Am J Psychiatry /38:1, January 1981 1

THE AMERICAN JOURNAL OF PSYCHIATRY

The Symptoms of Major Depressive Illness

BY J. CRAIG NELSON, M.D., AND DENNIS S. CHARNEY, M.D.

syndrome. The unipolar-bipolar distinction depends

The authors suggest that the common characteristics
on a history of mania rather than the character of the
ofvarious descriptions ofmajor depressive illness -
depressive syndrome. Primary affective disorder is
endogenous, endogenomorphic, psychotic, and
diagnosed when no preexisting nonaffective psychiat-
melancholic-are an autonomous course, a need for
nc illness is present. Symptoms were not considered
biological treatment, and a presumed alteration in
useful for making the primary-secondary distinction
neurochemistry. The literature is reviewed to
(1). Thus, while symptom criteria were provided,
determine those symptoms which best characterize
symptoms were less specific than the past history of
this syndrome. These symptoms may prove useful in
illness in defining that entity. Although both of these
developing valid diagnostic criteriafor major
diagnostic systems introduced an important dis-
depressive illness.
tinction, they are of limited value for defining the
symptom characteristics of major depressive illness.
Other investigators have suggested that the concept
S ince the introduction of the primary affective dis- of endogenous depression implies more than a lack of
order criteria (1 2), , interest in the use of explicit precipitant and is useful for defining the character of
criteria for the diagnosis of depressive illness has major depressive illness. Rosenthal and Kierman in
grown considerably. The development ofthe Research 1966 (7) stated that endogenous depression is charac-
Diagnostic Criteria (RDC) (3) and the operational cii-
terized by typical symptoms and signs, a relatively
teria for DSM-III (4) reflect this interest. Although the
stable premorbid personality, a course unaffected by
value of reliable criteria has been emphasized, the na- environmental events, and an onset that may be
ture of the depressive entity defined by the criteria has unassociated with a psychological precipitant. Klein
received less attention. (8) suggested that it is the morphology of endogenous
The characteristics of endogenous depression have depression which characterizes the syndrome and in-
been well described. However, the endogenous dis- troduced the term endogenomorphic for descriptive
tinction based on lack of precipitating stress was chal- clarity. Endogenous depression has also been associ-
lenged by the demonstration that many depressions ated with an altered physiologic state (9-12). Van
considered endogenous were preceded by stress Praag (13) indicated that the description of altered
(5). Therefore, recent diagnostic systems have avoided
physiologic mechanisms would be useful in estab-
this terminology. Two diagnostic systems widely rec- lishing the diagnostic validity of the depressive state.
ognized in the past two decades, the primary-second- Endogenous depression has also been assumed to
ary (1, 2) and unipolar-bipolar (6) distinctions, are both have a specific relationship to treatment, either phar-
diagnoses of exclusion with respect to the depressive macotherapy (14-17) or ECT (18-21). Thus, there ex-
ists in the literature a concept of major depressive ill-
ness characterized by a depressive syndrome unaf-
fected by environmental changes, associated with
Received May 14, 1979; accepted Aug. 7, 1979.
alterations in neurochemistry and requiring biological
From the Department of Psychiatry, Yale University School of
Medicine. treatment. The terms endogenous, psychotic,

Address reprint requests to Dr. Nelson, Department of Psychia- and endogenomorphic

have been used to refer to
try, Yale University School of Medicine, 333 Cedar St. , New this entity. Rosenthal and Gudeman (22) suggested
Haven, CT 06510.
Copyright 1981 American Psychiatric Association 0002-953X1 that the term autonomous depression, introduced
8 1/OI/0001/13/$OO.50. by Gillespie (23) in 1929, is most descriptive.
2 MAJOR DEPRESSIVE ILLNESS Am J Psychiatry 138:1, January 1981

In previous reports we examined the ability of the and provide a framework for the development of valid
criteria for primary affective disorder to identify pa- and useful descriptive criteria.
tients with an autonomous depressive illness and to
distinguish them from patients who appear responsive
to psychosocial intervention (24). In our sample, a!- MULTIVARIATE ANALYSIS STUDIES

though the criteria for primary affective disorder iden-

tified most patients with autonomous depression, they Numerous investigators have used factor analysis or
also included 45% of the patients with nonautonomous cluster analysis to determine if an endogenous entity
depression. It was clear that the patients identified by can be generated. In 14 factor analytic studies that ex-
these criteria were a heterogeneous group with regard amined a mixed group of depressed patients, evidence
to the autonomous-nonautonomous distinction. Re- of an endogenous factor was found (7, 18, 19, 22, 27-
cently, Feinberg and associates reported misdiagnosis 36). Eight studies that used various forms of cluster
ofendogenous depression based on the RDC for major analysis determined a cluster which appeared to be en-
depressive episode (25). Spitzer and associates (3) dogenous (16, 32, 37-42). As previous reviewers have
have also reported a lack of agreement between cri- concluded, the evidence for endogenous depression is
teria for major depressive disorder and a subset of cri- substantial (27, 43-47). What is less clear is whether
teria intended to identify patients with endogenous de- endogenous depression is 1) a discrete entity that can
pression. Although psychobiological research of a!- be distinguished clearly from other forms of nonen-
fective illness continues to focus on a concept of dogenous depressive illness, or 2) the severe end of a
endogenous depression, the diagnostic criteria for pri- spectrum of depressive illness. Regardless of that dis-
mary affective disorder and the RDC for major depres- tinction, an accurate symptomatic description of en-
sive episode do not identify a homogeneous group of dogenous depression would be useful.
patients with endogenous depression. However, the We reviewed those analytic studies that identified an
value of criteria that do identify such patients is obvi- endogenous factor or cluster for evidence ofsymp-
ous. toms commonly associated with endogenous depres-
The final revision of DSM-III (4) introduces an im- sion. The analytic techniques used in these studies are
portant change with respect to the diagnosis of depres- powerful tools, but one must also understand their tim-
sion. Recognizing the heterogeneity of patients meet- itations. Factor analysis determines dimensions that
ing criteria for major depressive
episode, DSM-III describe correlated groups of symptoms, while cluster
proposes a subcategory, termed melancholia, for analysis distinguishes groups of patients who have
patients who have more severe depression with symp- common features. Both forms of analysis can describe
toms characteristic ofendogenous depression and who only those patients initially selected for study. In addi-
are likely to require and respond to somatic treatment tion, symptoms selected for review must include items
(26). This concept of major depressive illness is consis- that are likely to distinguish endogenous from reactive
tent with the concept of endogenous or endogeno- depression. The method of assessing symptoms can at-
morphic depression described above and is helpful in so have an important influence on the outcome. A
defining the nature of the depressive entity for which common criticism of some of the earlier factor analytic
criteria can be developed. studies was the halo effect
that might result from

DSM-III does provide criteria for the diagnosis of clinicians expectations of associations among certain
melancholia. However, there has been little discussion symptoms (27). Although the influence ofclinician bias
of the evidence supporting the selection of specific is decreased in self-report rating scales, such in-
symptoms as criteria. In 1968 Mendels and Cochrane struments may be less reliable in severely depressed
(27) reviewed seven factor analytic studies to deter- patients (48) and thus may have limited usefulness in
mine symptoms characteristic of endogenous depres- the study of endogenous depression, which is usually
sion. Since that time the number of factor analytic severe. Recent studies have attempted to control for
studies of endogenous depression has more than dou- rater bias by using structured interviews or rating
bled. In addition, several studies employing cluster scales with established reliability.
analysis as well as discriminant function analysis have
Factor Analytic Studies
provided descriptive data characteristic of endogenous
depression. A few investigators have attempted instru- We reviewed 20 factor analytic studies of depressed
mental measurement of symptoms of endogenous de- patients. One study that included only patients with
pression. Finally, several treatment-response studies severe psychotic depression was not useful in examin-
have investigated the symptomatic characteristics of ing the endogenous-nonendogenous distinction (49).
depressed patients who require pharmacologic inter- Another study included anxious and depressed pa-
vention. tients and was more useful for examining the dis-
The purpose of our report is to review the descrip- tinction between those two groups (50). In two studies
tive literature on endogenous depression, determine factor analysis was used to derive several descriptive
the symptoms that best characterize the syndrome, dimensions rather than general factors (37, 51). Of the
Am J Psychiatry 138:1, January /98! J. CRAIG NELSON AND DENNIS S. CHARNEY 3

