Reliability and Validity of the Alcohol Use Disorders Identification Test-Consumption in Screening for Adults with Alcohol Use Disorders and Risky Drinking In Japan

Excessive alcohol consumption is associated with not only multiple health problems including cancer but also many social problems such as drunken driving, child abuse, and domestic violence; it also leads to substantial social expense (WHO, 2009; Osaki et al., 2012; de Menezes et al., 2013; Yaegashi et al., 2014). Screening tests have been proposed to identify alcohol use disorders. The Alcohol Use Disorders Identification Test (AUDIT) is one of the popular screening tests used in Western countries, and an intervention program based on the results of this test has been introduced (Babor et al., 2001). The test consists of 10 questions, and each response for the questions is assigned a specific score. The total AUDIT


Introduction
Excessive alcohol consumption is associated with not only multiple health problems including cancer but also many social problems such as drunken driving, child abuse, and domestic violence; it also leads to substantial social expense (WHO, 2009;Osaki et al., 2012;de Menezes et al., 2013;Yaegashi et al., 2014).Screening tests have been proposed to identify alcohol use disorders.The Alcohol Use Disorders Identification Test (AUDIT) is one of the popular screening tests used in Western countries, and an intervention program based on the results of this test has been introduced (Babor et al., 2001).
The test consists of 10 questions, and each response for the questions is assigned a specific score.The total AUDIT

Reliability and Validity of the Alcohol Use Disorders
Identification Test -Consumption in Screening for Adults with Alcohol Use Disorders and Risky Drinking In Japan Yoneatsu Osaki 1 *, Aro Ino 2, Sachio Matsushita 3 , Susumu Higuchi 3 , Yoko Kondo 1 , Aya Kinjo 1 score is calculated by adding up the scores from all of the questions.Based on the AUDIT score, respondents are classified into categories of alcohol use disorders (Babor et al., 2001).
The AUDIT was developed as a self-reported tool; however, in general health care settings, a substantial amount of time may be required to respond to all of the questions.Owing to the attention placed on social problems relating to alcohol use disorders globally, the Japanese government has recently started to address alcohol problems.In Japan, a brief intervention to reduce alcohol consumption in people with an alcohol use disorder at clinics or during health examinations has been expected in recent years.Therefore, a short screening test is required by busy medical facilities in Japan.
The AUDIT Consumption (AUDIT-C) includes the first 3 questions of the AUDIT and is proposed and used in Western countries.The reliability and validity of the AUDIT-C have been determined for use in Western countries.To detect problem drinking, the proposed cutoffs for the AUDIT-C are ≥5points for men and ≥3 points for women (Johnson et al, 2013).The present study aimed to assess the reliability and validity of the AUDIT-C for detecting suspected alcohol dependence, alcohol use disorders, or risky drinking in Japanese adults.To the best of our knowledge, this study is the first conducted in an Asian country to assess the reliability and validity of the AUDIT-C.

Assessment of the reliability of the AUDIT-C
The reliability of the AUDIT-C was assessed using a test-retest method to determine its reproducibility.The subjects were 113 employees working in psychiatry medical clinics (n=27), a nursing facility (n=51), a rehabilitation facility (n=21), and a health examination organization (n=14) in Mie Prefecture, Japan.We asked the subjects to complete the AUDIT-C twice with 2 weeks between the 2 tests.The survey was conducted in October 2011.

Assessment of the validity of the AUDIT-C
We assessed the validity of the AUDIT-C using data from a nationwide survey on adult drinking behaviors in Japan conducted in 2008.The nationwide survey was interview-based and conducted in the subjects' homes; subjects were selected randomly from resident register lists of the municipalities.Of the selected 6,956 subjects, 4,123 people responded to the survey (response rate=59.3%).Trained interviewers visited the subjects' houses and asked them to participate in the survey.
The questionnaire included the full AUDIT in addition to questions about alcohol use behaviors including the frequency of alcohol use, average alcohol consumption on a day that included drinking, and the frequency of binge drinking (pure alcohol consumption ≥60 g per drinking occasion).
Ethical approval was provided by the ethical committees of the Faculty of Medicine of Tottori University and the Kurihama Medical and Alcohol Center.

Statistical analyses
To determine the reliability, we calculated the k coefficient, Cronbach's coefficient alpha, Spearman's correlation coefficient, and intraclass correlation coefficient (ICC).
The validity was assessed by comparing the total score of the full AUDIT, which was considered the gold standard, with the AUDIT-C score.We calculated the sensitivity and specificity.We also determined appropriate cutoffs for the AUDIT-C in the Japanese population using the positive likelihood ratio and Youden index (Youden, 1950;Akobeng, 2007).Youden index is defined as "sensitivity+specificity -1".
We also assessed the effectiveness of the AUDIT-C for identifying risky drinking.Risky drinking was defined as heavy drinking (280 g pure alcohol/week for men and 168 g pure alcohol/week for women) or binge drinking (consumption of 60 g pure alcohol on ≥1 occasion per week).Statistical analyses were conducted using SPSS for Windows ver.18 (SPSS Inc., Chicago, IL, USA).

Assessment of reliability
The total AUDIT-C score and scores for each AUDIT-C question demonstrated a high level of agreement between the repeated surveys (Table 1).Each index of agreement was close to 1. Therefore, we observed a high reproducibility of the AUDIT-C.However, the reproducibility of questions 2 and 3 of the AUDIT-C was relatively low compared to that of question 1 (Table 1).

