Psychometric properties of the short version of the children of alcoholics screening test (CAST-6) among Swedish adolescents

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=ipsc20

Nordic Journal of Psychiatry

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/ipsc20

Psychometric properties of the short version of the children of alcoholics screening test (CAST-6) among Swedish adolescents

Tobias H. Elgán , Anne H. Berman , Nitya Jayaram-Lindström , Anders Hammarberg , Camilla Jalling & Håkan Källmén

To cite this article: Tobias H. Elgán , Anne H. Berman , Nitya Jayaram-Lindström , Anders

Hammarberg , Camilla Jalling & Håkan Källmén (2021) Psychometric properties of the short version of the children of alcoholics screening test (CAST-6) among Swedish adolescents, Nordic Journal of Psychiatry, 75:2, 155-158, DOI: 10.1080/08039488.2020.1812000

To link to this article: https://doi.org/10.1080/08039488.2020.1812000

© 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group

Published online: 27 Aug 2020.

Submit your article to this journal Article views: 432

View related articles View Crossmark data

Citing articles: 1 View citing articles

BRIEF REPORT

Psychometric properties of the short version of the children of alcoholics screening test (CAST-6) among Swedish adolescents

Tobias H. Elg
an

, Anne H. Berman

, Nitya Jayaram-Lindstr€om

, Anders Hammarberg

, Camilla Jalling

and Håkan K€allm
en

a

STAD (Stockholm Prevents Alcohol and Drug Problems), Stockholm, Sweden;

Centre for Psychiatry Research, Department of Clinical Neuroscience, Karolinska Institutet, & Stockholm Health Care Services, Region Stockholm, Stockholm, Sweden;

Department of Psychology, Uppsala University, Uppsala, Sweden;

National Board of Forensic Medicine, Huddinge, Sweden

ABSTRACT

Aims: The Children of Alcoholics Screening Test (CAST-6) is a brief screening instrument developed to identify children with parents having problematic alcohol use. The aim of this study was to evaluate the psychometric properties of the CAST-6 among adolescents aged 15 –18 years, and also to identify an optimal cut-off score for this age group.

Methods: A total of 3000 15 to18 year-olds were randomly selected from a register of postal addresses in Sweden. An invitation letter, including access information to the electronic questionnaire, was sent out by regular mail and 1450 adolescents responded with baseline data. Test-retest reliability within a 2 –3-week period was calculated based on the 111 respondents who answered the same questionnaire twice. To determine an optimal cut-off score, a small treatment-seeking sample ( n ¼ 22) was recruited from a support group agency to be used as a reference group.

Results: The six items of the CAST-6 screening test loaded onto one latent factor with good internal consistency (alpha ¼ 0.88), and excellent test-retest reliability (ICC ¼ 0.93, 95% CI 0.90–0.95). The opti- mal cut-off score among adolescents was 2 points with a sensitivity of 55% and specificity of 79%

(AUROC ¼ 0.71, 95% CI 0.58–0.83).

Conclusions: The CAST-6 has good to excellent psychometric properties among adolescents. The iden- tified optimal cut-off score of 2 points should be treated with caution due to study limitations. The CAST-6 can be used in various settings to identify a vulnerable at-risk group of children and adoles- cents that may be in need of support.

ARTICLE HISTORY Received 7 February 2020 Revised 9 July 2020 Accepted 13 August 2020 KEYWORDS

Parents with problematic alcohol use; factor structure;

internal consistency; test- retest reliability; optimal cut-off

Introduction

Estimates on the prevalence of children with parents having problematic alcohol use reveal that this is a widespread problem. For instance, recent estimates from the US show a figure of 23 percent [1], while figures from Denmark and Sweden reveal estimates ranging from 3.9 to 20.1 percent [2 –5 ]. The reason for this variation mainly depends on how alcohol problems among parents have been defined. For instance, studies reporting lower prevalence typically define alcohol problems more on the basis of dependency criteria as opposed to risky use.

