Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Oct 11.
Published in final edited form as: J Offender Rehabil. 2022 Oct 11;61(8):442–455. doi: 10.1080/10509674.2022.2128153

Analyses of the TCU Drug Screen 5: Using an Item Response Theory Model with a Sample of Juvenile Justice Youth

Amanda L Wiese a, Thomas B Sease a,*, Danica Knight a, Kevin Knight a
PMCID: PMC10263186  NIHMSID: NIHMS1853473  PMID: 37323828

Abstract

It is important to identify substance use disorders among youth who enter the juvenile justice system using a validated screener, such as the Texas Christian University Drug Screen 5 (TCU DS 5), so that necessary services can be provided to youth in need of treatment. While the TCU DS 5 is a valid, evidence-based screener, the use of an Item Response Theory model may better differentiate between mild, moderate, and severe forms of substance use disorders. The current study analyzed the feasibility and incremental value gained in using an Item Response Theory model to compute drug use severity scores as compared to its current scoring methodology. Results showed that while Item Response Theory may not be worthwhile as the standard method of scoring, item level analyses revealed there are benefits to using Item Response Theory to determine which items on a screener are most suggestive of severe substance use problems.

Keywords: adolescents, juvenile justice, substance use disorder, item response theory, psychometric evaluation

Introduction

Substance use (SU) has long been known to have an intimate role in youths’ involvement with the juvenile justice system. Youth involved with the juvenile justice system are nine times more like to develop a substance use disorder (SUD) when compared to their non-justice involved counterparts (Center for Behavioral Health and Quality, 2016). In fact, an estimated 45% to 65% of youth in the United States involved with the juvenile justice system meet the diagnostic criteria for a SUD (Dennis et al., 2009). The prevalence of SU in this population places these youth at a heightened risk for mental health difficulties (Tapia et al., 2016), Human Immunodeficiency Virus (HIV), other sexually transmitted infections (Donenberg et al., 2015), and recidivism (Henggeler et al., 2002). Thus, the juvenile justice system is uniquely positioned to prevent, identify, and treat SUDs in this vulnerable population.

According to the Juvenile Justice Behavioral Health Services Cascade (Belenko et al., 2017), the first step in identifying youth with SU problems is to administer an evidence-based screening instrument at intake. Put differently, every person who enters the juvenile justice system should be screened in a timely manner using a validated screener that provides clinically meaningful results about the client’s severity of SU. For people who score above a screener’s designated threshold, a more comprehensive diagnostic assessment/interview is then performed to further understand the client’s individualized treatment needs. Together, the information gathered from the screening and assessment process is used to inform the frequency, intensity, and type of treatment services provided to the client (Belenko et al., 2017). Unfortunately, of youth who enter the juvenile justice system, only 68%-71% are screened; of those, 48%-58% are identified as in need of SU treatment and approximately 15%-27% youth identified as needing treatment are referred to receive services (Dennis et al., 2019).

Ideally, it is important that SU screeners correspond with clinical diagnostic tools (ASAM, 2014), such as the Diagnostic and Statistical Manual of Mental Disorders 5 (DSM-5; American Psychiatric Association, 2013). The DSM-5 identifies different types of substance involvement (alcohol; cannabis; hallucinogens; inhalants; opioids; sedatives, hypnotics, or anxiolytics; stimulants; tobacco; other/unknown) that may constitute the diagnosis of a SUD. Moreover, there are 11 symptoms used to classify someone has having a SUD, which can be categorized as: 1) impaired control, 2) social impairment, 3) risky use, and 4) pharmacological criteria (American Psychiatric Association, 2013). The presence of 0–1 symptom indicates no substance use problem, 2–3 symptoms indicate a mild SU problem, 4–5 symptoms indicate a moderate SUD, and greater than six symptoms indicate a severe SUD (American Psychiatric Association, 2013). Although the DSM-5 provides a standardized framework for understanding the severity of SU, juvenile justice agencies commonly use screening assessments that are not intended to diagnose SUDs (Vincent et al., 2012) because of a lack of funding, the length of most formalized assessments, or practical concerns around the administration and scoring of these instruments. Nevertheless, this practice in juvenile justice programs is troublesome considering a SU diagnosis affords treatment providers the information needed to intervene with evidence-based services and lower youths’ risk of involvement with the justice system in the future (Farabee et al., 2001).

