Content uploaded by Alex Gamma
Author content
All content in this area was uploaded by Alex Gamma on Dec 03, 2017
Content may be subject to copyright.
Research report
Diagnostic conversion from depression to bipolar disorders:
results of a long-term prospective study of hospital admissions
Jules Angst*, Robert Sellaro, Hans H. Stassen, Alex Gamma
Epidemiological Research, Zurich University Psychiatric Hospital, Lenggstrasse 31, P.O. Box 68, 8029 Zurich, Switzerland
Received 26 March 2003; accepted 3 April 2003
Abstract
Objectives: To analyse the time course and some risk factors for a diagnostic change from major depression to bipolar
disorders (BP) over an average of 20 years from the onset of the disorders. Methods: Patients (406) with major mood disorders
hospitalised at some time between 1959 and 1963 were followed-up until 1985. The analysis also included the course prior to
hospitalisation. Survival analyses and Cox regression models were applied. Results: A diagnostic change from depression to
bipolar I occurred in about 1% of the patients per year and to bipolar II disorders in about 0.5% per year. Risk factors for a
change from depression to BP-I disorder were male sex and an early onset of the disorder; risk factors for a change from
depression to BP-II disorder were female sex, a later onset of the disorder and a positive family history of mania. Conclusions:
Across the entire lifetime, every new episode of depression brings a new risk for mania; more than half of our severe mood
disorder cases became bipolars. The risk of depression developing into bipolar disorder remains constant lifelong. Limitations:
The diagnostic classification of ICD-9 met RDC criteria for bipolar disorder in only 90% of cases. Part of the data collected in
retrospect may be less reliable; the prospective data were only collected every 5 years from 1965 to 1985 using multiple
sources; mild manifestations between the follow-ups may have been partially missed. The sample of subsequent hospital
admissions for major depression and mania represents a severe group of patients and generalisations to ambulatory cases may
not be possible. Not all risk factors for diagnostic conversion described in the literature could be assessed in this study.
D2003 Elsevier B.V. All rights reserved.
Keywords: Diagnostic change; Risk factors; Depression; Bipolar disorders
1. Introduction
The homogeneity and unity of manic-depressive
disorders (Kraepelin, 1898) were disproved in the late
1960s by three monographs (Angst, 1966; Perris,
1966; Winokur et al., 1969) and the distinction
between bipolar disorder (BP) (Falret, 1851) and
depression was re-introduced. Since then, a persistent
problem has been the uncertain diagnosis of depres-
sive disorder, since it is clear that the group of recurrent
depressives includes hidden bipolar subjects whose
depression may develop into hypomania or mania at
any time. One consequence of this is the long time
elapsing between the onset of depression and a
diagnosis of BP. This period was more than 7.5 years
in the retrospective study of Ghaemi et al. (1999) and
8 years in a national survey of the NDMDA (1993) in
0165-0327/$ - see front matter D2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0165-0327(03)00195-2
* Corresponding author. Tel.: +41-1-384-2611; fax: +41-1-384-
2446.
E-mail address: jangst@bli.unizh.ch (J. Angst).
www.elsevier.com/locate/jad
Journal of Affective Disorders 84 (2005) 149 – 157
the United States. Underdiagnosis of minor bipolar
syndromes is the rule.
Although there have been many reports on the rates
of diagnostic changes from depression to bipolar
disorder (reviewed by Angst, 1988), few of them
have taken the length of follow-up into account. On
the basis of four large studies (Kinkelin, 1954; Mar-
neros et al., 1991; Coryell et al., 1995; Angst and
Preisig, 1995a), we estimated a mean diagnostic
change rate of 1% per year of observation (Angst,
2000).Coryell et al. (1995), for instance, found 5.0%
diagnostic changes from depression to hypomania and
5.2% to mania over 10 years of follow-up; 7.5% of
bipolar II patients developed mania.
The rates recently reported by Goldberg et al. (2001)
in a 15-year follow-up of 74 initially hospitalised
patients were higher, with the diagnosis changing from
depression to hypomania in 26% and from depression
to mania in 19% of patients. Their survival analyses
suggested a constant (linear) risk of diagnostic change.
