Literature Review

tray15

instruments_measuring_dsm_iii_and_1.pdf

Home >Psychology homework help >Literature Review

Journal of Personality Disorders. 1(3). 220-240, 1987

O 1987 The Guilford Press

INSTRUMENTS MEASURING DSM-III AND

DSM-III-R PERSONALITY DISORDERS

James H. Reich, MD, MPH

Four self-report and 10 semistructured interviews for

diagnosing DSM-III or DSM-III-R personality disorders are

described in terms of their diagnoses, reliability, and validity

data. Evidence is presented for the validity of the DSM-III

personality disorder clusters. The distorting effects of state

depression and anxiety on personality measurement are

determined to be the most difficult methodologic hurdle facing these instruments at present.

Allen Frances (1985) has commented that the publication of DSM-III created

a new cottage industry the development of instruments to measure DSM-

III-defined personality disorders. At the time of DSM-III there was already

evidence that some lifetime psychiatric diagnoses could be made reliably

(Andreasen et at, 1981 ; Mazure & Gershom, 1979), some requirements for

measuring normal personality were shortly thereafter delineated (Buss &

Craik, 1983: Mischel & Peake, 1982; Rushton, Brainerd, & Presseley, 1983),

and antisocial personality had long been measured reliably (Guze & Good

win, 1971; Spitzer, Endicott, & Robbins, 1975). So rapid have been advan

ces in this field that it is now necessary to update an earlier review of these

instruments written 2 years ago (Reich, 1985). Reviewed here are self-report and semistructured interviews that measure disorders closely related to the

DSM-III or DSM-III-R concepts ofpersonality disorders . Two summary tables

(Table 1 for self-report and Table 2 for interview) are included so that the

salient points of the different instruments can be easily compared.

I. SELF-REPORT MEASURES

PERSONALITY DIAGNOSTIC QUESTIONNAIRE

The Personality Diagnostic Questionnaire (PDQ) (Hyler et al., 1982) is a

152-item, self-administered, forced-choice, true/false diagnostic instrument

measuring all 1 1 DSM-III, Axis II personality disorders. It has a scoring

From the University of Iowa College of Medicine. Requests for reprints should be sent to the

author at Department of Psychiatry, University of Iowa College of Medicine, Iowa City, Iowa

52242.

Support by a University of Iowa Department of Psychiatry institutional grant is acknowledged.

220

INSTRUMENTS MEASURING PERSONALITY DISORDERS 221

sheet, a subscale to measure response set, and significant-other and patient versions; and it takes 30 minutes to administer. Its criteria are identical to

DSM-III, Axis II. It was developed by S. Hyler, R. Reider, R. Spitzer, and J.

Williams.

Test-retest on psychiatric outpatients reliability over a 1-month period in a group selected for presence of Axis II personality disorders resulted in

the following kappa values: paranoid .59, schizotypal .56, antisocial

.74, borderline .63, avoidant .70, compulsive .75, narcissistic .11,

histrionic .30, dependent .04, and passive-aggressive -.25. When

compared blindly to a clinical interview for borderline personality disorder, the PDQ's sensitivity was .69 and its specificity was .63 (Hurt et al., 1984). The PDQ will fail in its goal of measuring all Axis II personality disorders

until all diagnoses have adequate reliability. It will be important that future

research determine the effect of state anxiety and depression on personality measures. As the PDQ tends to diagnose more disorders than are diagnosed

clinically (Hurt et al., 1984), it has possible clinical use as a screening instrument for those diagnoses where its reliabilities are adequate. Past versions of the PDQ have measured DSM-III criteria (Hyler et al.,

1982). A new version is now available measuring DSM-III-R criteria includ

ing the new diagnosis of masochistic personality disorders (Hyler et al.,

1983). The updated versions are similar to past versions in many respects, but they have not yet undergone separate reliability and validity testing.

MILLON CLINICAL MULTIAXIAL INVENTORY

The Millon Clinical Multiaxial Inventory (MCMI) is a computer-scored, 175-

item, forced-choice, true/false, self-report instrument taking 20 minutes to

administer (a hand scoring system is now also available). It was developed

by T. Millon (1982).

Personality diagnoses test-retest reliability at 4 to 6 weeks ranged from

.77 to .85 in psychiatric patients (Millon, 1982). The validity of the MCMI

has been examined by comparison with clinically evaluated MCMI person

ality types, the MMPI, SCI-90, PSI, and by factor analytic techniques (Millon,

1982). Although designed for use by psychiatric patients, norms are avail

able for both psychiatric patients and normals. In addition to generating

personality scales, the MCMI generates nine scales relevant to DSM-III, Axis

I diagnoses, response set scales, and a validity scale.

One important way in which Millon's concept of personality disorders

differs from the DSM-III is his separation of personality disorders into two

groups (Millon, 1981). One group he labels "basic personality patterns" and

the remainder (borderline, schizotypal, and paranoid) are labeled "patho

logical personality disorders." The pathological personality disorders have

a much more negative prognosis. Another difference in the MCMI approach is its use of statistically unique combinations of symptoms instead of one-

to-one symptom description. This allows the same symptom responses to

be present in different clusters and greatly improves the diagnostic economy of a set of symptoms. This approach is unique in scales reviewed here.

Recently data have become available comparing MCMI diagnoses to those

of clinicians (Millon. personal communication). Over 300 clinicians rated

2,679 patients, all of whom were given at least one personality disorder

222 REICH

X X

s If)

o N

x C

Q O co c co <v

x ID

-3

x C CO 3

CO Gj o

X XI

OJ c or co

c o ~

g c CU oo-o "3 en CO

O 3 5 X oo

X X J

X X

ca o.

a-o

CO CO

"3 CO

E u

-a

CO cfl

CO c

co'

n rt

CO' & X

0. s

C o o N _ o CO

OJ c a c o

CO*

C cd u

c CO a a

o "O CD CD

xi cfl 3 na -a

a o TJ O x:

CO 3 C c CQ o a. o ^H CO CO

CO c 2 0. oj a; 3 <u

'0 u

l> a] ^C V, > a en

co . "3

co rt +-;

a u -2

C X c a^ o G

c -o x H co o 7- h o >-> F*-;>.'-

* CO O o K co u

W a c z x a:

5 CO*

* X3

co ir <J O

C co

ii3

m a "5

Outp Axis norm

c - a

F - -

5S5o c

J= -5 s ^ D.

c a

'C s s v if) If)

.2 t X J

> j=

5? 3 - Q Q CO

O < a co

INSTRUMENTS MEASURING PERSONALITY DISORDERS 223

X X X X

X X

E a in

X XX in

X X r- 6 =3

X X m

o (N

a o

J3 .5

2 5>

C 'o

_co a -g 13 : c c x o

CO 1)

a a

E E o o

x S CO

71 "--

E o

>.a;

V -a

1 1; t l >,"a ^r X X aj co G Q Q Q TD 0. M

a CJ

E E E

CO o

c o

Qfj 1)