Children growing up with parental alcohol problems have an elevated risk for several adverse health outcomes, includ- ing early alcohol use [6 –11 ]. It is therefore important to have validated and reliable screening instruments available, in order to identify children at risk so that they can be offered timely support. For this purpose, single questions [12] and more extended questionnaires such as the 13-item short ver- sion of the modified Michigan Alcoholism Screening Test (M/

F-SMAST) [13] have been utilized. Another available screen- ing questionnaire is the six-item Children of Alcoholics Screening Test (CAST-6, Table 1) [14]. The CAST-6 was devel- oped from the original 30-item CAST instrument using three distinct adult samples: substance use disorder outpatients, psychiatric outpatients and medical students [15]. The CAST- 6 has shown high internal consistency (Cronbach ’s alpha ¼ 0.86 –0.92), and concurrent validity (r ¼ 0.93) when compared to the original CAST among adults [12,14], and good test- retest reliability ( r ¼ 0.78), albeit within a large time frame of one year, when rated by 6th and 7th graders (i.e. 11 to 13 year-olds) [16]. Two alternative cut-off scores have been proposed, one a more inclusive cut-off at 2 points and another more conservative one at 3 points [14,16,17].

The CAST-6 has the potential to play an important role for identifying children in need of support. To our know- ledge, the psychometric properties of the CAST-6 among adolescents have not been explored to a greater extent.

Research is particularly scarce among children in late adoles- cence (i.e. 15 to 19 year-olds) when cognitive development

CONTACT Tobias H. Elg
an tobias.elgan@ki.se STAD, Centre for Psychiatry Research, Norra Stationsgatan 69, SE-113 64, Stockholm, Sweden ß 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group

This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way.

2021, VOL. 75, NO. 2, 155 –158

https://doi.org/10.1080/08039488.2020.1812000

related to decision making, memory, and emotional reactions is at a peak [18]. The overall aim of the current study was to explore the psychometric properties of the CAST-6 among children in their late adolescence aged 15 –18 years. More specifically, this study examined the factor structure, internal consistency, test-retest reliability of the CAST-6 up to three weeks, and also identified an optimal cut-off score for adolescents.

Methods

Participants and procedure

A total of 3000 adolescents aged 15 –18 years were randomly selected from a register of all persons having a postal address in Sweden. Data was collected in October through December 2015 by a marketing company who invited indi- viduals via regular post. The invitation letter included access information to the electronic questionnaire along with an individual password. Two reminders were sent, with a two- week interval, to those who did not respond, yielding a total of 1450 respondents. Test-retest reliability was calculated based on those 111 respondents who consented and answered the same questionnaire a second time, two to three weeks after their first response. To calculate optimal cut-off scores, a research assistant recruited a small treat- ment-seeking sample from a support group agency in Stockholm, Sweden, that targets children who have parents with substance use problems. We intended to have 50 indi- viduals in this sample but succeeded in recruiting only 22.

Typically, children are referred to this agency by parental request or via referral by social services. The electronic ques- tionnaire contained background questions such as the respondent ’s age, sex, and country of birth, along with the CAST-6 (Table 1) and some additional screening instruments

for own substance use, of which the results have been pre- sented elsewhere [19]. Each item in the CAST-6 could be answered by a ‘Yes’ (1 point) or a ‘No’ (0 point). The total score of the CAST-6 was achieved by adding the points from each item together. The Regional Ethical Review Board in Stockholm approved this study (2015/391-31/4).

Statistics

Construct validity was examined using exploratory principal axis factor analysis (EFA). Internal reliability was calculated using Cronbach ’s alpha and test-retest reliability by intraclass correlation coefficients (ICC), using the recommended value of below 0.50 as poor, 0.50 –0.74 as moderate, 0.75–0.90 as good, and above 0.90 as excellent [20]. Concurrent validity and the optimal cut-off score were calculated using receiver operator characteristics (ROC). Youden ’s index was used to indicate the optimum cut-off score [21]. All analyses were conducted using SPSS (version 25).

Results

The mean age among the general population and support group samples was 16.7 (SD ¼ 1.2) and 16.2 (SD ¼ 1.2) years, respectively. Using the more conservative cut-off score of 3 points, approximately 17 percent of the girls and 12 percent of the boys in the general population sample had CAST-6 scores indicating that at least one parent had alcohol prob- lems (Table 2; using more inclusive cut-off score, these fig- ures were 25% and 17%). The CAST-6 scores between sexes were significantly different ( U ¼ 223824, p < .001). Eight indi- viduals in the support group sample had a CAST-6 score of zero. The mean CAST-6 score among the general population sample was 0.9, while the same value was 2.9 in the support group sample.

