OUP user menu


Jean-Bernard Daeppen, Tom L. Smith, Marc A. Schuckit
DOI: http://dx.doi.org/10.1093/alcalc/34.5.767 767-772 First published online: 1 September 1999


The self-rating of drinking habits was compared to DSM-III-R diagnoses of alcohol abuse and dependence in 181 men with an average age (± SD) of 38.7 ± 1.91 years. Results indicate that the 150 subjects without alcohol-related diagnosis (Group 1) rated themselves as ‘non-problem drinker’, in categories from ‘non-drinker’ to ‘heavy drinker’. Among the 15 individuals with alcohol abuse (Group 2), none rated their drinking pattern as ‘problem drinker’. Two (12.5%) subjects in the group of 16 individuals with alcohol dependence (Group 3) rated themselves as ‘problem drinker’, while most did not consider their drinking patterns as problematic. Within subjects who identified themselves as the same type of drinker (e.g. ‘infrequent drinker’, ‘moderate drinker’, etc ...), the quantity, frequency, and number of alcohol-related problems were higher in Groups 2 and 3, compared to Group 1. The self-rating of drinking habits using a single question failed to identify over 90% of the subjects diagnosed with alcohol use disorder (100% of those with alcohol abuse and 87.5% of those with alcohol dependence), and did not differentiate between levels of alcohol intake and number of alcohol-related problems for subjects who identified as a particular drinking type.


The identification of alcohol-related problems and alcohol dependence is a major issue in clinical practice. High alcohol intake predicts future medical complications, including gastrointestinal, neurological, and cardiovascular disease (Ashley et al., 1981), and social, psychological and legal problems (Schuckit et al., 1995Schuckit et al., 1999).

Given the relatively limited time clinicians have available to address numerous medical problems, screening for problematic drinking should be brief and effective. A number of screening questionnaires [e.g. Alcohol Use Disorders Identification Test (AUDIT), Cut-down–Annoyed–Guilty– Eye-opener (CAGE), and Michigan Alcoholism Screening Test (MAST)] have demonstrated moderate to high sensitivities and specificities in identifying individuals with harmful drinking levels (Bush et al., 1987; Cherpitel, 1995; Bohn et al., 1995; Mackenzie et al., 1996), although it appears that the shortest of these questionnaires, i.e. the CAGE, is not used systematically in clinical practice (Wenrich et al., 1995). Indeed, many clinicians ask general, descriptive questions about drinking, but it is difficult to know the effectiveness of these questions. A review of the literature (Medline 1966 to 1997) identified only one study examining the validity of self-ratings of alcohol consumption patterns (Garett and Bahr, 1974). Results indicated that correlation between subjects' description of their consumption (abstainer, light, moderate, or heavy drinker) and measures of quantity and frequency were in the range from 0.58 to 0.64 in men, and 0.82 to 0.89 in women. The study, however, did not analyse separately problematic and non-problematic drinkers, which certainly limits the extent of the findings.

The aim of the present study was to evaluate how high functioning men in their 30s rate their alcohol consumption. Self-ratings were compared to a detailed history of subjects' drinking patterns and to DSM-III-R diagnoses of alcohol abuse and dependence.


The sample reported here was originally recruited between 1979 and 1983 among 18- to 25-year-old drinking but non-alcohol-dependent male students or non-academic staff at the University of California, San Diego, excluding subjects with DSM-III-R diagnoses of antisocial personality disorder, schizophrenia, or mania. The main topic of the study started in 1979 was related to the genetics of alcoholism (Schuckit and Smith, 1996). The data reported are from the first set of subjects interviewed as part of the ongoing 15-year follow-up started in March 1994. Data regarding alcohol consumption and alcohol-related problems reflect the 5-year period preceding the 15-year follow-up.

The 15-year follow-up used a structured interview adapted from the Schedule for Affective Disorders and Schizophrenia (SCID) (Schuckit et al., 1988). The instrument included DSM-III-R criteria to review abuse and dependence on alcohol (American Psychiatric Association, 1987) incorporated into a list of 29 life problems related to chronic heavy drinking (Schuckit et al., 1993). Subjects were also requested to estimate their average quantity and frequency of alcohol use over the last 6 months and the 4.5 preceding years (total of 5 years). Self-reports of alcohol intake and problems were corroborated through an identical face-to-face structured interview administered to an additional informant, usually the spouse. The worst case scenario (the highest alcohol consumption and problems) was adopted to select information to be used for analysis when information provided by the subject was discordant from that of the resource person. On the basis of prior work conducted by the same research group, the rate of agreement between subject and resource person regarding the occurrence of alcohol-related life events is estimated to be 90% or higher.

