1. Context
Individuals with mental health disorders, such as schizophrenia and bipolar disorder, have diagnosable psychiatric conditions that significantly interfere with their occupational performance across various domains, thereby limiting inclusion and participation (1). According to a 2021 estimate, 49.5% of adolescents aged 13 to 17 had mental disorders, and 22.8% of the adult population manifested mental disorders (2, 3). Adolescents with developmental disabilities and a genetic predisposition often face a negative social environment, leading to a high comorbidity with mental illness. Evidence examining psychopathology and intellectual impairment found a 28% aggression rate in individuals with mental health disorders (1). This behavioral difficulty can interfere with learning new skills, maintaining attention at school, and developing effective social skills, which impedes successful interaction in various settings, such as school, and can lead to peer rejection. Consequently, adolescents may lose self-esteem, engage in alcohol and substance abuse, and become delinquent (4). Moreover, difficulty following societal rules and a tendency toward irritability and aggression can result in ongoing legal issues, potentially leading to incarceration, which negatively affects mental and physical health and the ability to return to school or work (5). Violence and aggression towards self, others, or objects in inpatient psychiatric units can harm the individual, others, and staff, increase care expenses, and disrupt the therapeutic environment (6). When seeking community integration for adult psychiatric patients with aggressive behavior, their challenging behavior often hinders their ability to reintegrate. This results in diminished self-control, self-direction, social competencies, and a lower quality of life (7). Challenging behaviors can stem from environmental or biological factors, necessitating both pharmacological and behavioral modification methods (6, 7). These methods include de-escalation, time-outs, increased observation and support, and medication use with the patient's consent to calm them down in a non-coercive manner. If the individual does not cooperate, staff may resort to restraining the patient geographically, physically, or chemically (rapid tranquillization) as a last resort for safety reasons. However, since coercive methods can have harmful complications, particularly for mentally ill patients with physical comorbidities, and because reducing restraint and seclusion leads to shorter and fewer hospitalizations and decreased use of sedatives and hypnotics, identifying alternative behavioral plans for individuals with aggressive behavior is crucial (8). Intervention strategies known as Contingency Management (CM), based on operant conditioning, provide contingent reinforcement of exhibited desirable behavior with exchangeable tokens and have been noted to be effective in addressing behavioral difficulties (9). The Token economy (TE), a CM intervention, has historically been successfully employed as a behavior-management and motivational tool in residential, inpatient, school, and correctional settings for diverse populations. These populations include chronic psychiatric patients, students with Learning Disorders, individuals with Autism spectrum disorder (ASD), individuals with Intellectual Disability Disorder (IDD), children with Attention Deficit Hyperactivity Disorder (ADHD), individuals with substance abuse issues, and those with head injuries (10, 11). This approach has been effectively applied to various behaviors, such as reducing disruptive behavior and increasing attentive behavior at school, promoting adaptive behaviors, targeting symptomatic behaviors in inpatient settings, and improving pro-social behaviors in community-based settings (12). In 2016, a review article discussed the use of the Token economy (TE) for treating symptoms of mentally ill patients admitted to hospitals. After analyzing literature from 1999 to 2013, Glowacki et al. concluded that TE is an economically friendly intervention that can decrease negative behaviors in inpatient psychiatric settings. They also noted that, due to the limited number of studies on the efficacy of TE for reducing negative symptoms, there is a need for further in-depth research to explore its effectiveness (13). In another review study conducted by Austin, TE was implemented to manage the behavior of students with emotional and behavioral disorders. He noted that TE is an effective intervention within the classroom; however, due to the small number of participants, there were no significant differences between pre-test and post-test in most of the reviewed studies. He recommended conducting the study again with more participants and modifications (14).
2. Objective
Therefore, this literature review aims to analyze research published within the last 20 years (2003 - 2023) that used Contingency Management in the form of TE in mentally ill adults and adolescents. The goal is to explore its impact on aggressive behavior and identify factors that increase the effectiveness of the TE program.
3. Evidence Acquisition
3.1. Search Strategy
Electronic searches were conducted by the first and second authors, limiting the results to English-language publications from the period 2003 - 2023. The search was carried out in PubMed, Google Scholar, ProQuest, Scopus, and Web of Science. The MeSH terms used included: Token Economy, Token Reinforcement, Mental Disorders, Aggression, and Violence. Other terminologies found in systematic reviews, such as Contingency Management, Behavioral Management, Psychiatric Disorders, Challenging Behavior, and Disruptive Behavior, were also applied to ensure all related articles were included. Additionally, references cited in relevant literature were examined (Figure 1).