16 studies designed to determine the presence of an the most appropriate analytic technique for testing an
endogenous factor in a population of depressed pa- actual taxonomy or classification system of depressive
tients, 14 determined such a factor and 2 did not (52, illness; however, the 9 cluster analytic studies exam-
53). One ofthe studies produced a factor described by med are difficult to compare because of variability in
the authors as endogenous; however, that factor dif- the diagnostic populations sampled, the ratings used,
fered descriptively from the factors reported in other the symptoms included, and the quantification of the
studies (27). In that study several items other than data. Nevertheless, these studies do provide some
symptoms (e.g., age, personality traits) were used to symptomatic description of the clusters, which can be
derive the factor. The endogenous factors identified in reviewed. Studies using inverse factor analysis are dis-
the other 13 studies were reasonably similar (7, 18, 19, cussed with those using cluster analysis because both
22, 28-36). Data from these 13 studies are presented in attempt to identify groups of patients rather than
table 1. groups of symptoms.
It can be seen that the sample in each study was Of the 9 cluster analytic studies, 8 identified an en-
limited to depressed patients, so subtypes of de- dogenous cluster. The ninth study (54) identified a
pressed patients could be distinguished. The studies somewhat similar group of patients with severe de-
included both inpatients and outpatients and men and pressed mood and retardation. However, this group
women. The most recent study (36) included data from was particularly characterized by chronicity and ad-
three separate samples. However, one sample, a group vanced age and other variables not usually associated
of volunteers, did not appear comparable with other with endogenous depression. The specific symptoms
patient samples studied. Therefore, we have included characteristic of the endogenous clusters identified in
only data from the other two samples, both taken from the 8 studies are shown in table 3. The data presenta-
populations of depressed outpatients. tion in these studies ranged from bar graphs to raw
Table 1 also presents the factor loadings for the 13 score means to cluster loadings, so it is not possible to
studies described. The factor loadings are presented in present comparable numerical data. Positive findings
this form so that it is possible to review the association are indicated in the table, however. In the 1961 study
of individual symptoms with the endogenous factor by Grinker and associates (37), an endogenous factor
across studies. It is also possible to inspect the magni- was not found in the factor analysis, but the authors
tude of the association in the studies. Because studies reported that an analysis of patterns formed by pa-
vary in item definition, rating methods, and sample tients identified a group which appeared similar to en-
size, the actual symptom loadings are not directly dogenous depression. Overall and associates (38), in
comparable. However, finding an association of a 1966, presented the results ofa profile analysis that
symptom with the endogenous factor under these dif- were useful in predicting response to treatment. One
ferent conditions actually strengthens the general- of the profiles that was specific in terms of response to
izability ofthe finding. Symptom variables are present- antidepressant treatment was termed a retarded
ed only if they were reported in at least 4 studies. For cluster or profile. Because the authors used the Brief
purposes of comparison, the factors are presented so Psychiatric Rating Scale (55) to describe the cluster,
that symptoms associated with endogenous depression only a few of the items were specific for depression.
have a positive sign. In addition, while there were mi- However, the BPRS items of emotional withdrawal
nor interstudy variations in descriptions of a given and blunted affect seem similar to lack of responsivity
symptom, items have been grouped together under the and loss of interest, respectively. In another study of
relevant symptom heading. depressed outpatients, Fahy and associates (32) de-
Table 2 summarizes data for symptoms examined in scribed a cluster similar to those described previously
at least 4 studies and indicates the number of studies in for endogenous depression, although the authors in-
which the factor loading for the symptom was either dicated only a limited number ofsymptoms descriptive
low (<.30), moderate (.30-.49), or strong (.50). It is of the cluster.
interesting that low factor loadings were found for a In 1969 Pilowsky and associates (39) reported the
group of symptoms which have been used as criteria results ofa study employing numerical taxonomy, a
for the diagnosis of major depressive illness (3, 4)- form of cluster analysis. The authors used a 57-item
suicidal thinking or attempts, weight loss, difficulty questionnaire that included several items related to the
falling asleep, and midnight awakening. Loss of appe- endogenous-reactive distinction. In that study there
tite and loss of energy, which are included in the cri- were two samples-one composed of patients with en-
teria for primary affective disorder, were included in dogenous and reactive depression and a second of pa-
only two studies. tients with other psychiatric diagnoses. The relation-
ship found between the actual clinical diagnosis and
Cluster Analytic Studies
the statistical cluster was greater than predicted by
Nine studies have used various forms of cluster chance, but by no means a perfect fit.
analysis to generate groups of patients with similar de- Everitt and associates (40) described the results of a
pressive features (16, 32, 37-42, 54). Clusteranalysis is collaborative London-New York study of 500 patients
4 MAJOR DEPRESSIVE ILLNESS Am J Psychiatry 138:!, January 1981

TABLE 1
Symptom Loadings for the Endogenous Factor in 13 Studies of Depressed Patients

Hamilton Kiloh Hordern Carney Rosenthal Rosenthal Kay Fahy

and and and and and and and and
Sample Characteristic White Garside Associates Associates Kierman Gudeman Kendell Associates Associates
andSymptom (28) (29) (30) (18) (7) (22) (19) (31) (32)
Sample
Number 64 143 137 129 50 100 696 104 126
Status Inpatients Outpatients Inpatients Inpatients Inpatients Inpatients Inpatients Inpatients Outpatients
and
outpatients
Sex Male Mixed Female Mixed Female Female Mixed Mixed Mixed
Symptom
Depressedmood .76 37 #{149}79 .65 .53 .61
Distinct quality of mood .39 .42 .61 .70
Guilt, self-reproach .73 .20 .45 .67b .50 .54 49b
.71 .26
Self-pity -.32 -.55 -.22 -.31 -.48
Retardation .68 .55 .63 .55 .57 .44 .64
Agitation -.03 .17 .41 .44 .53 .34 <.2
Lackofreactivityt .60 .57 .26 .67 .64
a.m. worsening .53 .06 .08 .12 -.32 .40
Suicidal ideation or
attempts .53 .23 .25 .22 .10 .43 .19 .45 <.2
Lossofinterest .46 .77 .42 .51 .35
Lossofweight .35 3je .18 .36 .45 .25 .30
Difficultyfallingasleep .21 -.24 .19 -.31 .37 .26 -.13 <.2
Mid-night awakening .28 - . 14 .27 .70 .52
Early morning awakening .34 .69 .30 .23 .59 .42 - .29 .38
Anxiety -.37 -.07 .25 -.69 .17 .37 .20 <.2
Gastrointestinal complaints .28 .33 .44 .53
Hypochondriasis .16 -.12 .08 -.62 .10 .06 .06 <.2
Difficulty concentrating
orperplexity .26 .55 .53 .36
Irritability - . 17 - .47 - .20 - .04 . 12 < .2
Delusions of guilt,
retribution, nihilism .59-.67 .44 .49
ajp of depressed mood.
bIUSjOflj guilt.
tltems rated as reactivity changed to lack of reactivity with change in sign.
dBoth ideas and attempts rated, loading for attempts listed.
CLOSS of more than 7 pounds.
Range of loadings for types of delusions (e.g. . somatic, nihilistic).

