Journal of Psychosomdc Research, Vol. 30, No. 1, pp. 17-91, 1986 0022-3999/86 $3.004 ~0.

00
Printed in Great Britain. 0 1986 Pergamon Press Ltd.

THE PSYCHOSOCIAL ADJUSTMENT TO ILLNESS

SCALE (PAIS)

LEONARDR. DEROGATIS
(Received 9 October 1984; accepted in revised form 6 June 1985)

Abstract-The current report introduces the Psychosocial Adjustment to Illness Scale (PAIS) and its self-
report version the PAIS-SR. The PAIS is a multi-dimensional, semi-structured clinical interview designed
to assess the psychological and social adjustment of medical patients, or members of their immediate
families, to the patient’s illness. The report reviews and discusses the concepts that form the foundation
for the development of the PAIS. In terms of psychometric characteristics, both internal consistency and
interrater reliabilities are presented and a series of predictive and convergent validity studies are reviewed.
Six normative illness groups are described and discussed, and contributions to construct validity for the
scale from factor analytic and other studies of internal structure are presented.

INTRODUCTION

DURING the past several decades there has been an intensified interest in the concept
of psychosocial adjustment in medicine. Weissman and her colleagues [l-3] note that
in psychiatry increased concern for the social reintegration of patients flows from
the current re-emphasis on the community care of psychiatric patients. In other
medical specialities, technologic and treatment advances have resulted in large
numbers of patients who would have previously suffered mortality now surviving
with more or less chronic disorders. The enhanced demands made by such illnesses
on the coping skills, psychological integrity, and social supports of these patients has
further emphasized the significance of psychosocial adjustment.
In two of her reviews on the measurement of psychosocial adjustment, Weissman
[l, 31 has identified approximately two dozen instruments developed to assess the
psychosocial functioning of various psychiatric patients. These measures range from
highly structured interviews taking up to 2 hr to administer, to self-report inventories
requiring 20 min to complete. For the medical patient, however, no such catalogue
of assessment techniques is available, in spite of the fact that the patient’s psycho-
social adjustment has been consistently observed [4-61 to be just as important to the
illness as the status of the physical disease. This observation has served as a major
impetus for the research program described in the present report.

DOMAINS OF THE PAIS

Just as intrapsychic functioning is typically viewed as consisting of distinct com-

ponents, e.g. mood and affect, intellectual processes, and memory, so psychosocial
adjustment is usually conceived of as being multidimensional. Many of the principal
dimensions of psychosocial adjustment are comprised of, or strongly associated
with, salient role behaviors. For example, the vocational role as worker, domestic
roles of spouse and parent, and social-leisure time roles all represent significant
dimensions of psychosocial functioning. It follows from this observation that

Johns Hopkins University School of Medicine.

Reprint requests should be addressed to Dr Leonard R. Derogatis, Division of Medical Psychology,
Adolf Meyer Building, Room 200, The Johns Hopkins Hospital, 601 N. Wolfe Street, Baltimore,
Maryland 21205, U.S.A.
 realistic measurement of psychosocial adjustment should represent it as being com-
prised of independent but related domains.
Although numerous factors relevant to psychosocial adjustment can be identified,
issues of cost-efficiency and time constraints demand that only the most central
domains be included in an instrument designed to serve as an operational defi-
nition of the construct. Consistent with this awareness, the PAIS was developed to
reflect adjustment in seven principal psychosocial areas, all of which have been
shown to have high relevance for adjustment to medical illness, both in our own
research, and in the experience of other investigators. These domains, outlined
below, are assessed via a semi-structured interview that requires approximately
20-30 min to complete, and may stand alone or be imbedded in a more compre-
hensive evaluative session.
The first domain of the PALS is Health Care Orientation. This domain is primarily
concerned with the nature of the respondent’s health care posture, and whether it
will function to promote a positive or negative adjustment to the illness and its treat-
ment. Patient attitudes about health care in general, perceptions of health care
professionals, quality of health information, and the nature of the patient’s expect-
ancies about his/her disorder and its treatment are all assessed.
The second domain of the PAIS is termed Vocational Environment. This section
is designed to reflect the impact that a medical disorder may have on vocational
adjustment, where ‘vocational’ is flexibly defined to indicate work, school, or
home, whichever is most appropriate. Items in this section assess perceived quality
of vocational performance, vocational satisfaction, lost time, vocational interest,
and a number of other variables that are associated with quality of adjustment in
the work sphere. Domestic Environment is the third domain of the PAIS. Measure-
ment in this domain is oriented toward illness-induced difficulties that arise pri-
marily in the home or family environment. It is designed to assess problems in
adaption experienced by the patient and the patient’s family unit in response to the
patient’s illness. The items comprising this section evaluate a number of aspects of
family living, including financial impact of the illness, quality of relationships,
family communications, and effects of physical disabilities.
The fourth domain is Sexual Relationships, and is designed to provide a quanti-
tative estimate of any changes in the quality of sexual functioning or relationship
associated with the patient’s illness or sequelae of the illness. Items are designed in a
progressive sequence, beginning with a focus on quality of interpersonal relation-
ship, and moving toward more specific issues of sexual functioning. Items assessing
sexual interest, frequency, quality of performance and level of satisfaction are
included in the measurement of this domain. Extended Family Relationships repre-
sents the fifth domain of the PALS. This section is devoted to measuring any dis-
ruption or derangement in relationships with the extended family constellation that
arise associated with the illness experience. The framework of typical interactions
with extended family is used to assess any negative impact of the illness upon
communication, quality of relationships, interest in interacting with family, and
other variables reflective of this life domain.
Social Environment is the sixth domain of adjustment measured by the PAIS, and
reflects the patient’s current social and leisure time activities, as well as the degree to
which the patient has suffered impairment or constriction of these activities as a
 Psychosocial adjustment to illness scale (PAIS) 19

