P300 Abnormalities in Cocaine, Methamphetamine, Heroin Addiction, and Alcoholism
The P300 component of the ERP, occurring 300–600 ms post-stimulus, is the most widely used ERP in psychiatry and other clinical applications (Polich et al. 1994; Polich and Herbst 2000; Pritchard 1981, 1986; Pritchard et al. 2004). The amplitude of the P300 reflects the allocation of attentional resources, while the latency is considered to reflect stimulus evaluation and classification time (Katayama and Polich 1998; Polich and Herbst 2000). The P300 is usually obtained in an oddball paradigm, wherein two stimuli are presented in a random order, one of them frequent (standard) and another one rare (target) (Polich 1990). A modification of the oddball task has been used where a third, also rare stimulus (distracter), is presented along with standard and target stimuli. It was reported that these infrequent distracters elicit a frontocentral P300, so called P3a, whereas the rare targets elicit a parietal P300, so called P3b (Katayama and Polich 1996, 1998). The P3a is recorded at the anterior scalp locations and has been interpreted as reflecting frontal lobe activity (Gaeta et al. 2003; Knight 1984). Though the P300 response in general is thought to represent ‘‘context updating/closure,’’ in a three-stimuli oddball task the P3a is interpreted as ‘‘orienting,’’ and the P3b is viewed as an index of the ability to maintain sustained attention to target (Na¨a¨ta¨nen 1990). The anterior P3a indexes the contextual salience of the rare stimuli, whereas the posterior P3b is indexing task-relevance of the stimuli (Gaeta et al. 2003).
A robust finding in ERP studies on alcoholism is that alcoholics as well as individuals at high risk to develop alcoholism have been shown to have a low P300 amplitude in various task paradigms (Cohen et al. 2002; Hada et al. 2000; Porjesz et al. 2005; Porjesz and Begleiter 1998). Kouri et al. (1996) examined the P300 component in patients who were dually dependent on cocaine and heroin. The results showed no P300 amplitude differences between the patients and healthy non-drug-dependent volunteers when patients presented for detoxification. However, after the course of detoxification, the P300 amplitude was significantly smaller in the cocaine-and heroin-dependent group than in the non-dependent control group. In a study by Bauer (2001b) the P300 did not differentiate among patients characterized by histories of either cocaine, or cocaine and alcohol, or heroin dependence. Across all the patient groups, the P300 was significantly reduced in amplitude relative to the P300 ERPs recorded from individuals with no history of alcohol or drug dependence. This study also demonstrated that continued abstinence from heroin and from cocaine and alcohol is also associated with a trend toward normalization of the P300. In a recent study of Papageorgiou et al. (2004) the P300 component was evaluated during the anticipatory period of a short memory task in 20 patients characterized by a past history of heroin dependence (6 months abstinence), in 18 current heroin users and in 20 matched healthy subjects. Abstinent heroin addicts exhibited a significant reduction of the P300 amplitude at the central frontal region, relative to the other two groups.
The results of early work examining the effect of cannabis use and THC administration on visual and auditory ERPs have been inconclusive (Rodin et al. 1970; Roth et al. 1973). Later studies of Patrick et al. (1995, 1997) could not find P300 latency differences in audio and visual oddball tasks between THC users without psychiatric problems and controls. Although THC users displayed reduced auditory and visual P300 amplitudes in this study, when age differences between THC users and controls were removed, all significant P300 amplitude differences were removed as well.
Acute and chronic use of cocaine exerts neuropharmacological effects on amplitude and latency of both anterior and posterior P300 ERP components (Biggins et al. 1997; Fein et al. 1996; Herning et al. 1994a; Kouri et al. 1996; Polich 1990). Longer P300 (P3b) latency without abnormalities in amplitude was reported in several studies on cocaine withdrawal (Herning et al. 1994a; Lukas 1993). Noldy and Carlen (1997) demonstrated effects of cocaine withdrawal on the latency of the P300 in an auditory oddball task. In cocaine-dependent patients, P3a amplitude decrements over frontal areas are persistent even after long periods of abstinence (Bauer 1997). The latency of the P3a was delayed and the amplitude was reduced to novel non-targets in cocaine and alcohol-dependent subjects compared to controls (Biggins et al. 1997; Hada et al. 2000)in auditory and visual three-stimuli oddball tasks.