This paper will not deal with the risk of depressive
episodes switching into hypomania or mania, and will
therefore not discuss the hypothesis that antidepres-
sants are risk factors for such a switch (Goodwin and
Jamison, 1990). Nor will it analyse the time-point of
such switches within a depressive episode.
The goal of this paper is to specifically describe the
risks of long-term diagnostic conversion from depres-
sion to bipolar disorder (bipolar I and bipolar II) as a
function of time, and in relation to age of onset,
number of depressive episodes, sex and family history
for bipolar disorder. In contrast to earlier analyses of
the course (Angst and Preisig, 1995a), this paper
applies survival techniques for the description of the
time course of diagnostic changes over lifetime,
including 26 years of prospective follow-up.
2. Methods
2.1. Sample and assessments
The sample consists of 406 patients (186 unipolar
and 220 bipolar depressive or manic) who were
admitted to the University Psychiatric Hospital of
Zurich between 1959 and 1963 with a diagnosis of
mania or endogenous depression, endo-reactive de-
pression, manic-depressive disorder or affective disor-
der with mood-congruent or mood-incongruent
psychotic features (hallucinations or delusions) includ-
ing schizo-affective disorder. Sixty-one percent of the
patients met the criteria for psychosis at least once over
their lifetime, underlining that we are dealing with a
hospitalised group of seriously ill patients.
Psychopathology was assessed between 1959 and
1963 in 331 patients who were depressed at admis-
sion and discharge. The unstructured clinical inter-
views were systematically guided by a comprehensive
rating scale for psychopathology (Angst et al., 1964),
which included 44 psychological items and the syn-
dromes agitation, retardation, hypochondriasis, sui-
cidal thoughts and suicide attempts. In addition, data
from significant others were collected in all cases. A
family history of psychiatric disorders and suicidality
was taken from the probands and from at least one
first degree relative. Prospectively psychopathology
was documented by a list of 10 syndromes (Angst et
al., 1968).
Bipolarity was assumed as soon as hypomania
occurred for a few days, regardless of whether it
seemed to be drug-induced or not. The diagnosis of
bipolar I vs. bipolar II disorder was made by ICD-9
criteria approximating the original criteria of Dunner
et al. (1976), applying hospitalisation as a classifier:
bipolar II disorder required hospitalisation for depres-
sion, bipolar I disorder required hospitalisation for
mania. In a methodological study of a random sub-
sample of 152 cases, the diagnostic criteria of Feigh-
ner et al. (1972) were applied (Grigo, 1981). There
was diagnostic agreement in 89% of unipolar depres-
sion and 90% of bipolar disorders (specificity 0.8;
sensitivity 0.93). With the exception of two bipolar II
cases, whose hypomania lasted only 1 week, all
bipolars met the 2 weeks minimum criterion. A
detailed description of the sample, all of whom
qualify for DSM-III-R criteria for major depression
and mania or hypomania, has been published (Angst
and Preisig, 1995a).
Follow-ups were carried out in 1963, 1965, 1970,
1975, 1980 and 1985. For each episode, the onset and
end were dated, and the treatment status (none,
ambulatory, hospitalisation) and the status between
episodes (no symptoms, residual symptoms, prophy-
lactic treatment) were recorded. Patients’ previous
history was reconstructed as fully as possible on the
basis of information from numerous sources, includ-
J. Angst et al. / Journal of Affective Disorders 84 (2005) 149–157150
ing other institutions, family doctors and significant
others (Angst and Preisig, 1995b).
The statistical analysis will mainly focus on the
entire course of the disorders (including the retrospec-
tive period). In addition, a subsample followed purely
prospectively will be analysed comparatively for
methodological reasons.
2.2. Statistics
Analyses were carried out in STATISTICA 5.5 and
STATA 7.0 for Windows. Non-parametric survival
analyses according to Kaplan and Meier (1958) with
log-rank tests were used to compare survivorship
functions across groups. Multivariate Cox models
(MULCOX, Wei et al., 1989) were used to assess
the effects of explanatory variables (sex, age of onset,
number of episodes, family history of mania) on the
rates of diagnostic change (hazard ratios) over the
study period. Continuous variables were dichotomised
at the median or polytomised if they were not linear in
the log hazard function. Linearity was assessed by
fitting first- and second-order fractional polynomial
models to the variable. If there was a significant
statistical gain from a higher-order compared to the
linear model, the variable was assumed not to be
linear. Variables that violated the proportional hazard
assumption (as assessed using STATA’s ‘‘stphtest’’
command) were excluded from the models.