result of the current illness and/or its sequelae. Activities are partitioned into
‘individual’, ‘family’, and ‘social’ categories, with items focused upon interests
and actual behavior in each category.
The final section of the PAIS covers Psychological Distress, and is designed to
measure dysphoric thoughts and feelings that are associated with the patient’s dis-
order, or are a direct result of the illness and its sequelae. Principal affective indi-
cators of psychological distress such as anxiety, depression and hostility, as well as
reduced self-esteem, body image problems, and inappropriate guilt are evaluated in
this domain. The format of the Psychological Distress domain is somewhat distinct,
from the others in that in addition to a suggested question to the patient for each
item, there is also a question to the interviewer that must be answered (see Appendix
A).
The PAIS interview is comprised of 46 items in total, each with a suggested
question to the interviewer designed to obtain the requisite information for that
item. Appendix A provides sample items from each of the seven principal domains
of the PAIS so that the reader may develop a more tangible appreciation of the
instrument.

THE PAIS-SR: A SELF-REPORT VERSION

The PAIS was designed as a semi-structured clinical interview primarily because

it was felt that the interview format with its enhanced flexibility and depth would
provide the most sensitive and valid estimate of the patient’s adjustment status. It is
obvious, however, that patients are not always available for personal interviews, and
in some research (e.g. large, community based studies), the costs associated with
professional interviewers can easily become prohibitive. The self-report modality of
measurement, although it sacrifices some flexibility, still retains much to recommend
it: self-report scales are relatively inexpensive, simple to administer, and eliminate
error due to rater biases. They also represent phenomena as experienced by a unique
source, the respondent himself, and on many occasions represent the only viable
alternative to no measurement at all.
For these reasons and others of a psychometric nature, we developed a self-report
version of the PAIS, the PAIS-SR [7]. In designing the PAIS-SR an attempt was
made to match the self-report instrument as closely as possible to the PAIS. The
self-report PAIS is also comprised of 46 items, and both instruments share a time
referent of ‘the past 30 days including today’. For each of the 46 PAIS items, there
is a matching item on the PAIS-SR. In most instances the self-report item is simply a
rephrased version of the original PAIS item; however, in a few instances, it was
necessary to rewrite the item entirely because of transparency and social desirability.
A parallel program of psychometric studies for the PAIS-SR, including the develop-
ment of a library of norms [S], is ongoing, and it is probably most realistic to
consider it as a companion instrument that measures the same constructs via a dif-
ferent modality.

NORMATIVE SAMPLES

Most psychological measuring instruments address a discrete number of primary

reference groups of interest, which in turn highly determine the development of
80 LEONARD R. DEROGATIS

specific norms for the test. With the PAIS, the typical normative strategy is imposs-
ible to implement since the scale measures adjustment to illness, and there are an
extremely large number of potential disorders and resulting illness groups. Primarily
for this reason, the normative development of the PAIS has been approached
through the concept of a normative library. This strategy involves the programmatic
development of norms for the PAIS and PAIS-SR for distinct illness categories as
representative cohorts of patients become available. Currently, six norms (four for
the PAIS and two for the PAIS-SR) have been completed, and are presented in
detail elsewhere [8].
The four normative groups available for the PAIS interview consist of a sample of
lung cancer patients (n = 120), a cohort of renal dialysis patients (n = 272), a group
of acute burn patients (n = 231), and a sample of essential hypertensives (n = 447).
Norms for the PAIS-SR are available for samples of cardiac bypass patients (n =
170), and heterogeneous cancer patients (n = 114). Table I provides a demographic
outline of each normative group.

