Full text of DOI:10.1093/geront/41.5.643

The Gerontologist
Vol. 41, No. 5, 643–651
Copyright 2001 by The Gerontological Society of America
Profile of Discrete Emotions in Affective
Disorders in Older Primary Care Patients
Larry Seidlitz, PhD,1 Jeffrey M. Lyness, MD,1 Yeates Conwell, MD,1
Paul R. Duberstein, PhD,1 and Christopher Cox, PhD2
Purpose: This research examined whether the frequencies of
specific emotions are associated with major and minor de-
pression in older primary care patients. Design and Meth-
ods: Older primary care patients (N = 146), prescreened
with a depression questionnaire, completed a diagnostic in-
terview and an emotions questionnaire. Results: Control-
ling for age, sex, and other psychiatric and medical illnesses,
major depressives differed from nondepressed controls in
nine emotions; minor depressives differed from controls in
four emotions. Major depressives differed from the controls
more in sadness, joy, and interest—but not anger, fear, or
guilt—than in comparison sets of emotions. Minor depres-
sives differed from the controls more in sadness and inner-
directed hostility—but not guilt, anger, fear, joy, or interest—
than in comparison sets of emotions. Implications: The
frequencies of discrete emotions are differentially associated
with major and minor depression; future research is needed
to determine their specific diagnostic and treatment implica-
tions.
and mood disorders is useful, as it may help bridge
the gap between theories of normal human emotion
and theories of late-life psychopathology.
Differential Emotions Theory (DET; Izard, 1977), a
general biopsychosocial theory that emphasizes dis-
crete emotion states, suggests that maladaptive link-
ages of specific types or patterns of emotions with
cognitive or behavioral systems that normally help
their regulation may lead to depressive disorders. In
addition, particular emotions, such as sadness, may
become linked maladaptively with others, such as an-
ger or guilt. DET suggests that depressive states tend
to be characterized by particular patterns of discrete
emotions. This viewpoint contrasts with the perspec-
tive that affective experience can be effectively ac-
counted for by two broad dimensions, positive and
negative affect, and that distinctions among particular
emotions within each dimension are relatively un-
important (Watson & Kendall, 1989). Izard (1977)
characterized the emotion patterns of adult depres-
sives in terms of the relative magnitudes of the re-
ported intensities of various discrete emotions, but did
not compare these patterns with those of control sub-
jects. Later studies, in which the emotion patterns of
depressed children and adolescents were compared
with nondepressed control subjects, tended to corrob-
orate the earlier patterns found in depressed adults,
though irritability was more prominent in the children
and adolescents (Blumberg & Izard, 1985, 1986;
Izard & Schwartz, 1986). The present study extends
this line of work by examining the patterns of emo-
tions in elderly people with major and minor depres-
sion and comparing them with nondepressed controls.
A more detailed description of the phenomenology
of emotions in depressed older persons may contrib-
ute to earlier detection and treatment, which in turn
may decrease morbidity and improve long-term
course (Keller et al., 1992). Enhanced assessment is
particularly needed because late-life depression is un-
derdiagnosed and undertreated (Freeling, Rao, Paykel,
Sireling, & Burton, 1985; Ormel et al., 1990). More-
over, the phenomenology of depressive disorders may
be less apparent in older than in younger persons (Al-
exopoulos et al., 1995; Gallo, Anthony, & Muthen,
1994; Garvey & Schaffer, 1994; Koenig, Cohen, Blazer,
Krishnan, & Sibert, 1993).
Key Words: Major depression, Minor depression,
Comorbidity, Late-life psychopathology
Disturbances in emotion are central components of
major and minor depressive mood disorders across
the life span, yet few studies have specifically exam-
ined the emotional experiences of older, depressed
people. Studying the relationships between emotions
This work was supported in part by National Institute of Mental
Health Grants K08 MH001408, K07 MH01113, K07 MH01135, and
T32 MH18911. We are grateful to the patients, staffs, and clinicians in the
practices of Drs. Judith Allen, Russell Maggio, Bruce Peyser, Marc Ber-
liant, Margaret Bergin, Robert Greene, Ralph Pincus, Mary Labanowski,
Charlene Conners, and the University of Rochester Family Medicine Cen-
ter at Highland Hospital for their selfless support of this project. We also
thank Cynthia Doane, Tamson Kelly Noel, and Adria Wisner for their co-
ordination of the study; Carrie Irvine, Holly Stiner, Aaron Gleason,
Gerard Kiernan, and Ziggy Yoediono for technical assistance; the support
staffs of the Program in Geriatrics and Neuropsychiatry and the Depart-
ment of Biostatistics; and Dr. Caine for his support and critique of the
manuscript.
Address correspondence to Larry Seidlitz, PhD, 2621 West Highway
12, Lodi, CA 95242. E-mail: Larry_Seidlitz@hotmail.com
Departments of 1Psychiatry and 2Biostatistics, University of Rochester
Medical Center, New York.
Vol. 41, No. 5, 2001
643

Another reason for examining discrete emotions
in late-life depression is that the frequencies of dis-
crete emotions may help distinguish normal mood
states from major and minor depressive disorders.
Although evidence suggests that subsyndromal de-
pression is associated with morbidity and mortality
(Broadhead, Blazer, George, & Tse, 1990; Wells et
al., 1989) and higher risk for subsequent major de-
pression (Horwath, Johnson, Klerman, & Weissman,
1992; Wells, Burnam, Rogers, Hays, & Camp,
1992), the proposition that minor depression is a dis-
tinct psychopathological condition remains contro-
versial (Angst & Ernst, 1993; Remick, Sadovnick,
Lam, Zis, & Yee, 1996; Skodol, Schwartz, Dohren-
wend, Levav, & Shrout, 1994). In gerontology, these
issues are especially salient because major depression
is less prevalent in geriatric populations (Blazer,
Hughes, & George, 1987), but older persons en-
dorse some depressive symptoms on questionnaire
measures more frequently than younger persons
(Mirowsky & Ross, 1992; Newmann, 1989; Zung,
Broadhead, & Roth, 1993), indicating age differ-
ences in clinical symptoms that are insufficiently re-
flected in the diagnostic criteria. Diagnosing depres-
sion in older medical patients is complex in part
because the criteria include symptoms that may be
caused by medical illness (Koenig, George, Peterson,
& Pieper, 1997). A more detailed understanding of
the emotional experience of older depressed and
nondepressed persons can help identify depression-
related symptoms in older patients that are not di-
rectly attributable to medical problems.
