1 University of Pittsburgh School of Medicine, Pittsburgh,
2 Childrens Hospital of Pittsburgh, Pittsburgh, PA, USA and
3 Royal Melbourne Hospital, Victoria, Australia
Received 19 September 2001; in revised form 1 February 2002; accepted 17 February 2002
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Substance-use screening instruments specifically designed for adolescents are available (Leccese and Waldron, 1994; Blum, 1997
), but are generally too long and time-consuming to be practical in the hectic environment of a hospital ED. Shorter scales have been developed for adolescents, but there is scant research on such instruments among adolescents who report a range of alcohol involvements. Knight et al. (1999) developed a 6-item screening instrument (the CRAFFT) for problematic drug and/or alcohol use by extracting items from several other, lengthier adolescent self-report instruments. These investigators found that the instrument possessed adequate internal consistency. Bastiaens et al. (2000) recently reported on a 5-item instrument, the RAFFT, and found it to possess good sensitivity and specificity for diagnosing substance use disorders among adolescents seen in an emergency room or ambulatory care centre. However, a selection bias was operating in both of these studies, as many of the subjects had apparent symptoms of substance misuse or had been referred for treatment of substance misuse pathology.
In contrast to these instruments developed for adolescents, there is a variety of measures designed to screen for problematic alcohol use in adults that are relatively short, can be used in EDs, and have established reliability and validity. It seems reasonable to investigate the potential of these instruments for the early detection of problematic alcohol use among adolescents at high risk for injury and illness.
There are only a few reports on the utility of existing adult-validated alcohol screening instruments for use with adolescents in EDs. Shope et al. (1997) used the CAGE in an ED study of 263 injured adolescents (mean age 16.9 years), and found that CAGE scores correlated positively with age, alcohol misuse (negative consequences of drinking), and overall alcohol involvement. However, this study did not involve a comparison of the CAGE with other alcohol screening measures. Chung et al. (2000) described an ED study of adolescents 1319 years old that compared the same three screening instruments for problematic alcohol use that are used in the current study: (1) the Alcohol Use Disorders Identification Test (AUDIT, Saunders et al., 1993); (2) the TWEAK (Russell, 1994
); (3) the CAGE (Ewing, 1984
). Chung et al. (2000) used DSM-IV (American Psychiatric Association, 1994
)-defined alcohol use disorders as the criterion measure, and found that the AUDIT performed best in predicting problematic alcohol use, the TWEAK performed satisfactorily at a cut-off score of 2, and the CAGE was relatively inefficient in its sensitivity and specificity for alcohol use disorders among the adolescents. Chung et al. (2000) excluded alcohol-positive adolescents and focused on the criterion validity of the instruments. Our method includes investigation of internal consistency and construct validity and includes alcohol-positive adolescents. As a result, we are able to compare alcohol-positive with alcohol-negative adolescents and to assess the criterion validity of the screening instruments using a standard that was not available in the Chung et al. (2000) investigation.
The AUDIT was developed by the World Health Organization as a measure of alcohol consumption, alcohol dependence, and alcohol-related problems. The TWEAK (Russell, 1994) was developed to assess potentially hazardous drinking patterns among women and has demonstrated reliability and validity in these samples. The CAGE (Ewing, 1984
) focuses on behaviours consistent with alcohol dependence. The objectives of the present study were to determine the reliability (internal consistency) and validity of these instruments in this sample, and to determine which of these instruments worked best in screening adolescents for problematic alcohol use in an ED.
![]() |
SUBJECTS AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Medical staff members informed the research technicians of patients meeting the above criteria. Parents or guardians of those aged <18 years were approached first, and the research project was explained. Parents were told that they would not be informed of the study findings for their adolescent child. If they gave written consent for their child to participate, the adolescent was then approached to obtain his/her written, informed assent. Following enrolment, an alcohol breathalyser test was administered using the Alco-Sensor IIITM intoximeter (Intoximeters Inc., St Louis, MO, USA) to determine if the adolescent had been drinking within several hours of arrival at the ED. The medical staff were not informed of any results and there was no entry of this research data item into the medical record.