Figure 1.
Identification of studies via databases/other methods
3.2. Criteria for Including Studies
Studies including randomized controlled trials, pre-test and post-test designs, pilot studies, and case reports that employed CM in the form of TE alone or in combination with other interventions on adults and adolescents with mental, behavioral, or neurodevelopmental disorders who demonstrated aggressive behavior were included for review.
3.3. Criteria for Excluding Studies
Research including child participants, specific study types such as systematic reviews, meta-analyses, anecdotal and descriptive articles, and articles without measurable outcomes were excluded.
3.4. Study Identification
Data extraction was performed by identifying relevant studies through electronic searches. Titles and abstracts were reviewed separately by the first and second authors after eliminating duplicate articles. Articles that did not meet the inclusion criteria were subsequently excluded.
3.5. Study Selection
Predetermined information such as authors, year of publication, study design, study setting, number of participants, interventions and their duration, target behaviors, results, and study limitations were screened to eliminate irrelevant studies. After combining the reviewed articles from both authors, the full manuscripts were read and assessed for final eligibility. Articles with data discrepancies were included only if both authors agreed on their importance and relevance to the purpose of this study.
3.6. Quality Assessment
The Quality Assessment Tool for Quantitative Studies (QATQS) was utilized to evaluate the quality of articles by the first and second authors. Articles were graded based on the QATQS components of selection bias, design, blinding, methods of data collection, confounders, and withdrawals and drop-outs (15). In addition, the level of studies was determined according to the Oxford Centre for Evidence-Based Medicine Levels of Evidence (OCEBM, 2011). Articles were classified according to their design on a level of 1 - 5, where RCTs as the strongest evidence are considered level 1 and the weakest studies like mechanism-based reasonings are level 5 (16).
3.7. Data Synthesis
The obtained data were summarized, depicting the frequency of outcomes measured in numbers or percentages descriptively, or reporting statistical information, including average test scores, statistical significance, and effect size in changes or differences.
4. Results
Of the 589 articles identified, 570 were eliminated due to duplication and irrelevance to the criteria. This left 19 studies, of which the full texts were reviewed, except for one study that was not accessible. Data were extracted from 12 studies, each involving a reward and incentive-based Token economy (TE) for adults or adolescents over 12 years old with aggressive behavior in various settings, to determine the effectiveness of TE (Table 1). After quality assessment using QATQS, two studies were rated as strong, one as moderate, and the remaining nine as weak (Table 2). According to the OCEBM (2011) levels of evidence, there were three level 2 studies, one level 3 study, and eight level 4 studies (Table 1). Most of the reviewed articles had low methodological quality, including very small sample sizes (N = 7) (10, 17 - 20, 23, 25), lack of long-term follow-up on the impacts of TE (N = 9) (4, 10, 17 - 20, 22 - 24), and results that were not generalizable (N = 9) (4, 6, 10, 17 - 19, 21, 23, 24). None of the studies, except for the Park and Lee study (6), included randomization and control groups. Most of the studies had only male participants, with Taylor and Mudford's study being the only one that included both sexes (17).