C a

o E E

u 3 o

CO Z c_>

.a o

S -G

224 REICH

CO <

to 2 Q cj

</> CO o. a

55 X X x 5

^, Q. C C n. G

Q X X

x x x 3

Q < CO

Q-C

a CO W)

s8* 2 coG ^o . en m .D co aj

E Sia 5 5- 6 8

c o H a a 2 E

S .5 E

5 o

2 loll p, m 3 u aao

S 55 g a&2

o J" C

tD o

-C to c

"2 Qfi

-a

,o 55 o

"3 -a C u o >- CO CD 1- >

si u CO

c CD O Cu CJ CJ

E c : o

co a

Q O

r ci ft

11* g" "

i n v

< Q. CO to

INSTRUMENTS MEASURING PERSONALITY DISORDERS 225

c O

CD OS

E U o CD 8

1 OS

u o

X 00

CN E <a CO o

o CJ

X CN <3

2 OS

= CJ

CO C eg "

O v CO x

9 os in "

P S

& c

"aj -o

ts . g

e C - 2

C ~ ^ ,rt C O c

- -bSococ^co

- -

i E ^ - ._ >. " S3 2 o S

= i? 8 c H =g Q c Q aQ

to a. w ra

en cj

E E

-2 o- E E

_ O OJ

c/3 co z: o

226 REICH

diagnosis. Although 18% of the clinicians were not blind, the results are

essentially the same when they are not included. Table 3 lists the MCMI

diagnoses and the highest clinical DSM-III diagnosis associations among a

subset of patients in whom diagnostic assignments were independently derived by two clinical judges. As can be seen, the highest DSM-III associ

ation for each MCMI diagnosis is its corresponding DSM-III diagnosis. The

high overlap of diagnoses is not uncommon in DSM-III measurement in

struments. In any given case, this may be due to either the DSM-III criteria

themselves or the measurement instrument. The MCMI is a well validated

instrument with good reliability and clearly has a valid relationship to the

DSM-III personality disorders. Its series of scales provide information that

is of much clinical value.

The MCMI scales D (dysthymia) and A (anxiety) are highly associated with

many of the personality scores. Both A and D have an association above .60

in 4 of 12 personality scales. The extent to which some of these personality scales are state dependent awaits further research.

The MCMI is presently in the process of being updated to a version that

will be closely aligned with DSM-III-R personality disorder criteria. Named

the MCMI-II, it will also have 175 items; a three-point weighted scoring

system will be applied to each item. Notably, special "corrections" have been

built in to reduce the effects ofconcurrent affective states such as dysthymia and anxiety.

BORDERLINE SYNDROME INDEX

The Borderline Syndrome Index (BSI) is a 52-item, yes/no, forced-choice,

self-report instrument measuring borderline personality disorder that takes

20 minutes to administer. It was developed by H. R. Conte et al. (1980). A

table giving the cumulative percentages of normals and of borderline per

sonality disorders scoring at each BSI level is available (Conte et al., 1980).

Although internally consistent (KuderRichardson = .95, p < .001), no

reliability data are reported. BSI scores were able to discriminate borderline

personality disorder from two other groups also diagnosed by DSM-III

schizophrenia and depression and also differentiated normals (Conte et

al., 1980).

Table 3. Association of MCMI and DSM-III Personality Diagnosis

MCMI Highest DSM-III clinical associations

Schizoid Schizoid 82.8, schizotypal 74.2, avoidant 69.9

Avoidant Avoidant 87.2, schizotypal 75.2. schizoid 71.9, borderline 70.7,

dependent 64.6. Passiveaggressive 64.6

Dependent Dependent 88.6, borderline 69.7, avoidant 66.9

Histrionic Histrionic 84.6, narcissistic 76.8, antisocial 67.2

Narcissistic Narcissistic 85.9. antisocial 72.7, paranoid 72.7, histrionic 69.6

Antisocial Antisocial 84.4, paranoid 74.3. narcissistic 68.2

Compulsive Compulsive 80.5

Passive-aggressive Passiveaggressive 86.2, borderline 77.1, avoidant 73.2

Schizotypal Schizotypal 85.2, schizoid 67.9. avoidant 73.2

Borderline Borderline 85.2, avoidant 74.3. dependent 70.7.

Passiveagressive 70.7, paranoid 65.8

Paranoid Paranoid 76.5. antisocial 67.6

INSTRUMENTS MEASURING PERSONALITY DISORDERS 227

In another study Edell (1984) found that although he BSI could discrim

inate DSM-III borderline and schizotypal patients from early schizophrenics and normals, it could not discriminate borderline from schizotypal person

ality disorders. This finding, plus the finding that the BSI correlated highly with virtually all scales of the MMPI, led Edell to speculate that the BSI might be tapping a broad measure of psychopathology. A key item in this instrument's development would be test-retest data.

It would also be useful to see how state anxiety and depression affect BSI

scores and to have further studies examining which personality disorders,

if any, the BSI discriminates.

BELL OBJECT RELATIONS SELF REPORT SCALE

The Bell Object Relations Self Report Scale (Bell) is a 45-item, true/false

questionnaire developed by Morris Bell (Bell, 1981; Bell, Metcalf, & Ryan, 1979, 1980). It was derived by factor analytic techniques and consists of

four scales Alienation, Insecure Attachment, Egocentricity, and Social

Competence. Internal consistency is high, with Cronbach's alpha ranging from .78 for Egocentricity to .90 for Alienation.

This scale is of interest due to claims that it can differentiate DSM-III-

defined borderline patients with a high degree of accuracy (Bell et al., in

preparation). The instrument can distinguish borderlines from normals

with a specificity of 96.7% and a sensitivity of 91.9%. It can distinguish borderlines from psychiatric inpatients (depressed and schizophrenic mixed

group) with a specificity of 88.6% and a sensitivity of 83.8%.

This is an instrument with promise at an early stage of its development. Still needed are replication test-retest studies and investigations as to how

it is affected by state depression and anxiety, and how well it distinguishes borderline from other DSM-III personality disorders.

INTERVIEW INSTRUMENTS

THE SCHEDULE FOR INTERVIEWING BORDERLINES

The Schedule for Interviewing Borderlines (SIB) is a 70-item, semistructured

interview for the diagnosis of borderline personality disorder (Schedule for

Borderline Personalities [SBR]) and for schizotypal personality disorder

(Schedule for Schizotypal Personalities [SSP]), and it takes 50 minutes to

administer. It was developed by M. Baron et al. (Baron, 1981; Baron, Asnis,

& Gruen, 1981). The SIB has a five-point scale for each question, and each

section has a statement score, scaled score, rating of severity, and an age

of onset.

No reliability or validity data are available for the SBR. The SSP was

developed on first-degree relatives of patients diagnosed as schizophrenic.

Test-retest at 6 months for the SSP yielded reliabilities between r = .61

and r = .91. The kappa for presence or absence of schizotypal personality

disorder using two raters was .88 (Baron et al. 1981). These high relia

bilities have been confirmed by Perry, O'Connell, and Drake (1984). SSP

norms for first-degree relatives of schizophrenic patients are available (Baron

etal, 1981).

