Neurocognitive Characterizations of Russian Heroin Addicts without a Significant History of Other Drug Use

1.1 Neuropsychological Deficits in Heroin Addicts

Numerous studies have examined the functional correlates of heroin’s potential neurotoxic actions in the form of neuropsychological performance. However, because heroin is a “late-stage” drug in most countries (i.e., most drug users “graduate” to heroin use only after extensive use of other drugs and alcohol [Zinkernagel et al., 2001]), no studies have isolated specific correlates of heroin addiction (Rogers and Robbins, 2001), with one exception. Fields and Fullerton (1975) studied Vietnam veterans who reported being dependent primarily upon heroin (averaging 4.9 yrs) with no history of excessive use of other drugs or alcohol. Using the Halstead Battery, the veterans performed as well as the control groups and unlike brain-damaged patients, with the exception of the abstract reasoning subtest on which heroin addicts performed less well than the controls. This study is relevant given that these subjects tended to be single drug users and were less likely to have an existing predisposition to drug abuse that might play a role in another environment or that may have predicted neurocognitive impairment (Robins et al., 1980). The latter scenario is often reported for those with a family history of alcoholism which significantly predicts cognitive dysfunction, as well as earlier age of drug use onset, for heroin addicts (Rodriguez, 1994).

Early studies were discrepant in their findings, with some reporting no remarkable differences in neuropsychological measures between heroin addicts and comparison groups, even in the context of long-term use of high doses of heroin (Brown and Partington, 1942; Strang and Gurling, 1989) while others found significant perceptual-motor, memory, attention and concentration, and intelligence deficits relative to controls (Ahmad et al., 1989; Wolf et al., 1979; Korin, 1974; Rounsaville et al., 1981). Due to the timing of these studies, however, more traditional and less sophisticated tests of neuropsychological function were used, comparison groups were not always appropriate for this population (e.g., brain damaged patients), and concomitant use of other drugs and alcohol were not accounted for. More recent studies have utilized test batteries with greater sensitivity to dimensions of ECFs and greater specificity in terms of the neuroanatomical loci associated with these functions.

1.2 Aims of the Current Study

In order to elucidate the neurocognitive characteristics of heroin addicts that may exist independently of alcohol and other drugs, it is necessary to examine a population with a history of single drug dependence. Thus, the present study was conducted in St. Petersburg, Russia where a unique pattern of drug use exists. Heroin addicts tend to have tried alcohol early in adolescence and then begin experimenting, rapidly developing a preference for heroin and discontinuing all other drug use (typically alcohol, marijuana and ephedrine). Another much larger group of Russians who become alcoholics tend to have initiated alcohol in their early teenage years and persist in using only alcohol throughout their addiction career. They do not use other drugs. A third group is comprised of heroin addicts who enter treatment and eventually abstain from heroin only to develop alcoholism. The inclusion of this group enables a determination of whether there is an additive “effect” on cognitive function when both heroin and alcohol are abused, as demonstrated in cocaine and cocaine+alcohol abusers (Bolla et al., 2000). Thus, the present study included these groups with the addition a demographically similar control group without a history of any drug or alcohol abuse or addiction.

A battery of sensitive and specific instruments was assembled to measure dimensions of ECF previously tested in heroin addicts with polysubstance abuse to begin to isolate correlates of heroin addiction independent of other drug use. Task selections also considered brain regions either most highly concentrated with heroin or that were most active after acute administration (e.g., prefrontal cortex, anterior cingulate, hippocampus, nucleus accumbens, and ventral tegmental area [(Chang et al., 1997; Galynker et al., 2000; Hakan et al., 1989; Kiyatkin, 1994; Koob et al., 1986; Tanda et al., 1997; Wise, 1989]), as well as studies cited above using tasks with some neuroanatomical specificity to provide direction with respect to identifying the functional correlates of heroin addiction. Thus, response shifting, strategy formation and execution, risky decision making, error monitoring, visual and spatial memory, and complex visual pattern recognition, as well as basic motor acuity and general intelligence, were measured. Although these functions are thought to be affected by psychoactive drugs, the present study sought to determine whether they are differentially or disproportionately affected in the heroin addicts relative to the other study groups.

