Addictive Behaviors, Vol. 16, pp. 223-233, Printed in the USA. All rights reserved.
DETECTION
1991 Copyright
OF ALCOHOL BARBARA
PROBLEMS
IN A HOSPITAL
L. NILES and BARBARA
Rutgers,
The State University
0306-4603/91 $3.00 + .OO c 1991 Pergamon Press plc
SETTING
S. McCRADY
of New Jersey
Abstract - Three hundred and six adult medical and family practice inpatients at a universityaffiliated teaching hospital were assessed for problem drinking using: (1) a short questionnaire containing the CAGE (a 4-question, self-report screening measure), and (2) a standardized review of medical records (an examination of specific blood chemistries and the physicians’ and nurses’ notes). Each patient completed one of two randomly assigned questionnaires containing the CAGE within the first 48 h of admission: a “Health Habits Questionnaire” or an “Alcohol Questionnaire.” No significant differences were found between the two questionnaires with regard to the number of CAGE items endorsed. Overall, 16.9% of the subjects endorsed two or more CAGE items. Using the chart screening method, 11.4% of the total subjects were considered “probable” or “definite” problem drinkers. The screener’s reading of the physicians’ notes was the variable most important in determining overall chart screening assessment. The correlation between the two screening measures was relatively low (r = .38, p < .OOl), suggesting that the two measures identify different populations of problem drinkers. INTRODUCTION
Although physicians have long been aware of the harmful physical effects of excessive alcohol use, the medical community is still struggling with how to detect and treat drinking problems. Research suggests that about 25% of the patients in a general hospital population may be dependent on alcohol (Allen, Eckardt, & Wallen, 1988; Lewis & Gordon, 1983) with some estimates as low as 5% (Keglar & Clark, 1979) and others over 50% (Gomberg, 1975; Panepinto & Kohut, 1971). These higher prevalence rates generally are found in urban or Veterans’ Administration hospitals (Decker, Fann, Giradin, Miller, & Kanas, 1979; Gomberg, 1975; McCusker, Cherubin, & Zimberg, 1971; Panepinto & Kohut, 1971). There is a substantial discrepancy between the prevalence of alcoholism found through rigorous screening and that reported by physicians through diagnoses (Keglar & Clark, 1979). Many physicians do not properly assess alcohol use nor consider its possible contribution to clinical problems unless the patient’s presenting symptoms are blatantly alcohol-related (e.g., cirrhosis, alcohol withdrawal) (Moore, Bone, Geller, Mamon, Stokes, & Levine, 1989). Alcohol-related physical disorders such as acute gastritis are typically diagnosed without documentation of the potential underlying alcohol problem (Stainback, 1989). As alcohol problems go undetected or unnoted, so they go untreated. This lack of treatment of drinking problems leads to the use of a disproportionate share of hospital resources by problem drinkers. The average problem drinker remains in the hospital longer than the nonproblem drinker (McCusker, Cherubin, & Zimberg, 197 1; Panepinto & Kohut, 1971). Given the amount of hospital resources being used for the treatment of the complications of alcoholism, there should be a great impetus for hospitals to develop programs to directly identify and manage alcoholism. To improve the management of alcohol problems in hospitals, there is a need for continued and increased physician education, specialized alcohol consultation, and systematic patient screening. Substantial research has accrued on the topic of early detection and screening of alcohol problems (Babor, Kranzler, & Lauerman, 1989; Moore et al., 1989); however, most hospitals do not routinely screen patients for alcohol problems. The Requests for reprints should be sent to Barbara L. Niles, 36 Evergreen 223
#2, Boston, MA 02130,
224
BARBARA
L. NILES and BARBARA
S. McCRADY
present study has attempted to improve the process of large-scale, routine screening of medical inpatients. A number of screening questionnaires have been developed to facilitate identification of alcoholism. Probably the most commonly used is the Michigan Alcoholism Screening Test (MAST) (Selzer, 1971), a 25item self-report questionnaire. Two shorter versions of the MAST have also been developed: the IO-item Brief MAST (Pokomy, Miller, & Kaplan, 1972) and the 13-item SMAST (Selzer, Vinokur, & van Roojen, 1975). Other similar instruments developed are the Self-Administered Alcoholism Screening Test (SSAST) (Swenson & Morse, 1975) and the Veterans Alcoholism Screening Test (VAST) (Magruder-Habib, Harris, & Fraker, 1982). All versions of the MAST have performed well in validation studies (Miller, 1976). However, the high face validity of some of the questions (e.g., “Do you feel you are a normal drinker?“) has led some investigators to question its clinical usefulness (Goldberg, 1974). Another useful screening instrument which addresses patients’ subjective experiences is the CAGE (Mayfield, McLeod, & Hall, 1974), which can be administered as an interview or in written form. The CAGE is an acronym for four questions: (1) Have you ever felt you should Cut down on your drinking?; (2) Have people Annoyed you by criticizing your drinking?; (3) Have you ever felt bad or Guilty about your drinking?; and (4) Have you ever had a drink first thing in the morning . ..? (Eyeopener). The CAGE has also been shown to be an effective screening instrument in hospital populations with sensitivity and specificity over 80% (Bush, Shaw, Cleary, Delbanco, & Aronson, 1987; Murray, 1974; Mayfield et al., 1974) when a positive test result is defined as affirmative responses to two or more questions. Ewing (1984) has suggested that the CAGE may be used to determine a “high index of suspicion” and that even one positive reply calls for further inquiry. Because the CAGE is so simple to administer and score, since it can be administered in written or interview form, and because the questions are relatively nonthreatening and thus minimally hinder clinical rapport, this instrument is ideal for hospital screening and could easily become incorporated into all history-taking. Although the CAGE can only be used as a gross screening measure and alone is not sufficient for diagnosis, investigators have examined its effectiveness in combination with biochemical markers (Bemadt, Mumford, Taylor, Smith, & Murray, 1982; Bemadt, Mumford, & Murray, 1984) and determined that a combination of selected laboratory tests and a self-report questionnaire are most effective in identifying alcoholism. Some of the most commonly used laboratory tests are the hepatic enzymes; elevated levels of serum glutamic oxalocetic transaminase (SGOT), serum glutamic pyruvic transaminase (SGPT) and gamma glutamyl transpeptidase (GGTP) are markers of excessive alcohol use with relatively high specificities (80~90%) but low sensitivities (20-70%) (Hays & Spickard, 1987; Rosalki, 1984; Paredes, 1982). Elevation of the mean corpuscular volume (MCV), an alteration in blood cells, has also often been used as a marker of alcohol abuse problems (Skinner, Holt, Schuller, Roy, & Israel, 1984). The MCV has a high specificity (89%) (Wu, Chanarin, & Levi, 1974) but low sensitivity (241%) (Cushman, Jacobson, Barboriak, & Anderson, 1984; Korri, Nuutineon, & Salaspuro, 1985). Combinations of these biochemical values have provided varying degrees of sensitivity and specificity. Watson, Mohs, Eskelson, Sampliner and Hartmann (1986) concluded in their review that sophisticated statistical analysis of groups of laboratory tests can separate the extremes; that is, laboratory tests can differentiate the patients with end-organ damage due to alcoholism from patients who rarely drink. But the distinction that needs to be made is much more subtle. For most effective treatment, patients need to be identi-
Detectionof alcohol problems
225
fied before they experience end-stage, alcohol-related diseases. Recent studies have suggested that biochemical tests alone are inadequate detectors of alcohol problems and that a self-report screening measure such as the CAGE should be used in combination with biochemical screening for more effective results (Babor et al., 1989; Bush et al., 1987; Watson et al., 1986; Hays & Spickard, 1987). The present study is an attempt to improve the process of patient screening and to develop practical methods for identifying alcohol problems by examining two different timeefficient screening methods: the CAGE in questionnaire form and review of medical records. In the administration of the CAGE, two self-report questionnaires each including the CAGE were compared; the questionnaires were labeled differently - “Alcohol” versus “Health Habits” - in order to explore the possibility that the labeling would elicit different response sets (i.e., higher versus lower self-disclosure) and produce different results. The questionnaires were given to medical inpatients and the results were compared against a standardized review of blood chemistries and other information from the medical record. METHOD Subjects Subjects were 306 patients admitted to the medicine and family practice units of a community hospital in a small city in the Northeastern United States. All patients admitted to the unit during the previous 48 h were approached and asked to participate in the study. Patients gave written informed consent before responding to the orally administered questionnaire. Subjects were not screened for psychiatric disorders or organic@; in order to establish generalizability, a heterogeneous sample was desired. From the admissions log book for each hospital unit, 471 patients were identified as eligible for the study. Reasons for nonparticipation included: (1) patient unable to speak (i.e., in a coma, asleep and did not wake up when approached, intubated, or highly medicated) (n = 119; 25.3%); (2) patient in severe pain and asked not to participate (n = 13, 2.8%); (3) patient not in his/her bed and could not be located by the investigators (n = 20, 4.2%); (4) patient could not speak English and no translator was available (n = 6, 1.3%); (5) patient with visitors and asked to participate at a later time (n = 31, 6.6%); (6) patient with a physician or nurse and could not be disturbed (n = 12, 2.6%); (7) krftient not allowed visitors due to radiation treatment or highly infectious diseases (n = 10, 2.1%); and (8) patient refused to participate in the study (n = 24, 5.1%). No patients refused after signing the consent form. Of the 306 patients surveyed, 48% were female and 52% were male with a mean age 58.98 years (SD = 16.88, range = 18-95). Measures Each subject was randomly assigned to complete one of two questionnaires in which the CAGE was embedded in a series of questions. In the Health Habits Questionnaire, other questions pertained to sleeping, smoking and caffeine habits; in the Alcohol Use Questionnaire, the other questions inquired further about alcohol use. The questionnaires were approximately equal in length, and the average patient took 5 min to respond. The Health Habits Questionnaire was developed in order to examine the effectiveness of a questionnaire which has questions about alcohol use (the CAGE) among more general questions about health, so that the subject would not be aware that the alcohol ques-
226
BARBARA
L. NILES and BARBARA
S. McCRADY
tions were singled out, but was answering the questions with the idea that he/she was providing general health information for the health care professionals. The Alcohol Questionnaire was developed to evaluate the CAGE in the context of a brief interview about alcohol use. Procedures The medical records of the 306 study subjects were screened for indicators of alcoholrelated problems using a standardized method. Specific blood chemistries associated with heavy drinking were examined, including MCV, total bilirubin, GGTP, SGOT, SGPT, Hct, MCH, potassium, albumin, and BUN. Three categories of assessments were made from the physician’s and nurse’s notes: physician’s assessment of alcohol problems, screener’s assessment of alcohol problems after reading physician history, physical and progress notes, and nurse’s assessment of alcohol problems. The screener’s assessment after reading the physician’s entries reflected the screener’s opinion after reading the history and physical rather than the physician’s opinion about the patients’s alcohol use. Screeners rated the physicians’ and nurses’ notes as containing no information regarding alcohol use, or information indicating no, possible, or definite alcohol problem. The screener then made an overall assessment about the patient’s drinking behavior (i.e., no, possible, probable or definite problem) using all the information available from the medical record. The screening of medical records was done by two staff members of the specialized alcohol consultation and treatment program at the medical center: the director, a clinical psychologist with a behavioral orientation; and a registered nurse, certified alcoholism counselor with a master’s degree in psychiatric nursing who has a disease-model approach. Screeners did not administer the CAGE and were blind to the results. Reliability for the four categorical responses which required a judgement to be made (nurse’s assessment, physician’s assessment, screener’s assessment of physician’s notes, and screener’s assessment of alcohol problem) was determined prior to the commencement of the study and there were reliability checks at three times during the study. At least 80% agreement between screeners was met for each of the four categorical responses. Specific guidelines for making these assessments, with an example for each possible category, were reviewed with the screeners prior to the first reliability check. Screeners were instructed to use all the available information from the medical record to determine the overall screener’s assessment of an alcohol problem. RESULTS
Incidence of alcohol problems CAGE. Of the total population of subjects, 52 (16.9%) endorsed two or more CAGE questions - 10 (6.8%) of the females and 42 (26.6%) of the males. Questionnaires. Among the 152 patients who answered the health questionnaire, 24 (15.8%) endorsed two or more items on the CAGE. Among the 154 patients who answered the alcohol questionnaire, 28 (18.1%) endorsed two or more items on the CAGE. There were no significant differences in the number of positive responses between the two questionnaires for any of the four CAGE questions or for the combined CAGE score. Screening assessment. The assessment from screening medical records concluded that 11.4% of the total subjects (6.1% of the females and 16.3% of the males) were probable or definite problem drinkers.
227
Detection of alcohol problems
Table 1. Association
between CAGE scores and information from medical records CAGE score
Negative or “possible” Positive “probable” or “yes” “no”
Medical records
Negative 0 or 1
Positive 2, 3, or 4
235
36
19
16
The physician provided no information regarding alcohol use for 118 (38.6%) patients. The physician’s assessments indicated that 21 (6.9%) had a possible drinking problem and 15 (4.9%) had a definite drinking problem. The screener’s assessment from the physician’s notes indicated more problem drinking than did the physicians’ assessments: for 30 patients (9.8%) a possible drinking problem was indicated and for 28 (9.2%) a definite drinking problem was inferred. The nurse’s assessment provided no information regarding alcohol use for 85 (27.8%) of the patients, indicated a possible drinking problem for 19 (6.2%) and a definite drinking problem for 6 (2.0%) of the patients. For female patients, the nurses indicated no definite drinking problems at all, and only 5 (3.4%) possible drinking problems. Association between CAGE and screener’s assessment. There was a significant association between the cumulative CAGE scores and screeners’ assessments when these variables were split into dichotomous categories, x2 (1, 306) = 14.72, p < ,001. A cumulative score of 0 or 1 on the CAGE was considered negative on the CAGE and 2, 3, or 4 was considered to be positive on the CAGE. A screening assessment of “no” or “possible” was considered negative medical record information and an assessment of “probable” or “yes” was considered positive medical record information. Table 1 shows the number of subjects falling into each category. The subjects who had negative medical record information, but were positive on the CAGE were significantly older (mean age = 59.78) than the subjects who had positive medical record information and were negative on the CAGE (mean age = 46.63), (t (53) = 2.95, p < .Ol). Among the patients who were positive on the CAGE, there were no statistically significant differences on any of the biochemical markers between those subjects who had negative medical record information and those subjects who had positive medical record information. However, there were some substantial absolute differences between these two groups on the liver enzymes, which probably were not statistically significant due to the very large variances. For example, among those subjects who were positive on the CAGE, the mean GGTP was 75.8 (SD = 104.8) for negative medical record information subjects and 143.3 (SD = 242.2) for the positive medical record information group, t (17) = 1.03, p = .317. There was a significant difference between these two groups in the amount of information provided by the physician. In the positive medical record information group, the physicians provided some information about alcohol consumption in every case, but in the negative medical record information group the physicians failed to provide any information about alcohol consumption in 44% of the cases, t (35) = 5.25, p < .OOl .
228
BARBARA
L. NILES and BARBARA
Table 2. r tests for biochemical
S. McCRADY
markers
Chart screen
Variable MCV Total bilirubin GGTP SGOT SGPT HCT MCH Potassium Albumin BUN
Negative Mean/(W) 89.15 (6.93) S870 (.716) 44.86 (59.2) 31.45 (39.7) 24.63 (29.8) 38.09 (6.4) 30.27 (2.5) 4.09 (.53) 3.59 (.59) 19.67 (13.1)
n 260 247 223 252 223 262 260 248 247 256
Positive Meani 92.76 (8.02) .9147 (.592) 151.66 (221 .O) 57.94 (55.6) 53.71 (52.9) 41.13 (7.6) 31.65 (3.0) 3.94 (.67) 3.58 (.94) 17.29 (8.2)
n
t/(dfI
34
3.00’ (292) 2.55’ (279) 5.96”’ (254) 3.47” (284) 4.69”’ (255) 2.55’ (294) 2.96” (292) 1.52 (280) 0.05 (279) 1.03 (288)
34 33 34 34 34 34 34 34 34
*p < .05; **p < .Ol; **-p < .001
Alcohol and health data Alcohol. For those patients who answered the alcohol questionnaire, the correlation between the reported number of drinks per week and the CAGE score was significant (1. = .4436, p < ,001). When the analyses were done separately by sex, some gender differences emerged. The correlation between the CAGE and the number of drinks per week was substantially greater for the females (r = .8179, p < .OOl), while the same correlation for the males was much lower (r = .3799, p < .OOl). When the frequencies of “drinks per occasion” were examined for the men who were negative on the CAGE, it was found that only 4.7% of them reported drinking more than 4 drinks per occasion, while 46.2% of the men who were positive on the CAGE reported drinking more than 4 drinks per occasion. Health. For those patients who answered the health questionnaire, the correlations between the CAGE and the health information also showed gender differences. The correlation between the CAGE and the number of packs of cigarettes per day was significant for all subjects (r = .2447, p < .OOl) and for the females (P = .3883, p < .OOl), but not for males. When the frequency of the smokers and the number of packs of cigarettes per day were examined for women who were negative on the CAGE, it was found that only 18% smoked cigarettes at all and 4% smoked more than 1 pack per day. In contrast, for women who were positive on the CAGE, 63.5% smoked and 25% smoked more than one pack per day. What determines a screening assessment t tests were done on each of the biochemical markers contributing to the screening assessment to determine if there were significant differences between patients who were
229
Detection of alcohol problems
Table 3. Regression
analyses of factors determining assessment for all subjects
the screening
Variable
? Change
Screener reading of MD notes GGTP Nurse assessment Diagnosis SGF’T MCV
s3577” .07835”’ .03337” .02332” .01825” .01172’
*p
DETECTION
1991 Copyright
OF ALCOHOL BARBARA
PROBLEMS
IN A HOSPITAL
L. NILES and BARBARA
Rutgers,
The State University
0306-4603/91 $3.00 + .OO c 1991 Pergamon Press plc
SETTING
S. McCRADY
of New Jersey
Abstract - Three hundred and six adult medical and family practice inpatients at a universityaffiliated teaching hospital were assessed for problem drinking using: (1) a short questionnaire containing the CAGE (a 4-question, self-report screening measure), and (2) a standardized review of medical records (an examination of specific blood chemistries and the physicians’ and nurses’ notes). Each patient completed one of two randomly assigned questionnaires containing the CAGE within the first 48 h of admission: a “Health Habits Questionnaire” or an “Alcohol Questionnaire.” No significant differences were found between the two questionnaires with regard to the number of CAGE items endorsed. Overall, 16.9% of the subjects endorsed two or more CAGE items. Using the chart screening method, 11.4% of the total subjects were considered “probable” or “definite” problem drinkers. The screener’s reading of the physicians’ notes was the variable most important in determining overall chart screening assessment. The correlation between the two screening measures was relatively low (r = .38, p < .OOl), suggesting that the two measures identify different populations of problem drinkers. INTRODUCTION
Although physicians have long been aware of the harmful physical effects of excessive alcohol use, the medical community is still struggling with how to detect and treat drinking problems. Research suggests that about 25% of the patients in a general hospital population may be dependent on alcohol (Allen, Eckardt, & Wallen, 1988; Lewis & Gordon, 1983) with some estimates as low as 5% (Keglar & Clark, 1979) and others over 50% (Gomberg, 1975; Panepinto & Kohut, 1971). These higher prevalence rates generally are found in urban or Veterans’ Administration hospitals (Decker, Fann, Giradin, Miller, & Kanas, 1979; Gomberg, 1975; McCusker, Cherubin, & Zimberg, 1971; Panepinto & Kohut, 1971). There is a substantial discrepancy between the prevalence of alcoholism found through rigorous screening and that reported by physicians through diagnoses (Keglar & Clark, 1979). Many physicians do not properly assess alcohol use nor consider its possible contribution to clinical problems unless the patient’s presenting symptoms are blatantly alcohol-related (e.g., cirrhosis, alcohol withdrawal) (Moore, Bone, Geller, Mamon, Stokes, & Levine, 1989). Alcohol-related physical disorders such as acute gastritis are typically diagnosed without documentation of the potential underlying alcohol problem (Stainback, 1989). As alcohol problems go undetected or unnoted, so they go untreated. This lack of treatment of drinking problems leads to the use of a disproportionate share of hospital resources by problem drinkers. The average problem drinker remains in the hospital longer than the nonproblem drinker (McCusker, Cherubin, & Zimberg, 197 1; Panepinto & Kohut, 1971). Given the amount of hospital resources being used for the treatment of the complications of alcoholism, there should be a great impetus for hospitals to develop programs to directly identify and manage alcoholism. To improve the management of alcohol problems in hospitals, there is a need for continued and increased physician education, specialized alcohol consultation, and systematic patient screening. Substantial research has accrued on the topic of early detection and screening of alcohol problems (Babor, Kranzler, & Lauerman, 1989; Moore et al., 1989); however, most hospitals do not routinely screen patients for alcohol problems. The Requests for reprints should be sent to Barbara L. Niles, 36 Evergreen 223
#2, Boston, MA 02130,
224
BARBARA
L. NILES and BARBARA
S. McCRADY
present study has attempted to improve the process of large-scale, routine screening of medical inpatients. A number of screening questionnaires have been developed to facilitate identification of alcoholism. Probably the most commonly used is the Michigan Alcoholism Screening Test (MAST) (Selzer, 1971), a 25item self-report questionnaire. Two shorter versions of the MAST have also been developed: the IO-item Brief MAST (Pokomy, Miller, & Kaplan, 1972) and the 13-item SMAST (Selzer, Vinokur, & van Roojen, 1975). Other similar instruments developed are the Self-Administered Alcoholism Screening Test (SSAST) (Swenson & Morse, 1975) and the Veterans Alcoholism Screening Test (VAST) (Magruder-Habib, Harris, & Fraker, 1982). All versions of the MAST have performed well in validation studies (Miller, 1976). However, the high face validity of some of the questions (e.g., “Do you feel you are a normal drinker?“) has led some investigators to question its clinical usefulness (Goldberg, 1974). Another useful screening instrument which addresses patients’ subjective experiences is the CAGE (Mayfield, McLeod, & Hall, 1974), which can be administered as an interview or in written form. The CAGE is an acronym for four questions: (1) Have you ever felt you should Cut down on your drinking?; (2) Have people Annoyed you by criticizing your drinking?; (3) Have you ever felt bad or Guilty about your drinking?; and (4) Have you ever had a drink first thing in the morning . ..? (Eyeopener). The CAGE has also been shown to be an effective screening instrument in hospital populations with sensitivity and specificity over 80% (Bush, Shaw, Cleary, Delbanco, & Aronson, 1987; Murray, 1974; Mayfield et al., 1974) when a positive test result is defined as affirmative responses to two or more questions. Ewing (1984) has suggested that the CAGE may be used to determine a “high index of suspicion” and that even one positive reply calls for further inquiry. Because the CAGE is so simple to administer and score, since it can be administered in written or interview form, and because the questions are relatively nonthreatening and thus minimally hinder clinical rapport, this instrument is ideal for hospital screening and could easily become incorporated into all history-taking. Although the CAGE can only be used as a gross screening measure and alone is not sufficient for diagnosis, investigators have examined its effectiveness in combination with biochemical markers (Bemadt, Mumford, Taylor, Smith, & Murray, 1982; Bemadt, Mumford, & Murray, 1984) and determined that a combination of selected laboratory tests and a self-report questionnaire are most effective in identifying alcoholism. Some of the most commonly used laboratory tests are the hepatic enzymes; elevated levels of serum glutamic oxalocetic transaminase (SGOT), serum glutamic pyruvic transaminase (SGPT) and gamma glutamyl transpeptidase (GGTP) are markers of excessive alcohol use with relatively high specificities (80~90%) but low sensitivities (20-70%) (Hays & Spickard, 1987; Rosalki, 1984; Paredes, 1982). Elevation of the mean corpuscular volume (MCV), an alteration in blood cells, has also often been used as a marker of alcohol abuse problems (Skinner, Holt, Schuller, Roy, & Israel, 1984). The MCV has a high specificity (89%) (Wu, Chanarin, & Levi, 1974) but low sensitivity (241%) (Cushman, Jacobson, Barboriak, & Anderson, 1984; Korri, Nuutineon, & Salaspuro, 1985). Combinations of these biochemical values have provided varying degrees of sensitivity and specificity. Watson, Mohs, Eskelson, Sampliner and Hartmann (1986) concluded in their review that sophisticated statistical analysis of groups of laboratory tests can separate the extremes; that is, laboratory tests can differentiate the patients with end-organ damage due to alcoholism from patients who rarely drink. But the distinction that needs to be made is much more subtle. For most effective treatment, patients need to be identi-
Detectionof alcohol problems
225
fied before they experience end-stage, alcohol-related diseases. Recent studies have suggested that biochemical tests alone are inadequate detectors of alcohol problems and that a self-report screening measure such as the CAGE should be used in combination with biochemical screening for more effective results (Babor et al., 1989; Bush et al., 1987; Watson et al., 1986; Hays & Spickard, 1987). The present study is an attempt to improve the process of patient screening and to develop practical methods for identifying alcohol problems by examining two different timeefficient screening methods: the CAGE in questionnaire form and review of medical records. In the administration of the CAGE, two self-report questionnaires each including the CAGE were compared; the questionnaires were labeled differently - “Alcohol” versus “Health Habits” - in order to explore the possibility that the labeling would elicit different response sets (i.e., higher versus lower self-disclosure) and produce different results. The questionnaires were given to medical inpatients and the results were compared against a standardized review of blood chemistries and other information from the medical record. METHOD Subjects Subjects were 306 patients admitted to the medicine and family practice units of a community hospital in a small city in the Northeastern United States. All patients admitted to the unit during the previous 48 h were approached and asked to participate in the study. Patients gave written informed consent before responding to the orally administered questionnaire. Subjects were not screened for psychiatric disorders or organic@; in order to establish generalizability, a heterogeneous sample was desired. From the admissions log book for each hospital unit, 471 patients were identified as eligible for the study. Reasons for nonparticipation included: (1) patient unable to speak (i.e., in a coma, asleep and did not wake up when approached, intubated, or highly medicated) (n = 119; 25.3%); (2) patient in severe pain and asked not to participate (n = 13, 2.8%); (3) patient not in his/her bed and could not be located by the investigators (n = 20, 4.2%); (4) patient could not speak English and no translator was available (n = 6, 1.3%); (5) patient with visitors and asked to participate at a later time (n = 31, 6.6%); (6) patient with a physician or nurse and could not be disturbed (n = 12, 2.6%); (7) krftient not allowed visitors due to radiation treatment or highly infectious diseases (n = 10, 2.1%); and (8) patient refused to participate in the study (n = 24, 5.1%). No patients refused after signing the consent form. Of the 306 patients surveyed, 48% were female and 52% were male with a mean age 58.98 years (SD = 16.88, range = 18-95). Measures Each subject was randomly assigned to complete one of two questionnaires in which the CAGE was embedded in a series of questions. In the Health Habits Questionnaire, other questions pertained to sleeping, smoking and caffeine habits; in the Alcohol Use Questionnaire, the other questions inquired further about alcohol use. The questionnaires were approximately equal in length, and the average patient took 5 min to respond. The Health Habits Questionnaire was developed in order to examine the effectiveness of a questionnaire which has questions about alcohol use (the CAGE) among more general questions about health, so that the subject would not be aware that the alcohol ques-
226
BARBARA
L. NILES and BARBARA
S. McCRADY
tions were singled out, but was answering the questions with the idea that he/she was providing general health information for the health care professionals. The Alcohol Questionnaire was developed to evaluate the CAGE in the context of a brief interview about alcohol use. Procedures The medical records of the 306 study subjects were screened for indicators of alcoholrelated problems using a standardized method. Specific blood chemistries associated with heavy drinking were examined, including MCV, total bilirubin, GGTP, SGOT, SGPT, Hct, MCH, potassium, albumin, and BUN. Three categories of assessments were made from the physician’s and nurse’s notes: physician’s assessment of alcohol problems, screener’s assessment of alcohol problems after reading physician history, physical and progress notes, and nurse’s assessment of alcohol problems. The screener’s assessment after reading the physician’s entries reflected the screener’s opinion after reading the history and physical rather than the physician’s opinion about the patients’s alcohol use. Screeners rated the physicians’ and nurses’ notes as containing no information regarding alcohol use, or information indicating no, possible, or definite alcohol problem. The screener then made an overall assessment about the patient’s drinking behavior (i.e., no, possible, probable or definite problem) using all the information available from the medical record. The screening of medical records was done by two staff members of the specialized alcohol consultation and treatment program at the medical center: the director, a clinical psychologist with a behavioral orientation; and a registered nurse, certified alcoholism counselor with a master’s degree in psychiatric nursing who has a disease-model approach. Screeners did not administer the CAGE and were blind to the results. Reliability for the four categorical responses which required a judgement to be made (nurse’s assessment, physician’s assessment, screener’s assessment of physician’s notes, and screener’s assessment of alcohol problem) was determined prior to the commencement of the study and there were reliability checks at three times during the study. At least 80% agreement between screeners was met for each of the four categorical responses. Specific guidelines for making these assessments, with an example for each possible category, were reviewed with the screeners prior to the first reliability check. Screeners were instructed to use all the available information from the medical record to determine the overall screener’s assessment of an alcohol problem. RESULTS
Incidence of alcohol problems CAGE. Of the total population of subjects, 52 (16.9%) endorsed two or more CAGE questions - 10 (6.8%) of the females and 42 (26.6%) of the males. Questionnaires. Among the 152 patients who answered the health questionnaire, 24 (15.8%) endorsed two or more items on the CAGE. Among the 154 patients who answered the alcohol questionnaire, 28 (18.1%) endorsed two or more items on the CAGE. There were no significant differences in the number of positive responses between the two questionnaires for any of the four CAGE questions or for the combined CAGE score. Screening assessment. The assessment from screening medical records concluded that 11.4% of the total subjects (6.1% of the females and 16.3% of the males) were probable or definite problem drinkers.
227
Detection of alcohol problems
Table 1. Association
between CAGE scores and information from medical records CAGE score
Negative or “possible” Positive “probable” or “yes” “no”
Medical records
Negative 0 or 1
Positive 2, 3, or 4
235
36
19
16
The physician provided no information regarding alcohol use for 118 (38.6%) patients. The physician’s assessments indicated that 21 (6.9%) had a possible drinking problem and 15 (4.9%) had a definite drinking problem. The screener’s assessment from the physician’s notes indicated more problem drinking than did the physicians’ assessments: for 30 patients (9.8%) a possible drinking problem was indicated and for 28 (9.2%) a definite drinking problem was inferred. The nurse’s assessment provided no information regarding alcohol use for 85 (27.8%) of the patients, indicated a possible drinking problem for 19 (6.2%) and a definite drinking problem for 6 (2.0%) of the patients. For female patients, the nurses indicated no definite drinking problems at all, and only 5 (3.4%) possible drinking problems. Association between CAGE and screener’s assessment. There was a significant association between the cumulative CAGE scores and screeners’ assessments when these variables were split into dichotomous categories, x2 (1, 306) = 14.72, p < ,001. A cumulative score of 0 or 1 on the CAGE was considered negative on the CAGE and 2, 3, or 4 was considered to be positive on the CAGE. A screening assessment of “no” or “possible” was considered negative medical record information and an assessment of “probable” or “yes” was considered positive medical record information. Table 1 shows the number of subjects falling into each category. The subjects who had negative medical record information, but were positive on the CAGE were significantly older (mean age = 59.78) than the subjects who had positive medical record information and were negative on the CAGE (mean age = 46.63), (t (53) = 2.95, p < .Ol). Among the patients who were positive on the CAGE, there were no statistically significant differences on any of the biochemical markers between those subjects who had negative medical record information and those subjects who had positive medical record information. However, there were some substantial absolute differences between these two groups on the liver enzymes, which probably were not statistically significant due to the very large variances. For example, among those subjects who were positive on the CAGE, the mean GGTP was 75.8 (SD = 104.8) for negative medical record information subjects and 143.3 (SD = 242.2) for the positive medical record information group, t (17) = 1.03, p = .317. There was a significant difference between these two groups in the amount of information provided by the physician. In the positive medical record information group, the physicians provided some information about alcohol consumption in every case, but in the negative medical record information group the physicians failed to provide any information about alcohol consumption in 44% of the cases, t (35) = 5.25, p < .OOl .
228
BARBARA
L. NILES and BARBARA
Table 2. r tests for biochemical
S. McCRADY
markers
Chart screen
Variable MCV Total bilirubin GGTP SGOT SGPT HCT MCH Potassium Albumin BUN
Negative Mean/(W) 89.15 (6.93) S870 (.716) 44.86 (59.2) 31.45 (39.7) 24.63 (29.8) 38.09 (6.4) 30.27 (2.5) 4.09 (.53) 3.59 (.59) 19.67 (13.1)
n 260 247 223 252 223 262 260 248 247 256
Positive Meani 92.76 (8.02) .9147 (.592) 151.66 (221 .O) 57.94 (55.6) 53.71 (52.9) 41.13 (7.6) 31.65 (3.0) 3.94 (.67) 3.58 (.94) 17.29 (8.2)
n
t/(dfI
34
3.00’ (292) 2.55’ (279) 5.96”’ (254) 3.47” (284) 4.69”’ (255) 2.55’ (294) 2.96” (292) 1.52 (280) 0.05 (279) 1.03 (288)
34 33 34 34 34 34 34 34 34
*p < .05; **p < .Ol; **-p < .001
Alcohol and health data Alcohol. For those patients who answered the alcohol questionnaire, the correlation between the reported number of drinks per week and the CAGE score was significant (1. = .4436, p < ,001). When the analyses were done separately by sex, some gender differences emerged. The correlation between the CAGE and the number of drinks per week was substantially greater for the females (r = .8179, p < .OOl), while the same correlation for the males was much lower (r = .3799, p < .OOl). When the frequencies of “drinks per occasion” were examined for the men who were negative on the CAGE, it was found that only 4.7% of them reported drinking more than 4 drinks per occasion, while 46.2% of the men who were positive on the CAGE reported drinking more than 4 drinks per occasion. Health. For those patients who answered the health questionnaire, the correlations between the CAGE and the health information also showed gender differences. The correlation between the CAGE and the number of packs of cigarettes per day was significant for all subjects (r = .2447, p < .OOl) and for the females (P = .3883, p < .OOl), but not for males. When the frequency of the smokers and the number of packs of cigarettes per day were examined for women who were negative on the CAGE, it was found that only 18% smoked cigarettes at all and 4% smoked more than 1 pack per day. In contrast, for women who were positive on the CAGE, 63.5% smoked and 25% smoked more than one pack per day. What determines a screening assessment t tests were done on each of the biochemical markers contributing to the screening assessment to determine if there were significant differences between patients who were
229
Detection of alcohol problems
Table 3. Regression
analyses of factors determining assessment for all subjects
the screening
Variable
? Change
Screener reading of MD notes GGTP Nurse assessment Diagnosis SGF’T MCV
s3577” .07835”’ .03337” .02332” .01825” .01172’
*p