The last portion of the interview included an item for subjects to rate their drinking pattern asking: ‘Over the last 5 years, how would you label your own drinking pattern overall?’. The possible answers were: (1) non-drinker, abstainer (‘non-drinker’); (2) infrequent, occasional, light social drinker (‘infrequent drinker’); (3) moderate social drinker (‘moderate drinker’); (4) frequent, heavy social drinker (‘heavy drinker’); (5) problem drinker, alcoholic (‘problem drinker’); (6) ‘recovering alcoholic’. As opposed to the other items in the interview, this question was scored only in the way the subject responded, regardless of the answer given by the resource person. The interviews were conducted by trained research associates and reviewed by a senior psychiatrist who assessed for DSM-III-R diagnoses of alcohol abuse or dependence. Diagnoses were based on a lifetime assessment with subjects with alcohol abuse and alcohol dependence being active drinkers (at least one alcoholic beverage/month) over the 5 years preceding the current investigation.

For the analyses presented below, the sample was split into three groups based on lifetime DSM-III-R diagnoses, including subjects with no alcohol abuse or dependence (Group 1), those with alcohol abuse (Group 2), and lastly those with alcohol dependence (Group 3). Within individuals who identified themselves as the same drinking style, problem drinkers (Group 2 + Group 3) were compared to subjects without alcohol-related diagnosis (Group 1) on drinking quantity, frequency, alcohol-related problems and DSM-III-R criteria of alcohol dependence. Because of the limited number of individuals within each drinking style, the Mann– Whitney U-test, a non-parametric alternative to the t-test comparing the median ranks between two groups, was used. Finally, the two time frames to assess for drinking quantity and frequency (last 6 months and last 6 months to 5 years) were compared using Cronbach-α indexes of reliability (Cronbach, 1951).


At the time of the present investigation, 185 men out of the 186 approached (99.5%) had completed interviews, of whom four subjects who rated themselves ‘recovering alcoholics’, were excluded because of the difficulty in interpreting data in the present context. Therefore, the data are based on the interviews of 181 men, with an average age (mean ± SD) of 38.7 ± 1.91 years, including 93.9% Caucasians, 3.4% Hispanics, 1.1% African-Americans, and 1.7% Asians, with 76.2% married or living as married, 9.4% separated or divorced, and 14.4% single. Subjects were generally highly educated, with 63.5% having completed postgraduate schooling, 30.9% who had graduated from college, and 5.5% who did not finish college. In addition, 97.8% were employed at the time of the interview with an average income of US $79 720 ± 60 389 per year.

Subjects' alcohol consumption was evaluated using two different time frames: (1) the 6 months preceding the interview (to reflect the current drinking) and (2) the previous 4.5 years (to provide a broader estimate of their drinking patterns). The measures of quantity and frequency over the last 6 months and last 5 years were compared using Cronbach-α coefficients, which indicated indices of reliability of 0.95 for ‘average days drinking/ month’, 0.81 for the ‘average drinks/drinking day’, 0.90 for the ‘maximum days drinking/month’, and 0.81 for the ‘maximum drinks/drinking day’. These high indices of reliability between the two time frames suggest that drinking measures over the last 5 years excluding the last 6 months offered an accurate estimate of the current drinking pattern. In order to use similar periods of reference for the self-rating of drinking habits (last 5 years) and drinking measures, the data reported reflected drinking quantity and frequency over the last 5 years (excluding the last 6 months).

The major question addressed by the data is the relationship between the self-rating of drinking habits and the diagnoses of alcohol abuse or dependence established from the structured interview. Table 1 indicates that, for the 150 subjects who did not qualify for an alcohol-related diagnosis (Group 1), five rated themselves as ‘non-drinkers’ (3.3%), 101 as ‘infrequent drinkers’ (67.3%), 42 as ‘moderate drinkers’ (28.0%), and two as ‘heavy drinkers’ (1.3%), whereas none labelled themselves as ‘problem drinker’. However, the data indicate that the men meeting criteria for alcohol abuse or dependence reported their alcohol use less accurately. For the 15 individuals with alcohol abuse (Group 2), none rated their drinking pattern as ‘problem drinker’, whereas six (40%) rated themselves as ‘infrequent drinkers’, and nine (60%) as ‘moderate drinkers’. Among the 16 subjects with alcohol dependence (Group 3), two men rated themselves as ‘problem drinker’ (12.5%), whereas most appeared to ignore their alcohol-related problems, with one rating himself as an ‘infrequent drinker’ (6.3%), seven as ‘moderate drinkers’ (43.5%), and six as ‘heavy drinkers’ (37.5%). Overall, there was a weak, albeit present, relationship between the self-rating of drinking habits and alcohol-related diagnosis, with 50% alcohol-dependent subjects rating themselves ‘heavy drinker’ or ‘alcoholic’, whereas 70% of those who did not qualify for an alcohol-related diagnosis perceived themselves as ‘non-drinker’ or ‘infrequent drinker’.