Hypotheses were threefold. First, based on previous studies including normal control comparisons, it was expected that the three drug groups would exhibit impairments on all neurocognitive measures relative to the control group. Support for this hypothesis would provide initial evidence for the relationship between heroin addiction and neurocognitive function independent of other drugs. It was further anticipated that deficits in heroin addicts relative to the other two drug groups would be specific to the prefrontal cortex. Thus, heroin addicts were expected to exhibit greater disadvantageous decision making and longer deliberation times during decision making (see Rogers et al., 1999a) and less efficient response shifting (e.g., Madden et al., 1997), related to studies of affective impulsivity, than the alcohol group. Also, we hypothesized that heroin addicts would show deficits in processing time, episodic memory/new learning, visual pattern recognition, problem solving, and cognitive flexibility that are less severe than those of alcoholics but that may have been masked by significant historical alcohol and polydrug use in previous studies. And finally, we expected the alcohol group to show significantly poorer performance on measures of motor acuity than the heroin group based on the literature showing generalized neurotoxic effects of alcohol. Thus, this study constituted a first attempt to isolate neurocognitive deficits that may be uniquely found in ‘pure’ heroin addicts relative to controls and alcoholics without the confound of a history of multiple drugs of abuse.

2.1 Participants

One hundred heroin addicts, 102 alcohol dependent patients and 60 heroin + alcohol dependent patients at the LRCA were tested; 160 control subjects were tested at Pavlov Medical University. Subjects were between the ages of 18 and 40. The LRCA is a 280-bed hospital for the treatment of alcoholics and drug addicts, 80% to 90% were occupied by alcoholics and 10% to 20% by heroin addicts. Typically, alcoholics and drug addicts remain in the hospital for one month for treatment of withdrawal and detoxification (approximately 7–10 days), any related psychiatric and somatic problems, and inpatient stabilization and initial rehabilitation. Substance abuse treatment included individual and group cognitive and behavioral relapse prevention therapy. Most alcoholic patients are assigned employment within the vicinity of the LRCA and work Monday through Friday. They leave the LRCA in the morning and must return by 5 PM each night, receiving a breathalyzer test upon their return. Heroin + alcohol dependent patients work inside the LRCA during the day. Both groups typically receive counseling sessions in the evenings. All patients from four departments of the LRCA (out of 7) were screened for eligibility. They were provided with the information about the study and only those who were interested and agreed to participate were recruited. All patients signed informed consent forms.

For this study, patients were assessed after three weeks of detoxification and stabilization, during the last week of their stay before discharge when they had been weaned from medications (only medications for somatic disorders had not been discontinued). Patients were excluded if LRCA evaluations indicated Axis I psychiatric disorders, severe head injury, or other sources of serious brain damage to avoid the significant confound introduced by the interaction between cognitive deficits associated with psychiatric disorders and severe head injury. Patients with Axis II disorders were not excluded, however, as this would reduce the subject pool significantly, limiting the feasibility of recruiting a sufficient number of patients. Also, given that AIDS has adverse effects on neurocognitive function, those known to be infected (as per LRCA records) were excluded (Everall et al., 1999).

Control subjects (N=160) without a history of significant drug or alcohol use were recruited from the community surrounding Pavlov Medical University and included factory workers, college students, and the unemployed. Flyers describing the study were distributed to staff and various employment sites and exhibited throughout this community; individuals either called or visited the University to express their interest. Because of the known age differences between patient populations of alcoholics and heroin addicts in Russia, control subjects were selected to match both the alcoholic and heroin addict groups according to age and other demographic characteristics and then combined for analyses. Thus, controls had a similar range of background characteristics to the addict groups in age, education (higher education after high school is generally received by all citizens in Russia), neighborhood of residence, and overall demographic mix. Control subjects received the same screening and test batteries administered to patients as described below and were subjected to the same exclusionary criteria.