RELIABILITY

In the context of psychological assessment, or any form of measurement, the

concept of reliability refers to consistency. Since there are multiple potential sources
of error in psychological measurement there are several different types of reliability.
Typically, internal consistency reliability, which derives from the domain sampling
model [9] and relates to the consistency of item representation, is one of the standard
forms calculated. Internal consistency reliability coefficients (coefficients CL)for the
PAIS and PAIS-SR have been calculated, and are given in Table II based upon three
different samples.
As can be seen from the tables, the subscales of the PAIS generally show high
internal consistency, with particularly uniform coefficients in the sample of kidney
dialysis patients. Coefficients are also high for the lung cancer sample, although the IX’S
for the Extended Family scale did drop to 0.12. In the original PAIS prototype, this
was the shortest scale of the instrument, consisting of only four items. Detailed
review of ratings distributions indicated almost no variation in the ratings on one of
the items, revealing the need to rewrite and expand this subscale.* Similar problems
were not apparent with the self-report PAIS-SR in our sample of cardiac patients;
however, the coefficient for Health Care Orientation did fall to 0.47.
When instruments are designed to elicit clinical judgements or ratings from clinical
observers, in addition to internal consistency reliability, reliability is also assessed
through evaluations of interrater agreement. Two estimates of interrater reliability
are given for the PAIS in Table III, based upon distinct sets of patients and clinical
raters.
As is evident from the magnitudes of the coefficients, agreement between clinical
judges on most PAIS domain scores was in the acceptable range, and in many
instances was quite good. A clinical psychologist and a nurse were the raters for the
small breast cancer sample, which represented a pilot cohort for a study of quality

*The developmental versions of the PAIS and PAIS-SR were comprised of a total of 45 items with four
items comprising the Extended Family domain. The final published versions of the scales consist of 46
items with a particularly ambiguous item from this domain rewritten as two distinct items.
 TABLE I.-DEM~CXAFWC CHARACTERISTICS
OF SIXNORMATIVEMEDICALSAMPLESFORTHE PAIS AND PAIS-SR

PAIS PAIS PAIS PAIS PAIS-SR PAIS-SR

Demographic lung renal acute essential cardiac mixed
information cancer dialysis burn hypertension cancer
(n = 120) (n = 272) (n=231) (n = 447) (n = 170) (n= 114)
n % n % n % n % n % n %
Sex
Male 120 (100) 153 (57) 189 (82) 237 (53) 154 (92) 41 (36)
Female 0 (0) 119 (43) 42 (18) 210 (47) 16 (8) 73 (64
Race
White 103 (86) 174 (64) - - 267 (60) 166 (98) 113 (99)
Black 17 (14) 85 (31) - - 165 (37) 2 (1) (1)
Other 0 (0) 13 (5) - - 15 (3) 2 (1) :, (1)
Social class
I 4 (3) 12 (4) - - 23 (5) - - (4)
II 12 (10) 19 (7) - - 14 (3) - - 1: (11)
III 25 (21) 26 (10) - - 58 (13) - - 41 (35)
IV 50 (42) 99 (36) - - 74 (17) - - 18 (15)
V 29 (24) 116 (42) - - 278 (62) - - 40 (35)
Married status
Married 89 (74) 140 (52) 155 (67) 294 (66) 132 (77) 90 (79)
Single 8 (7) 57 (21) 50 (22) 28 (6) (4) (4)
Other 23 (19) 75 (27) 26 (11) 125 (28) 3: (19) 2: (17)
Religion
Catholic - 100 (37) - - - - - - 32 (28)
Protestant - 145 (53) - - 66 (58)
Jew - - 10 (4) - - 10 (9)
Other - 6 (2) - - 2 (2)
None - - 11 (4) - - _ 3 (3)
Age
K 59.8 48.9 39.6 59.1 53.7 53.1
(I (8.38) (14.6) (12.1) (11.3) (9.6) (13.4)
82 LEONARD R. DEROGATIS

TABLE II.-RELIABILITY COEFFICIENTS(a) FOR THE PAIS AND PAIS-SR BASED ON RENAL
DIALYSIS,LUNG CANCER, AND CARDIAC PATIENTS

Coefficients a
Renal Lung
PAIS- dialysis cancer Cardiac
domain (PAIS: n = 269) (PAIS: n = 89) (PAIS-SR: n = 61)

1. Health orientation 0.63 0.83 0.47

II. Vocational environment 0.81 0.87 0.76
III. Domestic environment 0.87 0.68 0.77
IV. Sexual relationship 0.80 0.93 0.83
V. Extended family 0.66 0.12 0.62
VI. Social environment 0.78 0.93 0.80
VII. Psychological distress 0.80 0.81 0.85

TABLE Ill.-Two ESTIMATES OF INTERRATER RELIABILITIES FOR THE PAIS

BASED UPON 17 BREASTCANCER AND 37 HODGKINS PATIENTS

Intraclass r Intraclass r
PAIS breast cancer Hodgkins disease*
domain (n = 17) (n = 37)

I. Health orientation 0.74 0.70

II. Vocational environment 0.68 0.62
III. Domestic environment 0.61 0.52
IV. Sexual relationship 0.86 0.81
V. Extended family 0.56 0.33
VI. Social environment 0.82 0.72
VII Psychological distress 0.84 0.82
PAIS total score 0.86 0.83

*Previously reported in [ 131.

of life of patients treated through a regimen of mastectomy and adjuvant chemo-

therapy. In the Hodgkins sample, a heterogeneous group of interviewers, including
physicians, psychologists and social workers did the ratings on the patients.
Although there was discussion and information exchange concerning rating
standards in both instances, there were no formal instructive sessions, practice
interviews, or videotapes involved in either series.