with a variety of psychiatric disorders, 209 of whom 1974 (42) and another by Raskin and Crook in 1976
had a clinical diagnosis of depression. The impressive (16), are especially interesting. Both studies used large
aspect of this study is that in a large sample from dif- samples and produced more than one cluster of severe
ferent settings a group of patients with severe depres- depression. Overall described a study of 2,000 de-
sive illness was identified as a specific cluster by two pressed patients, predominantly inpatients, in which
different types ofanalysis. Thus the study gives strong three subtypes of severe depressives were generated.
support for the diagnostic grouping of severe or psy- The retarded subtype was similar to previous descrip-
chotic depressives. The study is less helpful in terms tions. The hostile subtype had the highest ratings for
of reviewing the specific symptoms associated with delusional thinking, guilt, hostility, and anxiety. The
that cluster. ratings appeared to indicate a more severe depression
Paykel (41), in 1971 , reported a cluster analysis than the BPRS scores for hostile depression in the
classification of depressed patients. He identified a 1966 report by Overall and associates (38). The third
group of severely depressed patients, which he called cluster identified agitated patients, who also had high
psychotic depressives.
This group appeared to have ratings for somatic concern, anxiety, and excitement.
the characteristics of endogenous depression. Paykel A fourth cluster of anxious patients with generally
also found evidence of three other groups that ap- lower scores was also described.
peared more characteristic of different types of neu-
Raskin and Crook (16) presented a review of 880 in-
rotic depression. His study was particularly useful patients included in a collaborative study. This report
because of its careful methodology and its inclusion of described an inverse factor analysis that identified four
symptoms thought to be descriptive of endogenous de- types of depressed patients. One group was consid-
pression. ered typical of endogenous depression. Another group
Two more recent studies, one reported by Overall in ofequally severe agitated depressed patients was iden-
Am J Psychiatry 138:1 , January 1981 J. CRAIG NELSON AND DENNIS S. CHARNEY 5

TABLE 2
Factor Analytic Studies: Association of Symptoms with the Endoge-
nous Factor
Garside Paykel KiIoh Lewinsohn
and and and and Number w ith Facto r Loading
Associates Associates Associates Associates
Positive Negative
(33) (34) (35) (36) Number
Including .49 to + .29 to .30 to
269 220 145 94 72 Symptom Item .50 .30 - .29 .49 .50

Inpatients Outpatients, Inpatients Outpatients Outpatients Retardation 12 11 1 0 0 0

day, and Lackofreactivity 10 9 0 1 0 0
inpatients Severity of
Mixed Mixed Mixed Mixed Mixed depressedmood 9 7 1 1 0 0
Lossofinterest 8 4 4 0 0 0
.66 .17 .62a Delusional
.71 .40 .55 .43 .45 thinking 4 2 2 0 0 0
.43 .24 .34 .36 .63 Distinct quality 9 4 5 0 0 0
.04 -.38 Guilt 14 6 5 3 0 0
.71 .50 .68 .68 .58 Early morning
.41 .51 awakening 13 3 6 4
.74 .57 .76 .71 .65 Difficulty
.45 .41 .50 concentrating 5 2 1 2 0 0
Agitation 9 2 4 3 0 0
- .02 .25d Gastrointestinal
.39 .76 .65 complaints 5 1 3 1 0 0
.39 .02 .40 Weight loss 10 0 7 3 0 0
-.07 .23 Morning
-.04 .23 worsening 9 2 3 3 1 0
.36 .24 .65 .06 .34 Mid-night
.10 .20 awakening 7 2 0 5 0 0
.37 Suicidal ideas or
- .37 .32 .22 thoughts 11 1 2 8 0 0
Difficulty falling
.27 asleep 10 0 1 8 1 0
-.50 -.54 Anxiety 10 0 1 7 1 1
Hypochondriasis 11 0 1 8 1 1
.51 Self-pity 7 0 0 2 4 1
Irritability 8 0 0 5 1 2

heavy factor loadings. These three symptoms received

strong support in the cluster analytic studies. The clus-
ter studies also provide strong support for three items
tified. A third group ofanxious depressed patients was that had more moderate loadings in the factor analytic
considered by the authors more typical of neurotic
studies-guilt, agitation, and delusional thinking. 0th-
depression. The fourth group consisted of younger er symptoms identified in the factor studies that also
patients with poor premorbid history. A review of received some support in the cluster studies were loss
these data indicates that both the agitated and endog- of interest, early morning awakening, morning wors-
enous groups had severe autonomous depression. ening, difficulty concentrating, and mid-night awak-
The agitated patients were more likely to be delusion- ening. Distinct quality of depressed mood and weight
al, guilty, and suicidal. Both groups received high loss, which were less often examined in the cluster
scores for severe depressed mood, lack of reactivity, studies, received support in only two reports. Loss of
agitation, morning worsening, difficulty concentrating, appetite and lack of insight were each supported by
midnight awakening, early morning awakening, and only one cluster study. Other items that had mixed
hypochondriasis. The endogenous group received (both positive and negative) or no association with the
higher ratings on weight loss and somatic concerns. It endogenous cluster were suicidal thinking, difficulty
is interesting that both of these large-sample studies falling asleep, anxiety, irritability, and somatic con-
produced at least two clusters of severe depressed pa- cerns.
tients, of which one was an agitated group.
Discriminant Function Studies
The cluster analytic studies as a group provide
strong evidence for the existence of subtypes of se- Discriminant function analysis is used to determine
verely depressed patients. The studies also seem to which features discriminate between two known diag-
agree on the symptoms that characterize this cluster. nostic entities. Thus symptoms differentiating known
In most of the factor analytic studies, severe de- groups of endogenous and neurotic depressed patients
pressed mood, retardation, and lack of reactivity had would best be determined by discriminant function
6 MAJOR DEPRESSIVE ILLNESS Am J Psychiatry 138:!, January 1981

TABLE 3
Symptoms of Endogenous Depression Identified by Cluster Analysis in 8 Studies

Grinker and Overall and Fahy and Pilowsky and Eventt and
Sample Characteristic Associates Associates Associates Associates Associates Paykel
and Symptom (37) (38) (32) (39) (40) (41)

Sample
Number 96 126 200 480 160
Diagnosis Depression Depression Mixed Mixed Depression
Status Inpatients Outpatients Outpatients Inpatients Outpatients
and inpatients and inpatients
Rating 47-item checklist BPRS Structured 57-item Present State Clinical interview
(feelings & behavior) interview questionnaire Exam (modified Hamilton)
Cluster name Pattern A Retarded type Cluster B Class B Psychotic Psychotic group
depression
Symptom
Retardation Strong Strong Strong Moderate Moderatea Strong
Lack of reactivity Strongb Strongc Strong Moderate Moderate
Depressed mood Strong Moderate Moderate Moderate
Loss of interest Strong Strong Moderate Slight
Delusional thinking Moderatee Strong
Distinct quality Moderate Moderate
Guilt Moderate Moderate Moderate Moderatee Moderate
Early morning awakening Moderate Moderate
Difficulty concentrating Strong Moderate
Agitation Moderate Moderate Strong
am. worsening Moderate Moderatea None
Loss of weight Moderate
Mid-night awakening Moderate None
Suicidal ideas or attempts Negative
aLondon sample higher on this item.
hlsolated and withdrawn. not responsive to others.
cEmotional withdrawal. not responsive to intervention.
Blunted affect. apathetic. no interest.
eBrooklyn sample higher on this item.