The frequencies and intensities of particular emo-
tions may have prognostic and therapeutic implica-
tions. For example, the Diagnostic and Statistical
Manual of Mental Disorders, 4th edition (DSM-IV;
American Psychiatric Association, 1994) criteria for
the melancholic features specifier for major depres-
sive disorder include either loss of pleasure in activi-
ties or lack of reactivity to pleasurable stimuli, along
with several other symptoms. In contrast, the atypi-
cal features specifier includes mood reactivity as its
primary symptom. Both specifiers have important
prognostic implications. A systematic investigation
of differences in the experience of a broad range of
emotions may help to refine the criteria for such sub-
types of depressive disorders and may help clinicians
predict the likely course of their patients’ depressive
illnesses. Such understanding may also contribute to
more effective, individually tailored psychotherapy
and pharmacotherapy for depression.
in reporting less happiness and contentment, and
more depression, sadness, and worry. A study of an
overlapping sample (Katz, Parmalee, & Streim, 1995)
classified the subjects as either major depressives,
dysphorics (defined as reporting significantly de-
pressed mood for 2 weeks or longer), or nonde-
pressed controls. Each group differed from the other
two groups in the Profile of Mood States (POMS;
McNair, Lorr, & Droppleman, 1971) factors of de-
pression, anger, vigor, fatigue, tension, and confusion.
These investigations help to provide a differenti-
ated description of the affective states associated
with major and minor depression in older persons.
Whereas diagnostic assessments are made partly on
the basis of the presence of particular emotions such
as sadness and guilt, the dimensional measurement
of their frequency is not synonymous with the cate-
gorical rating of their presence “most of the day,
nearly everyday, for 2 weeks or longer” (American
Psychiatric Association, 1994). Moreover, diagnoses
of major or minor depression do not require the
presence of even these two emotions, and the degree
to which they are present in older persons having ei-
ther of these disorders remains unclear. Further re-
search with older depressed patients is necessary to
determine the frequencies of other emotions (Ekman,
1993; Frijda, Kuipers, & ter Schure, 1989; Izard,
1977; Lazarus, 1991), and to compare and contrast
various emotions with each other in the magnitude
of their relationships with major and minor depres-
sion. Previous studies have not determined whether
particular emotions, relative to others, are differen-
tially correlated with these two disorders in older
persons. It is unclear whether all unpleasant emo-
tions are similarly elevated in older persons with ma-
jor or minor depression, or whether particular un-
pleasant emotions are especially prominent. It also is
uncertain whether one or both of the emotions of joy
and interest, which are treated jointly in one of the
diagnostic criteria for the two disorders, are lowered
in frequency in older depressed people.
In the present investigation, older primary care pa-
tients with either major or minor depression were
each compared with nondepressed patients in the
self-reported frequencies of 12 emotions. Most older
depressed persons are diagnosed and treated in pri-
mary care settings (Shepherd & Wilkinson, 1988).
DET (Izard, 1977) provided the general theoretical
framework and taxonomy. Basing its taxonomy on
the ability of observers to distinguish characteristic
facial expressions, DET distinguishes 11 basic emo-
tions and a 12th, labeled inner-directed hostility. The
latter composite emotion consists of a blend of an-
ger, disgust, and contempt directed against the self
and is posited to be prominent in depression. Finally,
to compare different emotions in the relative magni-
tudes of their relationships with major and minor
disorder, we examined whether differences between
the frequencies of specific emotions varied as a func-
tion of the subjects’ diagnosis. The associations be-
tween emotions and these diagnoses were examined
while controlling for the effects of confounding vari-
Previous research has shown that older residents
of a long-term care institution with major and minor
depressive disorders differed from nondepressed con-
trols in a variety of pleasant and unpleasant affec-
tive states (Lawton, Parmalee, Katz, & Nesselroade,
1996). Specifically, major depressives differed from
the controls in five pleasant (happy, content, warmth,
interest, and energy) and five unpleasant (sad, de-
pressed, irritated, annoyed, and worried) affects. The
minor depressives (defined as dysthymics or depres-
sive adjustment reactions) differed from the controls
644
The Gerontologist

ables: age, gender, and comorbid medical and psy-
chiatric conditions.
(84%) who were asked and were sufficiently fluent
in English to participate. Stratified sampling proce-
dures were used to include patients with a full range
of CES-D scores. Specifically, all subjects who scored
22 or above (n ꢀ 113) were asked to participate in
the study, as were a random sample (n ꢀ 323) of
those who scored below 22 (n ꢀ 836). Each week, a
random number table was used to select potential
subjects from those patients who scored below the
cutoff on the screening measure during the previous
week. The number that was selected depended on the
number of patients screened and the availability of
interviewer time. The cutoff of 22 was used because
a receiver operating curve analysis found 22 to be the
optimum cutoff for detecting major depression in
older primary care patients (Lyness et al., 1997). This
cutoff is consistent with unpublished data on the
CES-D from a mixed-age sample collected at the Uni-
versity of Pittsburgh (Herbert Schulberg, PhD, oral
communication, January 1994). Of 436 patients who
were selected as potential participants in the study,
216 (50%) agreed to complete a diagnostic interview
within a predetermined time-frame of one month.
Subjects above the cutoff who participated in the in-
terview (n ꢀ 56) did not significantly differ in CES-D
scores from those above the cutoff who did not par-
ticipate (n ꢀ 57). Similarly, those below the cutoff
who participated in the interview (n ꢀ 160) and
those below the cutoff who did not participate (n ꢀ
163) had comparable scores. Participants and non-
participants could not be distinguished on the basis
of age or sex (M ꢀ 71.50 years vs M ꢀ 71.37 years,
not significant (NS); 62% women vs 70% women,
NS; respectively).
