Pain, 32 (1988) 207-213 207

Elsevier

PA1 01150

Risk of misinterpretation of MMPI Schizophrenia scale elevations

in chronic pain patients

James E. Moore *3* * *, Miles E. McFall * * ’ * * *, Daniel R. Kivlahan * *’ * * *

and Fred Capestany **
* Virginia Mason Medical Center, 1100 Ninth Avenue, Seattle, WA 98101 (U.S.A.), * * VA Medical Center, 1660 South Columbian
Way, Seattle, WA 98108 (U.S.A.), and * * * Department of Psychiatry and Behavioral Sciences, University of Washington School of
Medicine, Seattle, WA 98195 (U.S.A.)

(Received 16 February 1987, revised received 13 March 1987, accepted 17 August 1987)

SummarY Seventy-three chronic pain patients with elevated MMPI Schizophrenia (SC) scale scores (T score > 70) were
compared with 55 psychotic and 87 non-psychotic psychiatric patients with elevated SC scores to examine group differences in item
content patterns on the Harris and Lingoes subscales. Chronic pain patients evidenced lower scores on all Harris and Lingoes SC
subscales, except for the Bizarre Sensory Experiences subscale on which they scored significantly higher than the psychiatric groups.
Results demonstrate that SC is elevated in many chronic pain patients because they endorse somatic symptoms and items suggestive
of depression and inertia, whereas psychotics endorse more items reflecting bizarre and disordered thinking, social alienation and
defective inhibition, and non-psychotic psychiatric patients endorse more depression, despair, thought disorganization and social
alienation. These data suggest that high SC scores of many chronic pain patients reflect symptoms and sequelae of their physical
problems, and do not necessarily reflect severe psychopathology.

Key words: Chronic pain patients; MMPI Schizophrenia scale; Group differences

Introduction The MMPI is a 566-item true/false personality

It is now commonly accepted that chronic pain
is influenced not by tissue damage alone, but also
by physical deconditioning and a variety of psy-
chological factors including the patient’s focus of
attention, emotional state, and past learning his-
tory [13]. Comprehensive evaluation of patients
with chronic pain, therefore, must include a psy-
chological assessment as well as a physical ex-
amination. In addition to an interview and behav-
ioral observation of the patient, a psychological
evaluation frequently includes a Minnesota Multi-
phasic Personality Inventory (MMPI) [ll].
Correspondence to: James E. Moore, Ph.D., Virginia Mason
Medical Center, 1100 Ninth Avenue, Seattle, WA 98101, U.S.A.
questionnaire developed using an empirical keying
approach to discriminate patients in various psy-
chiatric diagnostic groups from normal control
subjects. Eight clinical scales of the MMPI were
originally developed to measure hypochondriasis,
depression, hysteria, psychopathic deviancy,
paranoia, psychasthenia, schizophrenia, and
mania. Regardless of item content, only those
items that reliably discriminated between normals
and subjects in a particular diagnostic group (e.g.,
schizophrenia) were included in that clinical scale.
While the MMPI remains a valuable instrument
for psychological assessment, it has not been
accepted as an adequate diagnostic tool [8]. De-
spite several limitations, clinically useful inferences
and predictions about individual patients’ behav-

0304-3959/88/$03.50 0 1988 Elsevier Science Publishers B.V. (Biomedical Division)

