0.31, 1.48)	0.33	0.39 (0.16, 0.96)	0.041
	Caregiver BRIEF: GEC ≥65	110	1.26 (0.79, 2.02)	0.33	1.25 (0.74, 2.10)	0.40
	Caregiver BRIEF: BRI ≥65	110	1.11 (0.68, 1.82)	0.68	1.13 (0.65, 1.96)	0.67
	Caregiver BRIEF: MI ≥65	110	1.60 (1.04, 2.48)	0.034	1.50 (0.93, 2.42)	0.096
	Sensation Seeking (Per 10 pts)	130	1.31 (1.11, 1.55)	0.001	1.32 (1.11, 1.58)	0.002
	WISC: FSIQ (Per 15 pts)	176	1.10 (0.94, 1.27)	0.24	1.14 (0.97, 1.35)	0.12
	Perceptual Reasoning (Per 15)	176	1.03 (0.88, 1.20)	0.73	1.05 (0.89, 1.25)	0.54
	Processing Speed (Per 15 pts)	176	1.01 (0.87, 1.17)	0.92	1.06 (0.90, 1.25)	0.49
	Verbal comprehension (Per 15 pts)	176	1.13 (0.96, 1.32)	0.14	1.14 (0.96, 1.35)	0.14
	Working Memory (Per 15)	176	1.20 (1.03, 1.40)	0.018	1.21 (1.04, 1.42)	0.017
Marijuana	Interference Index (Raw Score)	126	1.08 (0.87, 1.33)	0.50	1.01 (0.81, 1.26)	0.93
	CCTT-1: Time T score	126	0.95 (0.93, 0.97)	<0.001	0.97 (0.95, 0.99)	0.008
	CCTT-2: Time T score	126	1.01 (1.00, 1.03)	0.082	1.02 (1.00, 1.04)	0.026
	Child BRIEF: GEC ≥65	116	1.28 (0.71, 2.03)	0.41	0.94 (0.49, 1.80)	0.85
	Child BRIEF: BRI ≥65	116	1.82 (1.03, 3.22)	0.039	1.22 (0.65, 2.30)	0.54
	Child BRIEF: MI ≥65	116	1.36 (0.76, 2.43)	0.299	1.02 (0.53, 1.97)	0.95
	Caregiver BRIEF: GEC ≥65	120	1.34 (0.87, 2.07)	0.18	1.02 (0.65, 1.60)	0.93
	Caregiver BRIEF: BRI ≥65	120	1.41 (0.90, 2.23)	0.14	1.29 (0.80, 2.09)	0.30
	Caregiver BRIEF: MI ≥65	120	1.57 (1.04, 2.39)	0.033	1.16 (0.75, 1.81)	0.50
	Sensation Seeking (Per 10 pts)	138	1.39 (1.18, 1.63)	<0.001	1.29 (1.07, 1.55)	0.007
	WISC: FSIQ (Per 15 pts)	172	0.98 (0.84, 1.14)	0.80	1.06 (0.89, 1.26)	0.52
	Perc Reasoning (Per 15 pts)	172	1.06 (0.90, 1.24)	0.48	1.09 (0.92, 1.30)	0.32
	Processing Speed (Per 15 pts)	172	0.98 (0.85, 1.14)	0.84	1.13 (0.96, 1.33)	0.15
	Verbal comprehension (Per 15 pts)	172	0.91 (0.78, 1.07)	0.26	0.95 (0.80, 1.14)	0.60
	Working Memory (Per 15 pts)	172	1.03 (0.89, 1.20)	0.68	1.07 (0.90, 1.25)	0.45
Other Substance	Interference Index (Raw Score)	41	1.07 (0.73, 1.59)	0.72	1.07 (0.72, 1.58)	0.75
	CCTT-1: Time T score	41	0.96 (0.93, 1.00)	0.053	0.96 (0.92, 1.00)	0.071
