A Twin Study of the Neuropsychological Consequences of Stimulant Abuse

THIS STUDY examined whether residual neuropsychological effects of stimulant abuse persist beyond a year of abstinence. Previous studies demonstrate that acute withdrawal from cocaine use is associated with deficits in memory, visuospatial abilities, and concentration.¹ In users with up to 45 days of abstinence, deficits remain in verbal memory, complex attention, concept formation,² executive functioning, spatial memory,³ sustained attention, learning, memory, problem solving, abstraction,⁴ and improvement of memory functioning over time.⁵ At 45 to 100 days of abstinence, deficits exist in short-term auditory recall,⁶ psychomotor slowing, sequential organizational skills, spatial construction skills, and motor strength,⁷ with increased resting tremor,⁸ a slower reaction time to visual and auditory discrimination tasks,⁹ and a slower reaction time on an auditory signal detection task.¹⁰ At 6 months of abstinence, Strickland et al¹¹ found deficits in attention, concentration, new learning, visual and verbal memory, word production, and visuomotor integration. Despite these findings, Selby and Azrin¹² found no significant deficits in former cocaine abusers who were abstinent for 36 months.

Cocaine and amphetamines have similar mechanisms of biological action,¹³ and thus, their consequences may also be similar. Of the few studies that examine the neuropsychological effects of amphetamine abuse, findings are similar to those in cocaine abuse, with deficits seen in memory^{14 - 15} and attention.¹⁶

Inconsistencies in study findings may result from methodological differences. Sample sizes were small (range, 8-61 subjects), nonrepresentative treatment samples were commonly used, and the test performance of abusers was compared with norms rather than with control subjects.^{11 ,17} Studies also differed in the presence of potential confounding factors, such as polydrug abuse^{1 ,6 ,18} and alcohol abuse.¹⁹ Cocaine abusers had a lower IQ compared with controls in some studies.^{4 ,20} They also differed in the prevalence of comorbid psychiatric disorders, such as antisocial personality disorder,²¹ which may independently influence neuropsychological performance,^{22 - 23} and attention-deficit/hyperactivity disorder (ADHD), which would result in preexisting attention problems.²⁴

The present study used a co-twin control research design to assess the presence of residual neuropsychological impairments after a minimum of 1 year of abstinence from heavy stimulant abuse. Given that twins share 50% (dizygotic[DZ]) or 100% (monozygotic [MZ]) of their genes and many environmental experiences, co-twins provide ideal controls for many factors. If a group of drug abusers is compared with an unrelated group of nonabusers, it is impossible to distinguish whether neuropsychological differences between the groups predated drug abuse or are consequences of it. Previous research on alcohol abuse suggests that familial vulnerability for substance abuse may be reflected in neuropsychological performance.²⁵ Our research design controls for such familial vulnerability.

Study participants were members of the Vietnam Era Twin Registry, a national registry consisting of male-male twin pairs in which both members served in the military during the Vietnam War era (1965-1975) and were born between January 1, 1939, and December 31, 1957. A complete description of the registry construction²⁶ and characteristics²⁷ and the method of zygosity determination²⁸ have been previously published.

Diagnostic data were collected in 1992 on 3234 MZ and DZ twin pairs for the Harvard Twin Study of Drug Abuse and Dependence. Registry participants were interviewed by telephone using the Diagnostic Interview Schedule, Version III–Revised (DIS-III-R).²⁹ Interviews were performed by telephone by the Institute for Survey Research at Temple University, Philadelphia, Pa. Interviewers were trained by one of us (M.J.L.), and their supervisor, who had attended a training course at Washington University, St Louis, Mo, conducted by the developers of the DIS-III-R. Telephone interviews were used instead of face-to-face interviews because of the geographical diversity of the sample. A number of studies^{30 - 32} have supported the comparability of telephone and face-to-face interviews. Lifetime diagnoses were made according to the criteria of the DSM-III-R.³³ We identified 124 twin pairs (69 MZ and 55 DZ) who met the following criteria: (1) 1 member used stimulants at least once weekly for a minimum of 1 year, and his twin sibling never used stimulants regularly (ie, on a weekly basis), and (2) the stimulant-abusing twin reported not using stimulants for at least 1 year before the interview.