Table 1.Data Extracted from 12 Included Studies
| Author, Year | Design, Level of Evidence | Setting, Population | N | Intervention | Duration | Objectives | Result | Limitation |
|---|---|---|---|---|---|---|---|---|
| Meyers et al, 2018 (18) | Quasi-experimental, 2 | Prison, SMI | 58 | CM + psychotherapy and psychoeducational group, individual counseling | 1 year based on the patients` function | Risk of self-injury, n* of requesting protective custody, decrease in crime steps, non-compliance, N of visits, minor or major violations, drug violations, N of physical attack | The N of requests for custody (M = 0.16, P < .05) and supervision (M = 0.66, P < 0.05) and the N of drug abuse decreased. The N of visits raised; But there was a slight increase in the scores of mental health condition (M = 3.63, P < 0.05) due to rises of minor violence on average between Time 1 (M = 0.54) and Time 4 (M = 1.04, P < 0.05). | Limited data available to assess program outcomes, the program is restricted to one facility, lack of control group, lack of individualized reinforcement, lack of procedural integrity |
| Travis, and Stornay, 2013 (19) | Single subject with ABA design, 4 | Residential forensic facility, adults with schizoaffective disorder and mild IDD | 3 | Skills training+ TE (Behavioral Skills Training | 3 weeks intervention and 3 months follow-up | Rate of aggressive and replacement responses that each participant shows after presenting 15 antecedent stimuli | During Aa, the average percentage of aggressive responses for each participant was 54%, 71%, and 64% which decreased to 12%, 10%, and 9% during Ba. The average percentage of replacement responses during Aa was 34%, 15%, and 26% for each participant, and during Ba, they increased to 79%, 78%, and 82%. During the generalization phase, the mean percentage of replacement responses was 74%, 71%, and 75%. During 3 months after the intervention, the N of community trips increased from 1 community trip for all participants to 5, 5, and 2. | Possibility of increased replacement behaviors due to the presentation of the experimenter in post-treatment sessions, lack of functional analysis, no follow-up, and the participants, themselves, did not evaluate the treatment efficacy in the social validity process |
| Park, and Lee, 2012 (6) | Quasi-experimental,2 | hospital, SMI | 44 | Short-Term TE | 8 weeks | Verbal and physical aggression, and damaging objects (Overt Aggression Scale) | After eight weeks, the experimental group showed a significant reduction in aggressive behavior for verbal attacks, from 87% to 63%, and for physical attacks and property damage, from 91% to 78%. However, the control group's scores showed increased aggressive behaviors and had significant differences (intervention group F = 194.46, P < 0.001). | Lack of randomization, lack of individualized reinforcement |
| Taylor, and Mudford, 2012 (17) | Single subject with ABAB design, 3 | Residential therapeutic community, youth with substance abuse and mental health disorders | 16 | TE (Thumbs Up) | 60 sessions | Decrease of inappropriate statements (verbal states that contain swear words, venting, neg-raving), increase of positive statements (showing approval, empathizing, describing desirable state or mood) | Positive statements increased from Aa to Ba for most residents (11/16). A sign test with an alpha level of 0.05 showed statistical significance (Z = 2.89, P = 0.006). All participants had a low rate of verbal statements in Aa and did not change during thumbs up (remained at near zero levels). | Limited time of intervention, no follow-up and generalization, limited data collection due to lack of control over participants entering and leaving the program, lack of individualized reinforcer |
| Tomaka, 2009 (10) | Single subject with ABAB design for 2 participants and AB design for 1 participant, 4 | Residential group homes, IDD with behavioral issues | 3 | A level system with built-in TE+ skills training | Nearly 3 months (it was different for each participant) | Frequency of problematic sexual behavior, assault, damaging objects, stripping, pulling the siren of the emergency system, and escaping | Two of the participants in an ABAB design could make it to and maintain the status of level 4 after approximately 78 and 90 days. The average replacement behavior per day for the first client was A1: 2.7, B1: 4.1, A2: 3.2, B2: 4.6, and for the second one was A1: 2.3, B1: 3.7, A2: 2.5, B2: 5.3. The third participant in an AB design who avoided performing the inappropriate behaviors, demonstrated more replacement behaviors in an average of 2.2 to 2.6 per day and made it to Level 3 after 95 days | Some vague descriptions of the topographies of the behaviors, limited time of fully training and monitoring staff implementing the level system. |