Several studies have investigated ERP changes associated with methamphetamine abuse and dependence. The P300 component of the auditory ERP was reported to show a prolonged latency in the oddball task in methamphetamine dependent subjects with a history of psychosis, compared to normal controls (Iwanami et al. 1994, 1998). In particular, the patients with methamphetamine dependence showed reduced P3a amplitude in the reading task and delayed P3b latency with normal P3b amplitude in the auditory oddball task. This was interpreted as indicating a prolonged central noradrenergic dysfunction due to earlier methamphetamine use.
In most ERP studies the P300 did not differentiate among patients characterized by histories of either cocaine, or cocaine and alcohol, or heroin dependence. Across all the patient groups, the P300 was significantly reduced in amplitude relative to P300 ERPs recorded from individuals with no history of alcohol or drug dependence. The latency of the frontal and parietal P300 was reported to be delayed, and the amplitude was reduced to novel non-targets in cocaine and alcohol-dependent subjects compared to controls in auditory and visual three-stimuli oddball tasks. Continued abstinence from heroin, cocaine, and alcohol was shown to be associated with a trend toward P300 normalization. Several studies have investigated ERP changes associated with methamphetamine abuse and dependence. In general, chronic psychoactive substance abuse and drug dependence are associated with delayed and attenuated cognitive ERP in auditory and visual oddball tasks.
qEEG and ERP Abnormalities in Addiction: Psychopharmacological Effects or Trait Markers?
Whether qEEG alterations and P300 decrements found in most of SUD are only a coincident ‘‘marker’’ of vulnerability or make a direct etiologic contribution to risk for substance dependence is still unknown (Bauer and Hesselbrok 2001; Carlson et al. 2002; O’Connor et al. 1994; Polich et al. 1994; Porjesz and Begleiter 1998). The P300 reduction and abnormal qEEG patterns are seen in mental disorders that often are comorbid with substance abuse, such as conduct disorder (Bauer and Hesselbrock 1999, 2001), ADHD (Bauer 1997; O’Connor et al. 1994), and bipolar or major affective disorder (Friedman and Squires-Wheeler 1994). Reduced P300 amplitude related to prefrontal brain dysfunction may suggest that a deficit in inhibitory control is an underlying mechanism shared by different psychopathologies (Bauer and Hesselbrock 1999; Clark et al. 1999; Tarter et al. 2003). According to Bauer (2002), certain ERP and qEEG abnormalities and impaired functioning on complex cognitive tests in patients formerly dependent on cocaine might not be proximately caused by drug use per se but be more related to comorbid alcohol use or another psychiatric condition. Taken together, the findings converge on the conclusion that there exists an inherited predisposition for an externalizing psychopathology that includes ADHD, conduct disorder, and substance abuse. PTSD seems to heighten the risk for addiction as well. Thus, the reviewed findings support the hypothesis that addicted subjects may manifest a P300 amplitude reduction and qEEG abnormalities as a trait reflecting the CNS disinhibition, which may be a predisposing factor for addiction liability, resistance to drug habit extinction, and relapse vulnerability.
Heritability and Neurotransmitter Considerations in Substance Use Disorders
There has been a consistent drift in addiction research between the psychosocial, cognitive and behavioral aspects of addiction and the biological and genetic emphasis. In much of the present data relating to genetics and animal models (Blum et al. 2006; Porjesz et al. 2005; Ryabinin and Weitemier 2006; Samochowiec et al. 2006), studies suggest that a genetic predisposition for SUD is an accepted concept. Much of the genetic research addresses the influence of alleles thought responsible in coding for genes that express phenotypic neurotransmitter production and distribution; mainly involving endorphins, dopamine and serotonin. These neurotransmitters, dopamine in particular, are also suspect in other appetitive and mood disorders and psychopathologies, of particular note, Reward Deprivation Syndrome (RDS). RDS is described as a dysfunction in the Brain Reward Cascade and proposes that abnormal craving behavior is a consequence of defects in the DRD2 and D1, D3, D4 and D5 dopaminergic receptor genes (Blum et al. 2006).
Blum and colleagues (1990, 1993, 1996) described this syndrome and identified the D2 dopamine receptor gene as a possible candidate for susceptibility to alcoholism in severe alcoholics (Blum et al. 1993) and proposed this gene’s association with dopamine production and distribution may produce a sevenfold increase in the likelihood of developing alcohol use problems (Uhl et al. 1993). This DRD2 dopamine receptor gene and polymorphisms within its genetic coding specific to addiction remain unclear due to its involvement in other disorders; including, obesity (Blum et al. 2006), Tourette’s syndrome (Comings et al. 1991) pathological aggression and violence, PTSD (Comings et al. 1996) and schizoid—avoidant disorder (Chen et al. 2005). SUD were classified as a subtype of RDS and treatment regimens for these disorders have been classified as inadequate (Blum et al. 2007) and research continues in developing possible genetic interventions that may produce dopamine and other neurotransmitter regulation in substance-induced rapid dopamine increase in limbic regions (Blum et al. 2007).