In 1995, we successfully recruited 62 of these pairs to participate in the current study. Participants were offered the opportunity to travel to one of 2 research sites at Harvard University, Boston, Mass, or Washington University, or to undergo evaluation in their hometown by one of the study's research assistants. Before the face-to-face evaluation, study research assistants explained the procedures and obtained written informed consent. The face-to-face evaluation included a questionnaire inquiring about drug use since the 1992 telephone interview.

We excluded 12 twin pairs after the evaluation, on the basis of the following criteria: (1) report of stimulant abuse in the past year (5 pairs);(2) regular use of marijuana since the previous survey and use of marijuana in the past year (6 pairs); and (3) presence of a major medical condition that would affect neuropsychological functioning (1 member of 1 pair displayed psychotic symptoms during the evaluation). Reasons for excluding recent marijuana users included: concerns that recent marijuana use might complicate the attempt to identify adverse effects of stimulants; a reasonable sample size remained after their exclusion; and the Harvard Twin Study had a separate focus on twin pairs discordant for marijuana use. The analyses for this article included 50 twin pairs (31 MZ and 19 DZ) discordant for heavy use of stimulants.

During the 1992 interview we assessed the number of days of stimulant use during the subjects' lifetime with separate sets of questions for cocaine (cocaine, crack, or coca leaves) and amphetamine abuse (amphetamines, khat, or ice). Questions included the following: Have you ever used [drug]? Have you ever used [drug] more than 5 times? Have you ever used [drug] regularly, ie, once a week or more? How many days per week did you use [drug] during your period of most frequent use? How old were you when you started your period of most frequent use? How old were you when you ended your period of most frequent use? and What was the longest period that you used [drug] almost every day? At the second assessment, subjects were asked whether they had used the drugs in the previous 6 years (since administration of the DIS-III-R). If so, the frequency questions were repeated to assess drug use in this intervening period. In addition, subjects were asked how recently they used the drugs.

For the 1992 data, the minimum number of days of stimulant use was the greater of the following calculations: (1) the total number of days of stimulant use during the period of regular use, (2) the total number of days of stimulant use during the period of most frequent use, or (3) the total number of days of stimulant use during the longest period of daily stimulant use. Additional days of stimulant use since the 1992 interview were added to this estimate of the minimum number of days of stimulant use. Total days of cocaine and amphetamine use were calculated separately, then total days of stimulant use were generated by summing these scores.

When making arrangements for the examination, all subjects were asked to abstain from alcohol use for at least 24 hours before the evaluation. On the day of testing, subjects were asked about relevant neuropsychological history, including education, special education for learning problems, diagnosis of a learning disability or ADHD, and history of head injuries.

Subjects were administered an extensive neuropsychological test battery. General intelligence was assessed with the Wechsler Adult Intelligence Scale–Revised (Verbal, Performance, and Full-scale IQ scores and subtest scaled scores).³⁴ General nonverbal problem solving and deductive reasoning were assessed with the Raven's Advanced Progressive Matrices–set I (scored by the number correct).³⁵ Reading achievement was assessed with the Wide Range Achievement Test–Revised (standard score).³⁶ The following specific neuropsychological tests and the scores obtained were also used: Cancellation test (time and organization scores for random verbal and nonverbal conditions)³⁷ ;Trail Making Test (time scores for parts A and B-A time scores); Continuous Performance Test (CPT) (mean reaction time, number correct, omissions, and number incorrect for AX and X-degraded visual conditions)³⁸ ;Wechsler Memory Scale–Revised (recall for immediate and delayed conditions of the Logical Memory and Visual Reproductions subtests)³⁹ ;California Verbal Learning Test (List A total trials 1-5, short-delay free and cued recall, long-delay free and cued recall, and recognition hits)⁴⁰ ; Rey-Osterrieth Complex Figure Test (accuracy and organization scores for copy, immediate and delayed recall conditions)^{41 - 44} ;Wisconsin Card Sorting Test (number of categories and perseverative errors)⁴⁵ ; Stroop test (color-word condition age-corrected T score)⁴⁶ ; Finger Tapping Test (mean number of taps across trials for dominant and nondominant hands)⁴⁷ ;and Grooved Pegboard Test (time for dominant and nondominant hands).⁴⁸