| Holmqvist, et al. 2007 (20) | Quasi-experimental, comparison of two interventions in two institute, 2 | Adolescents in residential care | 57 | Aggression replacement training+ TE compared to the relationally oriented treatment model | 10 weeks, twice a week and 1 hour for each session, 1 year follow-up | Rate of sentence register & the police suspension register (measured by penalty value score), admission and guilt feelings | The institutions had no significant differences in the rate of criminal activity. The values were: F = 0.69, P = 0.51 for the major sentence; F = 0.80, P = 0.46 for all sentences; F = 2.60, P = 0.09 for the major suspicion; and F = 0.22, P = 0.81 for the three major suspicions. 20% of the participants had no major crime after two years; 17 out of 56 had not been sentenced, while the remaining 6 had minor crimes. The effect size of all sentences was 0.71, while the score for the three major suspicions was 0.30. After B a, 9 adolescents were not reported as a criminal, and 5 participants were reported for minor crimes. | Lack of procedural integrity, limited N of participants, no control group, and randomization |
| Coogan et al. 2007 (21) | Single subject with ABAB design, 4 | School, adolescents with inappropriate behavior | 5 | Group CM + Self- Monitoring | 53 days | Any behavior that involves inappropriate physical contact, speaking, hostility, playfulness, distraction, noise-making, or leaving one's seat | The average percentage of disruptive behavior in student 1 was 25% in A1, 3.07% in B1, 21.06% in A2, and decreased to 5.71% in B2. For student 2 it was 30.44% in A1, 5.96% in B1, 20.33% in A2, and 8.17% in B2. In student 3, 30.35% in A1, 6.37% in B1, 25.33% in A2, and 8.95% in B2. For student 4, 38.25% in A1, 5.81% in B1, 32.33% in A2, and 12.6% in B2. In student 5, 34% in A1, 3.96% in B1, 19.89% in A2, and decreased to 4.96% in B2. Students 1, 2, and 3 had significant effect sizes of 1.88, 1.36, and 1.30, respectively. Whereas, students 4 and 5 had the largest effect sizes (2.24 and 2.26, respectively). | Lack of follow-up |
| Luby, 2007 (22) | Single subject with ABA design, 4 | Residential school, an adolescent with ASD | 1 | TE | 9 weeks | Frequency of assault, damaging objects, and inappropriate statements on the behavioral datasheet | The TE was an ineffective method to modify the participant's behavior. However, assault decreased from Stage I of 0 to 10 to Stage III of 0 to 9 (the highest frequency: 49th day & the lowest frequency: 42th - 45th day), and damaging objects decreased from Stage I of 0 to 6 to Stage III of 0 to 2 (the highest frequency: 45th day & the lowest frequency: 42th - 45th day); but these behaviors did not stabilize. Disruptive talk increased from Stage I of 0 to 7 to Stage III of 0 to 15. | Single subject design, the unwillingness of the participant to earn tokens, and the antecedent stimulus that triggered the target behavior were not addressed and most of the target behaviors occurred in the classroom, lunchtime, and cafeteria |
| Bisconer et al.2006 (23) | Single subject with AB a design, 4 | Psychiatric hospital, schizoaffective bipolar type, mild IDD | 1 | Social skill training+ contingent reinforcement | 42 months | Aggression towards others, personnel, and self, frequency of restrictive interventions usage | A reduction in the rate of aggression led to a lesser need for restrictive treatments to manage aggression and a reduction in staff injury. In addition, the TE has let the participant learn appropriate behaviors and skills, and to move to a less restrictive community home. | Lack of procedural integrity and follow-up sessions |
| Freeman, and Dexter-Mazza, 2004 (24) | Single subject with ABAB design,4 | Classroom of a Residential facility, ADHD, conduct disorder, adjustment disorder, and learning disorder | 1 | Self-monitoring, skill training + TE | 3 - 4 times per week, 24 sessions | Disruptive behavior and off-task behavior | Self-monitoring plus matching (adult feedback as a social reinforcer) was effective in preventing disruptive and off-task behavior in the classroom. The frequency of problem behavior decreased from 40% of intervals in the baseline to 12.06% when therapy was introduced. The mean percentage occurrence of targeted behaviors was M = 21.26 in A1, it decreased to M = 20.89 in B1, and in withdrawal and reimplementation of the matching process, they reached 28.6 and 13.2. | Significant discrepancy between the N of observations during Ba, with earlier phases with a greater N of observations, no follow-up, and contingency of reward with accurate self-monitoring made it difficult to determine which variable was responsible for changes |