	CCTT-2: Time T score	41	1.01 (0.99, 1.04)	0.35	1.01 (0.98, 1.04)	0.53
	Child BRIEF: GEC ≥65	35	0.42 (0.10, 1.83)	0.25	0.16 (0.02, 1.26)	0.082
	Child BRIEF: BRI ≥65	35	0.40 (0.09, 1.76)	0.23	0.14 (0.02, 1.11)	0.062
	Child BRIEF: MI ≥65	35	0.46 (0.11, 1.97)	0.29	0.18 (0.02, 1.44)	0.11
	Caregiver BRIEF: GEC ≥65	39	0.82 (0.35, 1.91)	0.65	0.73 (0.29, 1.87)	0.51
	Caregiver BRIEF: BRI ≥65	39	0.87 (0.35, 2.13)	0.76	0.74 (0.28, 1.97)	0.54
	Caregiver BRIEF: MI ≥65	39	0.88 (0.38, 2.05)	0.78	0.98 (0.40, 2.39)	0.96
	Sensation Seeking (Per 10 pts)	43	1.05 (0.78, 1.42)	0.75	0.97 (0.68, 1.38)	0.86
	WISC: FSIQ (Per 15 pts)	50	1.50 (1.09, 2.06)	0.013	1.43 (1.00, 2.05)	0.048
	Perc Reasoning (Per 15 pts)	50	1.45 (1.05, 2.00)	0.022	1.25 (0.88, 1.77)	0.21
	Processing Speed (Per 15 pts)	50	1.19 (0.89, 1.58)	0.24	1.13 (0.82, 1.56)	0.47
	Verbal comprehension (Per 15 pts)	50	1.38 (1.02, 1.87)	0.035	1.43 (1.02, 2.00)	0.038
	Working Memory (Per 15 pts)	50	1.33 (0.98, 1.79)	0.068	1.28 (0.91, 1.78)	0.15
Tobacco	Interference Index (Raw Score)	82	1.24 (0.96, 1.61)	0.10	1.19 (0.93, 1.53)	0.17
	CCTT-1: Time T score	82	0.96 (0.93, 0.98)	<0.001	0.97 (0.94, 1.00)	0.03
	CCTT-2: Time T score	82	1.01 (0.99, 1.03)	0.39	1.00 (0.98, 1.02)	0.86
	Child BRIEF: GEC ≥65	69	0.64 (0.23, 1.81)	0.40	0.9 (0.30, 2.64)	0.84
	Child BRIEF: BRI ≥65	69	1.00 (0.38, 2.61)	1.00	1.43 (0.51, 4.00)	0.49
	Child BRIEF: MI ≥65	69	1.01 (0.42, 2.39)	0.99	1.28 (0.50, 3.29)	0.60
	Caregiver BRIEF: GEC ≥65	77	0.97 (0.53, 1.78)	0.93	0.87 (0.47, 1.63)	0.66
	Caregiver BRIEF: BRI ≥65	77	1.52 (0.87, 2.68)	0.14	1.43 (0.78, 2.61)	0.24
	Caregiver BRIEF: MI ≥65	77	1.05 (0.59, 1.89)	0.86	0.99 (0.53, 1.83)	0.97
	Sensation Seeking (Per 10 pts)	87	1.29 (1.05, 1.59)	0.02	1.17 (0.93, 1.48)	0.19
	WISC: FSIQ (Per 15 pts)	119	1.06 (0.89, 1.28)	0.51	1.04 (0.84, 1.28)	0.73
	Perc Reasoning (Per 15 pts)	119	1.13 (0.94, 1.37)	0.19	1.10 (0.89, 1.36)	0.36
	Processing Speed (Per 15 pts)	119	0.99 (0.83, 1.19)	0.93	0.95 (0.77, 1.17)	0.64
	Verbal comprehension (Per 15 pts)	119	1.00 (0.83, 1.21)	0.98	0.98 (0.80, 1.21)	0.87
	Working Memory (Per 15 pts)	119	1.14 (0.95, 1.37)	0.16	1.13 (0.92, 1.39)	0.26