208
ioral tendencies can be made based on their pro-
file configurations [3,5,7].
Behavioral predictions based on MMPI profiles
of pain patients are based largely on prior em-
pirical research on behavioral correlates of MMPI
profiles with psychiatric patients. Without cross-
validation of these findings in a chronic pain
patient sample it remains undetermined whether
particular MMPI profiles will be predictive of the
same behavioral characteristics as they are in psy-
chiatric patients.
Several investigators [6,16] have observed that
the Hypochondria+ Depression and Hysteria
scales of the MMPI contain a large number of
items which assess somatic complaints, such as ‘I
have few or no pains,’ and ‘My sleep is fitful and
disturbed.’ These items are likely to be endorsed
by chronic pain patients for reasons other than the
presence of psychopathology. These patients, who
may have little or no psychological dysfunction,
are therefore at risk of being labeled as hypo-
chondriacal, depressed or hysterical. Taylor [18]
has demonstrated a similar mislabeling risk with
spinal cord injured patients.
Like other scales of the MMPI, the Schizo-
phrenia (SC) scale is heterogeneous in item content
because of the empirical keying approach guiding
construction. Recently, however, there has been
increasing emphasis on examining item content of
the MMPI scales to enhance the interpretive use-
fulness of the test [9]. One approach has been to
develop homogenous content dimensions of the
MMPI clinical scales on the basis of logical group-
ing of items with similar content into subscales
WI.
The usefulness of the Harris and Lingoes sub-
scales recently has been demonstrated by Prokop
[15], who found that pain patient MMPI profiles
with Hypochondriasis and Hysteria scales elevated
above a T score of 70, that is with a ‘conversion
V’ configuration, did not necessarily indicate the
actual presence of histrionic personality tenden-
cies. He examined the Harris and Lingoes sub-
scales of the Hysteria scale, and found that the
subscales thought to represent histrionic tenden-
cies, that is, Denial of Social Anxiety, Need for
Affection, and Inhibition of Aggression, were not
elevated in pain patients relative to norm,tl L~~~~~-
trols. The other subscales which assess concern
over physical dysfunction -- l.assitude-Malaise
and Somatic Complaints - were elevated in pain
patients relative to controls and accounted for
most of the elevation on the Hysteria scale. Pro-
kop’s research demonstrates that pain patients
who report fatigue and physical complaints may
be mislabeled as histrionic if the Hysteria scale
alone is examined.
Chronic pain patients also may be mislabeled
as psychotic based on an elevated Schizophrenia
scale (SC) (T score > 70). Several investigators
have cluster analyzed chronic pain patients’ MMPI
profiles, finding 3 discrete, homogeneous sub-
groups of male chronic pain patients [1,2]. In one
of these studies [l], 16% of chronic pain patients
fell into a profile group with a mean SC T score of
approximately 90. These patients had significantly
greater reports of pain intensity, sleep disturbance.
physical limitations, and psychosocial dysfunction
relative to the other two profile groups; however,
consistent with our clinical experience. none of
them evidenced significant thought disorder. Ncv-
ertheless, we have noticed that some clinicians.
particularly those who are inexperienced in the
area of chronic pain. tend to interpret MMPls
with such SC scale elevations as reflecting severe
psychopathology.
The present study was designed to investigate
the meaning of SC elevations in a chronic pain
patient sample. The Harris and Lingoes SC sub-
scales were explored as a means of studying the
content dimensions of the otherwise heteroge-
neous SC scale. We hypothesized that chronic pain
patients with SC elevations would have different
item response patterns than psychiatric patients
with similarly elevated Sc scales. Specifically, we
predicted that chronic pain patients would have
higher scores than both psychotic and non-psy-
chotic psychiatric patients on the Bizarre Sensory
Experiences (BSE) subscale because of items re-
flecting unusual physical sensations, but lower
scores than the psychiatric groups on the other SC
subscales reflecting severe cognitive or affective
disturbance.
 209