Following administration of the breathalyser test, a 29-item self-report questionnaire was administered in private to the adolescent. Although the questionnaire is in self-report format, the research technician remained with the subject to answer any questions or clarify issues related to completing the questionnaire. The alcohol screening items from the AUDIT, the TWEAK and the CAGE are embedded in the questionnaire. Participants were given a $10 gift certificate for their participation.
Sample
A total of 554 patients between the ages of 12 and 20 years were screened for eligibility. The total sample included 298 males and 253 females (data on gender is missing for three patients who were screened for eligibility but not enrolled in the study). A total of 416 patients (75%) were screened at the paediatric ED, and 138 (25%) were screened at the general hospital ED; 359 patients could be approached according to protocol guidelines (195 patients were under the age of consent and had no parent with them or were too ill or injured to approach); 282 (78.5%) patients agreed to participate. No statistically significant age, gender or racial differences were found between those who participated and those who refused to participate. One case was missing from this analysis (n = 358).
Of the 282 participants, 170 (60.3%) indicated that they never drank alcohol and nine (3.2%) provided inconsistent responses regarding their alcohol use. In all, 103 participants (36.5%) indicated that they used alcohol, i.e. more than a sip or taste on one occasion, and properly completed the alcohol screening questionnaire. These 103 alcohol-using participants (48 females, 55 males) are the focus of this report. There were 82 Caucasians (79.5%), 15 African-Americans (14.5%), five Asian-Americans (5%), and one Hispanic-American (1%) in the sample. The mean ± SD age of the sample was 17.5 ± 2.1 years (range 12.220.9).
Data analytical procedures
The scale properties of the three alcohol screening measures were determined through the use of the Reliabilities procedure in SPSS 8.0. This procedure was used to calculate Cronbach estimates of the inter-item reliabilities of the scales (Cronbach, 1951
). The procedure also computed item means and SD, a scale mean and SD, a correlation matrix among the items, an average inter-item correlation, the correlation between each item and a corrected item-total correlation. The Correlation procedure of SPSS 8.0 was used to calculate Pearson correlations among AUDIT, TWEAK and CAGE summative scales scores. Students t-test was used for group comparisons on the instruments. The assumption of homogeneity of mean differences was not met for some tests. For these tests, the t-statistic and degrees of freedom appropriate for an assumption of unequal variance are reported.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
|
|
CAGE.
Table 5 displays the CAGE items, their frequency distributions, and item-total correlations.
|
The estimate of internal consistency for the 4-item CAGE was = 0.66, which is satisfactory for such a short scale. The mean ± SD score on the CAGE was just 0.71 ± 1.03, reflecting the relative lack of alcohol problems as assessed by the CAGE in this young sample.
Concurrent validity of the screening measures
Concurrent validity of the instruments was assessed by correlations among the total scale scores. The AUDIT and TWEAK correlated at 0.83, the AUDIT and CAGE correlated at 0.43, and the TWEAK and CAGE correlated at 0.53. All three correlations were statistically significant (P < 0.001, two-tailed test).
Construct validity using known group differences on alcohol use
We investigated the construct validity of the instruments by comparing younger and older participants, males and females, and ethnic groups on the screening instruments. Epidemiological studies showed that older adolescents are generally more involved in problem drinking than younger adolescents (Barnes and Welte, 1986; Centers for Disease Control and Prevention, 1996
), that Caucasian adolescents are generally more involved in problem drinking than African-American adolescents (Rachal et al., 1982
; SAMHSA, 1997
) and that male adolescents are more involved in problem drinking than female adolescents (Jessor, 1987
; Windle, 1996
). Construct validity of these instruments for assessing problematic alcohol use among adolescents would be supported to the extent that these same known group differences are found on them.