228 REICH

The SIB lacks basic work on reliability and validity. The SSP has good reliability and some evidence for its validity. Norms on nonschizophrenic

populations are needed for both SSP and SBR. Work needs to be done to

see how well these instruments distinguish the syndromes they are meas

uring from other personality disorders and how they are affected by state

factors.

BORDERLINE PERSONALITY DISORDER SCALE

The Borderline Personality Disorder Scale (BPD Scale) is a 36-item, sem

istructured interview with questions in nine categories of behavior relevant

to borderline personality disorder, and it takes 90 minutes to administer

(Perry, 1982, personal communication; Perry & Klerman, 1980). BPD items

correlate .88 with DSM-III criteria.

A reliability correlation coefficient of .93 was reported using the videotape method (J. C. Perry, personal communication). Norms are available for t^e

following groups: borderline personality disorder, borderline traits, anti

social personality disorder, bipolar affective disease, alcoholism, schizo

phrenia, neurotics, and adjustment disorders. The scale can discriminate

borderline personality disorders from other nonpersonality disorders and

also from antisocial personality disorders. These interviews were not blind,

however. Specificity for definite borderline personality disorder does not go lower than 76% in three studies. A factor analytic study also provided sup

port for the diagnostic criteria used.

The PBD is a well developed scale with evidence for its reliability and

validity. Work showing the separation of borderline personality disorders

from other populations should be repeated blind and extended to other

personality disorders in addition to antisocial. The effect of state anxiety and depression on the scale should be investigated.

DIAGNOSTIC INTERVIEW FOR BORDERLINE PATIENTS

The Diagnostic Interview for Borderline Patients (DIB) is a 165-item, semi-

structured interview with a complete scoring system that takes 60 minutes

to administer (Gunderson, 1982a; Gunderson & Kolb, 1978; Gunderson,

Kolb, & Austin, 1981). The scale measures five areas of presumed impor tance to the diagnosis of borderline personality disorder and was developed

by John Gunderson et al.

Joint reliability was reported as .80 in one study (Perry & Klerman, 1980)

and as a .90, with a kappa of .62, in another (Kolb & Gunderson, 1980).

Other studies confirm these reliability findings (Cornell et al. , 1983; Frances

etal, 1984; Hurt etaZ., 1984: Kroll, Pyle et al, 1981). DIB and subsections

of the DIB have been validated against established psychiatric tests, in

cluding the MMPI (Kroll, PyleetaZ., 1981; Kroll, Carey etal., 1982; Loranger et al., 1984; Soloff, 1981a, 1981b), and also by factor analytic techniques (Gunderson & Kolb, 1978). Work has also been done validating the DIB

against clinical diagnosis of borderline personality disorder (Gunderson et

al, 1981), the SEG checklist rating (Soloff, 1981a), DSM-III criteria (Soloff,

1981a), and the borderline score of the structural interview method of as-

INSTRUMENTS MEASURING PERSONALITY DISORDERS 229

certaining borderline personality disorder (Kernberg et al., 1981; Koenigs-

berg, Kernberg, & Schomer, 1983). It appears probable that the DIB can

distinguish borderline personality disorders from affective disorders, schiz

ophrenia, and anxiety disorders (Soloff, 1981a). Two studies that compare DIB scores of borderline with DIB scores of

other personality disorders in the inpatient setting show that the DIB does

not discriminate between these two groups (Kroll, Sines et al., 1981; Kroll,

Carey et al., 1982). Barrash et al. (1983) were able to achieve differentiation

between borderline and personality disorders using the DIB. They are able

to do so by using a scoring system derived from cluster analysis of DIB

items. However, on testing the system on a second population, this method

was not found to be robust (Barrash et al., in preparation). A fourth study shows that in an outpatient population the DIB does have some ability to

distinguish borderline from other personality disorders; however, this study also indicates that in the patient setting the DIB loses some ability to dis

tinguish between psychotic and borderline personality patients (Soloff, 1981c). Frances et al. (1984) also find that the DIB can discriminate between bor

derline and other personality disorders in the outpatient setting. Using a

DSM-III clinical interview as the criteria, they find a DIB cut-off score of 7

yields a sensitivity of .73 and a specificity of .80 for this population (Frances

etal., 1984).

The DSM-III and the DIB criteria for borderline personality disorder were

designed to include a large amount of overlap. Three studies report the

expected high correlation between DIB scores and DSM-III clinical diagnosis (Kroll et al, 1982; Loranger et al, 1984; Soloff, 1981c). Although roughly

equivalent clinically, the DIB criteria may be slightly more broad than the

DSM-III criteria.

The DIB has become a standard ofcomparison for developing instruments

to measure borderline personality disorder. It has good reliability and va

lidity. As with some other instruments, it appears that the DIB may not be

able to distinguish borderline personality disorder clearly from other per

sonality disorders. There is no specific work on the effect of state on the

DIB.

SCHEDULE FOR AFFECTIVE DISORDERSRESEARCH DIAGNOSTIC

CRITERIA

The Schedule for Affective Disorders (SADS) is a structured interview that

diagnoses antisocial personality by Resarch Diagnostic Criteria (RDC) cri

teria as one of its diagnoses (Feighner et al., 1972; Spitzer et al., 1978).

The SADS antisocial personality disorder, although similar to DSM-III

(Stolzman et al, 1981), is slightly more stringent (Singerman et al., 1981).

Reliability for the RDC is generallly high (Spitzer et al., 1978) and is reported

as .72 for antisocial personality disorder by joint interview (Spitzer et al.,

1975). The SADS-RDC antisocial personality diagnosis is well developed,

extensively used, and easily scored.

Loranger et al. (1984) found that by modifying the SADS, a score equiv

alent to the DIB can be derived so it would be possible to use the SADS

for Axis I disorders, antisocial and borderline personality disorders, with

only slight modification.

230 REICH

NATIONAL INSTITUTE OF MENTAL HEALTH DIAGNOSTIC INTERVIEW

SCHEDULE

The National Institute of Mental Health Diagnostic Interview Schedule (NIMH-

DIS) is a structured psychiatric interview developed by Lee Robins et al. for

use by lay interviewers in community surveys (Robins et al., 1979). It in

cludes antisocial personality disorder as one of its diagnoses. Reliability for

the NIMH-DIS and its precursor is good (Helzer, Clayton, & Pambakian,

1982; Robins et al, 1981, 1982), and the present NIMH-DIS version has a

test-retest reliability kappa of .63 for the DSM-III antisocial personality

(Robins et al, 1981) and the SADS-RDC-generated diagnoses (Hesselbock

et al, 1982).

The inclusion of antisocial personality disorder in the DIS has allowed us

to obtain one of the few available prevalence estimates of a DSM-III person

ality disorder. Robins, Helzer, and Weissman (1984) estimated from Envi

ronmental Catchment Area (ECA) data that the lifetime prevalence of an

tisocial personality disorder is approximately 2.5%, with a male/female

predominance of approximately 5 to 1. Myers et al. (1984) estimated 6-

month prevalences of approximately 1.5% for men and 0.9% for women.