Method Procedure
All patients completed an MMPI as part of the
Subjects routine psychological assessment performed in
Three groups of subjects were selected from a each clinical setting. The MMPIs were scored to
sample of male veteran patients undergoing treat- obtain the 3 validity and 10 clinical scales, plus
ment at the Seattle Veterans Administration the Harris and Lingoes subscales for the SC scale.
Medical Center. The chronic pain patient group The Harris and Lingoes subscales for the SC scale
(N = 73) consisted of all patients meeting study are listed in Table I. The K-corrected MMPI T
inclusion criteria out of consecutive referrals to scores for each group are listed in Table II. Be-
the hospital’s pain clinic during 1984-1986. The cause there were significant differences among
average age of pain patients was 42.8 years, and groups for age and overall SC score, we performed
the average duration of chronic pain was 9.5 years. analyses of covariance (ANCOVA) on each Harris
Almost half (45%) of these patients reported pain and Lingoes SC subscale with age and raw SC score
in more than one site. The majority of patients as covariates. When an ANCOVA revealed signifi-
had low back pain (70%), with other pain sites cant differences among groups, pairwise compari-
reported in the following order: lower extremity sons of adjusted means also were performed. For
(30%), headache (l%), shoulder or arm (14%), these comparisons, regression analysis was used to
neck (12%) and abdomen (3%). Subjects in this test differences between unstandard~ed regression
condition had no history of psychosis or brain coefficient b’s for the coded vectors that corre-
injury. spond to the adjusted means 1141.
The psychotic patient group (N = 55, mean age
= 36.4 years) was comprised of consecutive in-pa-
tients on an acute care psychiatric unit during
1984-1986 who met inclusion criteria for the study.
These patients were included in the study of their TABLE I
medical chart documented the presence of a psy- HARRIS AND LINGOES SUBSCALES FOR SCHIZO-
chotic diagnosis and one or more of the following PHRENIA SCALE
characteristics: hallucinations, delusions, signs of
formal thought disorder, or treatment with anti- Subscale name High score descriptors a
psychotic medication. Social Alienation (SclA) Patients feel that they get a
The non-psychotic patient group (N = 87, mean raw deal, that others do not
care or understand, and that
age = 42.5 years) consisted of consecutive outpa- others are against them.
tient psychiatric patients meeting study criteria
during 1984-1986. These patients had mixed psy- Emotional Alienation (SclB) Patients feel despair, apa-
thy, and are ready to give
chiatric diagnoses, but no history of a psychotic
up.
diagnosis or any of the psychotic characteristics
listed above. Lack of Ego Mastery, Patients feel their cognitive
Cognitive (Sc2A) processes are unpaired.
Patients in the psychotic and non-psychotic
groups were included only if their medical charts Lack of Ego Mastery, Patients feel hopeless and
made no reference to the existence of organic Conative (SQB) unable to cope with life.
brain injury or chronic pain. To be included in the Lack of Ego Mastery, Patient feels out of control.
study, patients in all groups had to obtain T Defective I~bition
scores greater than 70 on the Sc scale of the W2C)
MMPI. Only valid profiles, that is those with F Bizarre Sensory Eqeri- Patients complain of un-
scale scores less than or equal to 90, were in- ences (Sc3) usual physical sensations,
cluded. disturbed thinking and/or
hallucinations.

a Condensed from Graham [S].

210

TABLE II the T score means prior to adjustment for co-

K-CORRECTED MMPI T SCORES BY GROUP variates (age and overall SC score) as well as after
adjustment for covariates for each group on each
Scale Pain Psychotics Non- Harris and Lingoes SC subscale. As demonstrated
patients psychotics in Table III the ANCOVAs revealed significant
L 51 53 49 differences among groups for all subscales except
F 65 78 69 Emotional Alienation. Pairwise comparisons of the
K 52 52 50
adjusted means on the Social Alienation subscale
HS 88 70 74
D 82 80 90 revealed that pain patients had significantly lower
HY 79 68 75 scores than each of the psychiatric groups. Be-
Pd 70 77 81 cause the ANCOVA for Emotional Alienation was
Mf 62 67 68 not significant, pairwise comparisons were not
Pa 66 76 69
performed. On both Lack of Ego Mastery
Pt 74 80 84
SC 81 91 87 (LEM)-Cognitive and LEM-Conative subscales,
Ma 63 67 62 pain patients scored significantly lower than the
Si 60 62 65 non-psychotic group, but did not differ from the
psychotic group. On LEM-Defective Inhibition,
pain patients differed significantly from the psy-
chotics but not from the non-psychotics. Despite
Results the apparent similarity of the psychotics and pain
patients on the Bizarre Sensory Experiences sub-
There were significant differences among groups scale, the adjusted mean of the pain patients was
for age, F (2, 211) = 6.5, P = 0.002, and raw SC significantly higher than that of each of the psy-
score, F (2, 212) = 4.6, P = 0.01. Table III shows chiatric groups.