We had the following hypotheses:
1. Consistent with related previous findings (Katz et
al., 1995; Lawton et al., 1996), persons with ei-
ther major or minor depression would differ from
nondepressed controls across a broad spectrum of
pleasant and unpleasant emotions.
2. Based on the fact that sadness, guilt, and inner-
directed hostility are reflected in the DSM-IV
criteria for major and minor depression, these
emotions would be related to major and minor
depression more strongly than the average of a set
of other unpleasant emotions (i.e., anger, fear,
disgust, contempt, shame, and shyness).
3. Based on related findings in previous studies (Gar-
vey & Schaffer, 1994; Katz et al., 1995), anger
and fear would be related to major and minor de-
pression more strongly than the average of a set
of other unpleasant emotions (i.e., disgust, con-
tempt, shame, and shyness).
4. Based on the fact that lower joy (i.e., diminished
pleasure) and interest are reflected in the DSM-IV
criteria for major and minor depression, these
emotions would be related to major and minor
depression more strongly than surprise, which is
not reflected in the DSM-IV criteria.
Methods
Participants (N ꢀ 146) aged 60 years or older
were recruited from private internal medicine offices
(n ꢀ 87) or from a freestanding, University-affiliated
Family Medicine Center (n ꢀ 59) using sampling pro-
cedures detailed below. The 146 participants were
classified into three groups: those with a current
major depressive disorder (n ꢀ 13); those with dys-
thymic disorder (n ꢀ 1) or current minor depres-
sion, defined as meeting the research criteria for mi-
nor depressive disorder in the appendix to DSM-IV
(n ꢀ 11); and the remaining participants who did not
meet criteria for current or past episodes of major
depressive disorder (n ꢀ 121). Two of the partici-
pants who were placed in the major depression
group had both current major depression and dys-
thymic disorder. Participants in this study overlapped
with those in other published studies (not cited to
preserve anonymity).
The final sample was selected from a larger num-
ber of patients from the primary care sites based on
sampling procedures and their completion of several
steps of data collection. To obtain an older sample
that was enriched with persons having significant de-
pressive symptoms, research personnel or office staff
administered the Center for Epidemiologic Studies–
Depression scale (CES-D; Radloff, 1977) to patients
aged 60 and older arriving for scheduled appoint-
ments. The CES-D was completed by 949 patients
The diagnostic interview was conducted using the
Structured Clinical Interview for Diagnostic and Sta-
tistical Manual of Mental Disorders, 3rd ed., revised
(SCID; Spitzer, Williams, Gibbon, & First, 1990), a
validated and reliable instrument that depends on
rater clinical judgment using all available sources of
data including the patient’s report, family report
when available, and medical records to arrive at Axis
I diagnoses. Raters, trained to administer all the mea-
sures used in the study, conducted the SCID and
rated the severity of depressive symptoms using the
highly validated, 24-item version of the Hamilton
Depression Rating Scale (Ham-D; Williams, 1988).
A physician–investigator reviewed the primary care
record for all subjects, along with all other avail-
able outpatient and inpatient medical and psychiatric
records. The interviewer presented the results of the
clinical interview at a weekly consensus conference
of investigators, raters, and other research personnel.
Consensus diagnoses on Axis I were assigned based
on all available data (primarily the SCID, Ham-D,
and record review). These diagnoses were used for
placing subjects into the major depression, minor de-
pression, or control groups for the present study. A
focus of the consensus conferences was determining
whether there was clear evidence that subjects met
the DSM-IV symptom criteria for the various de-
pression diagnoses, and on whether such diagnoses
Vol. 41, No. 5, 2001
645

should be ruled out because of evidence that the
symptoms had physical causes such as stroke or de-
mentia. Based on the medical record review and
other information collected in the SCID interview,
the physician–investigator assigned a rating of over-
all medical burden using the Cumulative Illness Rat-
ing Scale (CIRS; Linn, Linn, & Gurel, 1968). The
CIRS is a reliable and validated (Conwell, Forbes,
Cox, & Caine, 1993) measure of pathology based on
the sum of 5-point ratings of impairment for 13 inde-
pendent areas grouped under 6 body systems.
in age, sex, or Ham-D scores of depressive symptom
severity from patients with either of these disorders
who did not complete the DES-IV. Control subjects
who were included in the study did not differ in age
or sex from nondepressed patients who did not com-
plete the DES-IV, but they had slightly lower Ham-D
scores (M ꢀ 5.73, SD ꢀ 3.42 vs M ꢀ 8.00, SD ꢀ
4.84, t(137) ꢀ 2.48, p ꢁ .05). As 87% of the nonde-
pressed patients completed the DES-IV, however, the
mean of 5.73 was only slightly lower than that for
the entire pool of nondepressed patients (M ꢀ 6.02).
Control subjects who completed the DES-IV did not
significantly differ in their CIRS scores from poten-
tial control subjects who did not complete the mea-
sure, but participants in the study with major or mi-
nor depression had slightly lower CIRS scores than
the depressed patients who were excluded because
they did not complete the DES-IV (M ꢀ 7.16, SD ꢀ
2.27 vs M ꢀ 9.73, SD ꢀ 2.90, t(34) ꢀ 2.87, p ꢁ .01).
At the conclusion of the clinical interview, par-
ticipants were asked to complete and mail back in
a postage-paid envelope the Differential Emotions
Scale-IV (DES-IV; Izard, Libero, Putnam, & Haynes,
1993), a reliable and valid 36-item questionnaire
measure of the frequencies of 12 emotions recog-
nized in DET. In the instructions for the DES-IV,
participants were asked to consider their emotions
during the past week and indicate how often in their
daily life they experienced each emotion. Each of the
12 emotions is represented by 3 items, such as “feel
sad and gloomy, almost like crying?” (sadness) and
“feel scared, uneasy, like something might harm
you?” (fear). Items are rated on a 5-point scale:
“rarely” (1), “hardly ever” (2), “sometimes” (3), “of-
ten” (4), or “very often” (5). Ratings for the three items
representing each emotion are summed. In seven in-
stances in which participants omitted one of the three
items measuring an emotion, the score for the miss-
ing item was replaced by the average of the other two.
