Appendix A:

Assessment Feedback Procedures

by William R. Miller, PhD.


Preface
The instructions contained in appendix A refer to the assessment
feedback components of Motivational Enhancement Therapy, as prac
ticed in Project MATCH. It is not necessary, however, to use exactly
the same assessment instruments as were employed in Project
MATCH. The basic idea is to assess a range of dimensions, with
particular emphasis on those likely to reflect early problems or risk.
If you wish to replicate the exact procedures used in MATCH, infor
mation is provided at the end of this appendix for obtaining the needed
instruments. You may, however, construct your own assessment bat
tery and design a corresponding Personal Feedback Report (PFR)
based on normative data for the instruments you have chosen. The
PFR used in Project MATCH is reproduced following page 89.
In general, your assessment battery should sample a variety of poten
tial problem and risk domains. Here is a brief list of pertinent domains,
with examples of appropriate assessment approaches for each.


Alcohol              The volume of alcohol consumption is a primary dimension for assess
Consumption    ment, because all other risk and problem domains are related to the
quantity and frequency of use. There are four basic approaches for
quantifying alcohol consumption.


Quantity/              The simplest approach is to ask a few structured questions regarding
Frequency             the frequency (e.g., how many days per month does the person drink)
Questionnaire       and quantity of consumption (e.g., on a drinking day, how many drinks
does the person have on average). Such questions can be aided by
describing a standard drink unit (see Miller et al. 1991 for alternatives)
or asking separately about different kinds of beverages (beer, wine,
spirits, etc.). An advantage of this approach is that, unlike the others,
it can be administered by paper and pencil questionnaire. This method
appears to underestimate actual consumption, however, and reliabil
ity and validity parameters have not been established.

Grid Averaging A second approach is to reconstruct, by structured interview, a typical
drinking week and then account for episodes of drinking that deviate
from this pattern. This approach was introduced by Miller and Marlatt

Timeline                A third and still more detailed approach is to reconstruct drinking by
Followback            filling in an actual calendar for the past few weeks or months. Day by
day drinking data are obtained, taking advantage of the memory
prompting value of a calendar (Sobell et al. 1980). The Form 90
approach used in Project MATCH (see below) represents a hybrid of
the timeline and grid averaging methods.


Drinking Diary Finally, individuals can be asked to keep a daily diary of alcohol
consumption. These records can than be converted into quantitative
data. A freeware computer program for this purpose has been devel
oped by Markham, Miller, and Arciniega (see resource list at the end
of this appendix).

Alcohol-Related         As heavy drinking continues, life problems tend to accumulate. Some
Problems                   counting of such accumulation is a common measure of problem
severity. Measures such as the Michigan Alcoholism Screening Test
(MAST; Selzer 1971) combine life problems with other factors such as `
alcohol dependence symptoms and help seeking. Miller and Marlatt
(1984) attempted to differentiate between common problematic conse
quences of heavy drinking and other life problems, which may or may
not be alcohol related. The DRINC questionnaire (see below), devel
oped for Project MATCH, is intended as a purer measure of negative
consequences of drinking, apart from alcohol dependence signs.

Alcohol                      The alcohol dependence syndrome is currently a central diagnostic
Dependence              concept. Severity of dependence represents a third dimension to be
tapped in comprehensive assessment. A variety of alcohol dependence
scales have been published. Skinner's Alcohol Dependence Scale
(Skiller and Horn 1984) has been a popular instrument in North
America, with strong pyschometric characteristics.

Physical Health Heavy drinking also has predictable effects on physical health. The
most common evaluation approach in this domain has been a serum
chemistry profile, screening for elevations on variables commonly
affected by excessive drinking. These include liver enzymes (SGOT,
k
SGPT, GGT), mean corpuscular volume (MCV), and high-density
lipoprotein (HDL). Blood pressure can also be screened, because heavy
drinking contributes to hypertension.

Neuro-                      Knowledge of all of the above domains provides relatively little infor
psychological             mation about a person's cognitive functioning. Problem drinkers have
Functioning              been found to be impaired on a variety of neuropsychological tests
                                                (Miller and Saucedo 1983).