It is clear that heritability plays an important role in addictive disorders, however, to what extent environment, perception and synaptic permanency and plasticity influence the course of genetic adaptation or maladaptive traits requires further investigation. Suggested neuroanatomical substrates involved in SUD implicate mesolimbic and diencephalon regions; including the substantia nigra, reticular formation, medial forebrain bundle, nucleus accumbens, septum pediculum, olfactory tubercule and hippocampus and suggest that any concentration of alcohol exposure to these regions would make alcohol use virtually unavoidable (Myers and Privette 1989).
Studies of EEG Biofeedback in Substance Abuse Treatment
The Peniston Protocol (Alpha-Theta Feedback)
The early studies of Kamiya (e.g., Nowlis and Kamiya 1970) on self-regulation of alpha rhythm elicited substantial interest in the potential clinical applications of alpha biofeedback for SUD treatment. There were reported several uncontrolled case studies and conceptual reviews on alpha EEG training for alcohol (DeGood and Valle 1978; Denney et al. 1991; Jones and Holmes 1976; Passini et al. 1977; Tarbox 1983; Watson et al. 1978) and drug abuse treatment (Brinkman 1978; Goldberg et al. 1976, 1977; Lamontagne et al. 1977; Sim 1976), but the impact of alpha biofeedback training as a SUD therapy was not significant.
The bulk of the literature to date regarding EEG biofeedback of addictive disorders is focused on alpha-theta biofeedback. The technique involves the simultaneous measurement of occipital alpha (8–13 Hz) and theta (4– 8 Hz) and feedback by separate auditory tones for each frequency representing amplitudes greater than pre set thresholds. The subject is encouraged to relax and to increase the amount of time the signal is heard, that is to say, to increase the amount of time that the amplitude of each defined bandwidth exceeds the threshold. A variety of equipment and software has been used to acquire, process, and filter these signals, and there are differences in technique inherent with equipment and software.
Alpha-theta feedback training was first employed and described by Elmer Green and colleagues (Green et al. 1974) at the Menninger Clinic. This method was based on Green’s observations of single lead EEG during meditative states in practiced meditators, during which increased theta amplitude was observed following an initial increased alpha amplitude, then a drop off of alpha amplitude (theta/alpha crossover). When the feedback of the alpha and theta signal was applied to subjects, states of profound relaxation and reverie were reported to occur. The method was seen as useful in augmenting psychotherapy and promoting individual insight. It could be seen as a use of brain wave signal feedback to enable a subject to maintain a particular state of consciousness similar to a meditative or hypnotic relaxed state over a 30-or 40-min feedback session.
Goslinga (1975) gave the first description of the use of alpha-theta feedback in a SUD treatment program. This integrated program started in 1973 at the Topeka VA, and included group and individual therapies. Daily 20-min EEG biofeedback sessions (integrated with EMG biofeedback and temperature control biofeedback) were conducted over 6 weeks, resulting in free, loose associations, heightened sensitivity, and increased suggestibility. Patients discussed their insights and experiences associated with biofeedback in therapy groups several times a week, augmenting expressive psychotherapy. The first published clinical reports of efficacy of alpha-theta training at the Topeka VA were by Twemlow and Bowen (1976), who explored the impact of alpha-theta training on psychodynamic issues in 67 non-psychotic chronic male alcoholics in an inpatient treatment program. In this non-controlled study, they found that ‘‘religiousness’’ as a predictor of ‘‘self-actualization’’ may have increased as a result of imagery experienced in theta states. This was seen as positive to the program goal of augmenting Alcoholics Anonymous as a recovery philosophy. The high suggestibility of the method was acknowledged; ‘‘treatments such as brainwave training, which utilize abstract, ill understood techniques are potential repositories of magical projection and fantasy and would logically be more acceptable to alcoholics who are able to have ‘faith’ (devoutly or moderately religious)’’ (Twemlow and Bowen 1977). In another uncontrolled study at the Topeka VA, 21 alcoholics were reported to exhibit within and across session increases in raw theta amplitudes at occipital areas bilaterally measured by single lead EEG during the course of alpha-theta training, becoming more able to achieve deep states as manifested by EEG (Twemlow et al. 1977). These initial studies advanced the utility of biofeedback induced theta states in promoting insight and attitude change in alcoholics, with the assumptions that biofeedback-induced theta states are associated with heightened awareness and suggestibility, and that this heightened awareness and suggestibility would enhance recovery. Outcome data regarding abstinence were not reported.