Of the 216 patients who completed the diagnostic
interview, 175 met the diagnostic criteria for inclu-
sion in the study (no past or present schizoaffective
or bipolar disorder, no major depression in full or
partial remission). Of these, 146 (83%) returned the
DES-IV. Patients with major depression in full or
partial remission were excluded a priori from the
analyses because it seemed likely these diagnoses
might affect current emotional functioning, but their
categorization in either the major or minor depres-
sion groups seemed inappropriate. Participants with
current or past episodes of bipolar disorder or
schizoaffective disorder were excluded a priori for
the same reason. Presence of other non-mood Axis I
disorders was not an exclusionary criterion, partly
because comorbid psychiatric disorders are common
in late-life depression and excluding participants
with such comorbidities would result in groups that
were unrepresentative of the population of interest.
Within the group with major depression, two partici-
pants had dementia, two had somatoform disorder,
one had both sedative dependence and somatoform
pain disorder, one had sedative dependence, one had
delusional-paranoid disorder, and one had simple
phobia. Within the group with minor depression,
one participant had a somatoform disorder. Within
the nondepressed group, three participants had demen-
tia, one had organic personality disorder, one had so-
cial phobia, and one had somatoform pain disorder.
Participants included in the study with either a
major or minor depression did not significantly differ
Results
Demographic and Clinical Characteristics
of the Sample
Table 1 describes the three groups in terms of
mean age, the percentage who were male, the per-
centage with a non-mood Axis I disorder, and the
mean CIRS scores. In the analyses to be reported, co-
variates were included to adjust for differences be-
tween the groups in these factors.
Transformation of Emotion Scores and the Removal of
Statistical Outliers
In the analyses to be reported, a square root trans-
formation was applied to the raw emotion scores in
order to achieve acceptable homogeneity of variance.
In addition, outliers whose standardized residuals ex-
ceeded three were removed from all analyses. The
number of outliers removed from a given analysis
never exceeded two. Statistical outliers are inconsis-
tent with the analysis of variance (ANOVA) assump-
tion of normally distributed errors, and tend to dis-
proportionately influence the outcome of statistical
tests. They may derive from a variety of factors, but
it is difficult to know the reason with certainty in a
given case. Their removal affected only two findings,
Table 1. Demographic and Clinical Characteristics Controlled in
Analyses With Covariates
Major
Minor
Control
Group
Depression Depression
Covariate
n ꢀ 13
n ꢀ 12
n ꢀ 121
Male, % (n)
Age, M (SD)
CIRS score, M (SD)
Comorbid psychiatric
diagnosis, % (n)
46.15 (6)
58.33 (7)
43.80 (53)
68.85 (8.02) 74.75 (8.34) 72.32 (7.19)
7.15 (1.72) 7.17 (2.82) 5.75 (2.58)
61.54 (8)
8.33 (1)
4.96 (6)
Note: CIRS ꢀ Cumulative Illness Rating Scale (Linn et al.,
1968).
646
The Gerontologist

and in these instances they will be discussed in fur-
ther detail.
cated that there were significant differences among
the three diagnostic groups for 11 of the 12 emo-
tions. Only surprise, which may be hedonically posi-
tive or negative, was unrelated to depression diagno-
sis. Also shown in Table 2 are the results of the
Scheffé-adjusted group contrasts between the major
depressives and the controls and the minor depres-
sives and the controls.
The removal of outliers affected two of the find-
ings: The pairwise group comparisons between the
major depressives and controls for disgust and con-
tempt changed from marginally significant to signifi-
cant after their removal. One control subject, who
was removed from both analyses, had unusually high
scores on most negative emotions, but did not meet
the diagnostic criteria for a depressive disorder. This
subject appeared to have clinically significant depres-
sive symptoms that were not captured by the diag-
nostic criteria, but also tended to be inconsistent in
reporting emotions and symptoms across different
instruments. Another control subject, who had de-
mentia, was removed from the analysis of disgust.
Group Differences in Individual Emotions
The descriptive statistics for the 12 DES-IV raw
emotion scores for each of the three diagnostic groups
are shown in Table 2. Prior to testing group dif-
ferences in the frequencies of each of the individ-
ual emotions, a multivariate analysis of covariance
(MANCOVA) was conducted to test for group dif-
ferences in the set of 12 emotions (sadness, guilt, in-
ner-directed hostility, fear, anger, disgust, contempt,
shame, shyness, joy, interest, and surprise). In this
analysis, the 12 emotions were the dependent vari-
ables; depression diagnosis was the independent
variable; and age, gender, presence or absence of a
concurrent Axis I diagnosis, and the CIRS were covar-
iates. The effect of diagnostic group was significant
(Wilks’s lambda F(24,254) ꢀ 3.17, p ꢁ .0001).
Next, hypothesis 1 was tested by examining group
differences in the reported frequencies of each of the
12 emotions using separate analyses of covariance
(ANCOVAs). In addition to the overall effect of de-
pression diagnosis, Scheffé-adjusted t tests were con-
ducted to examine the pairwise differences between
the major depressives and the nondepressed controls,
and between the minor depressives and the nonde-
pressed controls, respectively. Due to the relatively
small size of the two depression groups, pairwise dif-
ferences between them were not examined. In each
analysis, one of the 12 transformed emotion scores
was the dependent variable, depression diagnosis
was the independent variable, and the four covari-
ates described previously were included.