Both Project MATCH and other checkupand feedback interventions have included neuropsychological test results (see Miller and Sovereign 1989; Miller et al. 1988), although interventions can also be effective without the inclusion of neuropsychological testing (Bien and Miller submitted; Brown and Miller submitted). Tests that commonly show impairment include the Block Design and Digit/Symbol subtests of the Wechsler Adult Intelligence Scale, the Wisconsin Card Sorting Task, and Halstead-Reitan subtests including the Tactual Performance Test, the Trail-Making Test, and the Categories Test.


Risk Factors Markers of high risk for alcohol problems can also be measured, apart
from the individual's current level of use and its consequences. Family
history of alcohoVdrug problems can be obtained by a variety of
methods (e.g., Cacciola et al. 1987; Miller and Marlatt 1984). Of
personality scales designed to detect correlates of risk for substance
abuse, the MacAndrew scale has fared best in research, though others
are.available (Jacobson 1989; Miller 1976). Beliefs about alcohol, as
assessed by Brown's Alcohol Expectancy Questionnaire, have also
been found to be predictive of risk (Brown 1985).


Motivation for          Various approaches are available for measuring the extent of an
Change                     individual's motivation for changing drinking. Some consist of simple
Likert scales assessing commitment to abstinence or other change
goals (e.g., Hall et al. 1990). Self-efficacy scales can be constructed to
ask about confidence in one's ability to change. Respondents can be
asked to rate the extent to which alcohol is helping or harming them
on a range of life dimensions (Appel and Miller 1984). Stages of change
derived from the Prochaska and DiClemente (1984) theoretical per
spective were used as the basis for construction of the University of
Rhode Island Change Assessment (Prochaska and DiClemente 1992;
DiClemente and Hughes 1990) and the alcohol-specific Stages of
Change Readiness and Treatment Eagerness Scale (SOCRATES;
Miller).


Comprehensive         Several questionnaires and structured interview protocols provide a
Assessment              range of quantitative scores that can be compared with normative or
Approaches               diagnostic standards. None of these taps all of the above dimensions,
but each provides a basis for judging status on several domains. The
Alcohol Use Inventory (AUI; Horn et al. 1987) is a widely used and
well-developed self-administered questionnaire that permits compar
ison of individual with normatived scores. The materials necessary to
administer, score, and interpret the AUI are available from National
Computer Systems, P.O. Box 1416, Minneapolis, MN 55440. The kit
includes the AUI manual, forms, client test book, hand-scored answer
key templates, and the AUI profile sheet, which summarizes the scores
and can be given to the client. Structured interviews include the
Addiction Severity Index (ASI; Cacciola et al. 1987), the Comprehens
sive Drinker Profile (CDP; Miller and Marlatt 1984, 1987), and the
Form 90 interview developed for Project MATCH (see below).
The crucial point is that the battery of assessment procedures to be
used as a basis for feedback can be tailored to the needs, time demands, "
and client characteristics of a program. What follows is but one
example-from Project MATCH-of how assessment feedback can bedone within the context of Motivational Enhancement Therapy.

 

The Project MATCH Assessment Feedback Protocol and Procedures for Completing The PFR

Prior to the first session with an MET client, the Personal Feedback

Report is prepared by obtaining the pertinent data from the client's

file. The following information from the Project MATCH assessment

battery is used:

 

* AUDIT score from the Quickscreen

* Form 90-I (Initial Intake)

* ASI family history section  

* MacAndrew scale score

* DRINC questionnaire  

* Serum chemistry profile

* Neuropsychological test results

* Alcohol Use Inventory

 

BACCuS, an IBM-PC software program, is used for converting alcohol

consumption data into standardized measures (Markham et al. submitted). i

 

Alcohol              The first datum to be presented to the client is the number of standard

Consumption    drinks consumed during a week of drinking. This calculation is avail- I,

able from Form 90-I, the Project MATCH interview protocol for E

quantifying alcohol consumption. Some degree of judgment is needed

here, but remember that the goal is to provide clients with a fair

picture of their alcohol consumption during a typical drinking week.