Assessment of validity
The Spearman's correlation coefficient for the relationship between each question and the total score of the full AUDIT was 0.93 for question 1, 0.61 for question 2, and 0.67 for question 3. Table 2 provides the sensitivity, specificity, positive likelihood ratio, and Youden index of the AUDIT-C categorized by cutoff values of the full AUDIT score (≥8 points, ≥12 points, ≥16 points, and ≥20 points).The AUDIT-C cutoffs are ≥5 points for men and ≥3 points for women when the tool is used in Western countries.The sensitivity and specificity of these cutoffs for an AUDIT score ≥8 points were 88% and 80%, respectively, for men and 98% and 78%, respectively, for women.The sensitivity and specificity of the same cutoffs for an AUDIT score ≥12 points were 96% and 72%, respectively, for men and 96% and 76%, respectively, for women.The sensitivity and specificity of the same cutoffs for an AUDIT score ≥16 points were 100% and 66%, respectively, for men and 83% and 76%, respectively, for women.
A decrease in the positive likelihood ratio was observed with a higher AUDIT score, which indicates a more serious drinking problem.When the cutoffs used in Western countries were applied, the specificity was relatively low, and a trend for increased false positives was observed.According to the Youden index, the appropriate cutoffs for the AUDIT-C were ≥5 points for men and ≥4 points for women in this Japanese sample.Table 3 provides the sensitivity, specificity, positive likelihood ratio, and Youden index for identifying risky drinking.The same appropriate cutoffs were observed for identifying heavy drinking or binge drinking.However, the AUDIT-C demonstrated high sensitivity and specificity in women 100.0

Discussion
To the best of our knowledge, this is the first study to determine the reliability and validity of the AUDIT-C in an Asian country.The high reproducibility of the AUDIT-C was confirmed in the present study.The validity of the AUDIT-C was also high in this Japanese sample.It has been reported that there are many people with alcohol dependence in Japan (Osaki et al., 2005).However, only a portion of these people see doctors according to the relatively low estimate of patients that visit medical facilities reported in a national patient survey (Osaki, 2013).However, the general belief is that the majority of patients with alcohol use disorders visit medical facilities or undergo health examinations because of prevalent physical and mental problems.With appropriate screening by physicians, brief interventions can be implemented with patients at risk of alcohol use disorders, resulting in reduced alcohol consumption.The availability of a simple screening test in busy clinics or health examination settings would enable brief intervention programs aimed at reducing alcohol consumption to be implemented with patients with suspected alcohol use disorders.The high reliability and validity of the AUDIT-C indicate its potential effectiveness, as a proxy questionnaire for the full AUDIT, for identifying alcohol dependence in the general population in Japan.
Furthermore, the AUDIT-C is also considered to be an effective screening test for identifying risky drinkers, owing to the observed reliability and validity of the AUDIT-C for identifying heavy drinking or binge drinking.The appropriate AUDIT-C cutoffs for identifying heavy drinking or binge drinking were the same as those

6574
for identifying alcohol use disorders with the full AUDIT.
The statistical analyses in the present study resulted in appropriate AUDIT-C cutoffs of ≥5 points for men and ≥4 points for women in Japan.
The cutoffs recommended for use in Western countries vary; however, the most commonly used cutoffs have been 4-6 points for men and 3-5 points for women (Bush et al., 1998;Nordqvist et al., 2004;Dawson et al., 2005;Gómez et al., 2006;Bradley et al., 2007;Rodriguez-Martos and Santamarina, 2007;Tuunanen et al., 2007;Frank et al., 2008;Kelly et al., 2009;Kaarne et al., 2010;Towers et al., 2011;Dawson et al., 2012;Crawford et al., 2013;Johnson et al., 2013).According to results from the United States and European countries, the AUDIT-C is a satisfactory screening tool for identifying alcohol dependence diagnosed by DSM-IV as binge drinking or heavy drinking with high sensitivity and specificity.One article indicated that the sensitivity of the AUDIT-C varies by race or ethnicity (Frank et al., 2008).The sensitivity and specificity of the AUDIT-C tend to be lower for specific populations, such as veterans or young patients in emergency services (Kelly et al., 2009;Crawford et al., 2013).
The results of the present study were similar to previous results.The sensitivity and specificity of the AUDIT-C for heavy drinking, binge drinking and the total AUDIT score in this Japanese sample were adequate compared to results from Western countries.Therefore, we could conclude that the use of the AUDIT-C as a screening test instead of the full AUDIT is appropriate in the Japanese population.The AUDIT-C can be used in primary health care and health examination settings to promote countermeasures for reducing alcohol-related disease burden and social problems in Japan.This indicates that AUDIT-C will be an useful tool for cancer prevention in Japan.

Table 2 . Validity of the Alcohol Use Disorders Identification Test -Consumption (AUDIT-C) for Identifying Alcohol Use Disorders Categorized by AUDIT Cut-offs
*AUDIT-C: Alcohol Use Disorders Identification Test -Consumption; Se: sensitivity; Sp: specificity; PPV: positive predictive value; LR: positive likelihood ratio; ROC: receiver operating characteristics; S.E.: standard error

Table 3 . Validity of the Alcohol Use Disorders Identification Test -Consumption (AUDIT-C) for Identifying Heavy Drinking and Binge Drinking
*M/F: Male/Female; **AUDIT-C: Alcohol Use Disorders Identification Test -Consumption; Se: sensitivity; Sp: specificity; PPV: positive predictive value; LR: positive likelihood ratio; ROC: receiver operating characteristics; S.E.: standard error for identifying high AUDIT scores and risky drinking, compared with men.The specificity of the AUDIT-C for identifying risky drinking was relatively low in men.