TABLE III

HARRIS AND LINGOES SC SUBSCALE UNADJUSTED (ADJUSTED) T SCORE MEANS BY GROUP

SC subs&e Pain patients Psychotics Non-psychotics F (2,212)

Social Alienation 55.1 67.0 65.0 12.9 ***

(58.4) a,b (64.4) (63.9)

Emotional Alienation 55.0 58.0 61.5 4.7

(57.0) (56.7) (60.7)

LEM-Cognitive 65.6 70.8 74.5 5.8 *

(68.8) b (68.9) (73.0)

LEM-Conative 67.3 70.3 77.7 11.9 ***

(69.7) b (68.8) (76.5)

LEM-Defective Inhibition 60.2 70.2 63.7 7.7 **

(63.6) a (67.4) (62.7)

Bizarre Sensory Experiences 73.4 72.0 64.2 8.9 ***

(75.9) a*b (70.5) (62.8)

a,b Pairwise comparisons of adjusted means: a Significantly different from psychotics (P < 0.05); b Significantly different from
non-psychotics (P < 0.05).
* P < 0.05; * * P < 0.01; * * * P < 0.001.
 211

Discussion greater social ~enation, more disturbances in

The results demonstrate that SC scale elevations
of chronic pain patients often may be the result of
a different item response pattern than similar SC
scale elevations of psychotic and non-psychotic
psychiatric patients. SC T scores greater than 70 in
pain patients tend to be the result of their en-
dorsement of somatic symptoms but also of items
reflecting depression, memory and concentration
problems, the feeling that life is a strain, and
trouble getting started on activities. Psychotic pa-
tients with similar SC elevations tend to have more
bizarre thought processes, social isolation, defec-
tive inhibition, and fewer somatic complaints than
do chronic pain patients. Non-psychotic psychi-
atric outpatients with similar SC scale elevations
differ from chronic pain patients by reporting
thinking, greater depression and despair, and fewer
somatic complaints.
The results strongly support our hypothesis that
it is the endorsement of unusual sensory experi-
ences that is the major contributor to SC elevations
of pain patients. We were curious, however, about
the absence of differences between pain patients
and psychotics on the LEM-Cognitive and Co-
native subscales and the similarity of the unad-
justed means for these groups on Bizarre Sensory
Experiences (see Table III). We therefore ex-
amined the individual items endorsed in these SC
subscales for 16 randomly selected patients in
each of the pain patient and psychotic groups to
determine whether they endorsed the same items
at the same frequency. Since this part of the study
was considered exploratory, we examined only a

TABLE IV
BSE, LEM-COGNITIVE AND LEM-CONATIVE ITEMS ENDORSED AT LEAST TWICE AS OFTEN BY EITHER PAIN
PATIENTS OR PSYCHOTICS

Frequency of endorsement a Item content

Pain Psychotics
patients
BSE
2 5 At times I have fits of laughing and crying that I cannot control.
5 13 I have had many peculiar and strange experiences.
1 6 At one or more times in my life I felt that someone was making me do things by hypnotizing me.
0 10 At times I hear so well it bothers me.
2 11 I often feel as if things were not real.

16 5 I have little or no trouble with my muscles twitching and jumping (false).

14 2 I have had no difficulty in keeping my balance in walking (false).
15 6 I have numbness in one or more regions of my skin.