View this table:
Table 1.

Relationship between the self-rating of drinking habits and three groups defined according to DSM-III-R diagnoses of alcohol abuse and dependence

CategoryGroup 1: No alcohol-related diagnosis (n = 150)Group 2: Alcohol abuse (n = 15)Group 3: Alcohol dependence (n = 16)
Figures in parentheses indicate number of respondents.
% Non-drinker3.3 (5)0.0 (0)0.0 (0)
% Infrequent drinker67.3 (101)40.0 (6)6.3 (1)
% Moderate drinker28.0 (42)60.0 (9)43.7 (7)
% Heavy drinker1.3 (2)0.0 (0)37.5 (6)
% Alcoholic0.0 (0)0.0 (2)12.5 (2)

The conclusion that the self-reported categories do not reflect accurately the drinking status for the subjects with abuse or dependence is supported by data in Table 2. These demonstrate the quantity, frequency, and alcohol-related problems between subjects with and without an alcohol-related diagnosis (Group 1 vs Group 2 + Group 3). When subjects rated themselves as ‘infrequent drinkers’, average and maximum quantity and frequency were higher in Groups 2 + 3, compared to Group 1. Similarly, subjects labelled ‘moderate drinkers’ reported higher maximum frequency in Groups 2 + 3 and approximately twice the average and maximum quantity in Groups 2 + 3, compared to Group 1. Finally, ‘heavy drinkers’ in Groups 2 + 3 revealed a larger average and maximum quantity, compared to Group 1.

View this table:
Table 2.

Self-rating with corresponding quantity, frequency, and alcohol-related problems in two groups

ParameterGroup 1: No alcohol-related diagnosis (n = 150) Means ± SDGroups 2 + 3: Alcohol abuse + alcohol dependence* (n = 31) Means ± SDMann–Whitney U-test Z
Figures in parentheses indicate number of respondents.
*DSM-III-R; aP < 0.05; bP < 0.01; cP < 0.001.
Average days drinking/month
  Non-drinker0.0 ± 0.07 (5)— (0)
  Infrequent drinker6.3 ± 6.59 (101)11.0 ± 5.45 (7)2.44a
  Moderate drinker16.3 ± 7.58 (42)17.7 ± 9.64 (16)0.30
  Heavy drinker27.5 ± 3.54 (2)22.3 ± 7.53 (6)1.03
  Alcoholic— (0)17.5 ± 9.19 (2)
Average drinks/drinking day
  Non-drinker0.2 ± 0.45 (5)— (0)
  Infrequent drinker1.6 ± 0.82 (101)3.0 ± 0.58 (7)4.05c
  Moderate drinker2.0 ± 0.73 (42)3.4 ± 1.36 (16)3.72c
  Heavy drinker2.0 ± 0.00 (2)5.5 ± 2.34 (6)2.06a
  Alcoholic— (0)3.5 ± 2.12 (2)
Maximum days/drinking month
  Non-drinker0.2 ± 0.45 (5)— (0)
  Infrequent drinker10.0 ± 7.76 (101)17.8 ± 8.57 (7)2.52a
  Moderate drinker21.3 ± 6.77 (42)26.4 ± 5.36 (16)2.65b
  Heavy drinker30.0 ± 0.00 (2)26.7 ± 6.05 (6)0.87
  Alcoholic— (0)22.5 ± 10.61 (2)
Maximum drinks/drinking day
  Non-drinker0.2 ± 0.45 (5)— (0)
  Infrequent drinker4.7 ± 3.05 (101)9.3 ± 2.21 (7)3.46c
  Moderate drinker6.9 ± 2.55 (42)10.6 ± 3.91 (16)3.41c
  Heavy drinker6.5 ± 0.71 (2)15.0 ± 3.58 (6)2.00a
  Alcoholic— (0)12.0 ± 2.83 (2)
Alcohol-related problems (29)
  Non-drinker0.2 ± 0.45 (5)— (0)
  Infrequent drinker0.3 ± 0.51 (101)2.3 ± 0.95 (7)4.90c
  Moderate drinker0.7 ± 0.99 (42)4.3 ± 2.15 (16)5.44c
  Heavy drinker0.5 ± 0.71 (2)9.0 ± 2.76 (6)2.02a
  Alcoholic— (0)7.5 ± 2.12 (2)
DSM-III-R criteria (9)
  Non-drinker0.0 ± 0.00 (5)— (0)
  Infrequent drinker0.0 ± 0.24 (101)1.1 ± 0.69 (7)6.59c
  Moderate drinker0.3 ± 0.54 (42)2.7 ± 1.39 (16)6.10c
  Heavy drinker0.0 ± 0.00 (2)5.3 ± 1.21 (6)2.04a
  Alcoholic— (0)4.0 ± 1.41 (2)