 

If the Steady Pattern Chart has been completed (page 6), use line 38

as the number of standard drinks per week. If no Steady Pattern Chart

has been completed, the client's drinking was too variable to provide

a consistent weekly pattern. In this case, consult the Summary Statistics sheet.

 

If the client abstained on fewer than 10 percent of days

during the 90-day window, multiply the "Average SECs per drinking 4 day" by 7 to obtain the number of standard drinks per week. Be sure you are examining the 90-day window and not the whole current period. If abstinent days exceed 10 percent, examine the calendar to determine whether these abstinent days mostly occurred within drinking weeks (e.g., no drinking on Monday through Wednesday) or whether they occurred in blocks in between periods of drinking (i.e., periodic drinker). In the former case, determine the typical number of drinking days in an average week and multiply this number of days by the Average SECs per drinking day (from the Summary Sheet) to obtain the number of standard drinks per week. In the latter case-a purely periodic drinker-determine from the calendar whether drinking episodes are normally at least 7 days in length. If so, use the same procedure as for the Steady Pattern Chart: multiply the Average SECs per drinking day by 7 to describe the number of standard drinks consumed during a typical week of drinking. If drinking episodes are typically shorter than 1 week (e.g., 3 days), multiply the average number of days in an episode by the Average SECs per drinking day (from the Summary Statistics). Again, remember that the guiding principle is to describe the number of standard drinks that the client consumed, on average, in a drinking week.

 

When you have obtained the client's average number of drinks per drinking week, use table 3 to obtain the client's percentile among American adults. Note the separate norms for men and women.


Estimated Blood      The second set of data presented to Project MATCH clients consists of
Alcohol                     computer-projected blood alcohol concentration (BAC) peaks, based on
Concentration          alcohol consumption patterns reported on Form 90-I. These projec
Peaks                       tions are computed by BACCuS and will normally have been com
                                pleted by the research assistant who conducted the Form 90-I
interview. Nevertheless, you should check these calculations using
BACCuS. Any projected peak over 600 mg% should be reported as 600
mg%. The reasoning here is that projections above this level are likely
to be overestimates, because actual BAC peaks above 600 mg%, though
possible, are relatively rare.

The BAC peak for a typical drinking week is obtained from line 39 of
Form 90-I. This is the highest intoxication peak from the typical
drinking week grid. Note that it may be necessary to use the BACCuS
program (Menu #3, BAC Peak for an Episode) to estimate BAC peaks
for several different days in order to determine which yielded the
highest BAC. It is not always obvious, from visual inspection, which
period will produce the highest BAC peak. Where a day contains at
least two periods of drinking separated by several hours (e.g., 6 drinks
from noon until 2:00 pin and then 8 drinks from 7:00-11:00 pm), it is
wise to try the BAC level for each period within the day, as well as for
the whole day. (In the above example, you would run 6 drinks in 2
hours, 8 drinks in 4 hours, and 14 drinks in 11 hours. The resulting
BAC projections for a 160-pound male would be 109, 124, and 152,
respectively. In this case, the BAC of 152, from 14 drinks in 11 hours, would be used.) If the Steady Pattern Chart was not completed on 90-I, leave this line blank.

 

The BAC peak for a heavier day of drinking is obtained from the Highest Peak BAC line of the Summary Statistics sheet. This represents the highest BAC peak reached during the 90-day period. This will never be lower than line 39 but may be the same as line 39. In this case, the number on both lines of section 2 would be the same.

 

Risk Factors The third feedback panel on the PFR reflects five risk factors. Higher

scores on these scales are associated with greater risk and severity of

alcohol-related problems.