THE LEYTON OBSESSIONAL INVENTORY

The Leyton Obsessional Inventory (Leyton) is a 67-item, structured interview

to measure obsessive symptoms and was developed by John Cooper (1970).

It has five subscales, a system of administration, and a scoring system; and

it takes 30 minutes to administer.

Test-retest correlation for the symptom scores subscale was .87 and for

the trait scores subscale was .91. Normal values are available for normal

males and females, females with obsessive traits, husbands of women with

obsessive traits, depressed patients during and after recovery from episode, and patients with "chronic obsessional illness." The Leyton successfully

distinguishes "chronic obsessional illness" from the other groups men

tioned here (Cooper, 1970). It is unclear how close the Leyton "chronic

obsessional illness" is to the DSM-III diagnosis of compulsive personality disorder, and comparisons should be made with caution.

This is a well-designed instrument with some evidence for its validity. Three of its subscales require reliability testing, as does the scale as a whole.

It would be useful to know whether the Leyton could distinguish compulsive

personality disorder from similarly presenting disorders (e.g., passiveag

gressive personality disorder and acute obsessional illness episodes). It would

also be useful to determine whether the instrument is influenced by state

anxiety or depression.

STRUCTURED CLINICAL INTERVIEW FOR DSM-III PERSONALITY

DISORDERS

The Structured Clinical Interview for DSM-III Personality Disorders (SCID

II) is a 120-item, comprehensive semistructured personality interview de

signed by Robert Spitzer, Janet Williams et al. (1985). Each item has a

four-point scoring scale (inadequate information, negative, subthreshold,

INSTRUMENTS MEASURING PERSONALITY DISORDERS 231

threshold), and specific probe questions are supplied. All DSM-III-R per

sonality disorders are included.

The SCID II occupies a separate niche from other instruments reviewed

here, for several reasons. First, it is the only comprehensive DSM-III-R per

sonality instrument that, in effect, is designed for the fastest possible de

livery. Some criteria are the same in different related disorders and therefore do not have to be repeated. If it is clear the patient will not qualify for a

given disorder, there are skip-outs to the next disorder. This is the instru

ment to use to determine the presence or absence of a personality disorder, when the personality disorders themselves are not under study. The SCID II is also unique in that it is the only instrument designed as an extension to an existing Axis I instrument. The SCID II is designed to be utilized by a skilled clinician.

Reliability and validity studies are ongoing and are being carried out by an experienced team, but results are not yet available.

PERSONALITY DISORDER EXAMINATION

The Personality Disorder Examination (PDE) is a comprehensive interview

for all DSM-III-R personality disorders developed by Loranger et al. (1985). It has 328 items, is designed for use by the experienced clinician, and takes

approximately 90 minutes to administer. The PDE is divided into five head

ings for ease of administration (work, self, interpersonal relations, affect,

and impulse control). Each item is scored on a three-point scale (no infor

mation, present below threshold, and clinically significant). The PDE can

be scored either of two ways: One method uses a predetermined statistical

threshold based on the nature and frequency data from preliminary clinical

trials. The second method relies more highly on the clinician's clinical judg ment. It is expected that the former method would have higher reliability. A significant-other version is in development. Simultaneous reliability information is available from two raters on 60

patients selected for the probability of having personality disorders. Inter-

rater agreement for presence or absence of a personality disorder is available

for five diagnoses for which there were enough subjects. These were: schi

zotypal .89, histrionic .87, borderline .98, antisocial .82, and com

pulsive .94.

This is a promising instrument that, due to the wealth of scored data

points, will probably also be useful in future research in the area of per

sonality disorders. It needs to have its reliability tested in test-retest fash

ion, its ability to distinguish normals and other psychiatric patients from

personality disorders should be examined, and its susceptibility to state

anxiety and depression should be determined. These studies are underway. It has been selected by the World Health Organization in some of its studies

involving personality disorders.

THE STRUCTURED INTERVIEW FOR THE DSM-III PERSONALITY

DISORDERS

The Structured Interview for the DSM-III Personality Disorders (SIDP) is a

semistructured, 160-item interview organized into sections relevant to per-

232 REICH

sonality styles; it covers all 1 1 DSM-III personality disorder diagnoses, has

anchor points for each question, and has a scoring manual. It was developed

by Pfohl, Stangl, and Zimmerman. (1982). The patient interview takes 60

to 90 minutes, but this assumes previous knowledge of the patient's Axis

I diagnosis. Usually the clinician takes additional time to interview signif icant others as well.

The SIDP joint interview kappa for presence or absence of a personality disorder is .71 (p < .001). Individual kappas for the more frequently diag nosed personality disorders were borderline .85, histrionic .75, schi

zotypal .62, avoidant .45, and dependent .90 (Stangl et al, 1985).

When those with and without SIDP-diagnosed personality disorders were

compared on MMPI there were significant differences in 7 scales (p < .05), as there also were on the MarkeNyman subscales (p < .05). Psychologist

interpretation for presence or absence ofpersonality disorder from the MMPI

agreed with the SIDP 71% of the time (Stangl et al., 1985). There are also

data showing that the SIDP is useful in distinguishing different subgroups ofpatients with major depressive disorder. Those with a personality disorder

have less frequent dexamethasone nonsuppression, poorer response to an

tidepressant medication, and a higher risk of depression, alcoholism, and

antisocial personality among first-degree relatives (Pfohl, Stangl, and Zim

merman, 1984). There are no published data on norms. The SIDP is highly consistent with DSM-III, Axis II criteria. For presence or absence of a per

sonality disorder, its kappa is .39 when compared with the PDQ (Pfohl et

al, 1984).

In terms of performing research on personality disorders, the SIDP ranks

between the PDE and SCID II instruments. It is more extensive than the

SCID II and does not have skip-outs. However, on the SIDP, only the DSM-

III criteria are rated, and not each individual question. The PDE, on the

other hand, scores each individual question, and this information is then

available for research purposes. Partly as a result of this, it appears the

SIDP might possibly take a few minutes less to administer.

The SIDP is well organized, has reliability for several personality disorders,

and has been validated in inpatients against the MarkeNyman and MMPI.

It will be necessary to publish reliability data on the remaining individual

personality disorder diagnoses. As the SIDP is being developed on a psy chiatric inpatient population (many of whom were depressed), it will be key to have test-retest data, with retest data occurring after the patient has

resolved the acute illness episode. It will also be helpful to examine a broader

range of diagnoses and outpatient populations. Data are also needed on

how well the SIDP differentiates between the different personality disorders

as well as on how it is affected by state anxiety and depressive factors.

PERSONALITY ASSESSMENT SCHEDULE

The Personality Assessment Schedule (PAS) is a semistructured interview

arranged in 24 topic areas that are scored on a 9-point scale, takes about

60 minutes to administer, and yields personality diagnoses that are roughly

equivalent to DSM-III personality disorder clusters. It was developed by Peter

Tyrer (Tyrer, Tyrer & Alexander, 1979; Tyrer et al, 1979, 1983) based on

a cluster analysis of personality traits of personality-disordered patients.