The number of alcohol-related problems and DSM-III-R criteria endorsed are also reported in Table 2, further supporting the results presented above. These data indicate that, when subjects characterized themselves as having the same drinking style, the number of alcohol-related problems and DSM-III-R criteria differed between Groups 2 + 3 and Group 1. For example, when subjects rated themselves as ‘moderate drinkers’, those in Group 1 endorsed, on average, less than one alcohol-related problem, whereas in Groups 2 + 3 they reported 4.3 problems (including 2.7 DSM-III-R criteria). Differences were even more impressive in subjects who rated themselves as ‘heavy drinkers’ with less than 1.0 alcohol-related problem in Group 1, and 9.0 alcohol-related problems in Groups 2 + 3 (including 5.3 DSM-III-R criteria).


The major aim of this study was to assess the relevance of the information provided by a question evaluating how a sample of men in the community rated themselves regarding their drinking patterns. First, data indicated that the self-rating item did not accurately identify individuals carrying an alcohol-related diagnosis, as none of those with alcohol abuse and only two (12.5%) of those with alcohol dependence were identified as ‘problem drinker’ by the self-rating item. Second, the data indicated that a wide range of drinking quantity, frequency, and alcohol-related problems were found within subjects who identified themselves as the same type of drinker, those with alcohol abuse and dependence revealing heavier alcohol consumption and more alcohol-related problems, compared to those in the same self-rating category but without an alcohol-related diagnosis.

The finding that the self-rating of drinking patterns failed to identify most individuals with alcohol abuse or dependence is particularly relevant in men in their 30s to early-40s, which is a critical period regarding harmful drinking and the development of subclinical alcohol-related problems (Schuckit et al., 1993). These data might dissuade clinicians from using a unique self-rating question to screen for problematic drinking, and warn them that among patients who claim that they are moderate or heavy social drinkers, there may be a fair proportion of individuals who are actually alcoholic. Consistent with these findings, Buchsbaum et al., (1995) reported results of a study evaluating the accuracy of screening for problematic drinking by asking questions related to the quantity and frequency of drinking, which indicated that the sensitivities of reported consumption declined with increasing actual consumption, from 100% for those who report zero to two drinks/week, to 21% for those reporting 24 drinks/week and over.

In interpreting these results, it is important to recognize some limitations of the data. First, the sample focused on relatively well-educated Caucasian men in their late-30s to early-40s, and results may not generalize to other age groups, females, or non-Caucasians. Results of the study by Garett and Bahr (1974) indicated differences in correlation between self-rating and drinking quantity and frequency in women (from 0.82 to 0.89) and men (from 0.58 to 0.64), suggesting that the present conclusions may not apply as well to women. Second, although the group of individuals who did not meet criteria for abuse and dependence was large (150), only 15 subjects were available for analysis in the group with alcohol abuse and 16 in the group with alcohol dependence. Finally, although efforts were made to optimize the accuracy of the data through a resource person to corroborate information and treatment records when appropriate, it is still possible that some subjects were mis-classified. Unfortunately, state markers of heavy drinking, such as α-glutamyltransferase or carbohydrate-deficient transferrin, were not useful because group membership was based on lifetime diagnoses of alcohol abuse and dependence, and according to the results of a prior study, not all subjects with alcohol abuse and dependence are actively drinking during the days preceding an interview (Schuckit et al., 1997).

In conclusion, this study demonstrated the low performance of a general question evaluating how subjects in the community rated their drinking patterns. This should serve as a warning to clinicians that the single question ‘How would you label your drinking pattern overall?’ did not discriminate between levels of alcohol intake or number of alcohol-related problems, and failed to identify 100% of the subjects with alcohol abuse and 87.5% of those with alcohol dependence.


This research was supported by NIAAA Grant #05526.


  • * Author to whom correspondence should be addressed at Unité d'Alcoologie, DMI, BH-10, CHUV-1011 Lausanne, Switzerland.


View Abstract