Reliable and clinically useful screening instruments have been developed to identify SUDs in clinical settings, however, their feasibility for juvenile justice agencies remain somewhat limited. The Adolescent Drug Involvement Scale (ADIS; Moberg & Hahn, 1991), for example, is a 13-item self-report measure designed to distinguish youth engaging in more problematic SU from youth experiencing minimal problems with SU. This assessment takes approximately five minutes to complete and is free to access. However, the ADIS is not tied to any specific assessment criteria, so there are no guidelines for how to interpret the scale beyond higher total scores meaning more serious SU involvement. The Drug Use Screening Inventory-Revised (DUSI-R; Tarter, 1990) is a 159-item self-report instrument that documents SU involvement with various substances and quantifies the severity of consequences associated with SU. This measure takes 20 to 40 minutes to complete, 20 minutes to score, and results are used to inform areas requiring a more comprehensive assessment. Each paper questionnaire costs $5.00, and the software license for computerized administration and scoring costs $250.00 per year. For a review and comparison of other SU screeners relevant to adolescents involved with the justice system see Pilowsky & Wu, 2013.

The Texas Christian University Drug Screen 5 (TCU DS 5) is a brief and cost-effective screening instrument that corresponds with the DSM-5 diagnostic criteria for SUDs (Knight et al., 2014; Knight et al., 2018). The TCU DS 5 was originally developed based on DSM-IV-R criteria showing strong internal reliability and validity (Broome et al., 1996; Knight et al., 2000). This measure was subsequently updated to align with changes made in the DSM-5, such as classifying substance use problems into none, mild, moderate, and severe. The TCU DS 5 (Knight et al., 2018) has demonstrated strong psychometric properties using Classical Testing Theory (Wiese et al., 2019) and has shown similar SU classification rates in adolescents involved with the juvenile justice system (Knight et al., 2018).

Another advantage of the TCU DS 5 is that it provides an additional set of items to help treatment staff gather more information about the client’s SU involvement. Namely, the TCU DS 5 includes a list of commonly used substances (e.g., alcohol, marijuana, prescription opioids) and asks the respondent to report the frequency at which they use each drug. This information can be used to pinpoint the substances most troublesome for clients and inform clinical practices designed meet their individual needs. The TCU DS 5 also includes a set of items measuring the extent to which youth perceive their SU as a problem, are motivated for treatment, and have received substance use treatment in the past. Sample items include, “How serious do you think your drug problems are,” “How important is it for you to get drug treatment now,” and “How many times before now have you ever been in a drug treatment program?” The final supplementary item (i.e., Item 16) asks respondents to report how many times in the past 12 months they have injected drugs with a needle. This specific item is useful for identifying youth that should also be screened for infectious diseases (e.g., HIV, Hepatitis C), which can be transmitted through intravenous drug use.

Although the TCU DS 5 has demonstrated acceptable psychometric properties using more traditional methods of scale evaluation (e.g., factor analysis, internal reliably scores), it is possible that the use of more advanced analytic models could improve the screener’s sensitivity to detecting SUDs. Item Response Theory (IRT) is a class of mathematical procedures that model the relationship between individual items on a scale and their assessment of a given latent variable (see Embretson & Reise, 2013 for a more comprehensive review of IRT procedures). This is to say, the application of IRT to the TCU DS 5 could be used to determine whether a statistically driven scoring algorithm improves the scale’s capacity to classify respondents’ severity of SU as compared to the original scoring method. In addition, item-level statistics generated through IRT can be used to determine which items on the TCU DS 5 are measuring low ability levels (i.e., low SU severity) as compared to high ability levels (i.e., high SU severity).

Current Study

The primary objective of this study was to compare classification rates of the TCU DS 5 using extant scoring procedures compared to IRT generated results. While the current methodology for scoring the TCU DS 5 uses a Classical Testing Theory approach, it is possible that identification of SU severity scores may benefit from the use of a statistically-driven, optimally-weighted scoring algorithm created through IRT. Stated differently, it is important to understand how the classifications of SU problems correspond to the estimate of SU severity. Of primary concern for the current study was whether an IRT approach better assessed SUD severity when compared to the current approach. Finally, this study also examined how IRT ability scores and Classical Testing Theory scores were related to the supplementary items of the TCU DS 5.

Method

Procedure and Participants

The present study used data collected as a part of a larger project assessing the effectiveness of an intervention designed to improve decision-making and reduce risky behavior among youth involved with the justice system (Knight et al., 2015). The total sample contained 312 males between the ages of 13 and 20 (M = 16.67, SD = 1.33) who were recruited from two correctional facilities located in the United States. Approximately two-thirds of the youth in this study were Black/African American (n = 196, 62.8%) and about a quarter were White (n = 72, 23.1%). Most people in this study were not Hispanic (n = 268, 85.9%).