3. Results
3.1. Status at first episode
As stated earlier, the sample consists of consecutive
hospital admissions for major depression or mania
over 5 years. It includes both first admissions and re-
admissions; hence, the data collected on a subject’s
first episode are in many cases retrospective: 118 of
406 (29.1%) patients were selected for study in their
first episode, 288 cases in later episodes. During their
first episode, 45 (11.1%) patients were not treated, 38
(9.4%) treated as outpatients, 270 (66.5%) treated as
inpatients; in 53 (13.1%) cases, data on treatment of
the first episode were missing. 309 (76.1%) of first
episodes manifested as depression, 27 (6.7%) as bipo-
lar disorders, 40 (9.9%) as manic and the remaining 30
(8.6%) as other psychotic syndromes (paranoid/cata-
tonic/hallucinatory/hebephrenic).
3.2. Diagnostic change from depression to bipolar
disorders: retrospective/prospective vs. purely
prospective subsample
Intuitively, one would expect the purely prospec-
tively observed subsample (P) to show more diagnos-
tic changes to bipolar disorder than the subsample RP,
observed retro- and prospectively. Surprisingly, this
was not the case: Group P (n= 97) manifested as
depression at first admission and was studied pro-
spectively; 15 cases (15.5%) converted from depres-
sion to bipolar disorders (7.2% BP-I, 8.25% BP-II)
over an average observation period of 13.8 years. In
the larger subsample RP (n= 212), who was studied
both retrospectively and prospectively, the conversion
rate was significantly larger. Over an average obser-
vation period of 23.4 years, 106 patients (50%)
switched from depression to bipolar disorder (32.1%
to BP-I and in 17.9% to BP-II disorders).
The higher conversion rate observed in subsample
RP cannot be explained solely by the longer observa-
tion, since there was also a significant difference
(p< 0.02 in the survival curves of diagnostic changes
(Fig. 1)) censoring for drop-outs.
The age of onset of the prospective subsample was
considerably higher (mean 47.9 years, s.d. 16.8) than
that of subsample RP (mean 33.5 years, s.d. 13.4). In
additional survival analyses (not shown in detail),
group P showed significantly (log rank test, p<0.05)
fewer diagnostic changes to bipolar I disorder than
group RP with its much earlier age of onset.
The survival curves of the changes from depression
to bipolar II disorders did not differ significantly
between the two groups P and RP.
3.3. Diagnostic change from depression to bipolar
disorders
The analysis will first deal with the sample as a
whole and study the change from depression to
bipolar disorder. The analysis describes the diagnostic
change starting with the first episode of the illness as a
function of years of follow-up.
A diagnostic change from depression to bipolar
disorder occurred in 39.2% of cases (121/309), in 75
J. Angst et al. / Journal of Affective Disorders 84 (2005) 149–157 151
of them (24.3%) changing to BP-I disorder and in 46
(14.9%) changing to BP-II disorder. 41.2% of all BP-
II cases (40/97) later developed into BP-I disorder.
Taking into account only prospectively observed epi-
sodes, there was 27.2% diagnostic changes to bipolar
disorder (15.7% to BP-I and 11.6% to BP-II); the
respective mean survival times were 13.0 (BP), 15.1
(BP-I) and 15.6 years (BP-II).
The survival analysis for the diagnosis of depres-
sion resulted in a mean survival time of 30.8
(s.d. = 1.09) years. The gradient of the survival curve
over the first 4 years is slightly steeper (change in
about 10% of cases), but from the 5th to the 50th year
of follow-up, it approximates a linear function (Fig.
2). This means that after the first 5 years, the risk of a
diagnostic change from depression to bipolar disor-
ders was constant over lifetime with a change rate of
about 1.25% per year.