INSTRUMENTS MEASURING PERSONALITY DISORDERS 233

The specific diagnoses yielded are: schizoid (roughly equivalent to the DSM-

III cluster I of schizoid, schizotypal, and paranoid, sociopathic (roughly

equivalent to the DSM-III cluster II of borderline, histrionic, narcissistic,

and antisocial), passive dependence (roughly equivalent to DSM-III depend ent, avoidant, and passiveaggressive), and anakastic (equivalent to DSM-

III compulsive). It is recommended that it be given to both the subject and

a knowledgeable informant; in this way it hopes to avoid the pitfalls of state

contamination of symptoms. It is not recommended that it be given during acute illness states. It is computer scored, but a hand scoring technique is

currently available. The PAS personality disorder categories are mutually exclusive. One of the major strengths of the PAS is the reliability of the

personality disorder diagnosis. A blind 2.75-year test-retest of psychiatric

patients yielded a weighted kappa of .64 for the presence or absence of a

personality disorder (Tyrer et al., 1983). A cross-national videotape relia

bility study demonstrated that the PAS can be as successfully utilized by American psychiatrists as their British counterparts (Tyrer et al. 1984).

There is some evidence for the validity of the diagnoses made by the PAS.

It was a strong predictor of success in a drug trial and predicted outcome

in an alcoholic treatment program, and the various diagnoses associated

differentially with different Axis I disorders (Griggs and Tyrer, 1981; Tyrer et al., 1983).

The PAS is a promising and well-validated instrument. The use of modified

DSM-III clusters is a reasonable approach (see the following section of this

paper). More information on the reliability of the individual diagnoses would

be helpful, however. Also, the PAS has a taxonomic problem in that where

personality disorders are severe and may fit multiple categories, the scoring

system forces the diagnosis into only one.

However, recent work on the PAS may correct these problems (Tyrer,

personal communication). The hand scoring technique now allows subdi

vision of personality disorders into 13 individual disorders, allows a rating of severity of disorder, and allows for the possibility of multiple diagnoses.

Reliability and validity work on these latest modifications is not yet pub lished. Comparison with comprehensive DSM-III interviews would help clar

ify further how closely each PAS diagnosis relates to the DSM-III diagnoses.

THE DSM-III PERSONALITY DISORDER CLUSTERS

It is clear to anyone working in the area of personality measurement that

the DSM-III personality disorder diagnoses are not mutually exclusive from

each other, and that their interrelations are of clinical and theoretical im

portance. There are several different methods for examining these associ

ations. Clusters can be derived from associations with Axis I disorders, by

factor analytic techniques or merely by observing their co-occurrence clin

ically. There are two theories of DSM-III personality association. DSM-III

divides the personality disorders into three clusters, whereas Millon (1981,

1982) divides them into two. The DSM-III clusters are: Cluster I schizoid,

paranoid, schizotypal; Cluster II narcissistic, antisocial, borderline, and

histrionic: Cluster III dependent, avoidant, passive-aggressive, and com

pulsive. The Millon clusters consists of: Cluster 1 schizotypal, borderline,

234 REICH

and paranoid; Cluster 2 all remaining personality disorders. Millon pos tulates that his Cluster 1 has an especially poor prognosis. The data on the relationship of personality disorder to Axis I disorders

have already been cited. The disorders in DSM-III Cluster I have an asso

ciation with schizophrenia, and two of those in DSM-III Cluster II (antisocial

and borderline) have an association with depression. Two of the Millon

Cluster 1 disorders (schizotypal and paranoid) have an association with

schizophrenia.

Recently Kass et al. (1985) performed a factor analysis on DSM-III-rated

personality disorders in 609 new outpatients admitted over a 14-month

period. They derived four clusters. These resembled the DSM-III clusters,

with the exception that compulsive personality formed a fourth cluster sep arate from Cluster III. Working from a different empirical and theoretical

base, Tyrer derived similar groupings (Tyrer, 1979; Tyrer & Alexander, 1979;

Tyrer et al. 1979). Further validity for the DSM-III cluster system comes

from work by Reich and Thompson (1987). Here the DSM-III personality clusters were able to distinguish between three separate emotionally ill pop ulations (competency to stand trial, disability, and chronic pain). The third line of approach to examining the cluster question is to examine

/how diagnoses associate clinically. Millon has associations between MCMI

personality scale scores and DSM-III consensually diagnosed personality disorders for a group of 877 patients. The highest associations for the

personality disorders in his Cluster 1 (excluding the association of the Millon

disorder with its corresponding DSM-III disorder) are as follows: for schi

zotypal schizoid 67.9 and avoidant 64.3; for borderline avoidant 74.3,

dependent 70.7, and passiveaggressive 70.7; for paranoid antisocial 67.6

and schizotypal 68.2. His associations for the DSM-III clusters are shown

in Table 4.

Pfohl et al. (in press), using the SIDP, cross-tabulated DSM-III-defined

personality disorders in 131 psychiatric inpatients. As only one paranoid and one schizoid were diagnosed, these cannot be of use. In terms of Millon's

categories, of those diagnosed schizotypal ( 12), 50% (6) were also diagnosed as borderline; and of those diagnosed as borderline (29), 21% (6) were also

diagnosed as schizotypal. Data for DSM-III Clusters II and III are shown in

Table 5.

Other clinical work relating to clusters include the follow-up study by Pope I et al (1983) of borderline personality disorder. In their population 73.7%

also were diagnosed as histrionic personality disorder and 9.1% were di

agnosed as antisocial personality disorder. Pfohl et al. (1984), in their study of personality disorders in patients with major depression, found that DSM-

III Cluster III patients were different from Clusters I and II on dexamethasone

suppression test results, response to treatment, and familial risk for depres sion and antisocial personality.

Examining these findings as a whole, we are not convinced that the evi

dence shows the personality disorders in Millon's Cluster 1 to associate

preferentially with each other. This may not be at variance with Millon's

position, however, as he feels that these disorders have an especially poor

prognosis but does not make claims for especially high levels of association

between the three. Only follow-up studies will answer questions of prog nosis.

INSTRUMENTS MEASURING PERSONALITY DISORDERS 235

Table 4. Association Between Millon Personality Factors and DSM-III Clinical Personality

Diagnoses Arranged by DSM-III Clusters (JV = 8771

Millon-defined

disorders DSM-III clinical ratings

Cluster I

Schizoid Paranoid Schizotypal

Schizoid

Paranoid

Schizotypal

82.8

37.4

67.9

37.4

76.5

49.2

Cluster II

74.2

49.2

85.8

Borderline Antisocial Histrionic Narcissistic

Borderline

Antisocial

Histrionic

Narcissistic

85.2

53.4

54.1

46.0

53.4

84.4

67.2

72.7

Cluster III

63.4

61.1

84.6

69.6

47.3

68.2

76.8

85.9

Dependent Avoidant Compulsive

Passive-

aggressive

Dependent Avoidant

Compulsive Passiveaggressive

88.6

64.6

58.2

61.8

66.9

87.2

46.8

73.2

45.8

39.4

80.5

34.3

62.8

64.6

47.6

86.2

The Axis I relationships, factor analyses, and clinical associations do in\

general tend to support the DSM-III cluster system. It does appear possible that compulsive personality should be separated from Cluster III into a

cluster of its own, however.