TCU Drug Screen 5

The TCU DS 5 is a valid, evidence-based screener that has demonstrated strong psychometric properties in adolescents and adults (Knight et al., 2014, 2018). The instrument can be administered as an independent self-report measure or in small groups with a proctor reading each item aloud. There are 17 items in total, and respondents take approximately 5–10 min to complete the assessment. For the first 11 items, respondents answer Yes (1) or No (0) to a series of questions asking about SU involvement in the past 12 months. Final scores are calculated by summing responses to the first 11 items. A final score of 0–1 indicates no SUD, 2–3 indicates a mild SUD, 4–5 indicates a moderate SUD, and six or more indicates a severe SUD. Items 12 through 17 are not included as part of the final TCU DS 5 total score; instead, these items are used to provide additional information about the client’s SU involvement.

Analytic Plan

Descriptive statistics were calculated, such as means and standard deviations for total scores. Next, to ensure IRT was an appropriate analysis for the dataset, the assumptions of IRT (e.g., monotonicity, unidimensionality) were explored. Category probability curves, which depict the relationship between the probability of a given category (i.e., Yes vs. No) as a function of a person’s ability (the construct being measured, in this case, severity of SUD), were evaluated to test the assumption of monotonicity. Principal Components Analysis was then used to test the assumption of unidimensionality.

Item Response Theory analyses evaluated how well each item discriminated between people with varying levels of SU severity. Importantly, the simple summative scoring method was compared to the IRT results wherein a statistically-driven, optimally-weighted scoring algorithm calculated TCU DS 5 scores. Correlational analysis examined the relationship between peoples’ ability levels computed via the IRT and total scores computed via Classical Testing Theory. A significant correlation would imply that the two methods for calculating TCU DS 5 scores were related and therefore equivalent. A one-way analysis of variance investigated whether there were significant differences between TCU DS 5 severity diagnoses on the ability levels computed via IRT. The R2 statistic elucidated the percentage of variation in ability scores accounted for by the TCU DS 5 SUD severity diagnoses. This is to say, a high R2 value would suggest there is a considerable amount correspondence between the two scoring methods. The R2 value coupled with the distributions of ability scores within SUD severity diagnoses determined whether the IRT scoring algorithm provided more information than that gathered by summing all scores. Finally, difficulty scores were generated for the supplementary items to determine which measured low ability levels (i.e., low SU severity), moderate ability levels (i.e., mild-moderate SU severity), and high ability levels (i.e., high SU severity).

Results

Total scores for the TCU DS 5 were calculated and ranged from 0 to 11 (M = 3.24, SD = 3.89). These scores were then converted into SUD severity scores (see Table 1), showing that most of the sample (51.6%) did not have a diagnosable SUD. The TCU DS 5 had a Cronbach’s alpha of 0.94.

Table 1.

Frequency and Percentage of TCU Drug Screen 5 SUD Severity Diagnoses

SUD Severity Frequency Percentage
None 161 51.6%
Mild 41 13.1%
Moderate 28 9.0%
Severe 82 26.3%
Open in a new tab

Note. SUD = substance use disorder.

To test the assumption of monotonicity, a category probability curve was generated for each of the 11 diagnostic items on the TCU DS 5. Critically, as a person’s SU severity increased, the probability of responding “no” to the items decreased, and the probability of responding “yes” increased. These results suggest that the assumption of monotonicity had been satisfied. Principal Components Analysis showed the TCU DS 5 contained one component (1st eigenvalue = 6.80) that explained 61.60% of the observed variance in participants’ responses. In other words, the TCU DS 5 was measuring a single component and the assumption unidimensionality was met.

Item fit statistics using IRT included infit and outfit mean-squared statistics, standardized weighted (infit) and unweighted (outfit) mean-squared statistics, and point-measure correlations (see Table 3). Infit and outfit mean-squared statistics describe how well items fit the overall model and are influenced by outlier cases. In both instances, misfit of items was determined using values less than 0.6 or greater than 1.4 (Kean et al., 2018). The infit and outfit mean-squared statistics identified Items 1, 2, 3, 9, and 11 as potentially misfit. These items correspond with the larger/longer, quit/control, time spent, physical/psychological problems, and withdrawal criteria of the DSM-5. Looked at differently, the standardized weighted and unweighted mean-squared statistics assessed misfit items using a cut-off less than −2.0 or greater than 2.0 (Kean et al., 2018). Again, these thresholds showed that Items 1, 2, 3, 9, and 11 were misfit, suggesting these items could be measuring a different latent variable (Kean et al., 2018). However, people did appear to be responding to the items similarly considering the point-measure correlations among the individual items exceeded the cut-off of 0.40.