The Cox model including sex, age of onset,
family history of bipolar disorder and number of
episodes (dichotomised at six episodes) showed that
patients with more than six episodes in their lives
changed diagnoses at 3.6 times the rate of those with
fewer episodes ( p< 0.0001). Women tended to have
a slower rate of change than men (hazard ratio = 0.7,
p= 0.1). The age of onset and family history were
not significantly related to the rate of diagnostic
change.
All Cox models are summarised in Table 1.
UP to BP (N=309)
p(log-rank)<0.02
Years since onset of first episode
Cumulative proportion surviving
0 1020304050607080
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Retrospective + prospective (N=212)
Prospective only (N=97)
Fig. 1. Diagnostic change from depression to bipolar disorders: comparison of prospectively studied subsample and also including retrospective
observations.
DE (UP) to BP (n=309)
conversion
Years since onset of first episode
Cumulative proportion surviving
01020304050
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Fig. 2. Diagnostic change from depression to bipolar disorders.
J. Angst et al. / Journal of Affective Disorders 84 (2005) 149–157152
3.4. Diagnostic change from depression to bipolar I
and bipolar II disorder
3.4.1. Change from depression to BP-I disorder
The survival curve (Fig. 3) demonstrates a linear
function similar to that in Fig. 2. The change rate was
about 1% per year. According to the Cox model,
patients with more than six episodes changed diagno-
sis at 3.2 times the rate of those with fewer episodes
(p< 0.0001). The rate of change in women was again
slower than that in men (hazard ratio = 0.5, p< 0.003).
An increase of 10 years in the age of onset was
associated with an 80% reduction in the rate of change
(p< 0.03).
3.4.2. Change from depression to BP-II disorder
Fig. 4 shows that the rate of change is about 0.5%
per year of follow-up; the rate of change during the
first year of follow-up was not elevated. In contrast to
the change from depression to BP-I disorder, the Cox
model showed that the rate of change from depression
to BP-II disorder was higher in women than in men
(hazard rate = 2.4, p< 0.06). The diagnosis of patients
with more than six episodes changed 2.5 faster than
that of patients with fewer episodes ( p< 0.01). Age of
onset was not linear in the log hazard and was
polytomised into 20-year bins. A 20-year increase in
age of onset resulted in a minimal, but significant,
increase in the rate of diagnostic change (hazard
ratio = 1.02, p< 0.03). A positive family history for
Table 1
Cox proportional hazard models for the diagnostic change between
subtypes of mood disorder
Hazard ratio 95% C.I. P
DE !BP
Sex 0.7 0.48 – 1.07 0.1
No. of episodes
a
3.6 2.30 – 5.57 0.0001
DE !BP-I
Sex 0.5 0.30 – 0.78 0.003
No. of episodes
a
3.2 1.75 – 6.01 0.0001
Age of onset
b
0.8 0.66 – 0.97 0.03
DE !BP-II
Sex 2.4 0.96 – 6.20 0.06
No. of episodes
a
2.5 1.25 – 5.03 0.009
Age of onset
c
1.02 1.00 – 1.05 0.03
FH + mania 3.0 0.92 – 9.74 0.07
BP-II !BP-I
Sex 0.5 0.25 – 0.91 0.02
DE = depression, BP = bipolar disorder.
a
Dichotomised at six episodes.
b
Calculated for a change of 10 years.
c
Polytomised into 20-year bins.
DE (UP) to BP-I (n=309)
conversion
Years since onset of first episode
Cumulative proportion surviving
01020304050
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Fig. 3. Diagnostic change from depression to bipolar I disorder.
DE (UP) to BP-II (n=309)
conversion
Years since onset of first episode
Cumulative proportion surviving
01020304050
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Fig. 4. Diagnostic change from depression to bipolar II disorder.
J. Angst et al. / Journal of Affective Disorders 84 (2005) 149–157 153
mania tended to increase the rate of diagnostic change
3.0-fold ( p< 0.07).