2.2 Screening Procedures

Two levels of screening all subjects prior to participation were performed by Russian study staff. First, for patients, a record check of the LRCA’s current files containing assessments conducted upon intake for present hospitalization indicated whether the patient was eligible in terms of current psychiatric status (no dual Axis I diagnoses) using the ICD-X, drug abuse history, indications of HIV status (negative, if available), age and head injury (none severe). The Raven’s Colored Progressive Matrices (RCPM, Raven et al., 1938, 1996) were also used to exclude those with mental retardation (<80 estimated IQ) and to estimate general intelligence for all participants. This test was chosen due to its simplicity, nonverbal format, and culture-fairness. Importantly, this task is often used for patients with difficulty understanding English. The RCPM is considered to be a reasonable measure of general intellectual ability and is adaptable to administration in non-English-speaking individuals.

Second, patients were administered two questionnaires to evaluate drug use histories. A questionnaire coined the “DATASI” was developed by combining relevant items from the Addiction Severity Inventory (ASI: McLellan et al., 1992), the Drug Abuse Treatment Outcome Study (DATOS: Hubbard et al., 2003) and additional study-specific questions (available scoring routines for these tests were not used). The Time Line Follow-Back (TLFB) instrument extensively probed about their prior drug use by assessing alcohol (in standard drinks or grams of pure ethanol) and substance use over a specified period of time (Sobell and Sobell, 1996). Drug use histories in the larger patient pool were then evaluated to screen out (i) heroin addicts with histories of significant alcohol or other drug consumption, (ii) alcoholics with significant use of any other drug, and (iii) controls with excessive use of alcohol (including binge drinking) or other illicit drugs. Drug use exclusion criteria for each group, based on information gleaned from these instruments, is as follows:

• Heroin addicts who have used alcohol extensively in their lifetime (>100 grams or 3.5 ounces per week on average and/or >100 grams in a single occasion more than once a month).

• Alcoholics who have used other illicit drugs more than 10 times during their lifetime.

• Heroin+Alcohol group excludes those who have used other illicit drugs more than 10 times during their lifetime (e.g., ephredrine, marijuana).

• Control participants cannot exceed alcohol intake of >100 grams pure ethanol per week (3.5 ounces) on average and/or >100 grams pure ethanol in a single occasion more than once a month and/or use of illicit drugs more than 10 times during their lifetime.

Subsequently, patients who met eligibility criteria were recruited and were informed of all aspects of the study prior to obtaining written consent.

2.3 Psychiatric, Psychological, and Contextual Measures

Several psychological and psychiatric instruments were included both for screening purposes and to provide characterizations of groups for further elucidation of differences. These tests are often conducted by LRCA at intake for various study purposes, and permission was granted by consenting patients to the US study team to confidentially review their files. Axis I disorders were excluded by the structured interview for ICD-X, an international analog of DSM-IV. The Brief Psychiatric Rating Scale (BPRS: Ventura et al., 1993) and the Cattell 16 Personality Factor Assessment was administered to the substance dependent groups and controls. Each participant’s general psychiatric functioning was assessed by a psychologist or psychiatrist using the DSM-IV Global Assessment of Functioning scale (GAF: DSM-IV-TR, 2000). Specific contextual factors extracted from the Drug Abuse Treatment Outcome Study (DATOS, Flynn et al., 1997) interview and incorporated into the DATASI were assessed, including demographic and other background information including health characteristics, education, employment, and mental and physical health status.

Immediately prior to neurocognitive testing, a motor screening test (MOT) was administered to familiarize participants with the use of a computerized task using a touch screen, and to identify any potential differences between groups in general motor acuity. Latency in this task is the primary variable and is defined as the time taken for the participant to touch a cross after it appeared in various locations on the monitor. The arithmetic mean was calculated from the latencies of the ten crosses presented to which the subject correctly responded. Latency was measured in milliseconds.