Table 3.

Item Fit Statistics

Item # Infit MNSQ Infit ZSTD Outfit MNSQ Outfit ZSTD PTMZ R Corr. PCA Loading Item Difficulty
1 1.40 3.77 1.75 3.90 0.71 0.67 0.52
2 1.36 3.26 1.57 3.14 0.71 0.69 0.62
3 1.39 3.82 1.83 3.89 0.71 0.66 0.44
4 0.88 −1.24 0.92 −0.45 0.80 0.82 0.53
5 0.83 −1.50 0.58 −1.97 0.79 0.82 0.80
6 1.04 0.47 1.06 0.45 0.77 0.78 0.58
7 0.87 −1.14 0.84 −0.83 0.79 0.82 0.70
8 0.89 −0.89 0.89 −0.37 0.77 0.80 0.82
9 0.74 −2.50 0.59 −2.42 0.81 0.85 0.68
10 0.83 −1.97 0.67 −2.02 0.82 0.82 0.35
11 0.58 −4.52 0.42 −4.00 0.84 0.89 0.63
Open in a new tab

Note. MNSQ = mean-square; ZSTD = standardized weighted (infit) and unweighted (outfit) mean-squared fit statistics; PTMZ = point-measure correlation; PCA = principal component analysis.

Item difficulty represents the average ability levels of persons wherein 50% of the sample is correctly endorsing the item. Items with higher difficulty scores suggest that respondents answering “yes” to these items are likely have more severe SUDs. Alternatively, items with lower difficulty scores indicate that people with more mild SUDs are also answering “yes” to the item. As shown in Table 3, Item 8 had the highest difficulty score (“Did you use drugs that put you or others in physical danger”) and Item 10 had the lowest difficulty score (“Did using the same amount of a drug lead to it having less of an effect as it did before”).

A one-way between-subjects analysis of variance was performed looking at Classical Testing Theory severity diagnoses (None, Mild, Moderate, vs. Severe) on IRT ability scores. There was a significant effect of SUD severity diagnosis, F(3, 308) = 819.77, p ≤ .001, R2 = 0.88, 95% CI [0.51, 0.55]. Follow up tests revealed that all groups were significantly different from each other, ps ≤ .001 (see Table 4). Eighty-eight percent of the variance in ability scores were explained by the current severity diagnoses. In other words, there is a high correspondence between the classification system underlying the TCU DS 5 (Classical Testing Theory results) and the estimate of true SU severity (IRT results).

Table 4.

Ability Scores by SUD Severity Diagnosis

Severity Diagnosis Descriptive Statistics
None −0.49 (0.02)
Mild 0.24 (0.04)
Moderate 0.51 (0.05)
Severe 1.10 (0.03)
Open in a new tab

Note. Values are means and standard errors (in parentheses).

A series of exploratory analyses tested how IRT ability scores and total scores were related to the supplementary items (Items 13–17) of the TCU DS 5. In line with how most of the sample did not have a diagnosed SUD, about half of the people in this study (47%) reported that no drug caused them the most problems in the past 12 months (Item 12). Marijuana was the drug most often (35%) identified as being the substance causing youth the most serious problem. Considering the uniformity in responses to Item 12, no additional analyses were conducted with this item. Correlational analyses examined the relationship between supplementary Items 13–17 and TCU DS 5 total scores and ability scores computed using IRT. All correlations were significant for total and ability scores except for the following items: 13c, 13d, 13f, 13g, 13k, 13o, and 14. Ability scores were significantly related to Item 13c (“cannabinoids—hashish”) and Item 13d (“synthetic marijuana”), ps ≤ .042, but total scores were not related to either, ps ≥ .063. The opposite pattern of results emerged for Item 13k (“bath salts”), which was not significantly related to ability scores, p = .141, but was significantly related to total scores, p = .048. See Table 5 for correlations between supplementary items and total scores.

Table 5.

Relationships Between Supplementary Items, Ability Estimates and Total Scores

Item # Ability Estimates (IRT) Total Scores (CTT)
13a 0.29** 0.19**
13b 0.18* 0.11*
13c 0.16* 0.11
13d 0.12* 0.10
13e 0.12* .012*
13f 0.06 0.06
13g 0.08 0.08
13h 0.14* 0.15*
13i 0.19** 0.20**
13j 0.15* 0.19**
13k 0.08 0.11*
13l 0.19** 0.17*
13m 0.12* 0.12*
13n 0.12* 0.11*
13o 0.08 0.08
13p 0.21** 0.21**
13q 0.12* 0.12*
13r 0.14* 0.14*
13s 0.13* 0.11*
14 0.05 0.03
15 0.34** 0.31**
16 0.11* 0.11*
17 0.33** 0.32**
Open in a new tab

Note. Numbers represent Pearson correlations. IRT = Item Response Theory; CTT = Classical Test Theory.