Construct validity

All six CAST-6 items loaded into one factor (Table 3) with an eigenvalue of 3.8, explaining 63.6% of the co-variance between items. The six items loaded into one factor for both boys and girls, with similar factor loadings (data not shown).

Items 3 and 4, that relate to a particular situation when a parent was drinking or was drunk, had the lowest fac- tor loadings.

Table 1. The six CAST-6 items.

1. Have you ever thought that one of your parents had a drinking problem?

Yes/No

2. Did you ever encourage one of your parents to quit drinking? Yes/No 3. Did you ever argue or fight with a parent when he or she was drinking?

Yes/No

4. Have you ever heard your parents fight when one of them was drunk?

Yes/No

5. Did you ever feel like hiding or emptying a parent ’s bottle of liquor?

Yes/No

6. Did you ever wish that a parent would stop drinking? Yes/No

Table 2. Descriptive statistics for CAST-6 in the general population and support group samples.

CAST-6 n Mean SD Median IQR Proportion having parents with alcohol problems

General population sample

Girls 773 1.1 1.8 0 1 17.3 (14.8 –20.1)

24.5 (21.5 –27.6)

Boys 647 0.7 1.6 0 1 12.1 (9.7 –14.7)

16.8 (14.1 –19.9)

Support group sample 22 2.9 2.7 2.0 6 45.5 (24.4 –67.8)

54.5 (32.2 –75.6)

Mean, standard deviation (SD), median, inter quartile range (IQR) and the prevalence (95% Confidence Interval) of the perception of ever having a parent with a drinking problem.

a

Calculated based on the more conservative CAST-6 cut-off score of 3 points.

b

Calculated based on the more inclusive CAST-6 cut-off score of 2 points.

c

Data on gender missing for n ¼ 30 respondents.

156 T. H. ELGÁN ET AL.

Internal consistency and test-retest reliability

The Cronbach ’s alpha value of the CAST-6 was 0.88 (same for both sexes) indicating a good internal consistency. The ICC used as a measure of test-retest reliability revealed an excel- lent level (0.93, 95% CI 0.90 –0.95) overall and for both boys (0.96, 95% CI 0.93 –0.98, n ¼ 48) and girls (0.92, 95% CI 0.86 –0.95, n ¼ 61).

Concurrent validity and optimal screening threshold To evaluate the optimal screening threshold for the CAST-6, a ROC analysis was conducted (Table 4, Figure 1(A)). Results showed that the ability of the CAST-6 to discriminate between those with and without a parent having problem- atic alcohol use was fair (AUROC ¼ 0.71, 95% CI 0.58–0.83).

With the aim of achieving a balance between sensitivity and specificity, the optimal cut-off score was 2 points with a sen- sitivity of 55% and a specificity of 79%. The same analysis was conducted excluding the eight individuals who scored zero on the CAST-6, resulting in the same optimal cut-off

score of two, but with a sensitivity of 86% and a specificity of 79% (AUROC ¼ 0.91, 95% CI 0.86–0.96).

Discussion

The aim of the study was to evaluate the factor structure, internal consistency, test-retest reliability of the CAST-6, and to also identify an optimal cut-off score for the instrument among adolescents. Results demonstrate good psychometric properties for the instrument. The CAST-6 loaded on one latent factor and had good internal consistency and excellent test-retest reliability, corroborating earlier findings among both adults [12,14] and adolescents [16]. These results dem- onstrate that the CAST-6 performs well among the adoles- cent population.

We found the more inclusive cut-off score of 2 points to be optimal among 15 –18 year-olds, as opposed to the origin- ally recommended conservative score of 3 points for adults [14,17]. This result was also confirmed in an analysis where the eight individuals in the reference group who scored zero on the CAST-6 were excluded, thus showing additional evi- dence for the stability of our suggested cut-off score. The conservative cut-off score of 3 points has been recom- mended among the younger population of 6

and 7

graders [16]. However, the inclusive cut-off score of 2 points has been identified as optimal within an adult psychiatric patient group

, and has also been found to perform better than the conservative cut-off score in a group of adults in substance abuse treatment [12]. Despite lacking consensus at this point between the different studies and between differ- ent populations, an inclusive cut-off score of 2 shows utility in the adolescent population.