Group Differences in Patterns of Emotions
We next examined the relative magnitude of the
associations of the emotions with depression diag-
nostic group. As preliminary tests, two repeated-
measures MANCOVAs were conducted to determine
whether diagnosis group was (a) associated with any
one unpleasant emotion more than another un-
pleasant emotion, and (b) associated with any one
pleasant emotion more than another pleasant or
hedonically neutral emotion. In the first analysis, the
unpleasant emotions (sad, guilt, inner-directed hos-
tility, fear, anger, disgust, contempt, shame, and shy-
ness) were treated as dependent variables, and in the
second analysis, the pleasant or neutral emotions
(joy, interest, and surprise) were treated as depen-
The ANCOVA F statistics for the effect of diag-
nostic group on the transformed emotion scores,
controlling for the four covariates, are shown in Ta-
ble 2. Supporting Hypothesis 1, the ANCOVAs indi-
Table 2. Alphas, Means, and SDs of the DES-IV Emotion Subscale Scores and Results of Analysis of Covariance (ANCOVA) Tests
ANCOVA Statistics
Major
Depression
n ꢀ 13
Minor
Depression
n ꢀ 12
Control
Group
n ꢀ 121
Emotion
Alpha
df
F
p
Sadness
Guilt
Inner-directed hostility
Fear
.86
.68
.78
.82
.77
.73
.65
.72
.78
.76
.71
.65
10.54 (1.56)c
8.00 (1.73)
8.23 (2.45)
7.31 (2.29)
8.92 (2.60)
7.00 (2.52)
6.62 (2.79)
6.92 (3.17)
7.38 (3.10)
8.69 (2.43)
8.77 (1.24)
8.50 (2.42)
7.92 (2.19)
6.92 (2.19)
7.25 (2.34)
5.54 (2.35)
6.67 (2.61)
5.67 (2.67)
4.75 (1.60)
6.17 (2.37)
6.17 (3.33)
9.25 (1.71)
9.17 (1.80)
6.33 (2.19)
5.32 (2.18)
5.41 (1.76)
4.79 (1.76)
4.36 (1.76)
5.72 (2.11)
4.72 (1.93)
4.88 (1.80)
4.95 (1.84)
4.75 (1.93)
11.70 (1.69)
11.08 (2.03)
7.43 (2.21)
138
139
137
138
138
137
138
137
139
138
139
139
18.73
7.35
18.17
8.81
5.51
5.61
3.46
3.79
5.27
14.51
6.44
2.82
.0001a,b
.0009a
.0001a,b
.0002a
.0050a
.0046a
.0343a
.0250
Anger
Disgust
Contempt
Shame
Shyness
Joy
Interest
Surprise
.0062a
.0001a,b
.0021b
.0633
Notes: DES-IV ꢀ Differential Emotions Scale-IV (Izard et al., 1993). ANCOVAs were based on transformed data with 0–2 outliers
removed, controlled for age, gender, comorbid Axis I disorder, and medical illness.
aScheffé-adjusted, significant difference between controls and major depressives.
bScheffé-adjusted, significant difference between controls and minor depressives.
cn ꢀ 12 due to missing data.
Vol. 41, No. 5, 2001
647

dent variables. In both analyses, diagnostic group
was the main predictor, and age, gender, CIRS score,
and presence or absence of a comorbid psychiatric
diagnosis were covariates. An interaction of diagnos-
tic group with the within-subjects effect for type of
emotion would indicate that at least one emotion in-
cluded in the analysis was more strongly related to
diagnostic group than another emotion. For exam-
ple, if sadness and fear were rated as having similar
frequencies in the control group, but as having differ-
ent frequencies in the major depression group, it
would be reflected by an interaction of depression
group with the within-subjects effect for type of emo-
tion, and would suggest that one of these emotions
was more strongly related to major depression than
the other. This interaction was significant both in the
analysis of the unpleasant emotions, Wilks’s lambda
F(16,262) ꢀ 2.17, p ꢁ .01, and in the analysis of the
pleasant emotions, Wilks’s lambda F(4,274) ꢀ 2.84,
p ꢁ .03.
Having obtained these interactions of diagnostic
group with type of emotion in the repeated measures
MANCOVAs, Hypothesis 2 was then tested by con-
ducting three separate ANCOVAs. In these analyses,
the dependent variable was the mathematical differ-
ence between either sadness, guilt, or inner-directed
hostility, respectively, and the mean of the remain-
ing set of six unpleasant emotions (fear, anger,
contempt, disgust, shame, shyness); the independent
variable was diagnostic group; and the covariates
were age, sex, CIRS score, and presence or absence
of a nonmood psychiatric diagnosis. The results are
shown in Table 3. They indicated that sadness and
inner-directed hostility, but not guilt, were more
strongly related to diagnostic group than the mean of
the other unpleasant emotions. Pairwise group com-
parisons revealed that sadness distinguished both the
major depressives (t[138] ꢀ 3.82, p ꢁ .0002) and the
minor depressives (t[138] ꢀ 3.67, p ꢁ .0003) from
the nondepressed controls more than the mean of the
other unpleasant emotions. Inner-directed hostility
distinguished the minor depressives (t[138] ꢀ 3.56,
p ꢁ .0005), but not the major depressives (t[138] ꢀ
0.79, NS), from the nondepressed controls more
than the mean of the other unpleasant emotions.
Guilt did not distinguish either the major or the mi-
nor depressives from the nondepressed controls more
than the mean of the other unpleasant emotions.
To test Hypothesis 3, two further ANCOVAs
were conducted in a parallel manner using difference
scores as the dependent variables. One of these tested
whether diagnostic group was more strongly related
to anger than to the mean of the set of emotions
comprising disgust, contempt, shame, and shyness.
This hypothesis was not supported. The second tested
whether diagnostic group was more strongly related
to fear than to the mean of the same set of four nega-
tive emotions. This hypothesis also was not supported.
To test Hypothesis 4, we conducted two more
ANCOVAs to examine whether joy and interest, rel-
ative to surprise, were more strongly associated with
diagnostic group. As shown in Table 3, diagnostic
group predicted the differences between both joy and
interest from surprise. Pairwise group comparisons
revealed that relative to the nondepressed controls,
the major depressives reported less difference be-
tween their frequencies of joy and surprise (t[138] ꢀ
3.46, p ꢁ .0007), and less difference between their
frequencies of interest and surprise (t[138] ꢀ 3.09,
p ꢁ .0024); however, the differences between these
pairs of emotions were similar for the minor depres-
sives and the nondepressed controls.