**

p ≤ .001,

*

p < .05.

Discussion

Youth involved with the juvenile justice system are an at-risk population for SU involvement, which contributes to impaired psychosocial functioning, interpersonal functioning, and increased risk for recidivism (Donenberg et al., 2015; Henggeler et al., 2002; Tapia et al., 2016). Consequently, validated assessment tools accurately measuring clients’ SU severity have an important role in pinpointing youth in need of SU treatment. The purpose of this study was to evaluate whether an IRT generated scoring method for the TCU DS 5 improved the accuracy in classifying SU severity as compared to the traditional scoring method using Classical Testing Theory. Results showed that while IRT provides valuable information about SU severity, it is likely too burdensome to be the standard scoring method. The TCU DS 5 was designed to be administered and manually scored by juvenile justice staff who may not be familiar with, or have the resources, to compute the IRT generated scoring method. The ability to quickly score the TCU DS 5 by summing up “Yes” responses provides an efficient means of identifying youth who are most at risk for SU difficulties at intake.

Item level analyses using IRT identified five items (Items 1, 2, 3, 9, and 11) as being potentially misfit. Item response theory results also showed that responses to Items 1, 2, and 3 were highly unpredictable whereas responses to Items 9 and 11 were highly predictable. While the PCA results verify that the items on the TCU DS 5 are measuring the same construct, the TCU DS 5 may benefit from adjusting the way these items are worded. Predictable items (9 and 11) could be replaced with more efficient items, although even items with very low mean-squared values may still be adding useful information (Martin-Löf, 1974), and therefore these items should be retained to maintain consistency with DSM-5. Additionally, the infit and outfit values for Items 9 and 11 may be affected by the unpredictable response patterns of Items 1, 2, and 3. It is recommended that items with very unpredictable response patterns be removed or re-worded (Martin-Löf, 1974). Note that since the mean-squared values for Items 1, 2, and 3 are less than 2.0, they are not too misfit that they could be distorting or degrading the instrument (Gustafsson, 1980). These items should therefore be re-worded but not removed entirely.

The IRT results also provided insight into how people with different SU severities respond to the individual items of the TCU DS 5. For example, Item 10 (tolerance criteria) had the lowest difficulty estimate, which means that tolerance is likely a common symptom for people with mild-to-moderate SUDs. Thus, if a client endorses this item but not the others, they would likely a good candidate for a preventative intervention that is designed preclude the further escalation of SU problems. In contrast, Items 5, 7, 8, and 9 had relatively high difficulty estimates. Endorsing these items suggests that the person likely has a severe SUD. Said differently, adolescents who report getting so high or sick from their drug use that they do not go to work or school (Item 5), spend less time at work, school, or with friends because of their drug use (Item 7), use drugs that put themselves or others in physical danger (Item 8), or continue to use drugs despite it causing physical or psychological problems (Item 9) would most likely benefit from a higher intensity SU treatment to adequately meet their needs.

Finally, total scores were evaluated in terms of their relationship with the supplementary Items 14–17. How many times a person has been in a SU treatment program (Item 14) was, as expected, not related to total scores for youth involved in this study. That is, regardless of how severe an adolescent’s SUD is, they may be less likely to have been in SU treatment because of their young age. This is consistent with literature showing that only 6.3% of adolescents between the ages of 12 to 17 in need of SUD treatment received clinical services (Lipari et al., 2016). Unexpectedly, Items 15 (“How serious do you think your drug problems are?”) and 17 (“How important is it for you to get drug treatment now?”) were related to total scores. Correspondingly, adolescents in this study appeared to have some level of awareness about how serious their SU problems were and the importance of receiving formalized treatment. Lastly, Item 16 (“During the last 12 months, how often did you inject drugs with a needle?”) was related to total scores. Collectively, only five youth reported that they injected drugs more than “never.” Of these five individuals, four had a severe SUD and one had no SUD. This converges with the notion that certain patterns of SU, such as injection drug use, are indicative of acute risk or harm that warrants immediate attention (Levy & Williams, 2016). Reporting any amount of injection drug use should therefore require a referral to intensive treatment, regardless of total SU severity scores.