3.5. Diagnostic change from BP-II to BP-I disorders
Bipolar patients (97) manifested first with BP-II
disorders. Forty of them (41.2%) later developed BP-
I disorder, with a mean survival time of 29.1 (s.d. =
2.59) years. Restricted to the prospective observation
period, 33.3% changes occurred over a mean survival
time of 13.2 years. Over 25 years of follow-up, the
diagnostic change rate was almost linear and the risk
was therefore constant, with a rate of change of 2% per
year. Only gender was a significant variable in the Cox
model. Women changed from BP-II to BP-I disorder
more slowly than men (hazard ratio = 0.5, p< 0.03).
4. Discussion
A change in diagnosis from depression to bipolar
disorders has practical clinical implications for the
prognosis and choice of treatment, and it helps define
homogenous subgroups of mood disorders for biolog-
ical research purposes.
In this naturalistic study, patients were recruited at
many different points of their illness; it was therefore
inevitable that for most patients, some data were
retrospective and per se less reliable than data on
the prospectively assessed episodes. To our surprise,
the diagnostic changes were significantly less frequent
in the subsample studied purely prospectively than in
the partly retrospectively studied group. We think that
the much higher average age of onset of (49 years) of
the prospective group explains its lower diagnostic
switch rate; bipolar disorder manifests in most cases
between the second and fourth decades of life. There-
fore, had we confined the analysis of the course to
purely prospective periods, it would have consider-
ably restricted the size of the sample and made it less
representative.
For many reasons, depressive disorders are over-
diagnosed (Ghaemi et al., 1999), and the diagnosis is
always loaded with uncertainty. Clearly every new
episode is accompanied by a certain risk of mani-
festing as mania or hypomania. Two questions arise:
how high is the recurrence risk and does it change
over a lifetime?
A careful analysis of the 10-year prospective data
of the NIMH Collaborative Study on the Biology of
Depression suggested a constant risk of recurrence of
bipolar disorder and depression (Lavori et al., 1996);
this finding was recently confirmed on the basis of a
40-year follow-up (both retrospective and prospec-
tive) of our patient sample (Angst et al., 2003). Frailty
models would even indicate that the risk of recurrence
increases slightly (Kessing and Andersen, 1999; Kess-
ing et al., 1999, in press).
Our survival analyses demonstrated that the risk of
depression converting into any bipolar disorder
remained constant over decades, which is compatible
with the findings of Goldberg et al. (2001) of a
constant recurrence risk. Our finding of a higher initial
risk of diagnostic change is unexpected and difficult
to interpret. It could mean that (1) the risk is really
higher over the first 2 – 3 years of depressive disor-
ders, or rather, that (2) the finding is merely an artifact
of unreliable first onset data of hypomania/mania
resulting in higher age of onset. We would favour
the second explanation.
The survival analyses showed that rates of diag-
nostic change varied according to the bipolar sub-
group: from depression to bipolar I, the change was
about 1% per year and to bipolar II disorders about
0.5% per year. The risk of a diagnostic change from
bipolar II to bipolar I disorder was approximately 2%
per year of observation (Fig. 5). These results are
compatible with the estimates in the literature men-
tioned in the Introduction.
The analyses of diagnostic changes from depres-
sion to bipolar disorders clearly showed that the
proportion of depressive disorders converting into
bipolar disorders is highly dependent on the length
of the follow-up. In our study, 39% of the depressives
developed bipolar disorder; we found the change in
diagnosis from depression to bipolar disorder in 50%
of cases over an average of 37 years.
Including cases which manifested mania already in
their first episode, more than half of our total sample
ultimately became bipolars.
The research into the risk factors for conversion
from unipolar to bipolar disorders was recently
reviewed by Goldberg et al. (2001). Risk factors
included young age (Coryell et al., 1995), an earlier
age of onset (Akiskal et al., 1983), sudden onset
(Strober and Carlson, 1982), high number of previous
J. Angst et al. / Journal of Affective Disorders 84 (2005) 149–157154
episodes (Angst, 1965; Angst et al., 1978), psychotic
features (Akiskal et al., 1983; Strober and Carlson,
1982; Strober et al., 1993; Winokur and Morrison,
1973), previous hypomanic swings (Angst, 1965) or
psychomotor retardation (Akiskal et al., 1983; Strober
and Carlson, 1982). A positive family history for
bipolar disorder was found in some (Akiskal et al.,
1983; Coryell et al., 1995; Winokur et al., 1993) but
not in all studies (Angst, 1985). According to Akiskal
et al. (1995), temperamental instability and mood
lability in major depressives were strong predictors
of bipolar II disorders. Not all of these risk factors
could be investigated in our study. Although in both
BP-I and BP-II, the diagnostic change was correlated
with a higher number of episodes, in our study, risk
factors for bipolar I and II disorders differed to some
extent. Male sex and an early onset of the disorder
were correlated with a diagnostic change to BP-I
disorder; female sex and a later onset of the disorder
correlated with a change to BP-II disorder, as did a
positive family history of mania. As a general rule,
men seem to have a higher propensity to mania and
women to hypomania.