Table 3. Analysis of Covariance Tests of Group Differences in the Relative Frequencies of Specific Emotion Contrasts
Hypothesis/Dependent Variable
df
F
p
Pairwise Group Contrasts
Hypothesis 2
Sadness minus mean of comparison emotions
2,138
2,138
2,137
12.63
6.41
1.00
.0001
.0022
NS
Major depressiona
Minor depressiona
Control groupb
Inner-directed hostility minus mean of comparison emotions
Major depressiona,b
Minor depressiona
Control groupb
Guilt minus mean of comparison emotions
Hypothesis 3
Anger minus mean of comparison emotions
Fear minus mean of comparison emotions
Hypothesis 4
2,138
2,139
0.17
1.27
NS
NS
Joy minus surprise
2,138
6.31
.0024
Major depressiona
Minor depressiona,b
Control groupb
Interest minus surprise
2,138
4.80
.0097
Major depressiona
Minor depressiona,b
Control groupb
Notes: Comparison emotions for Hypothesis 2 were fear, anger, disgust, contempt, shame, and shyness. Comparison emotions for
Hypothesis 3 were disgust, contempt, shame, and shyness. Covariates for each analysis were age, sex, physical illness burden (CIRS
score), and presence or absence of a non-mood psychiatric diagnosis. In the group contrasts, groups that differed significantly with re-
spect to the dependent variable have different superscripts; nonsignificance is denoted by overlapping superscripts. NS ꢀ not significant.
648
The Gerontologist

these “hostile” emotions may signal the development
of a major depressive episode, though further re-
search is necessary to confirm this possibility.
Discussion
Several limitations to this study should be kept in
mind when interpreting these results. First, the par-
ticipants may not be representative of the broader
population of older primary care patients, in part be-
cause a limited number of physicians’ practices were
sampled, and in part because the need to complete
several phases of data collection may have deterred
individuals with poorer emotional or physical health
from participating. On the other hand, this is the
only study to our knowledge to examine the emo-
tions of older clinically depressed patients in a pri-
mary care setting, where most depressed elders are
diagnosed and treated. The relatively small sample
sizes for the major and minor depression groups in
particular limited the representativeness of these
groups, gave insufficient power to test pairwise dif-
ferences between them, and may have limited our
ability to detect other group differences in individual
emotions or emotion patterns, though many hypoth-
esized group differences were detected. Limited sam-
ple sizes and multiple tests also increase the possibil-
ity of type one errors. The presence of comorbid
mental disorders, although typical in this population,
might also have affected the group differences in
emotions. Although this possible effect was con-
trolled through the use of a single covariate, this con-
trol was limited. A covariate was not used to ad-
just for antidepressant treatment, in part because it
would likely remove variance in emotion associated
with depression diagnosis. Therefore, there may have
been an uncontrolled effect of antidepressant treat-
ment on the emotions, though probably in the direc-
tion of reducing rather than increasing group differ-
ences. Limitations of the measure of emotions should
also be acknowledged. The DES-IV consists of only
three-item scales for each emotion. Whereas previous
findings (Izard et al., 1993), the alphas reported in
Table 2, and other findings presented in this article
suggest the scales have acceptable reliability and va-
lidity, replication of the findings with alternative self-
report and nonself-report measures is needed.
Whereas joy and interest significantly distinguished
both the major depressives and the minor depres-
sives from the nondepressed controls, these pleas-
ant emotions were by no means absent in the two
depressed groups. In fact, the major depressives re-
ported that joy and interest occurred at moderate fre-
quencies, and the minor depressives reported that
they occurred at least as often as any unpleasant
emotion. These findings should alert clinicians that
major and minor depressive disorders, at least in
older patients from primary care settings, are more
characterized by an increase in unpleasant emotions
than a decrease in pleasant emotions. Pleasant emo-
tions, although decreased in frequency, are unlikely
to be absent or rare. Whereas normal emotional
experience was characterized by high frequencies of
joy and interest and infrequent unpleasant emotions,
major depression was characterized by moderate fre-
quencies of joy and interest, and moderate to high
frequencies of unpleasant emotions. These observa-
tions do not rule out the possibility, however, that
there may be depressive subtypes that are character-
ized by diminished pleasurable emotions rather than
an increase in unpleasant ones. These findings do
suggest that the frequencies of the positive and nega-
tive emotions were relatively independent in their re-
lationships with major and minor depression. The in-
dependence of positive and negative affects has been
reported and discussed in other contexts (Diener &
Emmons, 1985; Larsen & Ketelaar, 1991; Watson,
1988).
The results also demonstrated that major and mi-
nor depression are not uniformly associated with
various pleasant and unpleasant emotions, respec-
tively, but are associated more strongly with particu-
lar emotions than with others. Specifically, sadness
distinguished both the major and minor depressives
from the control subjects more than a set of other
unpleasant emotions. Inner-directed hostility also
distinguished the minor depressives, but not the ma-
jor depressives, from the nondepressed controls bet-
ter than other unpleasant emotions. Whereas guilt,
anger, and fear were associated with having a depres-
sion diagnosis, these emotions were not more ele-
vated in the depressed groups than were comparison
sets of other unpleasant emotions. Because guilt is
specifically included in the DSM-IV criteria for ma-
jor and minor depressive disorders, the lack of a dif-
ferential elevation of guilt relative to other un-
pleasant emotions is noteworthy, particularly in
the context of other research suggesting the lack of
prominence of guilt in late-life depression (Wallace
& Pfohl, 1995). Whereas joy and interest were re-
ported to be more frequent than surprise by both the
nondepressed controls and the minor depressives,
this was not the case for the major depressives. This
suggests that joy and interest are differentially re-
duced in frequency in major depression but not in
minor depression.