Limitations and Future Directions

There are several limitations of this study that are worth mentioning. First, the sample consisted of all male juveniles, so the results of this study do not necessarily extend to females involved with the juvenile justice system or justice-involved adults. In addition, adolescents in the present study were incarcerated at the time the TCU DS 5 was administered, so the results may not generalize to people in a non-justice sample. The fact youth in this study were incarcerated may have also affected how respondents answered the questions. Put another way, respondents may have felt answering truthfully could negatively affect their supervision requirements, despite agency staff making it clear that this was not the case. The misfit of Items 1, 2, and 3 warrants caution when interpreting the item-level analyses presented herein given these misfit items may have been affecting the parameter estimates for the other items as well. Relatedly, the retrospective nature of the data set may have been affected by recall bias. The fact that these adolescents had SUDs may have introduced a systematic error wherein they do not accurately remember previous events or unintentionally omitted important details. If recall bias was present, it may have led to different results than if this study was conducted prospectively. Finally, the supplementary items were only analyzed using correlational analyses, so these results should not be interpreted as causal.

Future studies are needed to investigate whether the suggested changes to Items 1, 2, and 3 would improve fitness of the IRT model. Similarly, researchers may consider adding another scale item to determine veracity of statements and ensure respondents are carefully reading each question. This would be helpful in terms of providing correctional staff and treatment providers an easy way to identify which youth may not be taking the survey seriously. These youth should be followed up with using a more comprehensive diagnostic assessment/interview to accurately assess these clients’ SU severity. Finally, forthcoming studies may consider replicating these findings in a more diverse sample that includes female adolescents and adults involved with the justice system. Measurement invariance procedures could also be tested on the TCU DS 5 to ensure the scale is measuring the same construct in people belonging to different demographic backgrounds (i.e., male vs. females, White vs. Black Indigenous or People of Color, non-incarcerated vs. incarcerated).

In conclusion, for field applications, the traditional way of scoring the TCU DS 5 appears to be worth continuing given the accuracy of scoring results and ease of administration, but the IRT analysis presented in this study provide insights into the TCU DS 5 that would not have been apparent using the current scoring method alone. For example, it identified additional diagnostic items that were endorsed by people with severe SUDs. These items may serve as strong indicators that a person would potentially benefit from intensive SUD treatment. Agencies who administer the TCU DS 5 should be aware that adolescents who endorse these items are perhaps most at risk.

Table 2.

Average Ability Scores for Each Response Category of the TCU Drug Screen 5

Item # Response Option Ability Mean
1 No −2.74
Yes .99
2 No −2.65
Yes 1.22
3 No −2.82
Yes .86
4 No −2.81
Yes 1.49
5 No −2.52
Yes 2.21
6 No −2.74
Yes 1.46
7 No −2.63
Yes 1.88
8 No −2.49
Yes 2.16
9 No −2.65
Yes 2.00
10 No −3.04
Yes 1.18
11 No −2.72
Yes 2.07
Open in a new tab

Note. Ability scores reflect average total scores (relative to each item) of persons who responded with each rating scale category.

Acknowledgments

This project was funded by the National Institute on Drug Abuse (NIDA; grant R01DA013093).

Footnotes

We have no conflicts of interest to disclose.