Age differences. To test for age differences on the alcohol screening measures, adolescents in the sample were divided into two groups at the mean age of 17.5 years. There were 51 participants in the younger group and 52 in the older group. The groups differed in the expected direction on the TWEAK (younger group mean = 2.23 ± 1.5 vs older group mean = 3.10 ± 1.8; t = 2.1, df = 68, P < 0.04) and on the AUDIT (younger group mean = 5.02 ± 4.9 vs older group mean = 10.39 ± 8.0; t = 4.0, df = 82.7, P < 0.001). The groups did not differ significantly on the CAGE (younger group mean = 0.60 ± 0.85 vs older group mean = 0.84 ± 1.17; t = 1.2, df = 90, P = 0.22).
Gender differences. Males and females did not differ on the AUDIT or the CAGE. They did differ significantly only on the TWEAK (male mean = 3.9 ± 1.5 vs female mean = 2.1 ± 1.8; t = 2.8, df = 67, P < 0.009).
Ethnic/racial differences. Caucasian adolescents (n = 82) were compared to African American adolescents (n = 15) on each screening measure (other adolescents were not included). The two racial groups differed significantly only on the AUDIT, with Caucasian adolescents scoring higher than African-American adolescents (Caucasian mean = 8.56 ± 7.6 versus African-American mean = 5.07 ± 3.9; t = 2.7, df = 37, P < 0.02).
Criterion validity
Hazardous drinking.
In order to assess potentially hazardous drinking, the participants were assigned to groups based on whether they drank six or more drinks on one occasion less than once a month versus more frequently. Bohn et al. (1995) used a similar binge drinking criterion with adults, and defined hazardous drinking as drinking more than six drinks on one occasion weekly or more often. Given the younger age and greater inexperience with alcohol among adolescents, it seemed reasonable to set the criterion here at monthly or more often.
Based on the specified criterion, there were 73 non-hazardous drinkers and 30 hazardous drinkers in the AUDIT and CAGE analyses, and 42 non-hazardous drinkers and 28 hazardous drinkers in the TWEAK analysis. All three screening instruments significantly discriminated between potentially hazardous drinkers and non-hazardous drinkers (AUDIT, mean for hazardous drinkers = 15.9 ± 7.3 vs 4.3 ± 3.2 for non-hazardous drinkers, t = 8.4, df = 33.8, P < 0.001; TWEAK, mean for hazardous drinkers = 3.93 ± 1.5 vs 1.93 ± 1.3 for non-hazardous drinkers, t = 5.8, df = 68, P < 0.001; CAGE, mean for hazardous drinkers = 1.1 ± 1.3 vs 0.56 ± 0.85 for non-hazardous drinkers, t = 2.1, df = 39.6, P < 0.05). The criterion for hazardous drinking is also the third AUDIT item and may have inflated differences on this scale. Therefore, we removed this item and computed a 9-item AUDIT score to compare the groups. Hazardous drinkers were still higher on the AUDIT: mean = 14.3 ± 6.4 vs 3.8 ± 2.8 for non-hazardous drinkers (t = 8.7, df = 33.8, P < 0.001).
Alcohol use prior to presentation in the ED. Eleven participants (11%) scored positively on the alcohol breathalyser test. The alcohol breathalyser score (BrAC) represents a criterion for problematic alcohol use that is external to the scores the participants received on the screening instruments. The breathalyser-positive participants were compared to the breathalyser-negative participants on the screening instruments, and were found to differ significantly only on the AUDIT (BrAC positive mean ± SD: 15.6 ± 9.6 versus BrAC negative: 6.8 ± 6.2; t = 2.96, df = 11.04, P < 0.02; two cases were missing from this analysis, n = 101).
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Utilizing the same screening instruments, Chung et al. (2000) also found that the AUDIT performed best in screening for problematic alcohol use among adolescents treated in an emergency department. These investigators used DSM-defined alcohol use disorders as the criterion measure for assessing the performance of these instruments. However, Chung et al. (2000) studied only alcohol-negative adolescents. Our method included screening for alcohol use at the time of the participants presentation at the emergency department as an additional criterion for problematic alcohol use. Our findings on this criterion contribute additional evidence that the AUDIT performs best as a screening instrument among adolescents seen in emergency departments.