¹Cox proportional hazards regression, withsite specific baseline hazards and clustered on caregiver, was used to estimate hazard ratios, adjustment included substance caregiver type, sex, Hispanic ethnicity, race, baseline substance use (other than the outcome), and age. Odds ratios are displayed per test points (pts) for continuous variables, means are displayed on the raw scale. CCTT=Children Color Trials Test; BREIF=Behavior Rating Inventory of Executive Function; GEC= Global Executive Composite, BRI=Behavioral regulation index; MI=Metacognition index; WISC=Wechsler intelligence scale for children; FSIQ=Full scale intelligence quotient.

Associations with Prevalent Substance Use

Table 4 shows the results of logistic regression models for associations of each cognitive and behavioral measure with adjusted odds of reporting past SU, as well as descriptive statistics for the cognitive measures among youth with and without prevalent use of each substance. Cognitive tests: In adjusted analyses, faster (better) performance on Color Trails 2 was associated with significantly higher odds of reporting past alcohol use at the time of the baseline ACASI (OR=1.02, 95% CI = 1.00–1.05), and faster performance on Color Trails 1 was associated with reporting past marijuana use (OR=1.02, 95% CI = 1.00–1.04). Weak associations with SU and the Interference Index for the combined YPHIV and YPHEU analysis set were observed. Higher WISC Full-Scale IQ was associated with both past alcohol (OR=1.77, 95% CI = 1.33–2.35) and past marijuana use (OR=1.52, 95% CI = 1.13–2.05). Similar positive associations with alcohol and marijuana use were seen for all WISC index scores with the exception of marijuana and the Verbal Comprehension index score (see Table 3). The risk for tobacco use with increasing WISC FSIQ ranged from a small decrease to a moderate increase (95% CI =0.96–1.67). Prevalence cognitive test interactions with HIV status: Among YPHIV, higher risk of tobacco use was observed with increasing interference Index scores (OR=1.38, 95%CI 0.97–1.98), but lower risk for YPHEU (OR=0.57, 95%CI 0.31–1.08). No interactions were observed for alcohol or marijuana use. Behavioral measures: Higher odds of past alcohol (OR=1.85, 95% CI=1.48–2.30), marijuana (OR=1.43, 95% CI=1.14–1.80), and tobacco use (OR=1.51, 95% CI=1.20–1.89) were strongly associated with higher self-reported Sensation Seeking, and with greater likelihood of self-reported BRIEF General Executive Composite and Behavioral Regulation Index scores that were in the clinically significant range (i.e., ≥65; General Executive Composite: OR=1.97,95% CI=1.02–3.78; OR=2.21, 95% CI=1.12–4.37;OR= 3.72, 95% CI=2.02–6.85, for alcohol, marijuana, and tobacco, respectively; Behavioral Regulation Index: OR=2.18, 95% CI = 1.10–4.33; OR=2.65 95% CI = 1.32–5.32; OR=5.49 95% CI = 2.89–10.4 for alcohol, marijuana, and tobacco, respectively); alcohol and tobacco use were also associated with self-reported BRIEF Metacognition Index scores in the clinically significant range (OR=2.13, 95% CI = 1.08–4.19; OR=2.59, 95% CI = 1.37–4.90, respectively). Higher odds of past tobacco use were associated with caregiver BRIEF General Executive Composite (OR=1.79, 95% CI = 1.01–3.19) and Behavioral Regulation Index (OR=2.07, 95% CI = 1.16–3.69) in the clinically significant range. Prevalence behavioral measure interactions with HIV status: There were no interactions with HIV status for any behavioral measures in prevalence analyses.

Table 4.

Associations of cognitive and behavioral measures with substance use prior to study entry.