2.4 Neurocognitive Testing

Three neurocognitive tasks from the CANTAB (CeNeS PLC, Cambridge, UK; Robbins et al., 1994, 1998) and three programmed using E-Prime were chosen to measure frontal cortically-modulated functions that are conceptually consistent with the phenomenology of alcohol and other drug abuse. The CANTAB test battery (2004) included measures of visual recognition memory, pattern recognition, problem solving and strategic planning and the E-Prime battery included the Stroop Color/Word Interference, Cambridge Decision Making (CDMT), and the Stop Change Tasks (SCT) (described below). All tasks required behavioral, not verbal, responses to avoid culture and language bias and careful instructions were translated into Cyrillic and carefully recited and explained to participants prior to task initiation.

3.1 Neurocognitive Assessments

The data are presented as the group means and standard errors. We also show the effect sizes that are significant without Tukey multiple comparison adjustment and bolded effect sizes remain significant after the adjustment (Table 3). Because some tests involve more than one variable we have highlighted those that also remain significant after additional Bonferroni adjustment for multiple tests. Most of the significant effects are of medium size according to Cohen (1992); i.e., from about 0.5 to 0.8, although for SOC we observe significant effect sizes around 1.

4.1 Interpretations of Specific Task Performance Deficits

All three drug groups showed deficits in general intellectual ability, as measured by the RCPM, relative to control subjects. Of the drug groups, the heroin addicts showed the highest level of performance while the alcoholics performed substantially worse. Alcoholics’ scores were not, however, statistically different from the heroin+alcohol group. Age was a factor in level of intellectual functioning with the alcoholics being significantly older than the other drug groups, however including age as a covariate did not alter these results. The fact that, although alcoholics and the heroin+alcohol group did not significantly differ in intelligence, the alcoholics used and were dependent on alcohol for a greater number of years suggesting that the drug of choice may be more influential in our results rather than the duration of use. Given overall group differences, the RCPM was initially included in the model and not shown to influence the results in any way, suggesting that higher order cognitive function (ECF) remains significantly different between the groups irrespective of deficits in subservient intellectual functions.

Similarly, on the measure of general motor ability, each drug group performed worse than the controls, however, alcoholics exhibited the lengthiest response times which is reflective of motor incoordination. Heroin addicts and the heroin+alcohol group were not different in response times. Psychomotor slowing and motor retardation have been attributed to alcoholism (e.g., Ciesielski et al., 1995) but no studies of motor disturbances in heroin addiction, other than simple acute administration effects, could be located. This study suggests that heroin addiction is also associated with motor sluggishness.

All drug groups differed significantly from controls, and for latencies to each other, on our measure of perceptual matching, immediate and delayed visual memory (DMS) and in ability to form visuo-spatial associations (PAL). These related functions recruit structures within the medial-temporal lobe such as the hippocampus, with input from the dorsolateral prefrontal cortex (DLPFC)(Lamar et al., 2004; Monk et al., 2002; Schon et al., 2004; Tanabe et al., 2005), suggesting that our drug groups have some dysfunction in these interconnected regions. Of note, however, performance of these groups on the basic measure of pattern recognition (DMS) was poorer than the controls, suggesting that more basic deficits exist in functions that are otherwise supportive of working memory. These basic deficits do not necessarily extend to impairments in motor acuity given that adjustments for this measure did not alter the results for either the DMS or PAL.

Nuances in performance on these two tasks provide some interesting contrasts between heroin addicts and the other two drug groups. With respect to delayed visual memory, heroin addicts exhibited the shortest latencies in both trial types and, at the same time, had fewer correct responses. Given that there were no significant differences in correct responses between drug groups and that heroin addicts responded more quickly and less accurately than did the controls, one interpretation is that heroin addicts may tend toward impaired motor inhibitory responding that impedes performance on a working memory task. Longer latencies in the alcoholics, in contrast, suggest less cognitive efficiency. As discussed below, however, impairments found in the heroin addicts are unlikely to be solely attributable to inhibitory deficits given their superior performance on some measures of the Stop Change Task. Unfortunately, attentional mechanisms were not evaluated in this study which may have provided an indication of whether the inability to recognize patterns simultaneously presented was related to a lack of focus. For the central measure of visuospatial learning (“stages completed”), on the other hand, heroin addicts showed a tendency to perform better than alcoholics who completed fewer stages than the other groups. Taken together, the poor performance of these drug groups on both tasks supports previous findings of deficits in learning and memory that may be a function of damage from neurotoxicity to the hippocampal formation in the temporal lobe (Day et al., 2003) which is structurally altered by drug addiction (Nestler, 2002; Robbins and Everitt, 2002).