In the first reported randomized and controlled study of alcoholics treated with alpha-theta EEG biofeedback, Peniston and Kulkosky (1989) described positive outcome results. Their subjects were inpatients in a VA hospital treatment program, all males with established chronic alcoholism and multiple past failed treatments. Following a temperature biofeedback pre-training phase, Peniston’s experimental subjects (n = 10) completed 15 30-min sessions of eyes closed occipital alpha-theta biofeedback. Compared to a traditionally treated alcoholic control group (n = 10), and nonalcoholic controls (n = 10), alcoholics receiving brainwave biofeedback showed significant increases in percentages of EEG recorded in the alpha and theta rhythms, and increased alpha rhythm amplitudes (single lead measurements at international site O1). The experimentally treated subjects showed reductions in Beck Depression Inventory scores compared to the control groups. Control subjects who received standard treatment alone showed increased levels of circulating beta-endorphin, an index of stress, whereas the EEG biofeedback group did not. Thirteen-month follow-up data indicated significantly more sustained prevention of relapse in alcoholics who completed alpha-theta brainwave training as compared to the control alcoholics, defining successful relapse prevention as ‘‘not using alcohol for more than six contiguous days’’ during the follow-up period. In a further report on the same control and experimental subjects, Peniston and Kulkosky (1990) described substantial changes in personality test results in the experimental group as compared to the controls. The experimental group showed improvement in psychological adjustment on 13 scales of the Millon Clinical Multiaxial Inventory compared to the traditionally treated alcoholics who improved on only two scales and became worse on one scale. On the 16-PF personality inventory, the neurofeedback training group demonstrated improvement on seven scales, compared to only one scale among the traditional treatment group. This small n study employed controls and blind outcome evaluation, with actual outcome figures of 80% positive outcome versus 20% in the traditional treatment control condition at 4-year follow up.
The protocol described by Peniston at the Fort Lyons VA cited above is similar to that initially employed by Twemlow and colleagues at the Topeka VA and Elmer Green at the Menninger Clinic, with two additions, i.e., (1) temperature training and (2) script. Peniston introduced temperature biofeedback training as a preconditioning relaxation exercise, along with an induction script to be read at the start of each session. This protocol (described as follows) has become known as the ‘‘Peniston Protocol’’ and has become the focus of research in subsequent studies. Subjects are first taught deep relaxation by skin temperature biofeedback for a minimum of five sessions that additionally incorporates autogenic phrases. Peniston also used the criteria of obtaining a temperature of 94F before moving on to EEG biofeedback. Participants then are instructed in EEG biofeedback and in an eyes closed and relaxed condition, receive auditory signals from an EEG apparatus using an international site O1 single electrode. A standard induction script employing suggestions to relax and ‘‘sink down’’ into reverie is read. When alpha (8–12 Hz) brainwaves exceed a preset threshold, a pleasant tone is heard, and by learning to voluntarily produce this tone, the subject becomes progressively relaxed. When theta brainwaves (4–8 Hz) are produced at a sufficiently high amplitude, a second tone is heard, and the subject becomes more relaxed and according to Peniston, enters a hypnagogic state of free reverie and high suggestibility. (Although theta increase and alpha decrease are thought by Peniston to be associated with a deeply relaxed state where hypnagogic reverie is present, this may simply represent drowsiness) (Niedermeyer 1999). Following the session, with the subject in a relaxed and suggestible state, a therapy session is conducted between the subject and therapist where the contents of the imagery experienced is explored and ‘‘abreactive’’ experiences are explored (Peniston and Kulkosky 1989, 1990, 1991).
Saxby and Peniston (1995) reported on 14 chronically alcohol dependent and depressed outpatients using this same protocol of alpha-theta brainwave biofeedback. Following treatment, subjects showed substantial decreases in depression and psychopathology as measured by standard instruments. Twenty-one month follow-up data indicated sustained abstinence from alcohol confirmed by collateral report. These male and female outpatients received 20 40-min sessions of feedback.