One important question raised by these findings concerns the meaning of multiple personality diagnoses. It appears that some personality disorders

Table 5. SIDP Cross-Tabulation Data for DSM-III Personality Disorder Clusters II and III

(IV = 129)

Personality disorders Percentage of other personality disorders also diagnosed

Cluster II

Histrionic Narcissistic Antisocial Borderline

(JV = 30) Histrionic 13 10 66

IN = 5) Narcissistic 80 0 80

IN = 5) Antisocial 60 0 80

(N = 29) Borderline 69 14 14

Cluster III

Passive

Avoidant Dependent Compulsive Aggressive

IN = 15) Avoidant 20 13 33

(JV = 17) Dependent 18 6 18

(IV = 7) Compulsive 29 14 43

(N = 18) Passive-Aggressive 28 17 17

Source: After Pfohl et al.. Comprehensive Psychiatry (in press).

236 REICH

such as borderline may exist either by themselves or as final common path

ways due to other emotional or personality disorders. One might expect different characteristics of these two patients, who are both diagnosed as

borderline. More research will have to be done in the future on the prognosis and treatment responses of patients with different associations of person

ality disorders.

THE INFLUENCE OF STATE ON PERSONALITY MEASUREMENT

Frances concluded in 1980, after analyzing the reasons for the low reliability of personality disorders at the time, that contaminating effects of state

created one of the major obstacles to achieving reliability. There is much

research to support this view. There are many reports on the effects of

depressive state for the Maudsley Personality Inventory (MPI) (Biachi & Fer-

gusson, 1977; Coppen & Metcalfe, 1965; Hirschfeld & Klerman, 1979;

Hirschfeld et al, 1983; Ingham, 1966; Kerr et al, 1970), and although one

study reports overcoming this difficulty with specific instructions (Kendell

& DiScipio, 1968), this result is disputed (Liebowitz et al, 1979). It appears that the effects of depressive state are similar in depressed unipolars and

bipolars (Leibowitz et al, 1979). Overall the depressive state appears to affect

measures of emotional strength, interpersonal dependency, and extraver-

sibn. ft does not appear to affect measures of rigidity, level of activity, and

dominance.

There is one report available on the effect of state anxiety on personality measurement (Reich et al, in preparation). The effect of anxiety on per

sonality measurements was examined in 56 patients in a clinical trial of

alprazolam for the treatment for panic attacks. The battery of personality tests used was similar to that used by Hirschfeld (Hirschfeld & Klerman,

1979; Hirschfeld et al, 1983) with the exception that the Maudsley and the

MMPI ego strength scale were not used. In the 16 patients who did not

improve, there was no change in personality scores over a 6-week period. i For the 40 patients who improved 5 points or more on the Hamilton Anxiety

] Scale, there were significant changes in personality measures taken at 6 weeks. These measures showed increased emotional strength, less inter

personal dependence, and more extraversion. These results are similar to

those found for depressive state by Hirschfeld et al (1979, 1983).

It is clear that depressive state affects some personality measures. It ap

pears that state anxiety does so as well, although those findings are in need

of replication. This makes it prudent to measure state anxiety and depres sion when performing personality measures. It is possible that the devel

opers of newer instruments may be able to eliminate individual highly state-

influenced items. An attempt to do so on the Maudsley was not successful,

however, and Biachi and Fergusson concluded that depressive state caused

changes in scoring across the entire set of items. It may or may not be

feasible to do on other instruments. Alternately, it maybe possible to develop formulas to adjust for different levels of anxiety or depression. The use of

knowledgeable informants is another possible route. At any rate, it appears that some assessment of state anxiety and depression given concurrently with a personality measure is mandatory in ascertaining the validity of that

measure.

INSTRUMENTS MEASURING PERSONALITY DISORDERS 237

ISSUES RELEVANT TO FUTURE RESEARCH IN DSM-III

PERSONALITY DISORDERS

At this point probably the most pressing issue in the development of per

sonality instruments is to overcome to as large an extent as possible the

contaminating effects of state on personality measurement. Although there

are some personality disorders which exist without a concurrent Axis I

diagnoses, most clinical situations involving personality disorders also in

volve a combination ofAxis I and Axis II problems. There are several possible

approaches to solving this problem, as mentioned above. Whichever meth

ods prove successful, this problem will have to be overcome in order to

provide clinically important and methodologically sound information.

Many of the personality diagnoses in DSM-III and DSM-III-R, although

clinically reasonable, have not been appropriately validated by research stud

ies. Few follow up studies have been performed on Cluster III disorders, and

the interrelationship of Axis I and Axis II disorders needs to be further

delineated by appropriate family, treatment, clinical descriptions, and course^ of illness studies. One interesting challenge to be met is the possibility that/

in some cases certain personality disorders, such as borderline, represent a final common pathway of multiple disorders. If this is so, it would be

important to try to distinguish which of these personality disorders are^

secondary and which are primary. One way might be to look at the co

occurrence of personality disorders. For example, borderline as a sole dis-/

order might tend to be primary, whereas borderline in conjunction with

specific other personality or Axis I disorders might represent a secondary disorder.

Another interesting challenge lies in discovering the best system to meas

ure personality disorders polythetic or monothetic. The polythetic ap

proach assumes that scoring a minimum number of a list of criteria for a

disorder qualifies one for the diagnosis while the monothetic approach re

quires that every one of a set of criteria be present in order to qualify for

the diagnosis. The polythetic approach implies that certain variation from

a prototype is acceptable. As it has been shown that prototypes can be

identified for at least seven DSM-III personality disorders (Blashfield et al,

1985) this approach may have merit. On the other hand, one of the best

validated of the personality disorders (antisocial) has a monothetic format.

Further research is needed in this area.

Overall the field of standardized measurement of personality disorders is

showing rapid development with increasingly more complex instruments

available. It is likely that solutions to some of the difficult methodologic

questions in the field may be found in the near future.

REFERENCES

Andreasen. N. C. Grove, W., Shapiro, R. et

al. (1981). Reliability of lifetime diagno

sis. Archives of General Psychiatry, 38.

400-405.

Baron. M. (1981). Schedulefor interviewing

borderlines. New York: New York State

Psychiatry Institute.

Baron. M.. Asnis. L.. & Gruen, K. (1981).

The schedule for schizotypal personalities

(SSP): A diagnostic interview for schizo

typal features. Journal ofPsychiatric Re

search 4. 213-228.

Barrash, J.. Kroll. K.. Carey, K. etal. (1983).

Discriminating borderline disorder from

238 REICH

other personality disorders. Archives of General Psychiatry. 40. 1297-1302.

Barrash, J., Kroll. J.. Sines. L. et al (in

preparation). Diagnostic interview for

borderlines schizotypy checklist: A method

for identifying schizotypal borderlines.

Bell, M. (1981). Bell object relations self-re

port scale. West Haven, CT: Psychology Service, VA Medical Center.

Bell, M.. Cicchetti, D.. Billington, R. et al.