analysis. We reviewed studies that used discriminant retardation. Suicide attempts before admission had a
function analysis to differentiate characteristics of en- strong negative correlation with the endogenous dis-
dogenous and reactive depressives (19, 56-58). We a!- tinction. Other items characteristic of the psychotic
so reviewed two studies (59, 60) designed to dif- depressed patients (although with lower weights) in-
ferentiate depressed patients from anxious patients; cluded loss of insight, delusions of guilt, seif-depreca-
however, since the depressed groups were not clearly tion, and lack of energy. Fleiss (57) analyzed data on
endogenous, the data are not presented. 65 psychotic and 38 neurotic patients from the London
Of the four discriminant function studies of de- portion of the previous study by discriminant function
pressed patients, three used patient populations that analysis. Items he found useful in distinguishing the
had also been examined using factor or cluster analysis psychotic group were retarded movement, agitation,
described in previous sections. Thus the data from and somatic concerns.
these discriminant function analyses do not provide in- Bhrolchain and associates (58) described a discrimi-
formation from a new group of patients. However, nant function analysis of 1 14 women (73 inpatients, 41
they do provide a different form of analysis for review- outpatients), a population not previously reviewed in
ing symptoms. this paper. The subjects were apparently psychotic
Kendell (19), in his 1968 monograph, described a and neurotic depressed patients, although it is not
discriminant function analysis that involved 391 pa- clear whether patients with other diagnoses were ex-
tients with psychotic depression and 250 with neurotic cluded. A structured Present State Examination (61)
depression. The items that best distinguished the psy- was used to rate symptoms. The raters restricted their
chotic depressive group were depressive delusions and investigation to the patients symptomatic presenta-
psychomotor change (retardation or agitation). Severe tion to avoid possible bias from knowledge of life
insomnia and persecutory delusions also characterized events or social circumstances. Overall severity and
the psychotic group. In 1970 Kendell and Gourlay (56) retardation were found most useful in distinguishing
did a discriminant function analysis of the 178 de- psychotic from neurotic patients. Early morning awak-
pressed patients identified in the collaborative Lon- ening, severity of obsessive thinking, loss of appetite,
don-New York study (40). Symptoms with heavy load- slowness of thinking and speech, and severity of func-
ings for the endogenous group were auditory halluci- tional impairment were also more prevalent in the psy-
nations, marked agitation, ideas of reference, and chotic group.
Am J Psychiatry 138:! , January 1981 J. CRAIG NELSON AND DENNIS S. CHARNEY 7

distinct quality of depressed mood. Other predictors

Raskin and
Crook were early morning awakening, depressed mood, re-
Overall (42) (16) tardation, and weight loss. Early morning worsening,
guilt, and somatization did not distinguish between the
2000 880 two groups.
Mixed Depression and schizophrenia In a previous report Nelson and Charney (24) de-
Inpatients and outpatients Inpatients
scribed a study of symptoms in 76 inpatients. Of this
BPRS Several rating scales group, 36 were designated endogenous (autonomous
depression) and 40 reactive (i.e., responsive to psy-
Retarded Agitated Hostile Type I Type 3 chosocial intervention and lacking a need for treat-
(endogenous)
ment with antidepressant medication). Symptoms for
Strong Moderate which the difference in frequency between the two
Strong Moderatec Moderate Moderate groups was both statistically significant and reason-
Strong Moderate Strong Moderate Moderate
ably great were psychomotor change (either agitation
Strong Moderated
Strong Strong or retardation), self-reproach, difficulty with concen-
tration, and
depressive delusions. Loss of interest and
Moderate Moderate Strong Strong Moderate early morning or midnight awakening were significant-
Moderate Moderate
Moderate Moderate ly different in the two groups, but with a lesser magni-
Strong Strong Strong Moderate tude of difference. Appetite change, weight loss, and
Moderate Moderate loss ofenergy did not differ in the two groups. Suicidal
Strong
thinking was more likely to be present in the reactive
Moderate Moderate
Strong Moderate sample.
In a recent report Endicott and
Spitzer (64) de-
scribed the use of RDC for endogenous subtype in 90
patients who met the criteria for major depressive epi-
sode. Among the symptoms that differed in patients
identified as endogenous and nonendogenous, de-
creased mood and lack of reactivity were most stnk-
ing. Distinct quality of mood and morning worsening
were also different in the two groups. Other items that
SYMPTOM FREQUENCY STUDIES differed in the two samples and have been suggested
elsewhere as characteristic of endogenous depression
In this section studies reporting differences in symp- included decreased capacity for pleasure, decreased
torn frequency in endogenous and reactive depressed involvement, and decreased concentration.
patients are reviewed. Foulds (62), in 1960, described
the use of a sign and symptom inventory in 20 neurotic
depressed patients and 40 psychotic depressed pa- INSTRUMENTAL MEASURES
tients. The psychotic depressed patients were divided
into 20 patients over and 20 patients under 60. The items Most of the studies described above used either pa-
that characterized the psychotic depressed patients tients reports of symptoms or interview observations.
were feelings of worthlessness, guilt, depressive delu- A few studies of depressed patients have reported ob-
sions, agitation, hopelessness, somatic concerns, and jective instrumental measures of symptomatology,
apathy. The younger psychotic depressed patients had such as EEG recording of sleep patterns, telemetric
more prominent symptoms related to guilt or self-re- recording of psychomotor activity, and psychological
proach, while the older group described more sleep testing of cognitive function.
disturbance, including difficulty falling asleep and Early morning awakening has traditionally been
early morning awakening. Foulds questioned whether considered a marker for endogenous depression, while
the sleep changes were more related to age than psy- difficulty falling asleep has been associated with neu-
chotic depression. rotic reactive depression. Altered sleep pattern has
In 1966 Sandifer and associates (63) described a list been included as a symptomatic criterion for major de-
of signs and symptoms, developed on the basis of their pressive illness. However, sleep studies are practi-

review of the literature, for differentiating psychotic cally unanimous that the pattern of sleep awakening
and neurotic depression. The authors used this symp- is not useful for making the endogenous-reactive dis-
torn inventory to rate 5 1 depressed female inpatients tinction (65). This observation seems in accord with
and then reviewed symptom items to determine their the failure of direct observation of sleep disturbance to
ability to predict assignment of patients to the endoge- discriminate between endogenous and reactive depres-
nous or reactive group. Items most predictive of en- sion (66). There is evidence that a shortened REM la-
dogenous depression were lack of responsivity and tency may be useful in distinguishing endogenous from
8 MAJOR DEPRESSIVE ILLNESS Am J Psychiatry 138:1, January 1981