| Wolfe et al. 2003 (4) | Single subject with A-B1-BC1-B2-BC2 design, 4 | School, adolescents with Emotional and Behavioral Disorders | 3 | TE+ cooperative games | 35 sessions based on the school year (started in the second semester and ended with the termination of the school year) | Pro-social (compliment, supporting, thanking, handshakes, or high fives) and anti-social actions (swearing, physical threats, name-calling, assault) | Prosocial verbal and physical behaviors increased during BC1 and BC2. B1 and B2 had lower means compared to the means of BC1 and BC2. Antisocial verbal behaviors were higher in the A and B1 phases, but in BC1 these behaviors decreased. During B2 behaviors returned to baseline but with the re-initiation of the BC2 these behaviors again decreased and were virtually non-existent. During A and BC1 antisocial physical behaviors were high; but when BC1 was initiated these behaviors dropped to zero and remained there till the end. | A small period of observation, during B2 the primary observer was not able to attend 2/3 sessions which is a concern for inter-rater agreements. |
| Zlomke and Leland 2003 (25) | Single subject with ABCB design,4 | School and residential group homes, youth with bipolar and conduct disorder | 1 | TE + self- monitoring | 19 weeks | Decrease of minor behavior (non-compliance and being out of the seat), disruptive behavior ( excessive talking out loud and singing), and aggression | In the school setting, A significant reduction in challenging behaviors occurred from a baseline mean of 118 after implementing TE. The decrease continued during the TE phase (mean = 63). Introducing self-monitoring alongside the TE resulted in a significant decrease in challenging behaviors. (mean = 7.75). However, the occurrence of targeted behaviors did not change in the group home. During the maintenance phase (4 weeks) compared to the baseline, the rate of challenging behaviors remained low, in both the school and group home (mean = 13). | Lack of procedural integrity, antecedent stimulus that triggered the target behavior were not addressed |
Abbreviations: TE, token economy; CM, contingency management; SMI, severe mental illness; N, number; ADHD, attention-deficit/hyperactivity disorder; IDD, intellectual disability disorder; ASD, Autism spectrum disorder.
a A: Baseline phase, B: Intervention phase, BC: Combination of the interventions.
Table 2.Evaluation of the Quality of Included Articles a
| Study | Selection Bias | Study Design | Confounders | Blinding | Data Collection Methods | Withdrawals and Drop-outs | Study Quality |
|---|---|---|---|---|---|---|---|
| Park, and Lee, (2012) (6) | 2 | 2 | 1 | 2 | 1 | 1 | 1 |
| Wolfe et al. (2003) (4) | 3 | 2 | 3 | 2 | 1 | 1 | 3 |
| Tomaka, 2009 (10) | 3 | 2 | 3 | 2 | 1 | 1 | 3 |
| Freeman and Dexter-Mazza, (2004) (24) | 3 | 2 | 3 | 2 | 1 | 1 | 3 |
| Coogan et al. (2007) (21) | 3 | 2 | 3 | 2 | 1 | 1 | 3 |
| Luby, 2007 (22) | 3 | 2 | 3 | 2 | 1 | 1 | 3 |
| Zlomke and Leland (2003) (25) | 3 | 2 | 3 | 2 | 1 | 1 | 3 |
| Holmqvist et al. (2007) (20) | 2 | 2 | 1 | 2 | 1 | 1 | 1 |
| Taylor and Mudford (2012) (17) | 2 | 2 | 3 | 2 | 1 | 1 | 2 |
| Bisconer et al. 2006 (23) | 3 | 2 | 3 | 2 | 3 | 1 | 3 |
| Meyers et al, (2018) (18) | 2 | 2 | 3 | 2 | 3 | 1 | 3 |
| Travis and Sturmey (2013) (19) | 3 | 2 | 3 | 2 | 1 | 1 | 3 |
a 1, strong; 2, moderate; 3, weak.
4.1. Token Economy (TE) in People with Mental Illness of Different Age Groups
4.1.1. Early Adolescence
4.1.2. Youth
The reduction of physical aggression and property destruction by an autistic student in the classroom did not stabilize, thus not confirming the effectiveness of TE. Additionally, disruptive talk increased after program implementation. The author suggested that the primary reason might be the student’s reluctance to earn tokens in exchange for appropriate behavior (22). Token economy and self-monitoring decreased the frequency of aggressive, disruptive, out-of-seat, and non-compliant behavior of a youth student in the school setting; however, these changes were not observed in the home group setting, which served as a control situation (25). A study by Holmqvist et al. on delinquent youths in forensic facilities compared TE plus Aggression Replacement Treatment with relational-based treatment. They noted a reduction and omission of crime (rate of sentence register & police suspension register) in both groups. The obtained results did not differ significantly between the two groups. However, the authors suggested that relationally oriented treatment might be more manageable for adolescents (20). Another study in the youth population examined the effect of a token program named Thumbs Up (signifying the positive orientation of the intervention) on behavioral statements of adolescents with mental health disorders and substance abuse. In this study, positive statements increased, but inappropriate statements (verbal expressions containing swear words, venting, and negative raving) did not change significantly because they were socially accepted by most participants during the baseline phase and remained at a low level during the token phase. The authors stated that the decreased inappropriate behavior might be due to the presence of staff in all settings (17).