Substance Use Outcome	Exposure of Interest	N for the Adjusted Model	No Prior Use	Prior Use	Adjusted Odds Ratio (95% CI)1	p-value
			Mean or Percent
Alcohol	Interference Index (Raw Score)	547	1.3	1.3	0.98 (0.74, 1.31)	0.91
	CCTT-1: Time T score	547	38.2	40.5	1.02 (1.00, 1.04)	0.075
	CCTT-2: Time T score	547	36.9	37.1	1.02 (1.00, 1.05)	0.028
	Child BRIEF: GEC ≥65	498	10.20%	18.10%	1.97 (1.02, 3.78)	0.043
	Child BRIEF: BRI ≥65	498	8.20%	18.10%	2.18 (1.10, 4.33)	0.025
	Child BRIEF: MI ≥65	498	9.30%	16.70%	2.13 (1.08, 4.19)	0.029
	Caregiver BRIEF: GEC ≥65	485	17.60%	27.30%	1.59 (0.88, 2.87)	0.12
	Caregiver BRIEF: BRI ≥65	485	18.40%	24.80%	1.73 (0.95, 3.15)	0.073
	Caregiver BRIEF: MI ≥65	485	16.80%	20.70%	1.09 (0.59, 2.04)	0.78
	Sensation Seeking (Per 10 pts)	561	48.1	54.7	1.85 (1.48, 2.30)	<0.001
	WISC: FSIQ (Per 15 pts)	582	85.8	90.3	1.77 (1.33, 2.35)	<0.001
	Perceptual Reasoning (Per 15 pts)	582	90.2	96.3	1.95 (1.44, 2.64)	<0.001
	Processing Speed (Per 15 pts)	582	87.9	90.1	1.53 (1.17, 2.01)	0.002
	Verbal comprehension (Per 15 pts)	582	87.8	90.5	1.62 (1.21, 2.17)	0.001
	Working Memory (Per 15 pts)	582	88.5	91.9	1.33 (1.02, 1.73)	0.034
Marijuana	Interference Index (Raw Score)	546	1.3	1.3	1.11 (0.82, 1.51)	0.50
	CCTT-1: Time T score	546	38.3	41.1	1.02 (1.00, 1.04)	0.036
	CCTT-2: Time T score	546	37	36.8	1.02 (1.00, 1.04)	0.12
	Child BRIEF: GEC ≥65	498	10.50%	16.50%	2.21 (1.12, 4.37)	0.022
	Child BRIEF: BRI ≥65	498	8.50%	20.20%	2.65 (1.32, 5.32)	0.006
	Child BRIEF: MI ≥65	498	10.00%	16.50%	2.00 (0.98, 4.10)	0.057
	Caregiver BRIEF: GEC ≥65	484	18.30%	27.50%	1.21 (0.65, 2.26)	0.54
	Caregiver BRIEF: BRI ≥65	484	18.80%	25.30%	1.39 (0.74, 2.60)	0.31
	Caregiver BRIEF: MI ≥65	484	17.60%	18.70%	0.74 (0.65, 2.26)	0.54
	Sensation Seeking (Per 10 pts)	561	48.9	53.3	1.43 (1.14, 1.80)	0.002
	WISC: FSIQ (Per 15 pts)	581	86	90	1.52 (1.13, 2.05)	0.006
	Perceptual Reasoning(Per 15 pts)	581	90.7	94.9	1.47 (1.08, 2.00)	0.014
	Processing Speed (Per 15)	581	88	90.3	1.53 (1.14, 2.05)	0.005
	Verbal comprehension (Per 15)	581	88	89.8	1.27 (0.94, 1.72)	0.13
	Working Memory (Per 15)	581	88.4	93.7	1.52 (1.13, 2.05)	0.006
Other Substance	Interference Index (Raw Score)	547	1.3	1.4	1.31 (0.94, 1.82)	0.11
	CCTT-1: Time T score	547	38.6	40.1	1.01 (0.99, 1.04)	0.32
	CCTT-2: Time T score	547	37	36.3	1.00 (0.97, 1.03)	0.96
	WISC: FSIQ Per 15 Points	582	86.3	88.5	1.17 (0.85, 1.62)	0.33
	Child BRIEF: GEC ≥65	498	12.30%	13.50%	1.20 (0.51, 2.82)	0.68
	Child BRIEF: BRI ≥65	498	10.80%	13.50%	1.35 (0.58, 3.14)	0.49
	Child BRIEF: MI ≥65	498	10.80%	17.30%	1.99 (0.89, 4.44)	0.093
	Caregiver BRIEF: GEC ≥65	485	19.20%	27.70%	1.26 (0.59, 2.70)	0.55
	Caregiver BRIEF: BRI ≥65	485	19.40%	25.50%	1.36 (0.63, 2.95)	0.43
	Caregiver BRIEF: MI ≥65	485	16.90%	25.50%	1.33 (0.61, 2.92)	0.48
	Perceptual Reasoning (Per 15 pts)	582	90.9	93.6	1.15 (0.82, 1.61)	0.41
	Processing Speed (Per 15 pts)	582	88.4	86.7	1.00 (0.71, 1.39)	0.98
	Verbal comprehension (Per 15 pts)	582	88	90.4	1.21 (0.87, 1.68)	0.53
	Working Memory (Per 15 pts)	582	88.7	92.3	1.23 (0.89, 1.71)	0.21
Tobacco	Interference Index (Raw Score)	545	1.3	1.4	1.05 (0.79, 1.40)	0.72
	CCTT-1: Time T score	545	38.6	37.9	1.00 (0.98, 1.02)	0.67
	CCTT-2: Time T score	545	37.3	35.3	1.00 (0.97, 1.02)	0.67
	Child BRIEF: GEC ≥65	498	8.50%	26.60%	3.72 (2.02, 6.85)	<0.001
	Child BRIEF: BRI ≥65	498	6.20%	28.40%	5.49 (2.89, 10.41)	<0.001
	Child BRIEF: MI ≥65	498	8.70%	21.10%	2.59 (1.37, 4.90)	0.004
	Caregiver BRIEF: GEC ≥65	484	17.40%	32.20%	1.79 (1.01, 3.19)	0.048
	Caregiver BRIEF: BRI ≥65	484	17.60%	31.00%	2.07 (1.16, 3.69)	0.014
	Caregiver BRIEF: MI ≥65	484	16.10%	25.30%	1.33 (0.72, 2.47)	0.36
	Sensation Seeking (Per 10 pts)	560	48.8	53.5	1.51 (1.20, 1.89)	<0.001
	WISC: FSIQ (Per 15 pts)	579	86.4	87.8	1.27 (0.96, 1.67)	0.092
	Perceptual Reasoning (Per 15 pts)	579	90.9	93.3	1.31 (0.98, 1.75)	0.068
	Processing Speed (Per 15 pts)	579	88.4	87.6	1.12 (0.85, 1.48)	0.40
	Verbal Comprehension (Per 15 pts)	579	88.1	89.1	1.25 (0.94, 1.66)	0.13
	Working Memory (Per 15 pts)	579	88.8	91.1	1.24 (0.94, 1.63)	0.12