reactive patients (67), but there is no clear behavioral Two other distinctions were also important to our
correlate of this laboratory finding. investigation. Symptoms that identify patients likely to
Telemetric recording of activity has been used as a respond to a particular drug may be different than the
direct measure of psychomotor change in depressed symptoms that respond when the drug is given (75).
patients. Kupfer and associates (68) have reported that Thus we examined studies reporting symptoms that
retardation is the prominent motoric symptom of bipo- were present before drug treatment and which were
lar depressed patients, while unipolar patients appear characteristic of responsive patients.
agitated. Studies such as this tend to support the clini- The other issue relevant to this review is the speci-
Cal observation of psychomotor change in depressed ficity of tricyclic antidepressant drugs for the pharma-
patients. However, telemetric recordings have not yet cologic treatment of depression. Traditionally, tn-
been shown to distinguish endogenous from reactive cyclic antidepressants have been considered the pri-
patients. mary pharmacologic treatment of depression. Hol-
Decreased ability to concentrate has also been con- lister and Overall (76) have questioned this belief,
sidered characteristic of endogenous depression. Psy- pointing out that antipsychotic drugs also may be ben-.
chological testing has been used to confirm objectively eficial for the treatment of certain types of depression.
the presence of this deficit. However, Miller (69) con- Since the purpose of our review was to identify symp-
cluded from a review of studies on psychological test- toms characteristic of patients having an autonomous
ing in depression that there were few differences be- syndrome for which biological treatment was required,
tween various depressive subtypes and that the dif- evidence indicating active drug-placebo differences for
ferences noted were related more to the severity of either a tricyclic antidepressant or an antipsychotic
depression rather than the subtype. drug was reviewed.
The placebo-controlled study that provides the
greatest detail on symptom variables related to treat-
ment response was reported by Raskin and associates
TREATMENT RESPONSE STUDIES (15, 16). Data from a large-scale collaborative study
were used to compare response to imipramine, chior-
It has been suggested that endogenous depression promazine, and placebo in depressed patients. The
will be more responsive than neurotic depression to treatment response data were analyzed by DSM-II di-
treatment with antidepressant drugs or ECT. Such a agnosis and a cluster analysis grouping. Patients given
difference would not only be clinically important but the DSM-1I diagnosis of psychotic depression re-
would validate the entity of endogenous depression. sponded best to imipramine and did not respond well
Several studies have found such a difference in treat- to chiorpromazine or placebo ( 15). Neurotic depressed
ment response (14-21, 70, 71), while other studies patients did slightly better on placebo than on active
have failed to do so (72, 73). A major problem in inves- drug. The psychotic depressed patients were distin-
tigating this question is a methodologic issue raised by guished from the neurotic group in this analysis by
Klein (8). He suggested that while endogenous or high scores on depressed mood, loss of interest, sleep
endogenomorphic patients may have a positive re- disturbance, and psychomotor retardation (16). In this
sponse to active drug and not to placebo, the response sample, loss of interest and retardation were not only
of the neurotic group will depend on the constitu- predictive of the group responsive to imipramine but
tion ofthat group. Ifthe neurotic group is composed of were also the symptoms among all patients that im-
individuals with acute situational depression, the proved most with imipramine and actually were made
group will have a good response to both active drug worse by chiorpromazine. Raskin and Crook (16) also
and placebo. To the extent that the group is composed described the treatment response of four clusters of
of chronic neurotic or chronic dysphonc patients, it depressed patients based on an inverse factor analysis
will tend to show a poor response to both active drug (described above). These data were difficult to inter-
and placebo. The effect of duration of symptoms on pret since the clusters were not well differentiated by
response was demonstrated in one placebo-controlled treatment response. The Type 3 cluster they termed
study (74) and was independent of drug effect. Thus endogenous showed the greatest active drug-placebo
comparison of treatment response between a neurotic difference in response; however, the Type 2 cluster,
and an endogenous group requires the use of a placebo which was least ill and contained mainly neurotic
treatment group to determine if response was drug re- depressed patients, showed moderate active drug re-
lated. It would be expected that patients with endoge- sponse, as did the Type 1 cluster. No cluster was more
nous depression would show a differential response to responsive to one active drug than the other. Thus the
active drug versus placebo, while those with neurotic psychotic-neurotic grouping based on DSM-II diagno-
depression would not show such a difference. With sis was more useful for defining an imipramine respon-
this in mind, we examined placebo-controlled studies sive group. There was a tendency for the Type 1 clus-
for information regarding symptoms predictive of ac- ter, which was a severe agitated group with psychotic
tive drug response. symptoms, to show greater response to chiorproma-
Am J Psychiatry 138:1, January 1981 3. CRAIG NELSON AND DENNIS S. CHARNEY 9

zinc, although the advantage was slight. Again, the which provide data on symptoms predictive of drug
symptoms distinctive of this cluster-agitation and response. Studies that failed to demonstrate treatment
delusional thinking-were also the symptoms in all differences, which did not describe presenting symp-
patients showing the greatest chiorpromazine-placebo toms, or which primarily compared response to dif.
response differences. ferent tricyclic agents were not discussed. The studies
One other placebo-controlled study provided evi- presented provide very strong evidence that psycho-
dence regarding predictive symptoms. Downing and motor retardation identifies patients responsive to tn-
Rickets (74, 77) compared amitriptyline and placebo in cyclic antidepressants (15, 16, 38, 70, 74, 76-80).
treating depressed outpatients. Although much of their There is some evidence that loss of interest and emo-
work focused on social or demographic variables, their tional withdrawal also predict this response (15, 16, 38,
results did indicate that severity of depression, de- 76, 78, 79). The response of agitated or delusional de-
pressed mood, and psychomotor change were predic- pressed patients to antipsychotic drugs is also sug-
tive of response to amitriptyline but not to placebo. gested in prospective studies (15, 16, 38, 76, 78, 79)
Since few placebo-controlled studies provide infor- and appears consistent with the retrospective observa-
mation regarding symptom prediction of drug re- tions that delusional patients respond poorly to tn-
sponse, other studies were also examined. The study cyclic antidepressants (81) but favorably to combined
of Kiloh and associates (14) was one of the first to ex- antipsychotic-antidepressant therapy (82).
amine symptoms predictive of imipramine response. Our findings conflict with those of Bielski and Frie-
They found six predictive items, of which three were del (83), who found anorexia, weight loss, and middle
symptoms; however, the correlations of response with or late insomnia most predictive of tricyclic response.
two of the symptoms-weight loss (r=.1O) and early They cited studies of Paykel and associates (70, 80) to
waking (r= . 18)-were of low magnitude and not sig- support the selection of anorexia as a predictive symp-
nificant in a sample of97 at the .05 level. Distinct qual- torn. While the amitriptyline-responsive psychotic
ity, the third symptom, did appear significantly corre- cluster scored high on anorexia, so did the anxious
lated with response. The fact that duration ofone year cluster that was least responsive. This symptom was
or less had the highest correlation with response raises not useful in distinguishing the two clusters. Inclusion
the question of whether response was affected more by ofweight loss was based on studies of two groups (14,
duration or drug since no placebo was employed. 16). We have indicated that the correlation of weight
Hollister, Overall, and associates (38, 76) described loss and response in the study of Kiloh and associates
a differential response to imipramine and thioridazine (14) was not significant. Weight loss was a symptom
in patients grouped by the profile
analysis pre- distinctive for the Type 3 cluster described by Raskin
viously described. The retarded cluster that was and Crook (16); however, their reports suggest that re-
characterized by retardation, emotional withdrawal, tardation and loss of interest were more predictive of
blunted affect, and conceptual disorganization was imipramine responders. These same studies (14, 16,
more responsive to imipramine than thioridazine. 70, 80) were cited as supporting the inclusion of middle
Thioridazine was more useful in the anxious cluster, a and late insomnia as predictor symptoms. The reports
group that included agitated patients. Subsequent of Raskin and associates (15, 16) do suggest that sleep
studies (78, 79) by this group also demonstrated bene- disturbance characterizes the imipramine-responsive
ficial response of the retarded cluster to amitriptyline group; however, the other studies are more variable.
and the anxious group to perphenazine. While these The association of early awakening and response in
studies did not utilize a placebo comparison group, the one study (14) was low (r= . 18). In the other study (70,
latter studies (78, 79) did administer placebo during the 80) middle insomnia was scored highest in the least re-
first week, which eliminated a substantial number of sponsive anxious cluster. Late insomnia was dis-
placebo responders before active drug was given. tinctive for the drug-responsive psychotic cluster but
Paykel and associates (70, 80) described the re- was not useful in the regression analysis for predicting
sponse of patients grouped by cluster analysis to ami- drug response.
triptyline. The psychotic group was most responsive;
the anxious group was least responsive. Items were
examined to determine those useful in distinguishing DISCUSSION
the clusters as well as those predictive of drug re-
sponse in the entire sample. The one symptom that Our review has focused on the symptoms character-
was useful for both distinguishing the psychotic group istic of autonomous depressive states. Although age,
as well as being significantly correlated with response duration of illness, age of onset, and personality traits
was retardation (70). While no placebo was employed, have each been associated with endogenous depres-
the authors indicated that response differences were sion, we did not investigate these items in our review.
greatest among patient groups during the third and Bhrolchain (84) has discussed the methodologic prob
fourth week when drug effects might be expected. lems involved in grouping such items with symptom
In this review we have focused on those studies variables. Such grouping requires the assumption of
10 MAJOR DEPRESSIVE ILLNESS Am J Psychiatry /38:!, January 198!