4.1.3. Adults
In the adult population, three articles implemented Token economy programs in psychiatric hospitals and noted a decrease in verbal attacks, property damage, and physical attacks toward self, staff, and other patients, which led to reduced use of interventions to control behavior (6, 23). Meyers et al. concluded that Contingency Management (CM) in chronically mentally ill prisoners was not effective in maintaining targeted behaviors like clowning and wandering (18). There were two studies on the use of TE for challenging behaviors in adults with intellectual and developmental disabilities (IDD) that indicated reductions in sexual behavior, assault, damaging objects, stripping, and pulling the siren of the emergency system, while replacement behaviors increased (10, 19).Token economy alone (6, 20, 21), TE in combination with skills training (10, 19, 23), and TE alongside self-monitoring (20, 21, 24) were the most common interventions used in the included studies. Other therapies implemented in combination with TE were level systems (10), Aggression Replacement Training (20), cooperative game therapy (4), and psychotherapy (18). Most of the studies were conducted on the adult population, with only a few studies applying Token Economy interventions to young adolescents (Figure 2).
Figure 2.
Chart of the Token Economy studies in different age groups
4.2. Main Factors Mentioned in Studies
4.2.1. Functional Assessment
Behavioral plans based on functional assessments can lead to meaningful reductions in aggression and significant increases in replacement behaviors. Functional assessment includes: (a) determining the target behavior; (b) identifying antecedents and situations that may stimulate the occurrence of the targeted behavior; (c) understanding the issues that arise as a result of the targeted behavior; (d) identifying remaining functional behavior; (e) implementing environmental adaptations that decrease the targeted behavior; and (f) selecting the best strategies (23). Three studies performed individualized functional assessments for each participant with IDD and schizoaffective disorder, bipolar type (10, 19, 23).
4.2.2. Antecedent, Behavior, Consequences
The place where the challenging behavior occurs, the stimulus that triggers the target behavior, the topography of the target behavior, the timing of its occurrence, and the outcome of the behavior are all parts of the functional assessment (10). Manipulation of antecedents and environmental modifications led to significant positive changes in the three studies mentioned earlier (10, 19, 23).
4.2.3. Replacement Behavior
Differential reinforcement of alternative behaviors that encourage appropriate actions to reduce aggression has been implemented in some studies (23). Aggression Replacement Training, used in combination with Token Economy (TE) on young offenders with a history of crime, included aggression control training, interpersonal skills training, and socio-moral reasoning aimed at teaching prosocial behaviors to replace aggression. While inappropriate behavior decreased in this study, the display of prosocial behaviors was not measured (20). Taylor and Mudford observed increased positive statements, such as indicating approval, describing a desirable state or mood, empathizing, appreciating, and apologizing. However, inappropriate conversations, including swearing, venting, and negative ranting, did not significantly change from the baseline phase to the intervention phase and remained low (17). In three studies, participants avoided demonstrating undesirable behaviors while concurrently exhibiting more desirable behaviors (10, 19, 23).
4.2.4. Staff Training
Accurate staff training is an important step for the consistency of program implementation. Staff implementing the Token economy (TE) must be able to (a) identify specific settings, events, and antecedents that stimulate targeted behavior; (b) cue verbally and physically to the targeted behavior; (c) provide feedback; (d) reinforce the desired behavior contingently; and (e) offer verbal praise as a social reinforcer for desirable behavior and the absence of challenging behavior (23). All the studies included staff training components throughout their programs. In Tomaka's study, two participants were receiving attention for opposing their peers, which reinforced them more than earning tokens for appropriate behavior. Therefore, staff began ignoring these undesirable behaviors and increased social reinforcement for exhibited replacement behaviors. This change led to a decrease in the targeted behaviors (10).
4.2.5. Inter-observer Reliability
Two independent raters simultaneously scored the observed behavioral responses. They divided the rate of their agreements by the sum of the rates of agreements and disagreements, then multiplied it by 100 to obtain the final percentage of consensus (10). Nine studies used inter-observer agreement on behavioral difficulties, with overall agreement ranging from 67% to 100% (4, 10, 17, 19-22, 24, 25). Some studies also assessed staff adherence to the TE implementation method to ensure procedural integrity and identify the need for retraining to ensure correct use of the therapy (10, 19).