The TWEAK was too low on internal consistency to be considered an adequate measure of problematic alcohol use in the current sample. However, if the tolerance item is removed, internal consistency for the TWEAK increases to an = 0.62. Chung et al. (2000) pointed out that tolerance is quite variable and it is, therefore, difficult to develop a definition of tolerance that is appropriate for both adults and adolescents. These investigators concluded that an item that assesses high-volume drinking, e.g. frequency of drinking five or more drinks, may be preferable in this population to one that assesses tolerance.
The TWEAK was, however, the only instrument that differentiated male and female drinkers, with males scoring higher. Interestingly, the highest degree of variance between the genders on the TWEAK was on item 1, the tolerance item. Sixty-five per cent of the males who responded to this item indicated that they can hold six or more drinks without passing out, whereas only 39% of the females gave this response. However, many adolescents responded that they did not have enough experience with alcohol to know how to answer the tolerance item. Thus, the tolerance item may work well in discriminating between the genders, but only among those participants who know how to answer it.
The CAGE did not perform well as a screening measure for problematic alcohol use in our adolescent/young adult sample. This finding is consistent with that by Chung et al. (2000). Heck and Williams (1995) similarly found the CAGE to be a poor predictor of problem drinking in college students. They suggested that its emphasis on items indicative of alcohol dependence may have been responsible for its poor performance in their relatively young college sample. Fleming et al. (1991), on the other hand, found that the AUDIT had good internal consistency (
= 0.80) among college students, and that it correctly classified 78% of those diagnosed with alcohol misuse by DSM-III criteria. These investigators suggested that the AUDIT has advantages over the CAGE, because it assesses frequency of use and quantity and frequency of binge drinking along a continuum, which is an advantage for the early detection of drinking problems.
Our assessment of concurrent validity may have been affected by the use of some of the same items for constructing the various scales. Notwithstanding this, the analysis indicated that all the instruments were measuring alcohol misuse, but the varying sizes of the correlations suggest that the instruments are not measuring the same underlying construct(s). The high correlation between the AUDIT and the TWEAK may be related to increased variance in the scores, due to their larger scales (AUDIT range 140; TWEAK range 07). Furthermore, the AUDIT measures alcohol intake and the tolerance item on the TWEAK may be a proxy for intake, since it is based on number of drinks. In contrast, the CAGE focuses on items related to alcohol dependence and this may explain its lower correlation with the other instruments.
Cherpitel and Clark (1995) studied these instruments in an adult ED sample and found that a significantly higher percentage of African-American males scored positively on all three measures compared to African-American females. The small number of participants who reported any alcohol use in our study precluded testing for gender differences within the ethnic/racial groups, but we did find that African-Americans scored lower on the AUDIT. Consistent with our findings, Clements (1998) found that only the AUDIT differentiated between the races in a college sample, with Caucasians scoring higher than African-Americans. Our findings are also in line with a recent epidemiological study (Grant, 1997) that reported lower rates of alcohol misuse and dependence among African-Americans.
The present findings should not be generalized beyond adolescents treated in EDs. Regional variations in alcohol use which affect samples and results have been reported from one ED to another in adult samples (Cherpitel and Clark, 1995). However, it is not clear that geographic region impacts our findings, since Chung et al. (2000) reported similar findings in an ED-treated sample of adolescents from a different region of the country. The fact that the current sample was primarily Caucasian requires us to be cautious in applying our findings to African-American or other race participants.
It is expected that level of illness and injury will affect the sample in studies such as ours, and we were unable to approach as many patients at the adult ED due to serious illness or injury. It is possible that more seriously ill or injured adolescent patients are treated at adult facilities. As a result, the AUDIT, the TWEAK, and the CAGE may perform differently from what was found in the current study, if used with critically injured adolescents.
Limitations notwithstanding, we were able to examine the performance of three widely used alcohol screening instruments in a sample of ED-treated adolescents. The success of prevention and treatment efforts relies on the early identification of substance use problems (Winters et al., 1993). This is a population that is at high risk for problematic alcohol use and a setting in which adolescents are likely to present first for treatment of alcohol-related illness or injury. Therefore, the potential for early detection of problematic alcohol use among ED-treated adolescents is great and this can have a positive impact on prevention and treatment efforts.