4.1. Limitations
The obvious limitations of the study arise from
sample selection. Firstly, the findings refer to severe,
hospitalised cases of mood disorders and can there-
fore not be generalised to outpatient or community
samples. Secondly, the study unavoidably included
retrospective course data. These shortcomings could
be remedied by a lifelong prospective epidemiologi-
cal study following a cohort from childhood to death.
The assessment intervals of 5 years may have also led
to a loss of information on hypomania occurring
between them. On the other hand, the strength of
the study is its duration and the fairly precise pro-
spective dating of the diagnostic changes allowing
survival analyses.
4.2. Conclusions
The main implication of this study for psychiatric
practice and research is the high ratio of bipolar
disorder to depression (approximately 1:1), a finding
which is compatible with earlier estimates of diagnos-
tic error (Angst et al., 1978), with some findings in
patient samples (Akiskal and Mallya, 1987; Allilaire
et al., 2001; Benazzi, 2001; Benazzi and Akiskal,
2003) and with our prospective cohort study of a
community sample (Angst et al., 2003).
In practice, this implies checking every depressive
patient carefully for any personal history of hypoman-
ic symptoms (taking into account reports by signifi-
cant others) and for any family history of mania. The
aim would be to reduce the overdiagnosis of depres-
sion and to identify a suspected group of hidden
bipolars as a predictor of so-called drug-induced
hypomania or future diagnostic changes to bipolar
disorders.
References
Akiskal, H.S., Mallya, G., 1987. Criteria for the ‘‘soft’’ bipolar
spectrum: treatment implications. Psychopharmacol. Bull. 23,
68 – 73.
Akiskal, H.S., Walker, P.W., Puzantian, V.R., King, D., Rosenthal,
T.L., Dranon, M., 1983. Bipolar outcome in the course of de-
pressive illness. Phenomenologic, familial, and pharmacologic
predictors. J. Affect. Disord. 5, 115– 128.
Akiskal, H.S., Maser, J.D., Zeller, P.J., Endicott, J., Coryell, W.,
Keller, M., Warshaw, M., Clayton, P., Goodwin, F., 1995.
Switching from ‘‘unipolar’’ to bipolar II. An 11-year prospective
study of clinical and temperamental predictors in 559 patients.
Arch. Gen. Psychiatry 52, 114 – 123.
Allilaire, J.-F., Hantouche, E.-G., Sechter, D., Bourgeois, M.-L.,
BP-II to BP-I (n=97)
conversion
Years since first BP-II episode
Cum
02010 30
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Fig. 5. Diagnostic change from bipolar II to bipolar I disorder.
J. Angst et al. / Journal of Affective Disorders 84 (2005) 149–157 155
Azorin, J.M., Lancrenon, S., Cha
ˆtenet-Duchene, L., Akiskal,
H.-S., 2001. Fre
´quence et aspects cliniques du trouble bipolaire
II dans une e
´tude multicentrique francaise: EPIDEP. Ence
´phale
XXVII, 149 – 158.
Angst, J., 1965. Zur Prognose antidepressiver Behandlungen. Ver-
laufsuntersuchungen und genetische Studie (The prognosis of
antidepressive treatments. Longitudinal and genetic studies).
Anglo-Ger. Med. Rev. 2, 733 – 751.
Angst, J., 1966. Zur Aetiologie und Nosologie endogener depres-
siver Psychosen. Eine genetische, soziologische und klinische
Studie. Springer, Berlin.