TABLE 4
Association of Symptoms with Endogenous Depressiona

Type of Study
Discriminant Symptom Treatment
Symptom Factor Cluster Function Frequency Instrumental Response
Retardation Strong Strong Strong Moderate Strong
Agitation Slight Moderate Strong Slight Moderate
Lack of reactivity Strong Strong Moderate
Severe depressed mood Strong Strong Moderate Moderate
Depressive delusions Moderate Moderate Moderate Moderate Moderate
Self-reproach Moderate Strong Slight Moderate Moderate
Loss of interest Moderate Moderate Moderate Moderate
Distinct quality ofmood Moderate Slight Moderate
Decreased concentration Slight Slight Moderate None
Diurnal morning worsening Slight Moderate Slight
Weight loss Slight Slight Slight Slight
Early morning awakening Slight Moderate Slight Slight None Slight
Mid-night awakening Slight Slight None Slight
Suicidal thoughts attempts Slight None
Difficulty falling asleep None None
aDegree of association of endogen ous syndrome and s ymptom demonstra ted. e.g. . none = no association demonstrated.

covariance among the variables, which may not be val- This review of autonomous depression began with
Id. Determination of the symptom characteristics of the assumption that autonomous depression could be
endogenous depression is a reasonable first step in conceptualized as a single entity. However, the data
identifying the entity whose association with other from the most recent cluster analytic studies suggest
variables might then be studied. that there may be at least two autonomous states-a
We have suggested that the depressive entity often retarded anhedonic cluster and an agitated delusional
referred to in the literature as endogenous or psychotic type. The treatment response data seem to support
is an autonomous state associated with alterations in this distinction. Retarded depression is most respon-
neurophysiology. Carroll and associates (85) have sug- sive to tricyclic antidepressants and may worsen with
gested that change in cortisol secretion occurs particu- antipsychotic treatment. Agitated delusional depres-
larly in endogenous depression. Decreased 3-me- sion appears to respond poorly to tricyclic drugs alone
thoxy-4-hydroxyphenyl glycol (MHPG) excretion may but may benefit from antipsychotic agents used in
also be specific to endogenous depression; low MHPG combination with antidepressants. There is also sug-
values have been reported in bipolar depression gestive evidence that the two types may have neu-
(86), severe primary affective disorder (87), and rochemical differences (88, 89). The probability of two
delusional unipolar depression (88). If an alteration autonomous states is important for descriptive pur-
in neurophysiology is an essential feature of auton- poses: certain symptoms may be useful for the auton-
omous depression , one might wonder if this
literature omous-nonautonomous distinction, while others may
would be useful in identifying symptoms of endoge- be more useful for distinguishing subtypes of auton-
nous depression. Our review of this literature revealed omous depression. The studies reviewed previously in
few reports of specific symptoms associated with a bi- this article appear useful in determining these symp-
ological abnormality. There have been reports of an toms.
association between change in cerebrospinal homo- Table 4 is a summary of depressive symptoms and
vanhllic acid, the major metabolite of dopamine, and the degree to which their association with endogenous
both psychomotor activity (89) and delusional thinking depression has been established in the studies re-
(88). However, these reports require replication. Re- viewed. It includes symptoms characteristic of endog-
ports of an association between major depressive ill- enous depression and symptoms proposed as diag-
ness and both decreased urinary MHPG (86, 87, 90) nostic criteria. Psychomotor change is the symptom
and increased cortisol secretion (85, 91) have unfortu- most strongly associated with autonomous depression.
nately not discussed the specific symptoms that char- Retardation was especially consistent in the factor ana-
acterize patients who show such changes. Never- lytic and drug response studies. In other types of stud-
theless, changes in the excretion of MHPG and the jes, both agitation and retardation showed a strong
secretion of cortisol may prove useful in establishing consistent association with autonomous depression.
the diagnostic validity of autonomous depressive states Several types of studies have indicated a definite asso-
and in identifying patients in whom the specificity of ciation between autonomous depression and severity
diagnostic criteria for endogenous depression might be ofdepressed mood, lack of reactivity, depressive delu-
tested. sions, self-reproach, and loss of interest. Severity of
Am J Psychiatry 138:1, January /981 J. CRAIG NELSON AND DENNIS S. CHARNEY 11

depressed mood and lack of reactivity are associated with weight loss, but this association does not appear
with both agitated and retarded depression. Delusional as strong or as well documented as that for the symp-
thinking and self-reproach are especially prominent in toms described above.
agitated depressed patients, while loss of interest may While suicidal thinking has been suggested as a cr1-
be more prominent in retarded depression. As has tenon for major depressive illness, there is consid-
been noted previously, the most useful symptoms for erable evidence that this occurs frequently in nonau-
distinguishing autonomous depression are those de- tonomous depression. Loss of energy has also been
rived primarily from observation of the patient rather described as a diagnostic criterion, but study of this
than patients self-reports (24). symptom is quite rare in the literature on endogenous
There is indication-although with less supporting depression. Anxiety, self-pity, and irritability have
evidence-that autonomous depression is associated been studied often, but usually in association with
with distinct quality ofmood, diurnal morning worsen- nonautonomous depressive syndromes. Mendels and
ing, and difficulty concentrating. Each of these items Cochrane (27) found moderate relationship between
deserves further study. For example, distinct quality endogenous depression and visceral concerns
in

of mood is not clearly independent of severity of de- four early factor analytic studies. Review of this item
pressed mood or lack of responsiveness. Impairment is complicated by variability in terminology for appar-
of concentration is presumably testable by use of psy- ently related items such as constipation, visceral con-
chological tests, although such testingto date has not cerns, somatic anxiety, somatic concerns, and hypo-
demonstrated an association of cognitive impairment chondriasis. Our review indicates that visceral con-
with specific depressive subtypes. Diurnal morning cerns seldom have been studied since the early factor
worsening seems potentially verifiable through direct analyses. Somatic concerns and hypochondriasis have
observation of mood throughout the day. To our more frequently been described, but findings range
knowledge, such work has not been done. Further from a strongly positive association with endogenous
study of these three items is necessary before their as- depression (16, 57) to a clear relationship with a non-
sociation with autonomous depression can be consid- endogenous group (18, 36-38).
ered as strong as that described for the symptoms in The criteria for primary affective disorder (2) and
the previous group. major depressive episode (4) include sleep and appe-
Changes in sleep, appetite, and weight have long tite change, loss of energy, and suicidal thinking
been considered basic neurovegetative symptoms of among the symptom items. Our review suggests that
depressive illness. Although disturbance ofeating and these four symptoms are not distinctive for endoge-
sleeping is frequent in endogenous depression, these nous depression and may account for the hetero-
symptoms often occur in nonendogenous syndromes. geneity of the patient population defined by the cri-
Our review raises questions about the utility of these teria. The RDC (3) for the endogenous subtype are
symptoms in distinguishing autonomous from non- more similar to the symptom list generated by our re-
autonomous depression. Pilowsky and associates (39) view, although we question whether distinct quality of
have suggested that these symptoms may be part of a mood or diurnal morning worsening merit emphasis as
nonspecific response to stress that occurs in several major items. We would also question the inclusion of
psychiatric syndromes. Several studies of patients re- mid-night and early morning awakening or appetite
ports indicate at least a moderate relationship between and weight loss in the criteria. Delusional thinking,
early morning awakening and autonomous depression. which is not included, would seem more predictive of
However, neither sleep laboratory electroencephalog- an autonomous depressive state.
raphy nor nurses observations of sleep patterns have DSM-III (4) introduces criteria for the diagnosis of
confirmed this relationship. It is possible that patients melancholia. Our review supports the inclusion of lack
reports may have predictive value. However, if patient ofreactivity or inability to experience pleasure, exces-
reports are distorted by diurnal morning worsening as sive guilt, diurnal variation, distinct quality, and psy-
Hauri suggests (65), the latter symptom may be more chomotor change as criteria. We agree with the em-
useful as a predictor. Loss of appetite has been studied phasis on lack of reactivity or inability to experience
surprisingly little, and its relationship to autonomous pleasure as a major necessary symptom. Psychomotor
depression has not been well established. Many pa- change, particularly retardation, appears to be the
tients with autonomous depression experience appe- most predictive symptom when present. The value of
tite loss, but we have found that nonautonomous pa- early morning awakening and appetite or weight loss
tients also frequently report this symptom (24). Weight as items useful for distinguishing endogenous from
loss was examined in 13 of the factor analytic studies. nonendogenous patients is not supported by our re-
In no study did this item have a heavy loading; in half view.
of the studies a moderate association was found, and The development of valid criteria for major depres-
in the other half the association was weak or not signif- sive illness first requires definition ofthe nature of that
icant. Other types of studies have included occasional illness. We have suggested that the nature of major de-
reference to an association of autonomous depression pressive illness, most often discussed in the literature
12 MAJOR DEPRESSIVE ILLNESS Am J Psychiatry /38:1, January 1981