With respect to the “strength of the probability of an error following an error” variable, all four groups responded with a predominance of random errors not related to performance on previous trials, suggesting that error-proneness characterizes each of our groups. Controls responded with the most random errors followed closely by the heroin addicts and lastly the alcoholics and heroin+alcohol group. No group produced positive numbers, however, which would reflect a tendency to experience interference from the emotional reaction to making an error, leading to additional errors. In light of findings that controls performed with less response shifting on the SCT and also made the greatest number of random errors on the DMS, it appears that disinhibitory tendencies on the part of controls may be a result specifically of reflection impulsivity (see Clark et al., 2006).

All three drug groups showed further impairments relative to controls on the test of frontally-modulated (particularly DLPFC but also anterior cingulate) problem solving abilities (SOC). Heroin addicts exhibited better performance than the heroin+alcohol group in that they required significantly fewer moves. Alcoholics did not perform better or worse than the other two drug groups, but showed deficits relative to controls. These findings are consistent with evidence for prefrontal dysfunction in alcoholics and drug addicts (Ersche et al., 2005; Paulus et al., 2002), particularly within the DLPFC shown to modulate problem solving and spatial planning on similar tasks (Baker et al., 1996; Manes et al., 2002), and correspond well with performance decrements found in learning and memory (e.g., DMS and PAL) in the present study. These skills (as measured by stages completed) in heroin addicts, on the other hand, appear to be somewhat less dysfunctional than in the other drug groups.

Relative to our original hypotheses, perhaps the most noteworthy group difference was that heroin addicts showed a distinctive pattern of performance on the CDMT, a measure of affective impulsivity in decision making. In the present study, heroin addicts selected a significantly greater percentage of disadvantageous or risky options and exhibited the lengthiest reaction times during high risk choices relative to controls. The other two drug groups selected a greater number of risky choices than the control group, although differences had a small effect size. Control subjects showed the shortest RTs, a finding consistent with Rogers et al. (1999a). Intuitively, one would surmise that the longer the reaction time, the greater the time would be for advantageous deliberation of choices and consequences. But instead, heroin addicts who deliberated longer made riskier decisions and exhibited worse behavioral outcomes. Disadvantageous decision making in conjunction with lengthier reaction times may therefore be indicative of both the difficulty in and quality of decision making under these circumstances that may alter processing of salient stimuli.

Of relevance, two studies reported significant decrements on the CDMT in amphetamine users and MMT patients relative to opiate users (Ersche et al. 2005; Rogers et al. 1999a) while the present study found greater impairment in heroin addicts relative to alcoholics. Together, these findings suggest that decision making deficits may be a function of drug type and not recency of drug use. This notion that specific drugs of abuse may confer differential cognitive effects partially supports our findings of decision making performance deficits in heroin addicts greater than alcoholics; excessive alcohol intake appeared not to relate to risky decision making as did heroin abuse. This finding may be a function of either neurotoxic effects or predisposition characteristics that may differ between alcoholics and heroin addicts. In a culture where drinking is commonplace, underlying mechanisms of propensity to alcoholism may differ in their significance relative to those factors (e.g., genetics) that are strongly influential in the US.