We also found that more girls compared to boys had scores indicating that a parent had alcohol problems. This has been observed in previous studies [16] and suggests that it may be relevant to propose differing sex-specific cut-off scores.

However, our reference group sample included only four boys and we could therefore not determine sex-specific cut-off scores. Future studies should investigate if there are differences between genders with regard to optimal cut-off scores.

A major limitation with this study is the small reference group from the support group facility offering an extensive program for adolescents having parents with substance use problems. The sample comprised only 22 individuals of which eight had a CAST-6 score of zero. Given the setting, it is unlikely that these adolescents did not have a parent with substance use problems. Possible explanations for the zero score might be that the adolescent no longer lives with the problem parent, or has never lived with them; this may have

Table 3. Exploratory factor analysis of the six CAST-6 items, using principal axis factoring ( n ¼ 1450).

Items Factor loading

1. Have you ever thought that one of your parents had a drinking problem? 0.82 2. Did you ever encourage one of your parents to quit drinking? 0.76 3. Did you ever argue or fight with a parent when he or she was drinking? 0.71 4. Have you ever heard your parents fight when one of them was drunk? 0.63 5. Did you ever feel like hiding or emptying a parent ’s bottle of liquor? 0.76

6. Did you ever wish that a parent would stop drinking? 0.83

Table 4. Sensitivity, specificity and Youden ’s index for CAST-6 among adoles- cents ( n ¼ 1450) using an adolescent support group sample as criter- ion ( n ¼ 22).

Sensitivity Specificity Youden ’s index

1 0.64 0.69 0.33

2 0.55 0.79 0.34

3 0.46 0.85 0.31

4 0.46 0.88 0.34

5 0.46 0.92 0.38

6 0.32 0.95 0.27

0 0,2 0,4 0,6 0,8 1

0 0,2 0,4 0,6 0,8 1

Se nsi vi ty

1-Specificity

Figure 1. Receiver operator characteristic (ROC) curve of CAST-6 to determine discriminatory ability among adolescents in the general population ( n ¼ 1450).

The discriminatory ability was determined using an adolescent support group

sample as the reference criterion ( n ¼ 22).

created a distance to the parent and the problem, such that it was difficult to even relate to the CAST-6 items. Another possible explanation might be that since the CAST-6 items explicitly ask about alcohol and related problems, it could be that the parent has substance use problems other than alco- hol, that were not referred to in the CAST-6 items. However, this is merely a speculation since we have no information about this. The parent might also have a poly-drug use prob- lem where alcohol is a less pronounced problem compared to the other substances. A final limitation of this study is the rather modest response rate, which implies a possible selec- tion bias in the general population sample.

Conclusion

Our study demonstrates good to excellent psychometric properties of the CAST-6 and identifies an optimal cut-off score of 2 points among adolescents. The suggested cut-off score should be treated with caution due to study limita- tions. The CAST-6 can potentially play an important role in various settings, such as the school, social services, the child and adolescent psychiatric care and clinics for substance use- related problems, to identify this vulnerable at-risk group of children that may be in need of support.

Acknowledgements

The authors are grateful for the work conducted by Maria Thell in data collection for the support group sample.

Author contributions

THE, CJ, AHB, and HK planned the study, applied for funding and con- tributed to the collection of data. All authors contributed to the plan- ning of statistical analyses. HK and THE conducted the statistical analyses. All authors contributed to the interpretation of results. THE wrote the first draft of the manuscript. All authors contributed signifi- cantly to drafting the manuscript and responses to reviewers.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Funding

This work was supported by the Alcohol Research Council of the Swedish Alcohol Retailing Monopoly under Grants FO2014-0025, FO2015-0041, and FO2016-0065.