Angst, J., 1985. Switch from depression to mania—a record study
over decades between 1920 and 1982. Psychopathology 18,
140 – 154.
Angst, J., 1988. Clinical course of affective disorders. In: Helgason,
T., Daly, R.J. (Eds.), Depressive Illness: Prediction of Course
and Outcome. Springer Verlag, Berlin, pp. 1 – 44.
Angst, J., 2000. Course and prognosis of mood disorders. In: Gelder,
M.G., Lopez-Ibor Jr., J.J., Andreasen, N.C. (Eds.), New Oxford
Textbook of Psychiatry, vol. 1. Oxford Univ. Press, Oxford,
pp. 719 – 724.
Angst, J., Preisig, M., 1995a. Course of a clinical cohort of uni-
polar, bipolar and schizoaffective patients. Results of a prospec-
tive study from 1959 to 1985. Schweiz. Arch. Neurol. Psychiatr.
146, 5 – 16.
Angst, J., Preisig, M., 1995b. Outcome of a clinical cohort of uni-
polar, bipolar and schizoaffective patients. Results of a prospec-
tive study from 1959 to 1985. Schweiz. Arch. Neurol. Psychiatr.
146, 17 – 23.
Angst, J., Battegay, R., Poeldinger, W., 1964. Zur Methodik der
statistischen Bearbeitung des Therapieverlaufs depressiver
Krankheitsbilder. Methods Inf. Med. 3, 54 – 56.
Angst, J., Grof, P., Hippius, H., Poeldinger, W., Weis, P., 1968. La
psychose maniaco-depressive est-elle periodique ou intermit-
tente? In: de Ajuriaguerra, J. (Ed.), Cycles biologiques et psy-
chiatrie. Masson, Paris, pp. 339 – 351.
Angst, J., Felder, W., Frey, R., Stassen, H.H., 1978. The course of
affective disorders. I. Change of diagnosis of monopolar, unipo-
lar, and bipolar illness. Arch. Psychiatr. Nervenkr. 226, 57 – 64.
Angst, J., Gamma, A., Benazzi, F., Ajdacic, V., Eich, D., Ro¨ssler,
W., 2003. Toward a re-definition of subthreshold bipolarity:
epidemiology and proposed criteria for bipolar-II, minor
bipolar disorders and hypomania. J. Affect. Disord. 73,
133 – 146.
Angst, J., Gamma, A., Sellaro, R., Lavovi, P., Zhang, H., 2003.
Recurrence of bipolar disorders and major depression: a life-
long perspective. Eur. Arch. Psychiatry Clin. Neurosci. 253,
236 – 240.
Benazzi, F., 2001. Bipolar II depression in late life: prevalence and
clinical features in 525 depressed outpatients. J. Affect. Disord.
66, 13 – 18.
Benazzi, F., Akiskal, H.S., 2003. Refining the evaluation of bipolar
II: beyond the strict SCID-CV guidelines for hypomania. J. Af-
fect. Disord. 73, 33– 38.
Coryell, W., Endicott, J., Maser, J.D., Keller, M.B., Leon, A.C.,
Akiskal, H.S., 1995. Long-term stability of polarity distinctions
in the affective disorders. Am. J. Psychiatry 152, 385 – 390.
Dunner, D.L., Fleiss, J.L., Fieve, R.R., 1976. The course of devel-
opment of mania in patients with recurrent depression. Am. J.
Psychiatry 133, 905 – 908.
Falret, J.P., 1851. Marche de la folie. Gazette Ho
ˆpitaux 24,
18 – 19.
Feighner, J.P., Robins, E., Guze, S.B., Woodruff, R.A., Winokur,
J.G., Munoz, R., 1972. Diagnostic criteria for use in psychi-
atric research. Arch. Gen. Psychiatry 26, 57 – 63.
Ghaemi, S.N., Sachs, G.S., Chiou, A.M., Pandurangi, A.K., Good-
win, F.K., 1999. Is bipolar disorder still underdiagnosed? Are
antidepressants overutilized? J. Affect. Disord. 52, 135 – 144.