Bodenhamer-Davis and Calloway (2004) reported a clinical trial with 16 chemically dependent outpatients, 10 of whom were probationers classified as high risk for re-arrest. Subjects completed an average of 31 alpha-theta biofeedback sessions. Psychometrics demonstrated improvements in personality and mood. Follow-up at 74–98 months indicated 81.3% of the treatment subjects were abstinent. Re-arrest rates and probation revocations for the probation treatment group were lower than those for a probation comparison group (40% vs. 79%).
Fahrion (1995) gave a preliminary report (n = 119) on a large randomized study of alpha-theta training for addiction in the Kansas Prison System using group-training equipment. A report of the completed study (n = 520) (Fahrion 2002) showed little difference between the two groups overall at 2-year outcome. But, when results were analyzed for age, race and drug of choice, neurofeedback emerged as a more efficacious treatment for younger and non-white and non-stimulant abusing participants. Interestingly, this protocol was not effective for cocaine abusers. (Stimulant abusers will be discussed later in this article under the Scott–Kaiser modification of the Peniston protocol.)
The issue of alpha-theta biofeedback in culturally sensitive groups that have not responded to traditional modes of addiction treatment (such as confrontational group therapies) has been considered in an open case series reported by Kelly (1997). This three year follow-up study presented the treatment outcomes of 19 Dine’ (Navajo) clients. Four (21%) participants achieved ‘‘sustained full remission,’’ 12 (63%) achieved ‘‘sustained partial remission,’’ and 3 (16%) remained ‘‘dependent.’’ The majority of participants also showed a significant increase in ‘‘level of functioning’’.
Schneider et al. (1993) used slow cortical potential biofeedback to treat 10 unmedicated alcoholic patients in four neurofeedback sessions after hospitalization. Seven patients participated in a fifth session an average of 4 months later. Six out of these seven patients had not had a relapse at the follow-up. These results are similar to those reported for alpha theta training.
Several other studies using the Peniston protocol and its modifications reported cases with positive clinical effects (Burkett et al. 2003, DeBeus et al. 2002; Fahrion et al. 1992; Finkelberg et al. 1996; Skok et al. 1997). These studies suggest that an applied psychophysiological approach based on an alpha-theta biofeedback protocol is a valuable alternative to conventional substance abuse treatment (Walters 1998). Nevertheless, most of these results were reported at the society meetings, and only few of these studies were published in mainstream peer-reviewed journals other than The Journal of Neurotherapy.
A critical analysis of the Peniston Protocol is discussed at length in the previous reviews (Trudeau 2000, 2005a, b). Several controlled studies of the Peniston protocol for addictions, completed by Lowe (1999), Moore and Trudeau (1998), and Taub and Rosenfeld (1994), suggest that alpha-theta training for addictions may be non-specific in terms of effect when compared to suggestion, sham or controlled treatment, or meditational techniques. By contrast, Egner et al. (2002) showed that alpha-theta training results in an increase of theta/alpha ratios, as compared to a control condition. In an in depth critical analysis that examines inconsistencies reported in the original Peniston papers, Graap and Freides (1998) raise serious issues about the reporting of original samples and procedures in these studies. In their analyses, the results may have been due as much to the intense therapies accompanying the biofeedback as due to the biofeedback itself. The subjects may have been comorbid for a number of conditions, which were not clearly reported, particularly PTSD, which may have been the focus of the treatment. In his reply to these criticisms, Peniston (1998) acknowledges that it ‘‘remains unknown whether the temperature training, the visualizations, the ATBWNT (alpha-theta brain wave neurotherapy), the therapist, the placebo, or the Hawthorne effects are responsible for the beneficial results.’’ The criticism raised above by Graap and Friedes (1998) regarding Peniston’s papers could also be applied to earlier replication studies. Neither Peniston’s studies nor the replication studies provide sufficient detail regarding the specifics of the types of equipment used for alpha-theta feedback, including filtering methods for the EEG signal or other technical information, to permit exact reproduction of the feedback protocols with other equipment. Outcome criteria also vary in the replication studies, with varying measures of abstinence and improvement. An exception to these concerns is the report of Scott et al. (2005), which will be discussed later in greater detail.