 

Tolerance Level Tolerance level is inferred from the BAC peaks reached during the

90-day window. The rationale is that the higher the projected BAC

peak, the higher the individual's tolerance. Use the higher of the two

numbers in Section 2 to arrive at the classification:

 

0-60 mg% Low tolerance

61-120 mg% Medium tolerance

121-180 mg% High tolerance

181 mg% + Very high tolerance

 

Other Drug Risk Other drug risk is judged from the lifetime use of other drugs, as reported on page 10 of Form 90-I. The rationale is that more frequent use of other drugs, or any use of drugs with higher dependence potential, is associated with greater risk for serious consequences and complications. Use the following classification system:

 

HIGH RISK Any use of cocaine or crack

or Any use of heroin, methadone, or other opiates

or Frequent use (more than 3 months of at least once per week) of any

other drug class except tobacco:

Marijuana, Hash, THC

Amphetamines, Stimulants, Diet Pills

Tranquilizers

Barbiturates

 

MEDIUM RISK Any lifetime nonprescription use, but not frequent use (i.e., 3 months or

less of weekly use) of any drug class except tobacco, opiates or cocaine:

Marijuana, Hash, THC

Amphetamines, Stimulants, Diet Pills

Tranquilizers

Barbiturates

 

LOW RISK No use of other drugs (Code = 0 for all 10 drug classes except tobacco)

 

Family Risk Family risk is judged from the family history of alcohol and other drug

problems obtained in the ASI interview. The following weighting

system is used to arrive at a total Family Risk score. Assign the designated number of points for each blood relative indicated to be positive for alcohol/drug problems:

 

If father positive add 2 points

If mother positive add 2 points

For each brother positive add 2 points

For each sister positive add 2 points

For each grandparent positive add I point

For each uncle or aunt positive add 1 point

Risk levels are judged according to the following classification system:

 

Family Risk Classifications

 

0-1 Low risk

2-3 Medium risk

4-6 High risk

7+ Very high risk

 

MacAndrew Scale The MacAndrew Scale score can be obtained directly from this scale. The following classification system is used for risk:

 

MacAndrew Scale Risk Levels

 

0-23 Normal range; lower risk

24-29 Medium risk

30+ High risk

 

Age at Onset Age at onset is the fifth risk factor in this panel. The rationale is that

younger onset of problems is associated with a more severe course and

symptomatology. Age at onset is calculated by the following procedure,

using three items obtained from the DRINC (Drinker Inventory of

Consequences) scale.

 

Calculating Age at Onset

 

1. Record these three numbers, if applicable, and sum them

    (from page 7 of Drinker Inventory of Consequences)

 

Age of first regular intoxication (item 17): Age of first loss-of-control (item 18): +

 

Age of first alcohol problems (item 19): +

 

TOTAL 2. Divide by the number of ages used in step 1:

 

Age at onset =

 

NOTE: If an age item was not recorded for the client (e.g., the client had never experienced loss of control), the average is based on the other two age items (divide by 2). If only one age item was completed, this constitutes the age at onset.


Risk level is judged according to this classification system:

 

Under 25.0 Higher risk

25.0-39.9 Medium risk

40.0 + Lower risk

 

Problem Severity The AUDIT score is recorded directly from this scale within the Quickscreen. The DRINC alcohol severity score is recorded directly from this questionnaire and is the sum of scores for the 55 lifetime consequences. Print the client's raw score for each of these two scales under the corresponding severity range (e.g., a 19 on the AUDIT would be printed under the HIGH descriptor, below the 16-25 range designation.)

 

The other information reviewed in the fourth panel is the profile of results from the AUI. Use the AUI Profile form, published by National Computer Systems, for this purpose. Circle the client's raw scores for all scales and connect the circles with straight lines. Do not cross the solid lines that divide categories.

 

Serum Chemistry Obtain the client's serum chemistry scores on SGOT, GGTP, SGPT, uric acid, and bilirubin (total) from the lab report. Record these lab scores on the corresponding lines of the PFR. Interpretive ranges are shown on the PFR.


Neuro psychological - A 5-point performance scale is used to interpret neuropsychological
 test results:

Test Results

1 Well above average
2 Above average
3 Average
4 Below average
5 Well below average

The scoring systems below attempt to correct for effects of age and/or
education level, based on available norms. The Shipley-Hartford
Vocabulary test is used as a "hold" test that is less likely to be affected
by alcohol, thus providing an estimate of the level of performance that
would ordinarily be expected from an individual.