4.2.6. Procedural Integrity
4.2.7. Choosing Reinforcers
Selecting back-up reinforcers that are effective, meaningful, and age-appropriate for participants can increase the success rates and satisfaction of participants in the TE program (4). Based on the reviewed studies, rewards were determined either by the participants themselves (17, 23), by the staff's knowledge of the participants (4, 22), or cooperatively by both (10, 20).
4.2.8. Generalization
One major criticism of TE noted in the literature is the transferability and generalizability of the results to other situations (27). Taylor and Mudford extended the intervention (Thumbs Up) to new staff. When the new staff were trained, they used Thumbs Up in morning meetings of youth participants with substance abuse and mental disorders. The frequency of positive statements did not differ significantly compared to the researcher who introduced Thumbs Up for the first time to participants, supporting its transferability to other staff (17). Travis and Sturmy assessed generalization across novel staff, antecedents, and settings. In their study, new staff members triggered five new stimulating antecedents in a non-familiar vocational and residential environment. Desirable behaviors were exhibited in the generalization phase, and none of the challenging behaviors were displayed (19). However, Zlomke and Zlomke confirmed positive behavioral changes in the classroom after TE implementation but did not observe the same results concurrently in the residential home group. They noted that they did not intend to examine the generalization of the intervention (25).
4.2.9. Social Validity
It is imperative that aspects of the study be validated by individuals beyond the primary researchers. This validation can be obtained formally or informally regarding the appropriateness of objectives, the importance of the intervention, and its effectiveness from participants, staff, or families (4). Five studies mentioned social validity, and all of them indicated satisfaction with the importance and effectiveness of TE (4, 17, 19, 21, 24).
5. Discussion
This literature review aimed to determine the effect of TE programs on the aggressive behaviors of mentally ill adults and adolescents. Evidence shows the effectiveness of TE in managing aggression in psychiatric hospitals. One study reported significant decreases in verbal and physical attacks and patient-to-staff injuries after the implementation of TE (6, 23). In the school setting, five studies were found, four of which showed reductions in aggression (disruptive, aggressive, and off-task behavior, antisocial verbal and physical behaviors) after the interventions were implemented (4, 17, 20, 21). However, in Luby's study, while property destruction and physical aggression decreased, these changes did not stabilize (22).
Token economy programs implemented in criminal settings showed moderate effects on behavior management. The number of crimes and aggressive responses decreased, and replacement responses increased, but in one study, minor violence among severely mentally ill prisoners increased with TE (18, 19, 25). In substance abuse residential care, positive statements increased, but inappropriate statements did not change from the baseline to the intervention phase (17). Tomaka noted a decrease in inappropriate sexual behavior, physical aggression, and property destruction among home group residents with intellectual disability disorder after TE implementation (10).
5.1. Limitations and Recommendations
Based on our research, the latest study was published in 2020, which we could not access the full text of. Except for the Meyers et al. study in 2018, all other studies were conducted more than 10 years ago. Additionally, most of the studies focused on the male population, ignoring gender differences. While studies have noted that random assignment, especially with control groups, is difficult to conduct for program evaluation, considering these factors could yield different results (18). We recommend that further research with methodological improvements should be conducted on behavioral management using TE. Key recommendations include:
(1) Implementing random assignment and including control groups.
(2) Considering both sexes to address gender differences.
(3) Conducting functional assessments and analyses of participants.
(4) Providing comprehensive staff training.
(5) Utilizing inter-rater reliability to ensure program consistency.
It is also imperative to follow up on the results to maintain the effects and ensure they can be transferred to broader societal contexts.
5.2. Conclusions
This literature review supports the efficacy of TE as a non-coercive method for reducing aggressive behaviors in adolescents and adults in various settings. The findings can be useful for establishing TE programs in modern psychiatric contexts to manage behavioral difficulties using token reinforcement effectively. This approach can help reduce injuries and the use of restrictive interventions. Although few studies have investigated the efficacy of TE for reducing challenging behaviors, most determined that TE was effective. However, TE for aggression reduction should be used with caution until further research is conducted on the topic.