Tendencies among these heroin addicts toward poor decision making ability may reflect an affective impulsivity that specifically relates to reward salience. Support for this notion can be found in studies by Kirby and Petry (2004) and Kirby et al. (1999) who reported that heroin (and cocaine) addicts showed significantly greater discounting of the value of a future reward as the delay to that reward increased but alcoholics did not. As performance on the CDMT and similar reward-based tasks has been shown to activate the OFC, an area associated with the coordination between processing of emotional and cognitive stimuli (Bechara et al. 1997; Ernst et al., 2002; London et al. 2000; Rogers et al. 1999a), deficits observed in these heroin addicts may be a consequence of OFC damage either predating or resulting from chronic drug use. Further investigation of the nature and extent of OFC involvement in these deficits among heroin addicts requires more comprehensive and longitudinal assessment of risky decision-making.

With respect to measures of motor inhibition and response shifting generated by the Stop Change Task, the data produced mixed results. As expected, controls showed superior overall performance on the summary measure of this task; however, they were significantly different only relative to the heroin+alcohol group. Also, the heroin+alcohol group had greater difficulty than the other two drug groups on the moderately easy delay trial. The trend observed for this dual drug use group to respond poorly to both non-delay and delay trials on all measures relative to other drug groups, and in 4 out of 5 measures relative to controls, suggests a possible additive association between using heroin plus alcohol and response inhibition/flexibility.

In marked contrast to our original hypotheses, controls displayed less accuracy on the SCT than the drug groups, especially the alcoholics who showed the highest level of performance on the hardest delay trial. Given that controls performed better when all trials, non-delayed and delayed, were considered, it is possible that these particular subjects were more error-prone specifically for trials that involved response shifting and not for those simply requiring motor inhibition. Although not many studies of cognitive function in heroin addicts and no studies of those who have used almost exclusively heroin have been conducted, one study of heroin addicts with a history of polydrug use reported less impulsivity relative to users of other drugs not attributable to acute drug effects (Bornovalova et al., 2005). In contrast, the majority of previous studies find support for impulsivity in heroin abusers when compared with normal controls but they did not address their comparability with other drug use groups and distinguished between various types of impulsivity (Kirby et al., 1999, affective/reward salience; Lee et al., 2005, behavioral/motor; Pau et al., 2002, cognitive). This finding is also in apparent contrast with the primary DMS results in the present study showing increased errors and shorter response times in heroin addicts which appear to be related to disinhibitory tendencies. The SCT is a difficult and often frustrating task. It is possible that the inpatient setting, with addicts who have intrinsic motivation to show behavioral improvement, was more conducive of concentrated play.

Finally, heroin addicts and alcoholics demonstrated less cognitive flexibility and conflict monitoring on the Stroop Interference Task than the control group while the heroin+alcohol group did not significantly differ; alcoholics performed particularly poorly and effect sizes suggest that the decrements in heroin addicts may not be functionally significant. Also, these results may be influenced by impairments found in basic motor function; however, significant differences between control subjects and the alcoholics, who had the poorest motor acuity, remained intact after adjustments were introduced. Thus, although motoric sluggishness may play a role in these Stroop findings as well as in the central measure of motor responsivity, it likely does not provide a full explanation for difficulties the heroin addicts exhibited in shifting attention to salient stimuli and inhibiting an initial tendency to respond.

The overall pattern of findings in this study pertaining to inhibitory tendencies suggests that these heroin addicts are prone to affective impulsivity as measured by the CDMT while behavioral or motor responsiveness were variable depending on the task demands; i.e., in heroin addicts, relative deficits were found in pattern recognition that requires attentional focus (DMS) and ability to suppress irrelevant information and inhibit an initial response (Stroop) while response shifting in response to new information (SCT) was relatively in tact. It appears from our findings and those of additional studies discrepant from one another that further subtyping of the concept impulsivity, beyond motor and affective, may be warranted. Higher order cognitive abilities, and not simply motor control, are required for performance on all three tasks that may be more centrally responsible for these group differences.