Experienced investigators have also noted that the population under investigation is an important consideration in choosing the appropriate instrument (Maisto et al., 1995; Cherpitel, 1998
). The higher internal consistency for the AUDIT, as compared to the TWEAK and CAGE, as well as its more consistent performance in differentiating alcohol-using adolescents along known group lines and in discriminating between problematic and non-problematic alcohol users, lead us to conclude that the AUDIT performs best as an alcohol screening instrument among adolescents treated in EDs. The consistency of findings between our study and others, especially those from other EDs and studies of samples close in age to our subjects (e.g. college students) adds to the validity of our findings and supports this conclusion.
![]() |
ACKNOWLEDGEMENTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
FOOTNOTES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Arria, A. M., Dohey, M. A., Mezzich, A. C., Bukstein, O. G. and Van Theil, D. H. (1995) Self-reported health problems and physical symptomatology in adolescent alcohol abusers. Journal of Adolescent Health 16, 226231.[ISI][Medline]
Barnes, G. M. and Welte, J. W. (1986) Adolescent alcohol abuse: subgroup differences and relationships to other problem behaviors. Journal of Adolescent Research 1, 7994.
Bastiaens L., Francis, G. and Lewis, K. (2000) The RAFFT as a screening tool for adolescent substance use disorders. American Journal of Addictions 9, 1016.[ISI][Medline]
Blum, R. (1997) Adolescent substance use and abuse. Archives of Pediatric and Adolescent Medicine 151, 805808.[ISI][Medline]
Bohn, M. J., Babor, T. F. and Kranzler, H. R. (1995) The alcohol use disorders identification test (AUDIT): validation of a screening instrument for use in medical settings. Journal of Studies on Alcohol 56, 423432.[ISI][Medline]
Centers for Disease Control and Prevention (1996) CDC Surveillance Summaries. Youth Risk Behavior Surveillance United States, 1995. September 27, 1996. MMWR Morbidity and Mortality Weekly Report 1996, 45 (No. SS-4).
Cherpitel, C. J. (1998) Performance of screening instruments for identifying alcohol dependence in the general population, compared with clinical populations. Alcoholism: Clinical and Experimental Research 22, 13991404.[ISI][Medline]
Cherpitel, C. J. and Clark, W. B. (1995) Ethnic differences in performance of screening instruments for identifying harmful drinking and alcohol dependence in the emergency room. Alcoholism: Clinical and Experimental Research 19, 628634.[ISI][Medline]
Chung, T., Colby, S., Barnett, N., Rohsenow, D., Spirito, A., and Monti, P. (2000) Screening adolescents for problem drinking: performance of brief screens against DSM-IV alcohol diagnoses. Journal of Studies on Alcohol 61, 579587.[ISI][Medline]
Clark, D. B., Lesnick, L. A. and Hegedus, A. M. (1997) Traumas and other adverse life events in adolescents with alcohol abuse and dependence. Journal of the American Academy of Child and Adolescent Psychiatry 36, 17441751.[ISI][Medline]
Clements, R. (1998) A critical evaluation of several alcohol screening instruments using the CIDI-SAM as a criterion measure. Alcoholism: Clinical and Experimental Research 22, 985993.[ISI][Medline]
Conigrave, K. M., Hall, W. D. and Saunders, J. B. (1995) The AUDIT questionnaire: choosing a cut-off score. Addiction 90, 13491356.[ISI][Medline]
Cronbach, L. J. (1951) Coefficient alpha and the internal structure of tests. Psychometrika 16, 297334.[ISI]
Ewing, J. A. (1984) Detecting alcoholism the CAGE questionnaire. Journal of the American Medical Association 252, 19051907.[Abstract]
Fleming, M. F., Barry, K. L. and MacDonald, R. (1991) The Alcohol Use Disorders Identification Test (AUDIT) in a college sample. International Journal of the Addictions 26, 11731185.[ISI][Medline]
Grant, B. (1997) Prevalence and correlates of alcohol use and DSM-IV alcohol dependence in the United States: results of the National Longitudinal Alcohol Epidemiologic Survey. Journal of Studies on Alcohol 58, 464473.[ISI][Medline]
Hansell, S. and White, H. W. (1991) Adolescent drug use, psychological distress and physical symptoms. Journal of Health and Social Behavior 32, 288301.[ISI][Medline]
Heck, E. J. and Williams, M. D. (1995) Using the CAGE to screen for drinking-related problems in college students. Journal of Studies on Alcohol 56, 282286.[ISI][Medline]
Jennet, B. and Bond, M. (1975) Assessment of outcome after severe brain damage. Lancet i, 480484.