References

  1. American Society of Addiction Medicine (ASAM). (2014). The ASAM standards of care for the addiction specialist physician. American Society of Addiction Medicine, Inc. Retrieved from https://www.asam.org/docs/default-source/publications/standards-of-care-final-design-document.pdf [Google Scholar]
  2. American Psychiatric Association. (2013). Diagnostic and Statistical Manual of Mental Disorders: DSM-5. Arlington, VA: American Psychiatric Association. 10.1176/appi.books.9780890425596 [DOI] [Google Scholar]
  3. Andrews DA, Bonta J, & Hoge RD (1990). Classification for effective rehabilitation: Rediscovering psychology. Criminal Justice and Behavior, 17, 19–52. 10.1177/0093854890017001004 [DOI] [Google Scholar]
  4. Andrews DA, Bonta J, & Wormith JS (2011). The risk-need-responsivity (RNR) model: Does adding the good lives model contribute to effective crime prevention? Criminal Justice and Behavior, 38(7), 735–755. 10.1177/0093854811406356 [DOI] [Google Scholar]
  5. Belenko S, Knight D, Wasserman GA, Dennis ML, Wiley T, Taxman FS,… Sales J (2017). The Juvenile Justice Behavioral Health Services Cascade: A new framework for measuring unmet substance use treatment services needs among adolescent offenders. Journal of Substance Abuse Treatment, 74, 80–91. 10.1016/j.jsat.2016.12.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Broome KM, Knight K, Hiller ML, & Simpson DD (1996). Drug treatment process indicators for probationers and prediction of recidivism. Journal of Substance Abuse Treatment, 13(6), 487–491. 10.1016/S0740-5472(96)00097-9 [DOI] [PubMed] [Google Scholar]
  7. Center for Behavioral Health and Quality. (2016). 2015 National Survey on Drug Use and Health: Detailed Tables. Rockville, MD: Substance Abuse and Mental Health Services Administration. [Google Scholar]
  8. Center for Substance Abuse Treatment. (2012). Screening and Assessing Adolescents for Substance Use Disorders. Treatment Improvement Protocol (TIP) Series, No. 31. HHS Publication No. (SMA) 12–4079. Rockville, MD: Substance Abuse and Mental Health Services Administration. [PubMed] [Google Scholar]
  9. Dennis ML, White MK, & Ives ML (2009). In Leukefeld CG, Gullotta TP, & Staton-Tindall M (Eds.), Issues in children’s and families’ lives. Adolescent substance abuse: Evidence-based approaches to prevention and treatment. Individual characteristics and needs associated with substance misuse of adolescents and young adults in addiction treatment (pp. 45–72). New York, NY: Springer Science + Business Media. [Google Scholar]
  10. Dennis ML, Smith CN, Belenko S, Knight D, McReynolds L, Rowan G, Dembo R, DiClemente R, Robertson A, & Wiley T (2019). Operationalizing a behavioral health services cascade of care model: Lessons learned from a 33-site implementation in juvenile justice community supervision. Federal Probation, 83(2), 52–64. Retrieved from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8341285/ [PMC free article] [PubMed] [Google Scholar]
  11. Donenberg GR, Emerson E, Mackesy-Amiti ME, & Udell W (2015). HIV-risk reduction with juvenile offender on probation. Journal of Child and Family Studies, 24(6), 1672–1674. 10.1007/s10826-014-9970-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Embretson SE, & Reise SP (2013). Item response theory. Psychology Press. 10.4324/9781410605269 [DOI] [Google Scholar]
  13. Farabee D, Shen H, Hser Y, Grella CE, & Anglin MD (2001). The effect of drug treatment on criminal behavior among adolescents in DATOS-A. Journal of Adolescent Research, 16, 679–696. 10.1177/0743558401166009 [DOI] [Google Scholar]
  14. Gustafsson JE (1980). Testing and obtaining fit of data to the Rasch model. British Journal of Mathematical and Statistical Psychology, 33(2), 205–233. 10.1111/j.2044-8317.1980.tb00609.x [DOI] [Google Scholar]
  15. Harris KM, Griffin BA, McCaffrey DF, & Morra AR (2007). Inconsistencies in self-reported drug use by adolescents in substance abuse treatment: Implications for outcomes and performance measurement. Journal of Substance Abuse Treatment, 34(3), 347–355. 10.1016/j.jsat.2007.05.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Henggeler SW, Clingempeel WG, Brondino MJ, & Pickrel SG (2002). Four-year follow-up of multisystemic therapy with substance-abusing and substance dependent juvenile offender. Journal of the American Academy of Child and Adolescent Psychiatry, 41(7), 868–874. 10.1097/00004583-200207000-00021 [DOI] [PubMed] [Google Scholar]
  17. IBM Corp. Released 2017. IBM SPSS Statistics for Windows, Version 25.0. Armonk, NY: IBM Corp. [Google Scholar]
  18. Kean JK, Brodke DS, Biber J, & Gross P (2018). An introduction to item response theory and Rasch analysis of the Eating Assessment Toll (EAT-10). Brain Impairment, 19(1), 91–102. 10.1017/BrImp.2017.31 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kean J, & Reilly J (2014). Classical test theory. Handbook for Clinical Research: Design, Statistics and Implementation, 192–194. [Google Scholar]