VALIDITY

Although a number of authorities such as Nunnally [9-l l] have recently ques-

tioned the value of the continued use of the term ‘validity’, when we speak about
the validity of measurement we are basically concerned with the essence of what is
being measured. Particularly in the case of multidimensional tests, the confirmation
of the hypothesized dimensional structure of the instrument represents an important
step in construct validation. This usually involves a factor analysis of a represen-
tative set of individuals’ responses to the test and a comparison of the empirical
factor (dimensional) structure with the a priori rational-hypothesized structure of
the test. The degree of agreement between the rational and empirical structures
 Psychosocial adjustment to illness scale (PAIS) 83

provides a measure of the agreement between the theoretic and actual dimensions
of measurement, and thereby, of construct validity.
Table IV provides the results of a factor analysis of the PAIS, based upon clinical
assessments of the 120 patients comprising the lung cancer cohort. An initial prin-
cipal components analysis was accomplished, with subsequent rotation to a nor-
malized varimax solution [12]. The orthogonal rotated factor matrix is given in
Table IV.
TABLEN.-PAIS FACTORMATRIXROTATEDTOANORMALIZEDVARI~~AX CRITERIONBASEDONAN
ANALYSIS OF 120 LUNG CANCER PATIENTS*

Factor
Item no. PAIS domain I II III IV V VI VII

1 Health care orientation - - 0.41 - -

2 Health care orientation - - - 0.42 - - -
3 Health care orientation - - - 0.66 - -
4 Health care orientation - - - - 0.69 - -
5 Health care orientation - - - 0.83 - - -
6 Health care orientation - - 0.91 - -
I Health care orientation - - 0.87 - -
8 Health care orientation - - - 0.89 - - -
9 Vocational environment - 0.85 - - - - -
10 Vocational environment - 0.85 - - - -
11 Vocational environment - 0.83 - - -
12 Vocational environment - 0.52 - - - -
13 Vocational environment - 0.83 - - - -
14 Vocational environment - 0.63 - - -
15 Domestic environment - - - - - 0.54
16 Domestic environment - - - - 0.60
17 Domestic environment 0.74 - - - -
18 Domestic environment - - - - 0.54
19 Domestic environment - - - - - 0.62
20 Domestic environment - - - - - 0.49
21 Domestic environment 0.73 - - - - -
22 Domestic environment - 0.55 - - - -
23 Sex relations - - 0.72 - - -
24 Sex relations - - 0.86 - - -
25 Sex relations - - 0.86 - - - -
26 Sex relations - - 0.88 - - -
27 Sex relations - - 0.74 - - -
28 Sex relations - 0.62 - - - -
29 Extended family - - - - 0.73 -
30 Extended family - - - - 0.61 -
31 Extended family - - - - - -
32 Extended family - - - - 0.56 -
33 Social environment 0.78 - - - - -
34 Social environment 0.80 - - - - -
35 Social environment 0.81 - - - - -
36 Social environment 0.83 - - - - - -
37 Social environment 0.78 - - - - - -
38 Social environment 0.82 - - - -
39 Psychological distress - - - - - 0.74 -
40 Psychological distress 0.42 - - 0.59 -
41 Psychological distress - - - - - 0.40 -
42 Psychological distress - - - - 0.53 -
43 Psychological distress - - - 0.75 -
4.4 Psychological distress 0.42 - - - - 0.55 -
45 Psychological distress 0.69 - - - - - -

*Coefficients less than 0.35 were omitted from the table.