Shipley-Hart ford Use the age-adjusted score to obtain a normalized T-score, as specified
Vocabulary Test in the revised Shipley-Hartford manual. Then use the following table
(SV) to convert the T-score into our 1-5 scale:

>_ 63 1 Well above average

57-62 2 Above average

44-56 3 Average

38-43 4 Below average

<_37 5 Well below average

 

Shipley-Hartford    se the age-adjusted score to obtain a normalized T-score, as specified

Abstraction Test     in the revised Shipley-Hartford manual. Then use the following table

(SHVA) to convert the T-score into our 1-5 scale:

 

>_ 63 1 Well above average

57-62 2 Above average

44-56 3 Average

38-43 4 Below average

<_ 37 5 Well below average

 

Trail-Making The score is the number of seconds to complete Form A.

 

Test, Form A Age

 

(TMTA) 20-39 40-49 50-59 60-69

1 <_ 21 <_ 22 <- 25 <- 29

2 22-26 23-28 26-29 30-35

3 27-41 29-44 30-48 36-66

4 42-49 45-58 49-66 67-103

5 >- 50 >- 59 >_ 67 >_ 104

Based on Lezak 1976, Table 17-6, page 558. Cutting points represent the 10th, 25th, 75th, and 90th percentiles.

 

Trail-Making The score is the number of seconds to complete Form B.

 

Test, Form B Age

 

(THTB) 20-39 40-49 50-59 60-69

1 <_ 45 <- 49 <_ 55 _< 64

2 46-55 50-57 56-75 65-89

3 56-93 58-99 76-134 90-171

4 94-128 1Q0-150 135-176 172-281

5 >_ 129 >_ 151 >_ 177 >_ 282

                        Based on Lezak, 1976, Table 17-6, page 558. Cutting points represent the 10th, 25th, 75th, and 90th percentiles.

 

Symbol Digit The score for the Symbol Digit Modalities Test is the number of correct

 

Modalities Test digits associated with their respective symbols within the 90-second

(SYDM) written testing period.

 

Use this table if client has 12 years or less of education.

 

Age 1 2 3 4 5

18-24 >_ 67 63-66 47-62 42-46 <- 41

25-34 >_ 65 61-64 46-60 41-45 <_ 40

35-44 _> 64 60-63 44-59 39-43 S 38

45-54 >_ 62 57-61 39-56 33-38 :5 32

55-64 >_ 55 51-54 36-53 31-35 <_ 30

65+ >_ 47 42-46 25-41 20-24 <_ 19

 

Table 3. Alcohol consumption norms for U.S. adults, in percents

 

Drinks per week Total Men Women

 

0 35 29 41

 

1 58 46 68 l

 

2 66 54 77

 

3 68 57 78

 

4 71 61 82

 

5 77 67 86

 

6 78 68 87

 

7 80 70 89

 

8 81 71 89

 

9 82 73 90

 

10 83 75 91

 

11 84 75 91

 

12 85 77 92

 

13 86 77 93

 

14 87 79 94

 

15 87 80 94

 

16 88 81 94

 

17 89 82 95

 

18 90 84 96

 

19 91 85 96

 

20 91 86 96

 

21 92 88 96

 

22 92 88 97

 

23-24 93 88 97

 

25 93 89 98

 

26-27 94 89 98

 

28 94 90 98

 

K.

 

29 95 91 98

 

30-33 95 92 98

 

34-35 95 93 98 E'

 

F

 

36 96 93 98

 

37-39 96 94 98

 

40 96 94 99

 

41-46 97 95 99

 

47-48 97 96 99

 

49-50 98 97 99

 

51-62 98 97 99

 

63-64 99 97 >99.5

 

65-84 99 98 >99.6

 

85-101 99 99 >99.9

 

102-159 >99.5 99 >99.9

 

160+ >99.8 >99.5 >99.9

 

Source: 1990 National Alcohol Survey, Alcohol Research Group, Berkeley.

 

Courtesy of Dr. Robin Room