Jessor, R. (1987) Problem-behaviour theory, psychosocial development, and adolescent problem drinking. British Journal of Addiction 82, 331342.[ISI][Medline]
Kelly, T. M., Donovan, J. E., Lynch, K. G. and Clark, D. B. (2001) Alcohol use disorders and other risk factors for adolescent suicidal ideation and attempts. Suicide and Life-Threatening Behavior 31, 181193.[ISI][Medline]
Knight, J. R., Shrier, L. A., Bravender, T. D., Farrell, M., Vander Bilt, J. and Shaffer, H. J. (1999) A new brief screen for adolescent substance abuse. Archives of Pediatric and Adolescent Medicine 153, 591596.
Leccese, M. and Waldron, H. B. (1994) Assessing adolescent substance use: a critique of current measurement instruments. Journal of Substance Abuse Treatment 11, 553563.[ISI][Medline]
Maisto, S. A., Connors, G. J. and Allen, J. P. (1995) Contrasting self-report screens for alcohol problems: a review. Alcoholism: Clinical and Experimental Research 19, 15101516.[ISI][Medline]
Nilssen, O., Ries, R. K., Rivara, F. P., Gurney, J. G. and Jurkovich, G. J. (1994) The CAGE questionnaire and The Short Michigan Alcohol Screening Test in trauma patients: comparison of their correlations with biological alcohol markers. Journal of Trauma 36, 784788.[ISI][Medline]
Rachal, J. V., Guess, L. L., Hubbard, R. L. and Maisto, S. A. (1982) Facts for planning no. 4: alcohol misuse by adolescents. Alcohol Health and Research World 6, 6168.
Russell, M. (1994) New assessment tools for risk drinking during pregnancy: T-ACE, TWEAK, and others. Alcohol Health and Research World 18, 5561.
SAMHSA (1997) National Household Survey on Drug Abuse Main Findings 1995. US Department of Health and Human Services, Rockville, MD.
Saunders, J. B., Aasland, O. G., Babor, T. F., de la Fuente, J. R. and Grant, M. (1993) Development of the alcohol use disorders identification test (AUDIT): WHO collaborative project on early detection of persons with harmful alcohol consumption II. Addiction 88, 791804.[ISI][Medline]
Shope, J. T., Copeland, L. A., Male, R. F., Blow, F. C. and Brockmann, L. M. (1997) Correlates of alcohol involvement among injured adolescents (abstract). Alcoholism: Clinical and Experimental Research 21 (Suppl. to no. 3), 29A.
Spain, D. A., Boaz, P. W., Davidson, D. J., Miller, F. B., Carrillo, E. H. and Richardson, J. D. (1997) Risk-taking behaviors among adolescent trauma patients. Journal of Trauma: Injury, Infection and Critical Care 43, 423426.[ISI]
Spirito, A., Lewander, W. J., Levy, S., Kurkjian, J. and Fritz, G. (1994) Emergency department assessment of adolescent suicide attempters: factors related to short-term follow-up outcome. Pediatric Emergency Care, 10, 612.[ISI][Medline]
Windle, M. (1996) An alcohol involvement typology for adolescents: convergent validity and longitudinal stability. Journal of Studies on Alcohol 57, 627637.[ISI][Medline]
Winters, K. C., Stinchfield, R. D. and Henly, G. A. (1993) Further validation of new scales measuring adolescent alcohol and other drug abuse. Journal of Studies on Alcohol 54, 534541.[ISI][Medline]