LEM-Cognitive
5 13 I have had very peculiar and strange experiences.
4 8 There is something wrong with my mind.
2 10 I am afraid of losing my mind.
2 11 I often feel as if things were not real.
3 8 I have strange and peculiar thoughts.

9 3 My memory seems to be r&right (false).

~~-C~nutive
1 8 Most of the time I would rather sit and daydream than anything else.
6 3 I worry over money and business.
2 1 Most of the time I wish I were dead.

a Based on a random sample of 16 patients from each group.

random subset of 16 patients in each group.
The items endorsed most frequently by pain
patients and psychotics were strikingly different
on the BSE and LEM-Cognitive subscales but
similar for the LEM-Conative subscale. Table IV
lists the items that were endorsed at least twice as
often by one group than the other. Pain patients
tended to endorse items on the BSE subscale that
reflected physical symptoms, whereas psychotic
patients endorsed items that reflected thought dis-
order. On the LEM-Cog~tive subscale, pain pa-
tients complained of memory problems whereas
psychotics again complained of bizarre dis-
turbances in thinking. On the LEM-Conative sub-
scale few items distinguished the groups, with pain
patients complaining of worrying over finances
and psychotics complaining of daydreaming. On
this subscale, both groups frequently endorsed
items such as, ‘Life is a strain for me much of the
time,’ ‘I have had periods of days, weeks or months
when I couldn’t take care of things because I
couldn’t get going, ’ ‘Most of the time I feel blue,’
and ‘I have diffic~ty in starting to do things.’
To some extent, the Harris and Lingoes sub-
scales may clarify the meaning of a high SC score
in chronic pain patients. However, closer inspec-
tion of these data suggests that item response
patterns within the Harris and Lingoes subscales
must also be considered to fully understand the
meaning of the SC score. It may be useful to
consider more homogeneous subsets of items
within the subscales. For example, one group of
items from the BSE subscale appears to reflect
disordered thinking and truly bizarre sensory ex-
periences. A second group reflects somatic com-
plaints that could conceivably accompany physical
injury or illness. In all cases, it is important to
more clearly examine the basis for the elevated SC
scale using either individual items or subgroups of
items (e.g., Harris and Lingoes subscales). Fur-
thermore, the meaning of these responses must be
considered in light of the patient’s physical dis-
order.
These results combined with previous caveats
about misdiagnosis of chronic pain patients as
h~~hond~acal, depressed or hysterical [6,15,16]
indicate that extreme care should be used in inter-
preting MMPI profiles of pain patients. Clinicians
should avoid the automatic application of the
standard interpretive rules to chronic pain pa-
tients.
The present study was done retrospectively and
suffers from a number of limitations. We em-
ployed only male veterans as subjects, and the
results cannot be generalized to other groups of
chronic pain patients without cross-validation.
Prior research has found that there may be im-
portant differences among patients seen in differ-
ent pain clinic settings, and these could conceiva-
bly affect the pattern of responses on the MMPI
[12]. Furthermore, differences between male and
female chronic pain patients’ MMPI profiles have
been noted [2].
Another limitation of the study involves the
manner of arriving at a diagnosis in our psychi-
atric patients. Formal diagnostic criteria [4,17]
were not consistently used. For the psychotic pa-
tients, we retied on discharge diagnoses, since
these were always available in the medical record
and presumably arrived at after considerable ex-
perience with the patients. For the non-psychotic
patients, we carefully ruled out psychosis by a
thorough review of the medical record to eliminate
any patients who had been suspected of having
psychosis or who were reported to have evidenced
a symptom of psychosis even when a diagnosis
was not given. Although most of these non-psy-
chotic patients were depressed, they represent a
mixture of diagnoses and are not as homogeneous
as might be desirable.
In conclusion, many chronic pain patients have
high SC scores because they endorse items reflect-
ing somatic complaints and the sense that life is a
strain. It would be a gross disservice to contribute
to the suffering of these patients by inaccurately
labeling them as having psychopathology. Some
chronic pain patients may have severe psychopa-
thology in addition to pain, and the MMPI may
still be quite useful in revealing the nature and
extent of these problems. Overlooking a major
psychological problem is no less serious than mis-
diagnosing one when it is not there. We recom-
mend that if the MMPI is used to assess chronic
pain patients, extreme care should be taken to
examine in greater detail the items or subsets of
items that contribute to scale elevations. The util-