as endogenous, is a depressive state which once devel- 18. Carney MWP, Roth M, Garside RF: The diagnosis of depressive
syndromes and the prediction of ECT response. Br J Psychiatry
oped is autonomous, is associated with alteration in
111:659-674, 1965
neurophysiology, and requires biological intervention. 19. Kendell RE: The Classification of Depressive Illness. London,
Our review of the literature suggests that the symp- Oxford University Press, 1968
toms distinctive for endogenous depression are altered 20. Mendels J: Electroconvulsive therapy and depression. Br J Psy-
chiatry 1 1 1:682-686, 1965
psychomotor activity, lack of reactivity to environ- 21. Pilowsky I, McGrath MD: Effect of ECT on responses to a de-
mental changes, severe depressed mood, depressive pression questionnaire: implications for taxonomy. Br J Psychi-
delusions, self-reproach, and loss of interest in plea- atry 117:685-688, 1970
22. Rosenthal SH, Gudeman JE: The endogenous depressive pat-
surable activity. These symptoms may be useful for
tern. Arch Gen Psychiatry16:241-249, 1967
the generation of valid diagnostic criteria. Whether 23. Gillespie RD: The clinical
differentiation of types of depression.
they can be translated directly into reliable criteria is a Guys Hospital Reports 79:306-344, 1929
question not addressed by this report. While these 24. Nelson JC, Charney DS: Primary affective disorder criteria and
the endogenous-reactive distinction. Arch Gen Psychiatry
symptoms may be useful in determining the presence
37:787-793, 1980
ofmajor depressive illness, the absence ofother major 25. Feinberg M, Carrol BJ, Steiner M, et al: Misdiagnoses of endog-
disorders, particularly schizophrenia, may require ad- enous depression with research diagnostic criteria. Lancet
ditional exclusion criteria similar to those in the RDC 1:267, 1979
26. Spitzer RL, Williams JBW, Skodol AE: DSM-III: the major
(3). The diagnostic validity ofthe depressive syndrome achievements and an overview. Am J Psychiatry 137:151-164,
defined by these symptoms would be confirmed by the 1980
ability of these symptoms to predict course and treat- 27. Mendels J, Cochrane C: The nosology of depression: the endog-
enous-reactive concept. Am J Psychiatry 124: 1-11, 1968
ment response and by the demonstration of a relation-
28. Hamilton M, White JM: Clinical syndrome in depressive states.
ship of the syndrome to specific neurophysiologic J Ment Sd 105:985-998, 1959
changes or specific family patterns of illness. 29. Kiloh LG, Garside RF: The independence of neurotic depres-
sion and endogenous depression. Br J Psychiatry 109:451-463,
1963
REFERENCES 30. Hordern A, Burt CG, Holt NF, et al: Depressive States: A Phar-
macotherapeutic Study. Springfield, Ill, Charles C Thomas,
1. WoodruifRA, Murphy GE, Herjanic M: The natural history of 1965
affective disorders: I. Symptoms of 72 patients at the time of 31. Kay WK, Garside RF, Beamish P, et al: Endogenous and neu-
index hospital admission. J Psychiatr Res 5:255-263, 1967 rotic syndromes of depression: a factor analytic study of 104
2. Feighner JP, Robins E, Guze SB, Ct al: Diagnostic criteria for cases. Clinical features. Br J Psychiatry 1 15:377-388, 1969
use in psychiatric research. Arch Gen Psychiatry 26:57-63, 1972 32. Fahy Ti, Brandon S, Garside RF: Classification of depressive
3. Spitzer RL, Endicott J, Robins E: Research Diagnostic Criteria. illness.Proc R Soc Med 62:331-339, 1969
Arch Gen Psychiatry 35:773-782, 1978 33. Garside RF, Kay DWK, Wilson IC, et al: Depressive syn-
4. American Psychiatric Association: Diagnostic and Statistical dromes and the classification of patients. Psychol Med 1:333-
Manual of Mental Disorders, 3rd ed. Washington, DC, APA, 338, 1971
1980 34. Paykel ES, Prusoff B, Kierman GL: The endogenous-neurotic
5. Garmany 0: Depressive states: their aetiology and treatment. continuum in depression: rater independence and factor distri-
Br Med J 1:5092-5095, 1958 butions. J Psychiatr Res 8:73-90, 1971
6. Leonhard K: Aufteilung der Endogenen Psychosen, ed 2. Ber- 35. Kiloh LG,
Andrews G, Neilson M, et al: The relationship of the
un, Academie Verlag, 1959 syndromes called endogenous and neurotic depression. Br J
7. Rosenthal SH, Kierman GL: Content and consistency in the en- Psychiatry 212:183-1%, 1972
dogenous depressive pattern. Br J Psychiatry 112:471-484, 1966 36. Lewinsohn PM, Zeiss AM, Robert MA, et al: Endogeneity and
8. Klein DF: Endogenomorphic depression. Arch Gen Psychiatry reactivity as orthogonal dimensions in depression. J Nerv Ment
31:447-454, 1974 Dis 164:327-332, 1977
9. Kline NS: Depression: diagnosis and treatment. Med Clin North 37. Grinker RR, Miller J, Sabshin M, et al: Factor patterns, in The
Am 45:1041-1053, 1961 Phenomena of Depressions. New York, Paul 1961
B Hoeber,
10. Roth M: Depressive states and their borderlands: classification, 38. Overall JE, Hollister LE, Johnson M, et al: Nosology of depres-
diagnosis and treatment. Compr Psychiatry 4:135-155, 1960 sion and differential response to drugs. JAMA 195:946-948,
11. Politt JD: Suggestions for a physiological classification of de- 1966
pression. Br J Psychiatry 1 1 1:489-495, 1965 39. Pilowsky I, Levine S. Boulton DM: The classification of depres-
12. Van Praag HM, Uleman AM, Spitz JC: The vital syndrome in- sion by numerical taxonomy. Br J Psychiatry 115:937-945, 1969
terview. Psychiatr Neurol Neurochir 68:329-346, 1965 40. Everitt BS, Gourlay AJ, Kendell RE: An attempt at validation
13. Van Praag HM: The vulnerable brain: biological factors in the of traditional psychiatric syndromes by cluster analysis. Br J
diagnosis and treatment ofdepression, in Psychiatric Diagnosis. Psychiatry 119:399-412, 1971
Edited by Rakoff VM, Stancer HC, Kedward HB. New York, 41 . Paykel ES: Classification of depressed patients: a cluster analy-
Brunner/Mazel, 1977 sis derived grouping. Br J Psychiatry 1 18:275-288, 1971
14. Kiloh LG, Ball JRB, Garside RF: Prognostic factors in treat- 42. Overall JE: The Brief Psychiatric Rating Scale in psycho-
ment of depressive states with imipramine. Br Med J 1:1225- pharmacology research. Mod Probi Pharmacopsychiatry 7:67-
1227, 1962 78, 1974
15. Raskin A, Schulterbrandt JG, Reatig N, et al: Differential re- 43. Eysenck HJ: The classification of depressive illness. Br J Psy-
sponse to chiorpromazine, imipramine, and placebo. Arch Gen chiatry 117:241-250, 1970
Psychiatry 23:164-173, 1970 44. Macfayden HW: The classification, of depressive disorders. J
16. Raskin A, Crook TA: The endogenous-neurotic distinction as a Clin Psychol 31:380-394, 1975
predictor of response to antidepressant drugs. Psychol Med 45. Kendell RE: The classification of depressions: a review of con-
6:59-70, 1976 temporary confusion. Br J Psychiatry 129: 15-28, 1976
17. Simpson GM, Lee JH, Cuculic Z, et al: Two dosages of imipra- 46. Fowles DC, Gersh F: Neurotic depression: I. The endogenous-
mine in hospitalized endogenous and neurotic depressives. neurotic distinction, in The Psychobiology, of the Depressive
Arch Gen Psychiatry 33:1093-1102, 1976 Disorders: Implications for the Effects of Stress. Edited by Dc-
Am J Psychiatry 138:!, January 1981 J. CRAIG NELSON AND DENNIS S. CHARNEY 13