¹Generalized estimating equation clustered on site was used to estimate odds ratios, adjustment included substance caregiver type, sex, Hispanic ethnicity, race, and age. Odds ratios are displayed per test points (pts) for continuous variables, means are displayed on the raw scale. CCTT=Children Color Trials Test; BREIF=Behavior Rating Inventory of Executive Function; GEC= Global Executive Composite, BRI=Behavioral regulation index; MI=Metacognition index; WISC=Wechsler intelligence scale for children; FSIQ=Full scale intelligence quotient.

DISCUSSION

The analyses reported herein provide a detailed description of SU among YPHIV and YPHEU age 7–22, showing that about half had initiated alcohol or marijuana use by the end of study follow-up. Of concern, among those who had used marijuana recently, almost one in five were using it daily. Daily alcohol and tobacco use were less common. Up to 42% of participants who had used substances reported impacts of SU on their daily lives that were concerning, such as using substances alone, riding in a car driven by someone who was high, and friends/family suggesting that they cut back. In general, youth living with or exposed to HIV perinatally have, like the overall youth population, significant risk for some degree of SU and negative consequences thereof.

Consistent with Alperen and other studies, predictors of SU were largely consistent across YPHIV and YPHEU groups in adjusted analyses despite the known risks for cognitive impairment among YPHIV(16). Interactions of HIV status and behavioral or cognitive predictors, when present, were generally weak. However, unlike their uninfected same age peers, YPHIV have the potential for increased risk of disease-related complications and HIV transmission due to indirect effects of SU on medication and healthcare management, and possibly direct neurophysiological interactions with HIV. For example, Bucek et al., reported an association between SU disorders in YPHIV and unsuppressed viral loads over time(2), raising concern about the health and secondary HIV transmission risks that accompany lack of viral suppression. Thus, understanding predictors of SU, particularly those with intervention implications, is critically important for YPHIV and has implications for public health.