For the CPT, because hit rates (HR) and false alarm rates (FA) are biased by the perceived payoffs for correct identification and perceived penalties for incorrect identification, a nonbiased sensitivity score was also determined arithmetically from HR and FA using the following equation⁴⁹ :

Test scores were organized into 5 neuropsychological functions based on how the data would be organized and interpreted in a clinical neuropsychological examination.⁵⁰ The first function was general intelligence and achievement. To our knowledge, no previous study on stimulant abuse has included the complete Wechsler Adult Intelligence Scale–Revised; given the extensive research on this test, all of the subtests were included in the intelligence function. We also included the Raven's Advanced Progressive Matrices and Wide Range Achievement Test–Revised in this function; both of these tests are used as estimates of intelligence. The second function was executive skills, which consisted of organization scores from the Cancellation and Rey-Osterrieth Complex Figure tests, the Trail Making Test part B-A time score, and the Wisconsin Card Sorting and Stroop tests. These scores reflect organizational skills, mental flexibility and categorization, and the ability to inhibit selected responses. The third function was attention, which included time scores from the Cancellation test and Trail Making Test part A and all CPT scores. These scores reflect speed of visual scanning, sequencing, and visual vigilance. The fourth function is memory, which included accuracy scores from the Rey-Osterrieth Complex Figure Test and scores from the Wechsler Memory Scale–Revised and California Verbal Learning Test; these scores represent measures of explicit memory. The final function is motor skills, which included scores from the Finger Tapping and Grooved Pegboard tests of motor dexterity and speed.

The neuropsychological test battery was administered by research assistants with college degrees. They were trained in standard neuropsychological test administration by one of us who is a trained neuropsychologist (R.T.). The training involved didactic presentations, demonstration, practice, and direct observations of the administrator with subsequent feedback before they were approved for administration of tests to actual subjects. The research assistants also administered a structured interview to assess alcohol use and diagnostic criteria for alcohol dependence. They were trained in this interview by two of us (R.T. and M.J.L.). The training involved review of diagnostic criteria, practice, and direct observations of the interviewer with feedback. Ongoing quality control was accomplished through individual and group supervision and having the research assistants cross-check each other's work for scoring, judgment, and clerical errors.

We used unpaired t tests and χ² analyses to examine whether there was a participation bias in our recruitment of pairs. To examine group differences on demographics and available lifetime diagnoses, we used paired t tests for continuous variables and the McNemar χ² test for paired proportions. To control for multiple tests, we first examined differences on the 5 neuropsychological functions between abusing and nonabusing twin-pair members using a multivariate approach with a repeated-measures general linear model. This model used each pair as a case and the 2 members of the pair as the within-subjects factor. We followed this up with a univariate approach. First, we used the Kolmogorov-Smirnov test to determine whether the neuropsychological test scores were normally distributed. For those scores that were normally distributed, we compared groups using matched-pairs t tests. For those scores that were not normally distributed, we compared groups using Wilcoxin signed rank tests. Within abusers, we examined dose-response relationships between the number of days of drug use (not normally distributed) and performance on neuropsychological test scores with Spearman rank order correlation coefficients. Correlational analyses were first conducted with separate scores for cocaine and amphetamine use. We conducted Fisher z tests to determine whether significant differences existed between correlations with cocaine vs amphetamine use. Tests were 2-tailed and the significance level was set at P<.05.