84 LEONARDR. DEROGATIS

Seven substantive dimensions were identified in the analysis, accounting for

approximately 63% of the variance in the matrix. In the varimax solution, factor I
accounted for 18% of the variance, with the remaining dimensions being associated
with 10, 9, 8, 7, 7 and 5% respectively. Although several of the factors were associ-
ated with relatively small proportions of variance in this sample, in a structure-
confirming analysis of this kind the fact that the hypothesized dimensional structure
of the PAIS was affirmed represents a significant contribution to validation.
Factor I has principal loadings from the items of the Social Environment domain.
In addition, several items from the Domestic Environment and Psychological
Distress domains show correlations with this factor. All Social Environment items
correlate solidly with factor I strongly suggesting that the central construct being
measured is social adjustment. Additional items which load on this factor include
items 45, 44 and 42 from the Psychological Distress scale which have to do with
body image distortion, self-esteem, and depression respectively. The two high-
loading items from the Domestic Environment scale measure physical disability
and difficulty with household tasks. This pattern of additional loadings suggests
that among patients with cancer, physical debilitation, self-image problems, and
depression are characteristics associated with a marked constriction in the social
environment, and adjustment difficulties in this domain.
Factor II reflects adjustment in the Vocational Environment. All six items from
the vocational domain load heavily on this dimension, confirming it as a clear
measure of the construct. In addition, we also observed that item 22, which is
concerned with adequacy offinancial resources, also loads on this measure.
Factor III is very explicit in its loading pattern, and clearly confirms the Sexual
Relations scale of the PAIS. All six items comprising the scale have marked loadings
on this dimension, with no appreciable loadings from other items.
Factor IV in the analysis represents the Health Care Orientation scale. Although
it reveals substantial loadings from most health care items, items 3 and 4 do not load
on this factor but rather correlate with factor V. These items reflect the patient’s
attitudes toward medicine and physicians, both in general, and concerning his/her
experience of the present illness. Factor V reflects adjustment in the Extended
Family domain, with three of the four items well-represented.
Factor VI draws its saturated loadings from the items of the Psychological
Distress scale. All items with the exception of item 45 measuring body image dis-
tortion, reveal substantial loadings on this measure. As mentioned previously, two
Psychological Distress items also show secondary loadings on Social Environment.
The final dimension, factor VII, primarily reflects items from the Domestic Environ-
ment domain. Five of the seven items comprising the scale load strongly on this
dimension, with an absence of additional items correlating. As mentioned above,
three items from this conceptual domain loaded primarily on other factors; how-
ever, the dimension still emerges in a clearly identifiable fashion.
In an ideal confirmatory factor analysis, each item loads univocally on the
dimension it is hypothesized to comprise, with no drift or scatter. In the real world
of empirical measurement, scores and relationships are affected by many important
respondent parameters such as diagnosis and stage of illness, sex, age, and numerous
other characteristics. In any particular analysis it is very unlikely to obtain a perfect
empirical reproduction of the rational dimensional pattern. What one strives for is
 Psychosocial adjustment to illness scale (PAIS) 85

the emergence of the fundamental dimensions inherent in the hypothesized structure

of the test, a goal which appears to have been reasonably well met in this analysis.

INTERRELATIONSHIPS AMONG DOMAINS AND THE TOTAL SCORE

One ideal of multidimensional measurement is to identify domains of a particular

construct of interest which are relatively specific and unique in their representation
of the construct, i.e. represent uncorrelated components of the attribute in question.
The appeal of such a configuration resides in the fact that it tends to reduce measure-
ment redundancy while increasing the breadth and sensitivity of measurement.
In developing the PAIS and the PAIS-SR, a great deal of attention went into first
identifying domains of adjustment that were felt to be relatively specific, and then
constructing items that would retain that specificity in empirical measurement. That
we have been reasonably successful in this endeavor is borne out by the data given
in Table V below.
The coefficients in Table V represent correlations among the seven domain scores
of the PAIS and PAIS-SR, and also between the domain scores and the PAIS total
score. Three separate sets of coefficients are given in the table (two for the PAIS and
one for the PAIS-SR) developed from three distinct study cohorts. Samples A and B
were two of our normative groups, while sample C was developed from an earlier
pilot study with the PAIS [13]. The coefficients demonstrate a consistent pattern
of low intercorrelations among domain scores with concomitant high correlations
with total score. This is true for both the PAIS, where the mean correlation among
domains was 0.33 among lung cancer patients and 0.10 among Hodgkins patients,
and the PAIS-SR where the average domain intercorrelation was 0.28. In almost
every instance, domain-total correlations were substantial and in the moderate-high
range. This pattern of relationship between components of the test approaches a
highly desireable configuration, and suggests that the PAIS and its self-report
counterpart (PAIS-SR) possess promising measurement potential.

CONVERGENT VALIDATION: RELATIONSHIPS WITH

EXTERNAL CRITERIA

The importance of demonstrating convergent and discriminative relationships

between test scores and external criteria is an essential step in the validation pro-
grams of modern psychological measures [ 141. In one of our studies with breast
cancer patients [15] we administered the PAIS along with a number of other psycho-
logical measures at the same point in time. Principal measures involved psychological
symptoms, mood and affect, patients’ attitudes and expectancies, and global adjust-
ment. Correlations between these tests and rating scales and the domain and total
scores of the PAIS are given below in Table VI.
Taking the Global Adjustment to Illness Scale (GAIS) first, we observe very high
correlations between the PAIS Total Adjustment score and the score on this ther-
mometer-type global adjustment rating scale (r=0.81). Of the domain scores,
Health Care Orientation also correlates very highly, as do Psychological Distress
and Domestic Environment. These are primary aspects of overall adjustment to
illness that one would logically anticipate being salient in arriving at a clinical judge-
ment, a pattern clearly reflected in the correlations. The SCL-90-R, a measure of
 TABLE V.-CORRELATIONS AMONG DOMAIN SCORES AND TOTAL SCORE FOR THE PAIS AND PAIS-SR BASED ON THREE CLINICAL SAMPLES