Extension of Previous Studies of Emotion in
Late-Life Depression
The results of this study complement previous
studies by demonstrating that the reported frequen-
cies of 11 emotions, many of which have not been
examined previously in depressed older persons,
were associated with depression diagnoses. While
sadness was reported to be the most frequently expe-
rienced emotion in the major depressives, it often
was accompanied by high frequencies of self-directed
hostility, anger, disgust, and contempt. This suggests
that major depression in older adults is frequently
characterized by both inner- and outer-directed hos-
tility. It is notable, however, that the reported experi-
ences of outer-directed anger, disgust, and contempt
were not prominent in minor depression. In the con-
text of minor depression, therefore, an increase in
Vol. 41, No. 5, 2001
649

The findings in this older sample provided partial
support for the views of DET toward emotional ex-
perience in clinical depression (Izard, 1977), and sug-
gest a refinement of those views. As postulated by
DET, clinically depressed patients differed from the
nondepressed controls across a broad spectrum of
positive and negative emotions, and the groups dif-
fered in certain emotions more than others, most
prominently in sadness. Also consistent with previ-
ous findings, inner-directed hostility and anger were
notably elevated, though not necessarily significantly
more than other unpleasant emotions. The specific
pattern of secondary emotions involved in clinical
depression seems to vary depending on the age of the
sample, but further research is needed to bear this
out. The present findings suggest that the relative de-
gree of association of particular emotions with late-
life depression varied depending on whether the di-
agnosis was major or minor depression. For example,
inner-directed hostility was elevated relative to a
comparison set of emotions in minor but not in ma-
jor depression, apparently because several emotions
in the comparison set were substantially more ele-
vated in major depression than in minor depression.
Similarly, the positive emotions of joy and interest
were comparatively more reduced in frequency than
was surprise in major depression but not in minor
depression. These findings suggest that the severity
of depressive illness (i.e., minor vs major depression)
is associated differently with different emotions—in
other words, is associated with the quality and tone as
well as with the intensity of the overall emotional state.
pathways for late-life depression. For example,
studies could examine whether secondary depres-
sive disorders, those “due to a general medical con-
dition” (American Psychiatric Association, 1994),
are characterized by specific patterns of emotions
that are distinguishable from those patterns expe-
rienced in primary depressive disorders. A related
issue is whether early-onset and late-onset depres-
sive disorders are associated with distinct emotion
patterns (Baldwin & Tomenson, 1995; Conwell,
Nelson, Kim, & Mazure, 1989). As understand-
ing of the neurobiological and psychosocial mech-
anisms of mood disorders increases, individual
differences in emotional functioning are likely to
become important indicators of specific types of
underlying dysfunction.
References
Alexopoulos, G. S., Meyers, B. S., Young, R. C., Chester, J., Feder, M., &
Einhorn, A. (1995). Anxiety in geriatric depression: Effects of age and
cognitive impairment. American Journal of Geriatric Psychiatry, 3,
108–118.
American Psychiatric Association. (1994). Diagnostic and statistical man-
ual of mental disorders (4th ed.). Washington DC: Author.
Angst, J., & Ernst, C. (1993). Current concepts of the classification of af-
fective disorders. International Clinical Psychopharmacology, 8, 211–
215.
Baldwin, R. C., & Tomenson, B. (1995). Depression in later life. A com-
parison of symptoms and risk factors in early and late onset cases.
British Journal of Psychiatry, 167, 649–652.
Blazer, D., Hughes, D. C., & George, L. K. (1987). The epidemiology of
depression in an elderly community population. The Gerontologist,
27, 281–287.
Blumberg, S. H., & Izard, C. E. (1985). Affective and cognitive character-
istics of depression in 10- and 11-year-old children. Journal of
Personality and Social Psychology, 49, 194–202.
Blumberg, S. H., & Izard, C. E. (1986). Discriminating patterns of emo-
tions in 10- and 11-year old children’s anxiety and depression. Journal
of Personality and Social Psychology, 51, 852–857.
Future Research
Broadhead, W. E., Blazer, D. G., George, L. K., & Tse, C. K. (1990). De-
pression, disability days, and days lost from work in a prospective
epidemiological survey. Journal of the American Medical Association,
264, 2524–2528.
Conwell, Y., Forbes, N. T., Cox, C., & Caine, E. D. (1993). Validation of
a measure of physical illness burden at autopsy: the Cumulative Illness
Rating Scale. Journal of the American Geriatrics Society, 41, 38–41.
Conwell, Y., Nelson, J. C., Kim, K. M., & Mazure, C. M. (1989). Depres-
sion in late life: Age of onset as marker of a subtype. Journal of
Affective Disorders, 17, 189–195.
Diener, E., & Emmons, R. A. (1985). The independence of positive and
negative affect. Journal of Personality and Social Psychology, 47,
1105–1117.
Ekman, P. (1993). Facial expression and emotion. American Psychologist,
48, 348–392.
Freeling, P., Rao, B. M., Paykel, E. S., Sireling, L. I., & Burton, R. H.
(1985). Unrecognized depression in general practice. British Medical
Journal, 290, 1880–1883.
Frijda, W. H., Kuipers, P., & ter Schure, E. (1989). Relations among emo-
tions, appraisal, and emotional action readiness. Journal of Personality
and Social Psychology, 57, 212–228.
Although the present study clarifies the relations
of discrete pleasant and unpleasant emotions with
late-life major and minor depressive disorders, future
research is necessary to examine the implications of
these relations.
1. Future studies are needed to examine whether a
predisposition for “hostile” emotions such as an-
ger, disgust, and contempt are risk factors for ma-
jor depression or for longer duration, greater like-
lihood of recurrence, or greater risk for suicide.
Further, possible mediators of the effects of emo-
tions on clinical outcome variables, such as inter-
personal conflict or cognitive processes, may be
examined.
Gallo, J. J., Anthony, J. C., & Muthen, B. O. (1994). Age differences in
the symptoms of depression: A latent trait analysis. Journal of Geron-
tology: Psychological Sciences, 49, P251–P264.
2. Future studies are needed to test whether particular
patterns of emotions in depressive disorders are in-
dicative of the likelihood of responding to particu-
lar types of antidepressant medications or psycho-
therapeutic treatments or approaches. For example,
mood disorders accompanied by high levels of fear
or anxiety may respond relatively better to particu-
lar types of medications or psychotherapies than
those not characterized by anxiety.
Garvey, M. J., & Schaffer, C. B. (1994). Are some symptoms of depression
age dependent? Journal of Affective Disorders, 32, 247–251.
Horwath, E., Johnson, J., Klerman, G.L., & Weissman, M.M. (1992). De-
pressive symptoms as relative and attributable risk factors for first-
onset major depression. Archives of General Psychiatry, 49, 817–823.
Izard, C. E. (1977). Human emotions. New York: Plenum Press.