At the time of the neuropsychological assessment, the subjects' mean age was 45.9 years (range, 40-51 years; SD, 2.9 years). The groups did not differ on mean length of education (abusers, 13.8 years [SD, 2.3 years]; nonabusers, 14.0 years [SD, 2.3 years]; P = .56) or indicators of learning problems in school such as being in a special school (abusers, 2%; nonabusers, 2%; P>.99), being in a special academic class (abusers, 10%; nonabusers, 10%; P>.99), or getting special help for academic problems (abusers, 16%; nonabusers, 14%; P>.99). In the abuser group, no subject ever received a diagnosis of a learning disability and 1 subject received a diagnosis of ADHD. In the nonabuser group, 1 subject received a diagnosis of a learning disability, and 1 subject, ADHD. The groups did not differ in ever having had a head injury (abusers, 42%; nonabusers, 44%; P>.99). These percentages are high because subjects reported even minor blows to the head. As an indicator of severity, only 2 abusers (4%) and 2 nonabusers (4%) had a loss of consciousness for longer than 5 minutes.

We used unpaired t tests and χ² analyses to assess whether our participants differed from those subjects who met criteria and were recruited but declined to participate. We compared the abusing twin from examined pairs (n = 50) with the abusing twin from nonexamined pairs(n = 63; the 63rd pair was discovered after recruitment procedures to have met criteria). We did not include twins who were examined but excluded from the study (n = 12). We found no participation bias for age, race, educational level, or minimum number of days of cocaine or amphetamine use reported in the 1992 survey.

The presence of other Axis I lifetime diagnoses did not differ between stimulant abusers and their co-twins (Table 1).

Given the high prevalence of alcohol dependence, we explored further the amount and frequency of alcohol intake in the 2 groups. We found no difference(P>.99) between the 2 groups in the number of current regular drinkers, ie, those who had at least 1 drink a month for at least 6 months (33 abusers and 32 nonabusers). We also found no difference in the mean number of alcoholic drinks consumed in the past 12 months (abusers, 235.2[SD, 402.9]; nonabusers, 295.4 [SD, 392.0]; n = 50; t =−0.95; P = .35), the mean duration of the worst period of drinking (abusers, 44.8 months [SD, 80.1 months]; nonabusers, 41.1 months [SD, 54.5 months]; n = 50; t = 0.30; P = .77), and the mean number of drinks consumed during the worst period of drinking (abusers, 11 317.6 [SD, 40 759.3]; nonabusers, 8490.5 [SD, 23 909.7]; n = 50; t =0.42; P = .68). We also examined recent tobacco use, given the high prevalence of tobacco dependence. We found no difference in the mean number of cigarettes smoked in the previous 12 months (abusers, 4085.1[SD, 5371.1]; nonabusers, 4014.4 [SD, 7685.0]; n = 50; z = −0.94; P = .35).

Although some individuals with a history of cannabis abuse remained in the sample, as stated previously, we excluded individuals who reported regular use of marijuana since the previous survey and use in the past year. The stimulant abuser who met criteria for dependence on sedatives in the 1992 survey reported that since then, he had used sedatives more than 5 times, but did not use them regularly. His most frequent use of sedatives was 2 days a week for 1 month, and he had last used sedatives during a period of 1 month to 1 year before our examination. The stimulant abuser who met criteria for opiate dependence in the 1992 survey denied use of any opiates since that time. We also recorded subjects' self-report of recent use of other illicit substances, which was minimal (Table 2).

Within the group of index twins, 10 subjects abused only cocaine; 31, only amphetamines; and 9, both drugs (Table 3). Abusers of both drugs had a higher mean number of days of cocaine and amphetamine use than did abusers of only cocaine or only amphetamines. The variable that summed days of cocaine or amphetamine use ranged from 52 to 18 980 days (mean, 1145.64 days [SD, 2876.88 days]), and was not normally distributed. Within the group of co-twin controls, one had 8 instances of cocaine use and another had 12 instances of cocaine use.