ORCID

Tobias H. Elg
an http://orcid.org/0000-0001-6406-4685 Anne H. Berman http://orcid.org/0000-0002-7709-0230 Nitya Jayaram-Lindstr€om http://orcid.org/0000-0002-2678-3782 Anders Hammarberg http://orcid.org/0000-0002-9976-5227 Camilla Jalling http://orcid.org/0000-0001-5905-2110 Håkan K€allm
en http://orcid.org/0000-0001-9288-3864

References

[1] Thapa S, Selya AS, Jonk Y. Time-varying effects of parental alco- holism on depression. Prev Chronic Dis. 2017;14:E136.

[2] Bendtsen P, Damsgaard MT, Tolstrup JS, et al. Adolescent alcohol use reflects community-level alcohol consumption irrespective of parental drinking. J Adolesc Health. 2013;53(3):368 –373.

[3] Pisinger VSC, Hawton K, Tolstrup JS. Self-injury and suicide behavior among young people with perceived parental alcohol problems in Denmark: a school-based survey. Eur Child Adolesc Psychiatry. 2018;27(2):201 –208.

[4] Elg
an TH, Leifman H. Prevalence of adolescents who perceive their parents to have alcohol problems: a Swedish national sur- vey using a web panel. Scand J Public Health. 2013;41(7):

680 –683.

[5] Raninen J, Elg
an TH, Sundin E, et al. Prevalence of children whose parents have a substance use disorder: Findings from a Swedish general population survey. Scand J Public Health. 2016;44(1):

14 –17.

[6] Dube SR, Anda RF, Felitti VJ, et al. Growing up with parental alco- hol abuse: exposure to childhood abuse, neglect, and household dysfunction. Child Abuse Negl. 2001;25(12):1627 –1640.

[7] Johnson JL, Leff M. Children of substance abusers: overview of research findings. Pediatrics. 1999;103(5 Pt 2):1085 –1099.

[8] Rossow I, Felix L, Keating P, et al. Parental drinking and adverse outcomes in children: a scoping review of cohort studies. Drug Alcohol Rev. 2016;35(4):397 –405.

[9] Anda RF, Whitfield CL, Felitti VJ, et al. Adverse childhood experi- ences, alcoholic parents, and later risk of alcoholism and depres- sion. Psychiatr Serv. 2002;53(8):1001 –1009.

[10] Lieberman DZ. Children of alcoholics: an update. Curr Opin Pediatr. 2000;12(4):336 –340.

[11] Casswell S, You RQ, Huckle T. Alcohol ’s harm to others: reduced wellbeing and health status for those with heavy drinkers in their lives. Addiction. 2011;106(6):1087 –1094.

[12] Hodgins DC, Shimp L. Identifying adult children of alcoholics:

methodological review and a comparison of the CAST-6 with other methods. Addiction. 1995;90(2):255 –267.

[13] Sher KJ, Descutner C. Reports of paternal alcoholism: reliability across siblings. Addict Behav. 1986;11(1):25 –30.

[14] Hodgins DC, Maticka-Tyndale E, el-Guebaly N, et al. The cast-6:

development of a short-form of the Children of Alcoholics Screening Test. Addict Behav. 1993;18(3):337 –345.

[15] Jones JW. The children of alcoholics screening test and test man- ual. Chicago, IL: Camelot Unlimited; 1983.

[16] Havey JM, Dodd DK. Classifying COAs with three variations of the CAST: classification rates, stability, and gender differences. Addict Behav. 1995;20(4):501 –507.

[17] Hodgins DC, Maticka-Tyndale E, el-Guebaly N, et al. Alternative cut-point scores for the CAST-6. Addict Behav. 1995;20(2):

267 –270.

[18] Bundy DAP, de Silva N, Horton S, Disease Control Priorities-3 Child and Adolescent Health and Development Authors Group, et al. Investment in child and adolescent health and develop- ment: key messages from Disease Control Priorities, 3rd Edition.

Lancet. 2018;391(10121):687 –699.

[19] K €allm
en H, Berman AH, Jayaram-Lindstr€om N, et al. Psychometric properties of the AUDIT, AUDIT-C, CRAFFT and ASSIST-Y among Swedish adolescents. Eur Addict Res. 2019;25(2):68 –77.

[20] Koo TK, Li MY. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med.

2016;15(2):155 –163.

[21] Akobeng AK. Understanding diagnostic tests 3: receiver operating characteristic curves. Acta Paediatr. 2007;96(5):644 –647.

158 T. H. ELGÁN ET AL.