II III IV V VI VII PAIS total

a b ca b ca b ca b ca b ca b ca b c

I. Health care 0.15, 0.01, 0.03 0.32, 0.27, 0.05 0.16, 0.14, 0.19 0.30, 0.31, 0.24 0.39, 0.28, 0.14 0.22, 0.3 1) 0.05 0.56, 0.53, 0.46
orientation
II. Vocational 0.47, 0.03, 0.01 0.24, 0.03, 0.05 0.05, 0.1 I ( 0.01 0.45, 0.05, 0.04 0.35, 0.01 0.18 0.60, 0.12, 0.22 6
environment ?
111. Domestic 0.34, 0.39, 0.03 0.50, 0.59, 0.00 0.70, 0.52, 0.08 0.55, 0.58, 0.16 0.83, 0.79, 0.43
g
environment
?J
IV. Sexual 0.04, 0.23, 0.33 0.26, 0.41, 0.10 0.26, 0.30, 0.04 0.53, 0.60, 0.47 0
relationship E

V. Extended 0.51, 0.53, 0.22 0.25, 0.48, 0.01 0.47, 0.70, 0.08 :
family -1
z
VI. Social 0.55, 0.47, 0.09 0.83, 0.77, 0.34
environment
VII. Psychological 0.74, 0.76, 0.44
distress
PAIS total

*Sample a = 120 lung cancer patients-(PAIS); sample b = 148 kidney dialysis patients-(PAIS-SR); sample c = 37 patients with Hodgkins disease-(PAIS).
 TABLE VI.-CORRELATIONS BETWEEN PAIS DOMAIN AND TOTAL SCORES AND OTHER PSYCHOLOGICAL TEST SCORES BASED UPON
27 BREASTCANCER PATIENTS

(1) (11) (111) (IV w (VU (VII)

External PAIS Extended
criterion total Health care Vocational Domestic Sexual family Social Psychological
measure score orientation environment environment relations relations environment distress

Global adjustment
to illness scale
(GAIS) 0.81 0.80 0.54 0.76 0.46 0.40 0.12 0.70

SCL-90-R general
severity index 0.60 0.34 0.39 0.45 0.13 0.57 0.30 0.83

Affect balance
scale (ABS) index 0.69 0.31 0.46 0.67 0.42 0.58 0.31 0.77

Patient’s attitudes,
information and
expectancies
(PAIE) scale 0.64 0.88 0.28 0.52 0.40 0.06 0.14 0.42
88 LEONARD R.

psychological symptomatic distress, shows a different pattern of correlations.

Although the relationship between the SCL-90-R and the PALS total score is sub-
stantial (r=0.60), it is not as high as in the case of the GAIS, reflecting the fact
that psychosocial adjustment is comprised of more than simply an absence of
psychological symptoms. The maximal correlation between PAIS domain scores
and the SCL-90-R is with Psychological Distress (r= 0.83), which is highly consistent
with the stated construct. The Affect Balance Scale (ABS) is a measure of mood and
affect which also correlates in a consistent fashion with the PAIS. Highest ABS
correlation is with Psychological Distress (r=O.77), followed by the PAIS total
score (r=0.69), and Domestic Environment (r=0.67). Much as with the symptom
measure, correlations with measures such as Health Care Orientation and Social
Environment were much lower for the ABS. The PAIE scale was used as a simple
measure of the patient’s attitudes and expectancies concerning his illness, and the
information he/she possessed. It correlates substantially with the PAIS total score
(r= 0.64), but shows a particularly marked correlation with Health Care Orientation
(r=0.88). Correlations with other PAIS domain scores range from moderate to
low, a pattern in keeping with the construct definition of the PAIS domains.
Other investigators have also reported good convergent relationships for the
PAIS. In the pilot project alluded to earlier [13] a series of significant correlations
with external criteria for the PAIS domain scores were reported. Kaplan-DeNour
[16] has also reported significant positive relationships between the PAIS-SR total
score and the MAACL, and between physicians’ ratings of psychological impair-
ment and the Psychological Distress score on the PAIS-SR. Obviously, many more
convergent and discriminative studies will need to be accumulated to complete a
convincing relational network for the PAIS; however, the present investigations
represent a reasonable beginning.