The General Linear Model tests were significant for attention (F_13,32 = 2.5; P = .02) and motor functioning(F_4,44 = 2.7; P = .045). The General Linear Model tests were not significant for executive functioning (F_9,37 =1.9; P = .09), memory (F_14,35 = 1.1; P = .36), or general intellectual functioning (F_16,31 = 0.7; P = .77). We provide the results of the univariate tests for all 5 functions for the reader's information, although only 2 functions were significant between groups (Table 4). In the attention function, abusers had significantly slower time scores on the nonverbal condition of the Cancellation test and the Trail Making Test part A. In contrast, abusers performed better on a number of variables on the X-degraded CPT. Abusers were significantly more sensitive and had more correct responses and fewer omissions. In the motor function, on a motor test with a time limit (Finger Tapping Test), abusers got significantly fewer taps with their nondominant hand than did nonabusers, whereas on another motor test without a time limit (Grooved Pegboard Test), abusers took significantly more time to complete the dexterity test with their dominant hand.

We found no significant differences between the correlations of test variables with days of cocaine and amphetamine use. We conducted Spearman correlation coefficients of test variables with the summary score of stimulant use. Only the following 3 of these correlations were significant: the time score of the Grooved Pegboard Test for the nondominant hand (r = −0.28; P = .046); the visual reproduction subtest of the Wechsler Memory Scale, immediate recall (r = 0.29; P = .04); and the delayed recall condition organization score of the Rey-Osterrieth Complex Figure Test (r = 0.38; P = .006). The direction of the correlations indicates that more days of stimulant use are associated with improved performance. None of the correlations remain significant when using the Bonferroni correction for the level of significance (P<.0009).

We examined neuropsychological functioning in 50 twin pairs discordant for abuse of cocaine and/or amphetamines. We found evidence of long-term residual deficits in abusers on attention and motor functions. Although they were more impaired on some attention measures, abusers showed higher sensitivity than nonabusers on a test of vigilance by hitting a greater number of targets and having fewer omissions. Although we observed a dose-response relationship between increasing drug use and improved performance on measures of motor speed, visual memory, and organization, we examined many correlations, and these significant relationships could be due to chance.

Unlike those in other studies, our abusers were not impaired relative to their co-twins on verbal memory^{1 - 2 ,4 ,14 ,18} and verbal skills.⁵¹ Our findings are consistent with previous studies documenting deficits associated with shorter periods of abstinence from cocaine abuse in attention functions^{1 - 2 ,4 ,11} and extend the findings to an abstinence period of 1 year. Our findings are inconsistent with those of Selby and Azrin,¹² who found no impairment in incarcerated former cocaine abusers. However, their exclusion criteria included certain head injuries and Axis I diagnoses, which were not excluded in our study. These variables did not differ between our groups and probably do not account for our findings. Our sample is more representative of the general population in being of average intelligence and not incarcerated. Perhaps residual deficits continue to remit during a period of 1 to 3 years of abstinence, the time periods for the 2 studies.

Five variables that differed significantly in the group comparisons were on timed tests. In all cases, abusers were slower than nonabusers. This finding is surprising, given that stimulants would be expected to enhance speeded performance, at least in the short term. Indeed, we found that more stimulant abuse was correlated with faster reaction times using the nondominant hand on the Grooved Pegboard Test. However, our results are consistent with those of other studies that found psychomotor slowing⁷ and slower reaction times^{9 - 10} in cocaine abusers. Despite the relative impairment on motor skills, mean scores of the abusers were still within normal limits.