4.2 Final Remarks

In comparison with other drugs of abuse, there has been substantially less research into neuropsychological deficits in heroin addicts and existing studies are somewhat discrepant in their findings (Rogers and Robbins, 2001), in part due to variable comparison groups, from healthy controls to amphetamine abusers to MMT patients, and in part due to the focus on different functions. In most cases, the literature suggests that there is improvement in cognitive function over time with abstinence (Guerra et al., 1987), however, some reports suggest that this appears to be less true for heroin addicts (e.g., Papageorgiou et al., 2004). The present study, using a broad battery of neurocognitive tests including those considered to be sensitive measures of frontal lobe functioning as well as subservient functions such as intellectual and motor abilities, has shown that in most cases, heroin addicts are similar to alcoholics in their presentation of several marked deficits relative to controls. Thus, common mechanisms may be at play. A strong family history of alcoholism and drug abuse in all three drug groups, for example, is one reasonable explanation for these profound decrements in neurocognitive functioning. Alcoholics were also older, although the average age of these groups was lower than subjects generally included in US studies and controls for age did not alter the results. Interestingly, impairments were remarkable despite the relatively short duration of dependency on drugs and/or alcohol. Alcohol and heroin are similar with respect to several of their mechanisms of action and neurotoxic effects, including alterations in grey and white matter, structure and function of the hippocampus, amygdala, nucleus accumbens, anterior cingulate cortex, and OFC among others, and perturbations in neurotransmitter (particularly dopamine) responses and metabolism throughout the mesocorticolimbic system (Alexander-Kaufman et al., 2006; Daglish and Nutt, 2003; Koob and Ahmed, 2004; Schlaepfer et al., 2006; Schmidt et al., 2005). Thus, our results showing more similarities than differences between drug groups are perhaps not surprising.

On the other hand, deficits observed in alcoholics were more pronounced than in either drug group. Drug-specific explanations include the severity of alcohol intake over time from a relatively young age, the even higher percentage of alcoholics with first degree relatives with alcoholism relative to the other groups, and the possibility that recovery of function after abstinence from alcohol may be slower than for heroin. The one exception to the severity of impairments found in the alcoholics is our finding of greater risky decision making and reduced deliberation times in the heroin addicts relative to other groups. A few very recent studies were located that found differences in tasks that probe OFC activity in alcoholics (Fein et al., 2004; Mazas et al., 2000), and one study (Dao-Castellana et al., 1998) reported alterations in the function of other PFC structures (e.g., DLPFC) but specifically not OFC in alcoholics. It is important to note that most studies of addicts suggest that their drug use “affects” cognitive function even though baseline measures are not available. Such interpretations are, thus, inappropriate given that some level of impairment may antedate and perhaps contribute to the development of drug abuse. In the absence of longitudinal studies it is impossible to discern whether these differential deficits predate or result from use, particularly in light of evidence that ECF dysfunction is more prevalent in adolescents who are at high risk for substance abuse (Fishbein et al., 2006b; Tarter et al., 2003). This restriction on interpretation also applies to the present study. Nevertheless, the neurotoxic effects of heroin are likely to have a more direct relationship with the lifetime usage patterns and thus drug groups are likely to be distinguishable by level of neurocognitive functioning.

One limitation of this study is that patients were tested at three weeks post-detoxification, limiting our ability to generalize to longer periods of abstinence. There are conflicting findings about the duration of cognitive deficits and some functions may improve over time. On the other hand, a recent PET study conducted by the first author found persistent deficits in drug abusers after over 3 months of abstinence primarily in complex executive functions, but not simple abilities such as those measured by continuous performance tasks (Fishbein et al., 2005). It will also be instructive to assess the moderating influence of personality and psychiatric disorders as well as the potential for distinctive cultural influences on cognitive dysfunction among heroin addicts (Pau et al., 2002). Future studies may be particularly informed by the inclusion of Antisocial Personality Disorder (ASPD) which is prevalent in substance abusers and has been repeatedly associated with ECF deficits (Stevens et al. 2001). The possibility exists that inclusion of this construct may alter the results if group differences in ASPD are present. In conclusion, with replications and further elucidation of these results, the design of more effective treatments and perhaps even preventive interventions for heroin addiction may benefit from a focus on decision making abilities and affective impulsivity.