It should be noted that psychostimulant (cocaine, methamphetamine) addictions may require approaches and neurofeedback protocols other than alpha/theta training. Persons who are cocaine-dependent are cortically under-aroused during protracted abstinence (Roemer et al. 1995). qEEG changes, such as a decrease in high beta (18–26 Hz) power are typical for withdrawal from cocaine (Noldy et al. 1994). Cocaine abusers who are still taking this drug often show low amounts of delta and excess amounts of alpha and beta activity (Alper 1999; Prichep et al. 1999), whereas chronic methamphetamine abusers usually exhibit excessive delta and theta activity (Newton et al. 2003). Thus, cocaine and methamphetamine users may warrant a different EEG biofeedback protocol, at least at the beginning stages of neurofeedback therapy.
The Scott–Kaiser Modification of the Peniston Protocol
Scott and Kaiser (1998) describe combining a protocol for attentional training (beta and/or SMR augmentation with theta suppression) with the Peniston protocol (alpha-theta training) in a population of subjects with mixed substance abuse, rich in stimulant abusers. The beta protocol is similar to that used in ADHD (Kaiser and Othmer 2000) and was used until measures of attention normalized, and then the standard Peniston protocol without temperature training was applied (Scott et al. 2002). The study group is substantially different than that reported in either the Peniston or replication studies. The rationale is based in part on reports of substantial alteration of qEEG seen in stimulant abusers associated with early treatment failure (Prichep et al. 1996, 2002) likely associated with marked frontal neurotoxicity and alterations in dopamine receptor mechanisms (Alper 1999). Additionally, preexisting ADHD is associated with stimulant preference in adult substance abusers, and is independent of stimulant associated qEEG changes. These findings of chronic EEG abnormality and high incidence of preexisting ADHD in stimulant abusers suggest they may be less able to engage in the hypnagogic and auto-suggestive Peniston protocol (Trudeau et al. 1999). Furthermore, eyes-closed alpha feedback as a starting protocol may be deleterious in stimulant abusers because the most common EEG abnormality in crack cocaine addicts is excess frontal alpha (Prichep et al. 2002).
In their initial report, Scott and Kaiser (1998) described substantial improvement in measures of attention and also of personality (similar to those reported by Peniston and Kulkosky 1990). Their experimental subjects underwent an average of 13 SMR-beta (12–18 Hz) neurofeedback training sessions followed by 30 alpha-theta sessions during the first 45 days of treatment. Treatment retention was significantly better in the EEG biofeedback group and was associated with the initial SMR-beta training. A subsequent published paper (Scott et al. 2005) reported on an expanded series of 121 inpatient drug program subjects randomized to condition, followed up at 1 year. Subjects were tested and controlled for the presence of attentional and cognitive deficits, personality states and traits. The experimental group showed normalization of attentional variables following the SMR-Beta portion of the neurofeedback, while the control group showed no improvement. Experimental subjects demonstrated significant changes (p \ .05) beyond the control subjects on 5 of the 10 scales of the MMPI-2. Subjects in the experimental group were also more likely to stay in treatment longer and more likely to complete treatment as compared to the control group. Finally, the one-year sustained abstinence levels were significantly higher for the experimental group as compared to the control group.
The approach of beta training in conjunction with alpha-theta training has been applied successfully in a treatment program aimed at homeless crack cocaine abusers in Houston, as reported by Burkett et al. (2003), with impressive results. Two hundred and seventy (270) male addicts received 30 sessions of a protocol similar to the Scott Kaiser modification. One-year follow-up evaluations of 94 treatment completers indicated that 95.7% of subjects were maintaining a regular residence; 93.6% were employed/in school or training, and 88.3% had no subsequent arrests. Self-report depression scores dropped by 50% and self-report anxiety scores by 66%. Furthermore, 53.2% reported no alcohol or drug use 12 months after biofeedback, and 23.4% used drugs or alcohol only one to three times after their stay. This was a substantial improvement from the expected 30% or less expected recovery in this group. The remaining 23.4% reported using drugs or alcohol more than 20 times over the year. Urinalysis results corroborated self-reports of drug use. The treatment program saw substantial changes in length of stay and completion. After the introduction of the neurofeedback to the mission regimen, length of stay tripled, beginning at 30 days on average and culminating at 100 days after the addition of neurotherapy. In a later study the authors reported follow-up results on 87 subjects after completion of neurofeedback training (Burkett et al. 2005). The follow-up measures of drug screens, length of residence, and self-reported depression scores showed significant improvement. It should be noted that this study had limitations, because neurofeedback was positioned only as an adjunct therapy to all other faith-based treatments for crack cocaine abusing homeless persons enrolled in this residential shelter mission and was an uncontrolled study. Yet the improvement in program retention is impressive and may well be related to the improved outcome.