Izard, C. E., Libero, D. Z., Putnam, P., & Haynes, O. M. (1993). Stability
of emotion experiences and their relation to traits of personality. Jour-
nal of Personality and Social Psychology, 64, 847–860.
Izard, C. E., & Schwartz, G. M. (1986). Patterns of emotions in depres-
sion. In M. Rutter, C. E. Izard, & P. B. Read (Eds.), Depression in
young people (pp. 33–70). New York: Guilford Press.
3. Future studies are needed to investigate whether
emotional experience is indicative of distinct causal
Katz, I. R., Parmalee, P. A., & Streim, J. E. (1995). Depression in older pa-
650
The Gerontologist

tients in residential care: Significance of dysphoria and dimensional
assessment. American Journal of Geriatric Psychiatry, 3, 161–169.
Keller, M. B., Lavori, P. W., Mueller, T. I., Endicott, J., Coryell, W.,
Hirshfeld, R. M. A., & Shea, M. T. (1992). Time to recovery, chronic-
ity, and levels of psychopathology in major depression: A 5-year
prospective follow-up of 431 subjects. Archives of General Psychiatry,
49, 809–816.
Koenig, H. G., Cohen, H. J., Blazer, D. G., Krishnan, K. R. R., & Sibert,
T. E. (1993). Profile of depressive symptoms in younger and older
medical inpatients with major depression. Journal of the American Ge-
riatrics Society, 41, 1169–1176.
Koenig, H. G., George, L. K., Peterson, B. L., & Pieper, C. F. (1997). De-
pression in medically ill hospitalized older adults: Prevalence,
characteristics, and course of symptoms according to six diagnostic
schemes. American Journal of Psychiatry, 154, 1376–1383.
Larsen, R. J., & Ketelaar, T. (1991). Personality and susceptibility to
positive and negative emotional states. Journal of Personality and Psy-
chology, 61, 132–140.
Lawton, M. P., Parmalee, P. A., Katz, I. R., & Nesselroade, J. (1996). Af-
fective states in normal and depressed older people. Journal of
Gerontology: Psychological Sciences, 51B, P309–P316.
Lazarus, R. S. (1991). Emotion and adaptation. New York: Oxford Uni-
versity Press.
Linn, B. S., Linn, M. W., & Gurel, L. (1968). Cumulative Illness Rating
Scale. Journal of the American Geriatrics Society, 16, 622–626.
Lyness, J. M., Noel, T. K., Cox, C., King, D. A., Conwell, Y., & Caine, E. D.
(1997). Screening for depression in elderly primary care patients: A
comparison of the Center for Epidemiologic Studies–Depression scale
and the Geriatric Depression Scale. Archives of Internal Medicine,
157, 449–454.
McNair, D. M., Lorr, M., & Droppleman, L. F. (1971). Profile of mood
states. San Diego, CA: EdiTS/Educational and Industrial Testing
Service.
Mirowsky, J., & Ross, C. E. (1992). Age and depression. Journal of Health
and Social Behavior, 33, 187–205.
Newmann, J. P. (1989). Aging and depression. Psychology and Aging, 4,
150–165.
Remick, R. A., Sadovnick, A. D., Lam, R. W., Zis, A. P., & Yee, I. M.
(1996). Major depression, minor depression, and double depression:
Are they distinct entities? American Journal of Medical Genetics, 67,
347–353.
Shepherd, M., & Wilkinson, G. (1988). Primary care as the middle ground
for psychiatric epidemiology. Psychological Medicine, 18, 263–267.
Skodol, A. E., Schwartz, S., Dohrenwend, B. P., Levav, I., & Shrout, P.E.
(1994). Minor depression in a cohort of young adults in Israel. Ar-
chives of General Psychiatry, 51, 542–551.
Spitzer, R. L., Williams, J. B. W., Gibbon, M., & First, M. B. (1990).
Structured clinical interviews for DSM-III-R. Washington, D.C.:
American Psychiatric Association.
Wallace, J., & Pfohl, B. (1995). Age-related differences in the symptomatic
expression of major depression. Journal of Nervous & Mental Dis-
ease, 183, 99–102.
Watson, D. (1988). The vicissitudes of mood measurement: Effects of
varying descriptors, time frames, and responses found on measures of
positive and negative affect. Journal of Personality and Social Psychol-
ogy, 55, 128–141.
Watson, D., & Kendall, P. C. (1989). Understanding anxiety and depres-
sion: Their relation to negative and positive affective states. In P.C.
Kendall & D. Watson (Eds.), Anxiety and depression: Distinctive and
overlapping features (pp. 3–26). San Diego, CA: Academic Press.
Wells, K. B., Burnam, M. A., Rogers, W., Hays, R., & Camp, P. (1992).
The course of depression in adult outpatients. Results from the Medi-
cal Outcomes Study. Archives of General Psychiatry, 49, 788–794.
Wells, K. B., Stewart, A., Hays, R. D., Burnam, M. A., Rogers, W.,
Daniels, M., Berry, S., Greenfield, S., & Ware, J. (1989). The function-
ing and well-being of depressed patients. Results from the Medical
Outcomes Study. Journal of the American Medical Association, 262,
914–919.
Williams, J. B. W. (1988). A structured interview guide for the Hamilton
Depression Rating Scale. Archives of General Psychiatry, 45, 742–
747.
Zung, W. W. K., Broadhead, W. E., & Roth, M. E. (1993). Prevalence of
depressive symptoms in primary care. Journal of Family Practice, 37,
337–344.
Ormel, J., Van den Brink, W., Koeter, M. W. J., Giel, R., Van der Meer,
K., Van de Willige, G., & Wilmink, F. W. (1990). Recognition, man-
agement and outcome of psychological disorders in primary care: A
naturalistic follow-up study. Psychological Medicine, 20, 909–923.
Radloff, L. S. (1977). The CES–D scale: A self-report depression scale for
research in the general population. Journal of Applied Psychological
Measurement, 1, 385–401.
Received April 11, 2000
Accepted May 16, 2001
Decision Edition: Laurence G. Branch, PhD
Vol. 41, No. 5, 2001
651