pue RA. New York, Academic Press (in press) 70. Paykel ES, PrusoffBA, Kierman GL, et al: Clinical response to
47. Katz MM, Hirschfeld RMA: Phenomenology and classification amitnptyline among depressed women. J Nerv Ment Dis
of depression, in Psychopharmacology: A Generation of Prog- 156:149-165, 1973
ress. Edited by Lipton MA, DiMascio A, Killam KF. New 71. Deykin EY, DiMascio A: A relationship of patient background
York, Raven Press, 1978 characteristics to efficacy of pharmacotherapy in depression. J
48. PrusoffB, Kierman GL, Paykel ES: Concordance between din- Nerv Ment Dis 155:209-215, 1972
ical assessments and patients self-report in depression. Arch 72. Wittenborn JR, Kiremitci N: A comparison of antidepressant
Gen Psychiatry 26:546-552, 1972 medications in neurotic and psychotic patients. Arch Gen Psy-
49. Friedman AS, Cowitz B, Cohen HW, et al: Syndromes and chiatry 32:1172-1176, 1975
themes of psychotic depression. Arch Gen Psychiatry 9:504- 73. Abraham HC, Kanter VB, Rosen I, et al: A controlled clinical
509, 1963 trial of imipramine (Tofranil) with out-patients. Br J Psychiatry
50. Roth M, Gurney C, Garside RF, et al: Studies in the classifica- 109:286-293, 1963
tion of affective disorders. Br J Psychiatry 121:147-161, 1972 74. Downing RW, Rickels K: Predictors ofamitriptyline response in
5 1. Overall JE: Dimensions of manifest depression. Psychiatr Res out-patient depressives. J Nerv Ment Dis 154:248-263, 1972
1:239-245, 1962 75. Overall JE: Distinguishing between patterns of drug indications
52. McConaghy N, Joffe AD, Murphy B, et al: The independence of and patterns ofdrug effects. Psychopharmacol Bull 11:17, 1975
neurotic and endogenous depression. Br J Psychiatry 113:479- 76. Hollister LE, Overall JE: Reflections on the specificity of action
484, 1967 of antidepressants. Psychosomatics 6:361-365, 1965
53. Weckowicz TE, Cropley AJ, Muir W: An attempt to replicate 77. Downing RW, Rickels K: Predictors of response to amitripty-
the results ofa factor analytic study in depressed patients. J Clin line and placebo in three outpatient treatment settings. J Nerv
Psychol 27:30-31, 1971 Ment Dis 156:109-129, 1973
54. Weckowicz TE, Yonge KA, Cropley AJ, et al: Objective thera- 78. Hollister LE, Overall JE, Johnson MH, et al: Amitriptyline
py predictions in depression: a multivariate approach. J Clin alone and combined with perphenazine in newly admitted de-
Psychol 27:3-29, 1971 pressed patients. J Nerv Ment Dis 142:460-469, 1966
55. Overall JE, Gorham DR: The Brief Psychiatric Rating Scale. 79. Hollister LE, Overall JE, Shelton J, et al: Drug therapy of de-
Psychol Rep 10:799-812, 1962 pression. Arch Gen Psychiatry 17:486-493, 1967
56. Kendell RE, Gourlay AJ: The clinical distinction between psy- 80. Paykel ES: Depressive typologies and response to amitriptyline.
chotic and neurotic depression. Br J Psychiatry 117:257-266, BrJ Psychiatry 120:147-156, 1972
1970 81. Glassman AH, Kantor iS, Shostak M: Depression, delusions
57. Fleiss JL: Classification of the depressive disorders by numeri- and drug response. Am J Psychiatry 132:716-719, 1975
cal typology. J Psychiatr Res 9:141-153, 1972 82. Nelson JC, Bowers MB: Delusional unipolar depression. Arch
58. Bhrolchain MN, Brown GW, Hams TO: Psychotic and neurotic Gen Psychiatry 35:1321-1328, 1978
depression: II. Clinical characteristics. Br J Psychiatry 134:94- 83. Bielski Ri, Friedel RO: Prediction oftricyclic antidepressant re-
107, 1979 sponse. Arch Gen Psychiatry 33:1479-1489, 1976
59. Gurney C, Roth M, Garside RF, et al: Studies in the classifica- 84. Bhrolchain MN: Psychotic and neurotic depression: I. Some
tion of affective disorders. Br J Psychiatry 121:162-166, 1972 points of method. Br J Psychiatry 134:87-93, 1979
60. Prusoff B: Differentiating depressed from anxious neurotic out- 85. Carroll Bi, Curtis GC, Mendels J: Neuroendocrine regulation in
patients. Arch Gen Psychiatry 30:302-309, 1974 depression. Arch Gen Psychiatry 33:1051-1058, 1976
61. Wing JK, Cooper JE, Sartorius N: The Measurement and 86. Schildkraut ii, Orsulak PJ, Schatzberg AF, et al: Toward a bio-
Classification of Schizophrenic Symptoms. London, Cambridge chemical classification of depressive disorders. Arch Gen Psy-
University Press, 1974 chiatry 35:1427-1433, 1978
62. Foulds MA: Psychotic depression and age. J Ment Sci 106:1394- 87. De Leon-Jones F, Maas JW, Dekirmenjian H: Diagnostic sub-
1397, 1960 groups of affective disorders and their urinary excretion of cate-
63. Sandifer MYG, Wilson IC, Green L: The two-type thesis of de- cholamine metabolites. Am J Psychiatry 132:1141-1148, 1975
pressive disorders. Am J Psychiatry 123:93-97, 1966
88. Sweeney D, Nelson C, Bowers M, Ct al: Delusional versus non-
64. Endicott J, Spitzer RL: Use of the Research Diagnostic Criteria
delusional depression: neurochemical differences. Lancet
and the Schedule for Affective Disorders and Schizophrenia to
2:100-101, 1978
study affective disorders. Am J Psychiatry 136:52-56, 1979
65. Hauri P: Sleep in depression. Psychiatric Annals 4:45-62, 1974 89. Van Praag HM, Korfi, Lakke JPWF: Dopamine metabolism in
66. Costello CG, Selby MM: The relationship between sleep pat- depressions, psychoses, and Parkinsons disease: the problem
terns and reactive and endogenous depression. Br J Psychiatry of the specificity of biological variables in behavior disorders.
111:497-501, 1965
Psychol Med 5:138-146, 1975
67. Kupfer D: REM latency: a psychobiologic marker for primary 90. Beckmann H, Goodwin FK: Antidepressant response to tn-
depressive disease. Biol Psychiatry 11:159-175, 1976 cyclics and urinary MHPG in unipolar patients. Arch Gen Psy-
68. Kupfer Di, Weiss BL, Foster FG, Ct al: Psychomotor activity in chiatry 32:17-21, 1975
affective states. Arch Gen Psychiatry 30:765-768, 1974 91. Sachar EJ, Heliman L, Roffwarg HP, et al: Disrupted 24-hour
69. Miller WR: Psychological deficit in depression. Psychol Bull patterns of cortisol secretion is psychotic depression. Arch Gen
82:238.260, 1975 Psychiatry 28:19-24, 1973