PREDICTIVE VALIDITY: THE PRACTICAL UTILITY OF A SCALE

As has been noted previously [17], when most investigators and practitioners
make reference to the ‘validity’ of one or another psychological test, they are most
frequently referring to the predictive validity of the instrument. Predictive validity
refers to the capacity of the test to predict or discriminate among certain specific
definable outcomes, and defines the utility of the instrument for one or another
predictive application.
Predictive validity is highly specific to a particular status or criterion. Thus, a
test may have high validity as a screen for depressive disease, but be of little value in
screening for phobic disorders. Concerning the predictive validity of the PAIS, the
instrument has been utilized in a number of clinical research contexts that have
enabled us to test the predictive capacity of the instrument in relation to relevant
outcome criteria. The data in Table VII summarize the results of one such investi-
gation.
Table VII summarizes a comparison on the domain and total scores of the PAIS
between the 120 patients screened positive and confirmed for lung cancer in our
cohort, and 86 patients who were screened negative and served as controls. Assess-
ment took place at the second visit which varied from 1 to 3 months from the first.
 Psychosocial adjustment to illness scale (PAIS) 89

TABLE VII.-COMPARISON ON PAIS DOMAIN AND TOTAL SCORES BETWEEN

120PATIENTS SCREENED POSITIVE AND CONFIRMED FOR LUNG CANCER AND
86 PATIENTS SCREENED NEGATIVE

Screened positive Screened negative t-value P

PAIS domain (n = 120) (n = 86)
5T 0 3T 0

I. Health care 2.53 (2.99) 2.13 (2.40) 1.02 NS

orientation
II. Vocational 0.96 (2.75) 0.24 (1.08) 2.58 < 0.01
environment
III. Domestic 1.62 (2.69) 0.91 (1.60) 2.31 < 0.02
environment
IV. Sexual 1.10 (2.94) 0.93 (2.21) 1.58 < 0.10
relations
V. Extended 0.24 (1.03) 0.14 (0.69) 0.84 NS
family
VI. Social 1.32 (3.34) 0.48 (2.17) 2.19 < 0.05
environment
VII. Psychological 3.69 (3.53) 2.77 (2.92) 2.05 < 0.05
distress
PAIS
total score* 11.45 (13.55) 7.16 (7.32) 2.93 < 0.005

*PAIS total score as represented by a raw score sum rather than a sum of standardized
domain scores with a mean of 50 and SD of IO.

The comparison reveals that those with a confirmed diagnosis of lung cancer showed
poorer adjustment (higher scores) on all PAIS measures than the comparison
sample, with differences that were statistically significant on five of the seven
domain scores and the PAIS total. The total score was particularly sensitive to
differences between the groups (p < 0.005) with Vocational and Domestic Environ-
ment scores also being highly predictive. Social Environment and Psychological
Distress were somewhat less discriminating; however, differences remained statisti-
cally significant. The demonstration of the ability to discriminate between the
psychosocial adjustment of a seriously ill group of patients and a procedurally
similar control group is a convincing demonstration of predictive validity for the
scale.
Other researchers have also contributed to the predictive validity of the PAIS.
Kaplan-DeNour [ 161 contrasted the PAIS-SR scores of dialysis patients rated as
‘good’ and ‘bad’ adjusters by their treating physicians, and observed statistically
significant differences between the groups on PAIS-SR total and domain scores.
Working with a similar cohort of patients and the PAIS, Murphy [18] observed
consistent predictive relationships between PAIS domain scores and external
measures of adjustment. Lamping [19] has also done research with dialysis patients
supporting the validation of the PAIS. Her research demonstrated significant
correlations between the PAIS and appropriate scales from the POMS, MMPI,
and medical compliance measures. In addition, she was able to demonstrate that
PAIS scores could discriminate patients in terms of life events stress, work status,
severity of illness and coping styles.
 90 LEONARD R. DEROGATIS

These several investigations represent only the first few steps in building’ an
integrated and systematic network of studies that defines and sets limits for the
valid predictive relationships of the PAIS and PAIS-SR. As the instruments are
utilized with a greater number and more diversified illness groups, such definitions
should emerge more explicitly. For the moment, the PAIS and PAIS-SR have
proven reliable and valid in preliminary trials. Hopefully, they will continue to
fulfill this early promise and emerge as instruments of high utility for the assess-
ment of psychosocial adjustment in the medically ill.*