(in preparation). Prediction accuracy of the

Bell object relation self-report scale in the

diagnosis of borderline personality dis

order.

Bell. M.. Metcalf, J., & Ryan. E. (1979). Real

ity testing and object relations: A self-re

port instrument. Paper presented at the

87th Annual Convention of the American

Psychological Association, New York.

Bell, M.. Metcalf, J.. & Ryan. E. ( 1980). Real

ity testing object relations assessment

scale: Reliability and validity studies.

Paper presented at the 88th Annual Con

vention of the American Psychological As

sociation. Montreal

Biachi, G. N.. & Fergusson, D. M. (1977).

The effect of mental state on EPI scores

British Journal of Psychiatry, 131. 306-

309.

Blashfield, R., Sprock, J., Pinkston, K. el al.

(1985). Exemplar prototypes of personal

ity disorder diagnoses Comprehensive

Psychiatry 26, 11-21.

Buss, D., & Craik. K. (1983). The act fre

quency approach to personality. Psycho

logical Review, 90, 105-126.

Conte. H., Plutchik. R.. Karasu. T. et al.

( 1980). A self-report borderline scale: Dis

criminative validity and preliminary norms. Journal of Nervous and Mental

Disease. 168. 428-435.

Cooper. J. (1970). The Leyton obsessional

inventory. Psychological Medicine , 1, 48-

64.

Coppen. A., & Metcalfe. M. (1965). Effect of

a depressive illness on MPI scores. British

Journal of Psychiatry, 111, 236-239.

Cornell. G.. Silk, K.. Ludolf. P. et al. 1 1983).

The test-retest reliability of the diagnos tic interview for borderlines. Archives of General Psychiatry. 40, 1307-1311.

Edell. W. S. (1984). The borderline syndrome

index: Clinical validity and utility. Jour

nal ofNemous and Mental Disease, 172,

254-263.

Feighner, J., Robins, E., Guze, S. et al.

1 1972). Diagnostic criteria for psychiatric research. Archives ofGeneral Psychiatry,

35, 773-782.

Frances, A. (1980). The DSM-III personality

disorders section: A commentary. Ameri

can Journal of Psychiatry, 137, 1050-

1054.

Frances. A. (1985). DSM-III Axis II: Revi

sions and new methods of assessment.

Paper presented at issue workshop W52

at American Psychiatric Association. Dal

las, TX

Frances. A.. Clarkin, J.. Gilmore. M. et al.

( 1984). Reliability ofcriteria for borderline

personality disorder: A comparison of DSM-

III and DIB. American Journal of Psy

chiatry. 141, 1080-1084.

Griggs, S. M.. & Tyrer, P. J. (1981). Person

ality disorder, social adjustment and

treatment outcome in alcoholics. Journal

of Studies on Alcohol. 42. 802-805.

Gunderson. J. (1982a). Diagnostic inter

view for borderline patients (2nd ed.).

Belmont. MA: Harvard Medical School.

Gunderson. J. ( 1982b). Empirical studies in

the borderline diagnosis. In L. Grenspoon

(ed.), Psychiatry 1982. Annual Review.

Washington, DC: American Press.

Gunderson. J., & Kolb. J. (1978). Discrim

inating features of borderline patients.

American Journal of Psychiatry, 135,

792-796.

Gunderson, J., Kolb, J., & Austin, V. (1981).

The diagnostic interview for borderline

patients. American Journal of Psychia

try, 138, 896-903.

Guze. S. B.. & Goodwin. B. (1971). Diag nostic consistency in antisocial person

ality. American Journal ofPsychiatry, 128,

360-361.

Helzer, J.. Clayton, P., & Pambakian. R.

( 1982). Reliability of psychiatric diagnosis II: The test-retest reliability of diagnostic classification. Archives of General Psy

chiatry, 38, 674-677.

Hesselback. V.. Stabenau, J., Hesselback, M.

et al. 1 1982). A comparison of two inter

view schedules: The schedule for affective

disorders and schizophrenia lifetime

NIMH diagnostic interview schedule. Ar

chives of General Psychiatry, 39, 674-

677.

Hirschfeld. M. A.. & Klerman. G. L. (1979).

Personality attributes and affective dis

orders. American Journal of Psychiatry.

136, 67-70.

Hirschfeld, M. A., Klerman, G. L., Clayton, P. F. et al. (1983). Assessing personality: Effects of depressive state on trait meas

urement. American Journal of Psychia

try, 140, 695-699.

Hurt, S. W., Hyler, S. E., Frances, A. et al.

(1984). Assessing borderline personality disorder with self-report, clinical inter-

INSTRUMENTS MEASURING PERSONALITY DISORDERS 239

view, or semistructured interview. Amer

ican Journal of Psychiatry, 141, 1228-

1231.

Hyler, S., Reider, R., Spitzer, R. etal. (1982).

Personality diagnostic questionnaire

IPDQ). New York: New York State Psychi atric Institute.

Hyler. S., Reider, R.. Spitzer, R. etal. (1983).

Personality diagnostic questionnaire

tPDQ). New York: New York State Psychi atric Institute.

Ingham, J. G. (1966). Changes in M.P.I.

scores in neurotic patients: A three year

follow-up. British Journal of Psychiatry.

_112, 931-939.

/'Kass.'F., Skodol, A. E., Charles. M. A. et al. *

[4985). Scaled ratings of DSM-III person

ality disorders. American Journal of Psy

chiatry, 142, 627-630.

Kendell, R. E., DiScipio, W. J. (1968). Eyse- nek personality inventory scores of pa tients with depressive illness. British

Journal of Psychiatry. 114, 767-770.

Kernberg. O.. Goldstein, E.. Carr. A. et al.

(1981). Diagnosing borderline personal

ity: A pilot study using multiple diagnos tic methods. Journal of Nervous and

Mental Disease. 169, 225-231.

Kerr, T. A., Schapira, K., Roth. M. et al.

(1970). The relationship between the

Maudsley personality inventory and the

course of affective disorders. British Jour

nal of Psychiatry , 116, 11-19.

Koenigsberg, H., Kernberg, O.. & Schomer.

J. (1983). Diagnosing borderline condi

tions in an outpatient setting. Archives of

General Psychiatry, 40. 49-53.

Kolb. J.. & Gunderson, J. (1980). Diagnos

ing borderline patients with a semi-struc

tured interview. Archives of General Psy

chiatry. 37, 37-41.

Kroll. J.. Carey, K., Sines. L. el al. (1982).

Are there borderlines in Britain: A cross-

validation of U.S. findings. Archives of

General Psychiatry. 39. 60-63.

Kroll. J.. Pyle. R.. Zander, J. et al. (1981).

Borderline personality disorder: Interra-

ter reliabilitv of the diagnostic interview

for borderlines. Schizophrenia Bulletin,

7. 269-272.

Kroll. J., Sines. L.. Martin, K. et al. (1981).

Borderline personality disorder: Con

struct validity of the concept. Archives of

a General Psychiatry, 38. 1021-1026.

Liebowitz. M. R.. Stallone, F., Dunner, D. L.