As for youth in general, problems with self-regulation were associated with SU in our cohort. Survival analyses showed that self-reported sensation seeking strongly predicted incident alcohol or marijuana use among both YPHIV and YPHEU. Similarly, self-reported behavioral regulation problems in the clinical range predicted incident marijuana use within YPHIV. Prior use of alcohol, marijuana, and tobacco were also associated with self-reported sensation seeking, underscoring the strong role of self-regulation in SU initiation. These findings suggest that measuring self-regulation for screening and targeting appropriate self-regulation-based intervention tools may result in delaying SU initiation. In addition, broader mental health screening of youth with HIV is crucial. This is highlighted by some of the reasons endorsed for SU in our sample (e.g., to relax or feel better) and the known prevalence of mental health concerns among youth affected by HIV(14, 39–41).

Analyses of prior use of alcohol or marijuana showed that in our cohort it was largely associated with better, rather than worse, cognitive performance; this pattern was seen for a global IQ measure as well as for measures reflecting visual functioning, working memory/concentration, processing speed, and, for alcohol, verbal comprehension. Survival analyses also revealed that better cognitive functioning predicted higher risk of SU initiation; for example, youth with better working memory or fewer self-reported problems in metacognition were more likely to initiate alcohol use, and those with higher verbal comprehension or global cognitive functioning more likely to use illicit substances other than marijuana. In the case of tobacco, higher verbal comprehension was associated with incident use for YPHIV but not YPHEU. These cognitive findings, consistent with positive associations between SU and academic achievement previously reported from our cohort (13), are intriguing and warrant follow-up. Hypotheses to explore underlying this pattern could include, for example, a tendency for caregivers to grant greater autonomy to higher functioning youth and be more protective of those with impairments, or for higher functioning youth to display better social and peer functioning and thus have greater participation in unmonitored peer activities. In addition, youth with greater cognitive maturity may be more willing or able to engage in risk-taking behavior. Finally, higher functioning youth may have greater financial resources and thus access to substances; follow-up analyses including income as a confounder found that observed relationships were similar, although low income variability in our sample limited our ability to examine this factor. Future studies could examine associations between cognitive function, caregiver treatment, youth self-efficacy and SU among YPHIV to guide family-based interventions for SU prevention.

Although they may not be at higher risk of SU, youth with PHIV and cognitive problems warrant particular concern if they initiate SU. It is unknown whether a central nervous system compromised by current or past HIV disease progression during development is more vulnerable to harm from SU, which could have long-term consequences. Cognitive domains typically vulnerable to impact by both HIV and specific substances (e.g., memory functioning by marijuana) are particularly important topics for further research. In addition, cognitively impaired youth may be more likely to manifest maladaptive use patterns or interference by SU in daily activities; for example, a study of intellectually disabled youth found that, although they were less likely to use alcohol, those who did initiate use were at increased risk for problematic alcohol behavior(42). Future longitudinal studies are needed to determine whether youth with HIV-related cognitive compromise have a higher risk of negative consequences of SU, and whether there are differences by type of substance.

This study is not without limitations. A cohort of youth without HIV exposure was not available to examine its role in the relationships of interest. Despite the broad age range, numbers of youth above age 18 were low. Youth with severe cognitive impairment were excluded from analyses due to inability to complete the ACASI or questionable response validity. Caution is indicated in interpreting the self-report of individuals with global cognitive impairment; however, impairment when it exists is generally mild among our sample and research suggests that individuals with mild to borderline intellectual impairment are able to provide meaningful SU self-reports in comparison with collateral report and biomarker data(43).

CONCLUSIONS

This longitudinal study examined cognitive and behavioral predictors of self-reported SU among youth with PHIV or PHEU in order to inform prevention and intervention efforts. For both groups, problems with self-regulation such as high sensation-seeking and problems with everyday executive functioning strongly predicted initiation of alcohol and marijuana use as well as prevalent use. These results and the paucity of HIV group differences suggest that SU screening and interventions, including those focused on self-regulation, that have been developed for the general adolescent population might be effective for YPHIV, who may face HIV-specific negative consequences of SU. Intriguingly, increased risk for use of these substances was also associated with better cognitive functioning, a finding that warrants further examination. Adolescent SU involves social, cognitive and biological factors, and YPHIV and YPHEU may have additional complexities that warrant additional research and screening.

Availability of data and material:

All data support the claims herein and comply with field standards. Partial data from the Pediatric HIV/AIDS Cohort Study Adolescent Master Protocol (PHACS AMP) are available at https://dash.nichd.nih.gov.