REFERENCES

I. WEISSMAN MM.. The assessment of social adjustment: a review of techniques. Archs Gen Psychiat
1975; 32: 357-365.
2. WEISSMAN MM, BOTHWELL S. Assessment of social adjustment by patient self-report. Archs Gen
Psych& 1976; 33: 1111-1151.
3. WEISSMAN MM, SHOLOMSKASD, JOHN K. The assessment of social adjustment: an update. Archs
Gen Psychiar 1981; 38: 1250-1258.
4. KAPLAN-DENOUR A. Social adjustment of chronic dialysis patients. Am J Psychiat 1982; 139:
97-100.
5. MURAWSKI BJ, PENMAN D, SCHMITT M. Soclai support in healril d11d illness: the concept and its
measurement. Cu Nurs 1978; 1: 365-371.
6. ZYZANSKI SJ, STANTON BA, JENKINS CD, KLEIN MD. Medical and social outcomes in survivors of
major heart surgery. J Psychosom Res 198 1; 23: 213-221.
7. DEROGA~IS LR. Psychological Adjustment to Illness Scale. Baltimore: Clinical Psychometric
Research, 1977.
8. DEROGA~IS LR, LOPEZ M. Psychosocial Adjustmenr IO Illness Scale (PALS & PAIS-SR) Scoring,
Procedures & Adminisfrufion Manual - 1. Baltimore: Clinical Psychometric Research, 1983.
9. NUNNALLY JC. Psychometric Theory, 2nd Edition. New York: McGraw Hill, 1978.
10. NUNNALLY, JC. An overview of psychological measurement. Clinical Diagnosis of Mental Disorders:
A Handbook (Edited by WOLMAN B). New York: Plenum Press, 1978.
11. MESSICK S. Constructs and their vicissitudes in educational and psychological measurement. Psycho1
Bull 1981; 89: 575-588.
12. HARMAN H. Modern Factor Analysis, 2nd edition. Chicago: University of Chicago Press, 1967.
13. MORROW CR, CHIARELLO RJ, DEROGATIS LR. A new scale for assessing patients’ psychological
adjustment to medical illness. PsycholMed 1978; 8: 605-610.
14. CAMPBELL DT, FISKE DW. Convergent and discriminant validation by the multitrait-multimethod
matrix. PsycholBulll959; 56: 81-105.
15. DEROGATIS LR, ABELOFF MD, MELISARATOSN. Psychological coping mechanisms and survival time
in metastatic breast cancer. JAm MedAssoc 1979; 242: 1504-1508.
16. KAPLAN-DENOUR A. Psychosocial adjustment to illness scale (PAIS): a study of chronic hemodialysis
patients. J Psychosom Res 1982; 26: I l-26.
17. DEROGATIS LR, SPENCER PM. Psychometric issues in the psychological assessment of the cancer
patient. Cancer Supplement 1984; 53: 2228-2232.
18. MURPHY SP. Factors influencing adjustment and quality of life: a multivariate approach. Doctoral
Dissertat University of Illinois Medical Center, Chicago, 1982.
19. LAMPING DL. Psychosocial adaption and adjustment to the stress of chronic illness. Doctoral
Dissertat Harvard University, Cambridge, 1981.

*The PAIS and PAIS-SR are distributed by Clinical Psychometric Research, 1228 Wine Spring Lane,
Baltimore, MD 21204, U.S.A.
 Psychosocial adjustment to illness scale (PAIS) 91

APPENDIX A

Selected items from the principal PAIS domains

I. Health care orientation (S items).

Item example: ‘Can you describe for me your general attitude and approach to taking care of your
health?’
0= very conscientious and attentive to personal health;
1= reasonably conscientious to health needs-some omissions;
2= inconsistent about attending to health needs;
3= clearly inattentive to health needs-minimizes importance.

II. Vocational environment (6 items).

Item Example: ‘Is your work as important to you now as it was before your illness?’
0= equal or greater importance;
1= slightly less important than before;
2= clear loss of investment in work situation;
3= minimal or no investment remaining in work

III. Domestic environment (8 items).

Item example: ‘To what degree has your illness interfered with your duties and tasks around the house?
0 = no interference;
1 = slight interference, easily overcome;
2 = substantial impairment in some domestic tasks;
3 =marked impairment, affecting all or nearly all tasks.

IV. Sexual relationships: (6 ifems).

Item example:‘When som&people become ill, they report a loss of interest in sexual activities; have you
experienced a reduction of sexual interest associated with your illness?’
0= no loss of interest;
1= slight loss of interest;
2= moderate loss of interest;
3= significant or total loss of interest

V. Extended family relationships (S items).

Item example: ‘Do you normally depend on members of your extended family for physical or financial
assistance? Has this changed at all since your illness?’
0 = no physical or financial dependency on extended family;
1 =some physical or financial dependency on extended family, however, consistent with degree of
family commitment;
2 = some physical or financial dependency on extended family, beyond the level of family commitment;
3 =marked physical or financial dependency on extended family, clearly beyond the degree of family
commitment.

VI. Social environment (6 items).

Item examp/e: ‘Have you maintained your interests in social activites since your illness (e.g. social
clubs, church groups, going to the movies)?’
0= same level of interest as previously;
1= slightly less interest than before;
2= significantly less interest than before;
3= little or no interest remaining.

VII. Psychological distress (7 items).

Item example: ‘Does the patient behave in a manner that suggests his/her self-image and self-esteem
have fallen?’
Suggested question for the patient: ‘Have you been feeling less adequate, or helpless, or somewhat
down on yourself since your illness?’
0 = not at all;
1 = mildly;
2 = moderately;
3 = markedly.