ity of the Harris and Lingoes subscales are dem-
onstrated in this study, and the ready availability
of these subscale scores through standard com-
puter scoring services makes them an efficient tool
for examining MMPI item content patterns.
References
1 Armentrout, D.P., Moore, J.E., Parker, J.C., Hewett, J.E.
and Feltz, C., Pain patient MMPI subgroups: the psycho-
logical dimensions of pain, J. behav. Med., 5 (1982)
201-211.
2 Bradley, L.A., Prokop, C.K., Margohs, R. and Gentry,
W.D., Multivariate analysis of MMPI profiles of low back
pain patients, J. behav. Med., 1 (1978) 253-272.
3 DahIstriim, W.G., Welsh, G.S. and Dahhtriim, L.E., An
MMPI Handbook. Vol. 1. Clinical Interpretation, Univer-
sity of Minnesota Press, Minneapolis, MN, 1972.
4 Diagnostic and Statistical Manual of Mental Disorders
(DSM-III), American Psychiatric Association, Washington,
DC, 1980.
5 Duckworth, J.C. and Duckworth, E., MMPI Interpretation
Manual for Counselors and Clinicians, Accelerated Devel-
opment, Inc., Muncie, IN, 1975.
6 Franz, C., Paul, R., Bautz, M., Choroba, B. and Hildebrandt,
J., Psychosomatic aspects of chronic pain: a new way of
description based on MMPI item analysis, Pain, 26 (1986)
33-43.
7 Gilberstadt, H. and Duker, J., A Handbook for Clinical
213
and Actuarial MMPI Interpretation, Saunders, Phila-
delphia, PA, 1965.
8 Graham, J.R., The MMPI: a Practical Guide, Oxford Uni-
versity Press, New York, 1977.
9 Greene, R.L., The MMPI: an Interpretive Manual, Grune
and Stratton, New York, 1980.
10 Harris, R. and Lingoes, J., Subscales for the Minnesota
Multiphasic Personality Inventory, Mimeographed Materi-
als, The Langley-Porter Clinic, 1955.
11 Hathaway, S.R. and McKinley, J.C., The Minnesota Multi-
phasic Personality Inventory, The Psychological Corpora-
tion, New York, 1943.
12 HoIzman, A.D., Rudy, T.E., Gerber, K.E., Turk, D.C.,
Sanders, S.H., Zimmerman, J. and Kerns, R.D., Chronic
pain: a multiple-setting comparison of patient characteris-
tics, J. behav. Med., 8 (1985) 411-422.
13 Meback, R. and Wall, P.D., The Challenge of Pain, Basic
Books, New York, 1983.
14 Pedhazer, E.J., Multiple Regression and Behavioral Re-
search: Explanation and Prediction, Holt, Reinhardt and
Winston, New York, 1982, pp. 517-519.
15 Prokop, C.K., Hysteria scale elevations in low back pain
patients: a risk factor for misdiagnosis? J. consult. clin.
Psychol., 54 (1986) 558-562.
16 Smythe, H.A., Problems with the MMPI, J. Rheumatol., 11
(1984) 417-418.
17 Spitzer, R.L., Endicott, J. and Robins, E., Clinical criteria
for diagnosis and DSM III, Amer. J. Psychiat., 132 (1975)
1187-1192.
18 Taylor, G.P., Moderator-variable effect on personality-test
item endorsements of physically disabled patients, J. con-
sult. chn. Psychol., 35 (1970) 183-188.