Goldberg, J.F., Harrow, M., Whiteside, J.E., 2001. Risk for bipolar
illness in patients initially hospitalized for unipolar depression.
Am. J. Psychiatry 158, 1265 – 1270.
Goodwin, F.K., Jamison, K.R., 1990. Manic-Depressive Illness.
Oxford Univ. Press, New York.
Grigo, H.K., 1981. Zur Computer-Diagnostik affektiver und schizo-
affektiver Psychosen. Zurich University, Medical thesis.
Kaplan, E.L., Meier, P., 1958. Nonparametric estimation from in-
complete observation. J. Am. Stat. Assoc. 53, 457 – 481.
Kessing, L.V., Andersen, P.K., 1999. The effect of episodes on
recurrence in affective disorder: a case register study. J. Affect.
Disord. 53, 225 – 231.
Kessing, L.V., Olsen, E.W., Andersen, P.K., 1999. Recurrence in
affective disorder: analyses with frailty models. Am. J. Epide-
miol. 149, 404 – 411.
Kessing, L.V., Hansen, G.M., Andersen, P.K., Angst, J., 2004. Re-
currence of bipolar disorders and major depression: a life-long
perspective-II. Acta Psychiatr. Scand. 109, 339 – 344.
Kinkelin, M., 1954. Verlauf und Prognose des manisch-depressiven
Irreseins. Schweiz. Arch. Neurol. Neurochir. Psychiatr. 73,
100 – 146.
Kraepelin, E., 1898. Dementia praecox. (18. Versammlung d.su
¨d-
westdeutschen psychiatr. Vereins in Karlsruhe, 6.11.1897). Allg.
Z. Psychiatr. 55, 63 – 64.
Lavori, P.W., Dawson, R., Mueller, T.I., Warshaw, M., Swartz, A.,
Leon, A., 1996. Analysis of course of psychopathology: transi-
tions among states of health and illness. Int. J. Methods Psy-
chiatr. Res. 6, 321 – 334.
Marneros, A., Deister, A., Rohde, A., 1991. Affektive, schizoaffek-
tive und schizophrene Psychosen. Eine vergleichende Langzeit-
studie (Affective, Schizoaffective and Schizophrenic Disorders:
A Comparative Long-Term Study). Springer, Berlin.
National Depressive and Manic-Depressive Association NDMDA,
1993. National survey of NDMDA members finds long delay in
diagnosis of manic-depressive illness. Hosp. Community Psy-
chiatr. 44, 800 – 801.
Perris, C., 1966. A study of bipolar (manic-depressive) and uni-
polar recurrent depressive psychoses. Acta Psychiatr. Scand. 42
(Suppl. 194), 1 – 189.
Strober, M., Carlson, G., 1982. Bipolar illness in adolescents with
major depression. Clinical, genetic, and psychopharmacologic
predictors in a three-to four-year prospective follow-up inves-
tigation. Arch. Gen. Psychiatry 39, 549 – 555.
Strober, M., Lampert, C., Schmidt, S., Morrell, W., 1993. The
course of major depressive disorder in adolescents: I. Recovery
and risk of manic switching in a follow-up of psychotic and
J. Angst et al. / Journal of Affective Disorders 84 (2005) 149–157156
nonpsychotic subjects. J. Am. Acad. Child Adolesc. Psych. 32,
34 – 42.
Wei, L.J., Lin, D.Y., Weissfeld, L., 1989. Regression analysis of
multivariate incomplete failure time data by modeling marginal
distributions. J. Am. Stat. Assoc. 84, 1065 – 1073.
Winokur, G., Morrison, J.R., 1973. The Iowa 500: follow-up of 225
depressives. Br. J. Psychiatry 123, 543– 548.
Winokur, G., Clayton, P.J., Reich, T., 1969. Manic Depressive Ill-
ness. C.V. Mosby, Saint Louis.
Winokur, G., Coryell, W., Endicott, J., Akiskal, H., 1993. Further
distinctions between manic-depressive illness (bipolar disorder)
and primary depressive disorder (unipolar depression). Am. J.
Psychiatry 150, 1176 – 1181.
J. Angst et al. / Journal of Affective Disorders 84 (2005) 149–157 157