Our study is consistent with previous studies that found certain neuropsychological strengths in abusers,^{2 - 3 ,5 - 7 ,51} but we found an advantage on a vigilance task that has not been previously reported. Since amphetamines (eg, ritalin hydrochloride) improve attention deficits, attention may be improved among stimulant abusers. The subjects may have been dishonest about their recency of stimulant use, and that recent use may have directly improved their attention. Alternatively, the positive effects of stimulant use on vigilance may persist beyond cessation of use. Our results show that certain cognitive deficits persist beyond cessation of use, so advantages could also persist. Our twin study research design is unique among these studies that found a neuropsychological advantage for cocaine abusers. Because twins are ideal controls, if the nonabusing twin has a deficit in vigilance, that would argue for a deficit in vigilance existing in the abusing twin before stimulant use. Our nonabusers correctly responded to 55% of targets, and our abusers responded to 66%. In comparison, Tatman⁵² reported that normal subjects aged 30 to 43 years correctly responded to 77% of targets, and those aged 48 to 59 years correctly responded to 81% of hits. If an attention deficit existed before stimulant use, the abusing twins may have used stimulants to self-medicate or to normalize this deficit. Unfortunately, we cannot address this question directly without an additional control group. The fact that few of our twins reported having a diagnosis of ADHD sheds doubt on the self-medication hypothesis; however, we did not formally diagnose ADHD. Other investigators have found ADHD rates of 15%²⁴ and 35%⁵³ in cocaine abusers. King et al⁵⁴ found that 19% of opioid abusers entering treatment had ADHD and that the CPT indicated poorer attention in abusers with ADHD compared with those without ADHD. Our finding of improved vigilance among stimulants abusers needs to be replicated, and these alternative speculations need to be examined empirically before any conclusions are drawn.

Our study addresses many of the methodological limitations of earlier studies, primarily inadequate control groups. The use of a twin sample provided the ideal control group, and the large number of twins who participated in the original survey enabled us to identify a reasonably sized sample of pairs discordant for stimulant abuse. Despite this level of control, our study still has certain limitations. First, drug data were retrospectively self-reported, and we did not obtain objective verification of abstinence. Second, all subjects were male, and thus, these findings may not generalize to female subjects. We did not have enough subjects to examine more thoroughly whether effects associated with cocaine vs amphetamine abuse are similar or different. Finally, because our sample contained some DZ twin pairs, we cannot rule out the possibility that the observed group differences may be partially due to genetic effects.

We compared 50 twin pairs discordant for stimulant abuse on a battery of neuropsychological tests. Despite being abstinent for at least a year, abusers demonstrated neuropsychological impairments and selected advantages. These finding provide evidence of long-term residual effects of stimulant abuse.

Corresponding author and reprints: Rosemary Toomey, PhD, Department of Psychiatry, Massachusetts General Hospital, Bldg 149-9121E, 13th St, Charlestown, MA 02129 (e-mail: rosemary_toomey@hms.harvard.edu).

Submitted for publication November 20, 2000; final revision received February 20, 2002; accepted March 19, 2002.

This study was supported by grant DA04604 from the National Institute on Drug Abuse, Bethesda, Md (Dr Tsuang), and by the Department of Veterans Affairs Health Services Research and Development Service and the Cooperative Studies Program (study 992), Washington, DC.

The US Department of Veterans Affairs has provided financial support for the development and maintenance of the Vietnam Era Twin (VET) Registry. Numerous organizations have provided invaluable assistance in the conduct of this study, including: Department of Defense; National Personnel Records Center, National Archives and Records Administration; the Internal Revenue Service; National Opinion Research Center; National Research Council, National Academy of Sciences; the Institute for Survey Research, Temple University. Most importantly, the authors gratefully acknowledge the continued cooperation and participation of the members of the VET Registry and their families. Without their contribution, this research would not have been possible.

We thank Irv Gottesman, PhD, of the University of Virginia, Charlottesville, and Jag Khalsa, PhD, of the National Institute on Drug Abuse, who contributed to the successful completion of this project. We also thank Tom Doyle for his help in various aspects of the project, Zachary Warren, and Lisa James for help in manuscript preparation.