'et al. (19791 Personality features of pa

tients with primary affective disorders.

Acta Psychiatrica Scandinavica. 60. 214-

224.

Loranger. A. W., Oldham. J. M.. Russakoff.

L. M. et al. (1984). Structured interviews

and borderline personality disorder. Ar

chives of General Psychiatry, 41, 565-

568.

Loranger. A. W., Susman, V. L.. Oldham. J.

M. et al. (1985). Personality disorders ex

amination PDE: A structured Interview

for DSM-UTR personality disorders (Ver

sion May 15. 1985). White Plains, NY: The

New York Hospital-Cornell Medical Cen

ter, Westchester Division.

Mazurc, C, & Gershom. E. (1979). Blind

ness and reliability in lifetime psychiatric

diagnosis. Archives of General Psychia

try, 36, 521-525.

Millon. T. (1981). Disorders of personality .

Axis II. New York: John Wiley and Sons.

Millon, T. ( 1982). Millon clinical multiaxlal

inventory (2nd ed.). Minneapolis. MN: In

terpretive Scoring Systems.

Mischel, W.. &Peake, P. (1982). Beyond deja vu in the search for cross situational con

sistency. Psychological Review. 89, 105

126.

Myers. J. K.. Weissman. M. M., Tischler, G

L. et al. 1 19841. Six month prevalence of

psychiatric disorders in three communi

ties. Archives of General Psychiatn). -U.

959-967.

Perry. J. ( 1982). The borderline personality

disorder scale (BPD-Scale). Cambridge.

MA: Cambridge Hospital.

Perry, J., & Klerman, G. (1980). Clinical fea

tures of borderline personality disorder.

American Journal of Psychiatry. 137.

165-173.

'(|?erry, J. C. O'Connell. M. E.. & Drake, R. ( 1 984 ) . An assessment of the schedule for

schizotypal personalities and the DSM-III

criteria for diagnosing schizotypal per

sonality disorder. Journal ofNervous and

-Mental Disease, 172, 674-680.

' PfohO B.. Coryell, W., Zimmerman, M. et al. v

rfn press). DSM-III personality disorders:

Diagnostic overlap and internal consist

ency of individual DSM-III criteria. Com

prehensive Psychiatry.

Pfohl. B., Stangl. D., & Zimmerman, M. (1982).

The structured interviewfor DSM-III per

sonality disorders (SID-P). Iowa City, 1A.

University of Iowa Hospitals and Clinics.

: PfohT^ B.. Stangl, D.. & Zimmerman, M. (7984). The implications of DSM-III per sonality disorders for patients with major

depression. Journal of Affective Disor

ders. 7, 309-318.

Pope, H. G.. Jonas. J. M.. Hudson. J. I. et

al. (1983). The validity of DSM-III border

line personality disorder. Archives ofGen

eral Psychiatry, 40, 23-30.

240 REICH

Reich. J. H. (1985). The relationship be

tween antisocial behavior and affective ill

ness. Comprehensive Psychiatry , 26, 296-

303.

Reich, J. H.. Noyes. R.. Coryell, W. et al. (in

preparation). The effect ofstate anxiety on

personality measurement.

...Reich, J. H., & Thompson. W. D. (1987).

Differential assortment of DSM-III person

ality disorder clusters in three popula

tions. British Journal ofPsychiatry , 150.

471-475.

Robins, L. N., Helzer, J., Croughan, J. et al.

(1979). The National Institute of Mental

Health diagnostic interview. Rockville,

MD: NIMH.

Robins, L. N., Helzer, J., Croughan, J. etal.

(1981). National Institute of Mental Health

diagnostic interview schedule: Its history, characteristics and validity. Archives of

General Psychiatry, 38, 381-389.

Robins. L. N.. Helzer. J., Ratcliff, K. et al.

(1982). Validity of the diagnostic inter

view schedule, version II: DSM-III diag nosis. Psychological Medicine. 12. 855-

870.

Robins. L. N.. Helzer, J. E.. &Weissman, M.

M. (1984). Lifetime prevalence of specific

psychiatric disorders in three sites. Ar

chives of General Psychiatry. 41. 949-

958.

Rushton. J., Brainerd. C & Presseley, M.

(1983). Behavioral development and con

struct validity: The principle of aggrega tion. Psychological Bulletin, 94. 18-38.

Singerman. B., Stoltzman, K.. Robbins. L.

et al. (1981). Diagnostic concordance be

tween DSM-III. Feighner and RDC. Jour

nal of Clinical Psychiatry, 42, 422-426.

Soloff, P. (1981a). A comparison of border

line with depressed and schizophrenia pa tients on a new diagnostic interview.

Comprehensive Psychiatry, 22, 291-300.

Soloff. P. (1981b). Affect impulse, and psy chosis in borderline disorders: A valida

tion study. Comprehensive Psychiatry , 22,

337-350.

Soloff, P. (1981c). Concurrent validation of

a diagnostic interview for borderline pa

tients. American Journal of Psychiatry.

138. 691-693.

Spitzer, R., Endicott. J.. & Robbins. E.

(1975). Clinical criteria for psychiatric di

agnosis and DSM-III. American Journal

of Psychiatry. 132, 1187-1192.

Spitzer, R., Endicott. J. et al. (1978). Re

search diagnostic criteria: Rationale and

reliability. Archives of General Psychia

try. 35. 773-782.

Spitzer, R. L.. Williams. J. B., et al. (1985,

March 1) Structured clinical interviewfor

DSM-III personality disorders (SCID-I1).

(Available from Biometrics Research De

partment. New York State Psychiatric In

stitute, 722 West 168th St.. New York. NY).

Stangl. D.. Pfohl, B.. Zimmerman. M. et al.

(1985). A structured interview for DSM-III

personality disorders: A preliminary re

port. Archives of General Psychiatry. 42.

591-596.

^ Stolzman, K., Helzer, J., Robins, L. et al.

(1981). Does the DSM-III differ from the

systems on which it was built? Journal

of Clinical Psychiatry. 42. 411-421.

Tyrer, P. (1979). Personality assessment

schedule (PAS). (Available from Mapper-

ley Hospital. Porchester Road, Not

tingham NG36AA, U.K.)

Q Tyrer. P.. & Alexander, J. (1979). Classifi

cation of personality disorder. British

Journal of Psychiatry . 135. 238-242.

Tyrer, P. . Alexander, M. S.. Cichitti, D. etal.

(1979). Reliability of a schedule for rating

personality disorders. British Journal of

Psychiatry. 135. 168-174.

v^ Tyrer. P., Casey, P., & Gall, J. (1983). Re

lationship between neurosis and person

ality disorder. Bntish Journal ofPsychia

try. 142, 404-408.

*^ Tyrer, P.. Cichetti. D. V.. Casey. P. R. et al.

(1984). Cross national reliability study of

a schedule for assessing personality dis

orders. Journal of Nervous and Mental

Disease, 172. 718-721.

'*-, Tyrer, P., Strauss, J.. & Cichetti, D. (1983).

Temporal reliability of personality in psy chiatric patients. Psychological Medi

cine, 13, 393-398.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.