  20. Knight DK, Becan JE, Landrum B, Joe GW, & Flynn PM (2014). Screening and assessment tools for measuring adolescent client needs and functioning in substance abuse treatment. Substance Use & Misuse, 49(7), 902–918. 10.3109/10826084.2014.891617 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Knight DK, Belenko S, Wiley T, Robertson AA, Arrigona N, Dennis M, … & Leukefeld C (2015). Juvenile Justice—Translational Research on Interventions for Adolescents in the Legal System (JJ-TRIALS): a cluster randomized trial targeting system-wide improvement in substance use services. Implementation Science, 11(1), 1–18. 10.1186/s13012-016-0423-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Knight DK, Blue TR, Flynn PM, & Knight K (2018). The TCU drug screen 5: Identifying justice-involved individuals with substance use disorders. Journal of Offender Rehabilitation, 57(8), 525–537. 10.1080/10509674.2018.1549180 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Knight K, Simpson DD, & Morey JT (2000). An evaluation of the TCU Drug Screen, Final Report (NCJ 196682). Washington, DC: United States Department of Justice, National Institute of Justice. [Google Scholar]
  24. Levy SJ, & Williams JF (2016). Substance use screening, brief intervention, and referral to treatment. Pediatrics, 138(1), e20161211. 10.1542/peds.2016-1211 [DOI] [PubMed] [Google Scholar]
  25. Lipari RN, Park-Lee E, & Van Horn S (2016). The CBHSQ Report: American’s need for and receipt of substance use treatment in 2015. Rockville, MD: Center for Behavioral Health Statistics and Quality, Substance Abuse and Mental Health Services Administration. [PubMed] [Google Scholar]
  26. Linacre JM (2019). Winsteps® Rasch measurement computer program. Beaverton, Oregon: Winsteps.com [Google Scholar]
  27. Martin-Löf P (1974). The notion of redundancy and its use as a quantitative measure of the discrepancy between a statistical hypothesis and a set of observational data [with Discussion]. Scandinavian Journal of Statistics, 3–18. https://www.jstor.org/stable/4615544 [Google Scholar]
  28. Matsunaga M (2010). How to factor-analyze your data right: Do’s, don’ts, and how-to’s. International Journal of Psychological Research, 3(1), 97–110. [Google Scholar]
  29. Moberg DP, & Hahn L (1991). The Adolescent Drug Involvement Scale. Journal of Adolescent Chemical Dependency, 2(1), 75–88. 10.1300/J272v02n01_05 [DOI] [Google Scholar]
  30. Pilowsky DJ, & Wu LT (2013). Screening instruments for substance use and brief interventions targeting adolescents in primary care: A literature review. Addictive Behaviors, 38(5), 2146–2153. 10.1016/j.addbeh.2013.01.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. SAS Institute Inc. (2013). SAS® 9.4. Cary, NC: SAS Institute. [Google Scholar]
  32. Schwartz IM, Barton WH, & Orlando F (1991). Keeping kids out of secure detention: The misuse of juvenile detention has a profound impact on child welfare. Public Welfare, 49(2), 20–26. [Google Scholar]
  33. Seigle E, Walsh N, & Weber J (2014). Core principles for reducing recidivism and improving other outcomes for youth in the juvenile justice system. New York: Council of State Governments. [Google Scholar]
  34. Substance Abuse and Mental Health Services Administration (SAMHSA). (2014). Results from the 2013 National Survey on Drug Use and Health: Summary of National Findings, NSDUH Series H-48, HHS Publication No. (SMA) 14–4863. Rockville, MD: Substance Abuse and Mental Health Services Administration. [Google Scholar]
  35. Substance Abuse and Mental Health Services Administration (SAMHSA). (2016). Impact of the DSM-IV to DSM-5 Changes on the National Survey on Drug Use and Health [Internet]. Rockville, MD: Substance Abuse and Mental Health Services Administration. [PubMed] [Google Scholar]
  36. Tapia M, McCoy H, & Tucker L (2016). Suicidal ideation in juvenile arrestees: Exploring legal and temporal factors. Youth Violence and Juvenile Justice, 14(4), 468–483. 10.1177/1541204015579522 [DOI] [Google Scholar]
  37. Tarter R (1990). Evaluation and treatment of adolescent substance use: A decision tree method. American Journal of Drug and Alcohol Abuse, 16, 1–46. 10.3109/00952999009001570 [DOI] [PubMed] [Google Scholar]
  38. Vincent GM, Guy LS, & Grisso T (2012). Risk assessment in juvenile justice: A guidebook for implementation. Systems and Psychosocial Advances Research Center Publications and Presentations, 573. Retrieved from https://escholarship.umassmed.edu/psych_cmhsr/573 [Google Scholar]
  39. Wiese AL, Blue TR, Knight DK, & Knight K (2019). The validity of TCU Drug Screen 5 for identifying substance use disorders among justice-involved youth. Federal Probation, 83, 65–70. [Google Scholar]
  40. Winters KC, Tanner-Smith EE, Bresani E, & Meyers K (2014). Current advances in the treatment of adolescent drug use. Adolescent Health, Medicine and Therapeutics, 5, 199–210. 10.2147/AHMT.S48053 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Yang H, Chen F, Liu X, & Xin T (2019). An item response theory analysis of DSM-5 heroin use disorder in a clinical sample of Chinese adolescents. Frontiers in Psychology, 10, 2209. 10.3389/fpsyg.2019.02209 [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES