Screening medical patients for alcoholism with the MMPI: A comparison of seven scales

Abstract

Seven alcoholism scales for the MMPI were evaluated for effectiveness in screening medical patients. Sensitivity and specificity were assessed by use of five contrast samples: (1) 736 inpatients from an alcohol and drug dependence unit; (2) 4,923 unselected medical outpatients; (3) 7,044 selected medical outpatients; (4) 214 psychiatric inpatients; and (5) 1,408 contemporary normal persons. None of these scales can be recommended for use in screening medical patients. However, endorsement of MMPI item 215 should lead to further evaluation because classification was more accurate with this single item than with any of the scales evaluated.

The CAL: An MMPI alcoholism scale for general medical patients

“…Negative Predictive Values were 90% (87% to 94%) for the CAL and 89% (85% to 92%) for the MacAndrew scale. The primary difference between the two measures was the higher percentage of false-positives on the MacAndrew scale, a finding that we noted in prior investigations (Colligan et al, 1988b;Davis et al, 1987).…”

“…We started our investigation in an effort to devise a practical MMPI scale that would be useful in screening for alcoholism in a predominantly middle-class general medical patient population. Our prior research (ColIigan et a]., 1988a;1988b, Davis et al, 1987 led us to the conclusion that, when applied to our target population, existing alcoholism scales yield an unacceptable number of false-positive results when sensitivity is set at aThe percentage of persons in the respective samples who have CAL T-scores less than the value indicated. what we consider to be a meaningful level.…”

“…With new cross-validation samples, with sensitivity (proportion of alcoholics correctly identified) set at 90%, the specificity (proportion of nonalcoholics correctly identified) ranged from 90% to %% for selected contrast samples. We conclude that this scale shows promise as a screening device in similar populations.From previous studies (Colligan, Davis, Morse, & Offord, 1988a, 1988bDavis, Colligan, Morse, & Offord, 1987) we concluded that existing MMPI-derived alcoholism scales lack adequate specificity when sensitivity is set at what we consider appropriate levels for general medical samples. Furthermore, we noted that previous investigators typically did not include alcoholic women in their criterion groups, a serious omission in our view.Another issue is the method of scale construction used in the development of such measures.…”

“…From previous studies (Colligan, Davis, Morse, & Offord, 1988a, 1988bDavis, Colligan, Morse, & Offord, 1987) we concluded that existing MMPI-derived alcoholism scales lack adequate specificity when sensitivity is set at what we consider appropriate levels for general medical samples. Furthermore, we noted that previous investigators typically did not include alcoholic women in their criterion groups, a serious omission in our view.…”

“…Another issue, which we have commented on in the past (Colligan et al, 1988a(Colligan et al, , 1988bDavis et al, 1987), is that of the reliability of existing MMPI-derived alcoholism scales. We have found that internal consistency reliability is generally low, which raises concerns about the validity of such measures.…”