Course and Prognosis of Mood Disorders

The course and prognosis of mood disorders

Topics covered:

  • The importance of course and the limits of our knowledge
  • Stability of the diagnoses of mood disorders
  • Onset of bipolar disorder and depression
  • Length of episodes
  • Recurrence
  • Outcome
    • Incomplete remission
    • Long-term outcome
    • Mortality
  • Prediction
  • Course of other subtypes of mood disorders
    • Minor depression
    • Seasonal affective disorder
    • Rapid cycling mood disorder
  • Consequences for treatment
  • References

The importance of course and the limits of our knowledge

An understanding of the course of a disorder is fundamental for doctor and patient when deciding whether to start long-term prophylactic medication and, at a later stage, whether to stop a successful long-term treatment. Course is a crucial factor in estimating the social consequences, suicide risk, and mortality associated with mood disorders.

A distinction should be made between the natural history of a disorder, describing its spontaneous untreated course, and course as observed under treatment, whether of episodes or long term (both episodes and the intervals between them). Most mood disorders have a phasic course, characterized by multiple recurrences; a minority manifest as a single episode or have a chronic course.

The description of course includes the age of onset, episode length, recurrence of episodes, and residual symptoms between episodes and outcome (remission, chronicity, death). All these aspects are discussed in this article.

Our understanding of the course of affective disorders is limited for several reasons. The course of bipolar disorder and unipolar depression are markedly different; however, Kraepelin's (1) unification of bipolar disorder (2) with all affective disorders to form the single diagnosis ‘manic depressive insanity' resulted in there being very little investigation of the natural history of the two subgroups (bipolar and depressive disorders) before the introduction of modern pharmacotherapy. Moreover, modern studies are carried out on treated patients, and the effect of drug-induced changes on the natural history of disorders is difficult to estimate. Another methodological limitation of modern studies of course is the selection of samples. Traditionally, samples have comprised hospitalized psychiatric patients, with a minority including psychiatric outpatients; studies following patients in primary care or in the community have been rare. Yet another methodological problem is that of memory artefacts. Although a retest study on the age of onset over a time span of 6 months showed good reliability, over the longer term patients significantly forget or misreport the previous history of their disorders, especially the precise age of onset and the number of earlier episodes, and there are no lifelong prospective studies of course in representative samples selected from the community or general practice to compensate for such distortions.

For all these reasons our understanding of the natural history and the course of mood disorders remains limited, especially in milder cases.

Stability of the diagnoses of mood disorders

Ever since Kahlbaum (3) and Kraepelin (1) the course and outcome of mental disorders have played important roles as criteria and validators of psychiatric classification. Mood disorders can be roughly subclassified into bipolar disorder and depressive disorders. The two groups of disorders differ significantly as regards family history and course. (4,5 and 6)

Distinguishing between bipolar disorder and unipolar depressive disorder is hampered by the fact that the diagnosis of unipolar depression is always uncertain. A long-term follow-up study over 27 years showed an annual rate of diagnostic change from depression to hypomania of about 1 per cent. The risk of such change seems not to fade but rather to be renewed with each fresh episode during a recurrent course. Therefore most studies conducted on unipolar depressive subjects include an unknown proportion of bipolar cases. For the same reason the exact ratio of bipolar to unipolar depressive subjects is unknown. Modern estimates range from 1:5 to 1:1. Thus many depressives may be hidden bipolar patients; these would include depressives with a positive family history of mania, those who have at some time manifested minor mood swings of hypomania (1–3 days), those who have shown so-called ‘drug induced' hypomania, and depressive subjects with a hyperthymic or cyclothymic temperament or a cyclothymic mood disorder.  

Today, bipolar disorder is conceptualized as a wide spectrum embracing all the milder forms of the disorder, ranging from cyclothymia and mild and brief hypomania, through hypomania, bipolar II disorder, and bipolar I disorder, to ‘pure' mania. Bipolar II disorder shows good diagnostic stability over a 5-year period. (7) Nevertheless, over a patient's lifetime there may be frequent shifts across the diagnostic spectra of bipolar disorder and depressive disorders respectively. In the longer term the diagnosis of major depressive disorder remains stable in only 40 to 50 per cent of cases; in the remainder, diagnostic changes to dysthymia, recurrent brief depression, minor depression, and subthreshold depression are frequently observed, as well as recovery.

Onset of bipolar disorder and depression

In patients admitted to hospital between 1913 and 1940 and not treated by electroconvulsive therapy or modern psychotropic drugs, bipolar disorder clearly manifested earlier than unipolar depression; (8) this finding is confirmed by modern community studies. Bipolar disorder generally begins during adolescence but may manifest even earlier.

Epidemiological studies identify the age of onset of bipolar disorder as between 15 and 19 years (means), whereas studies of hospitalized patients date its onset in the early twenties or in the thirties. The age-of-onset curve is skewed, and therefore mean values are not representative. In a large Canadian general population study, (9) 95 per cent of manic cases manifested before the age of 26 (males) and 25 (females), and 95 per cent of major depressive episodes before the age of 55 (males) and 43 (females). There was a considerable time lag between the age at onset of the first impairing symptoms (15 years), diagnosis (19 years), first treatment (22 years) and hospitalization (26 years).

Bipolar I illness manifests earlier than bipolar II and psychotic bipolar disorder. Late-onset bipolar disorder is rare but does occur and may be associated with specific neuropathology.

Unlike bipolar disorder, depression may start at any time of life. There is no dichotomy but a continuum from early-onset to late-onset depression, with a systematic decrease in genetic vulnerability and an increase in precipitation by environmental factors. Late-onset depression is more often milder and chronic.

A two-peak distribution of the age of onset has sometimes been described for both bipolar disorder and depression, but there is no second peak linked with the menopause in women.

Prodromal hypomanic and depressive symptoms and subclinical syndromes frequently precede mania and major depression over many years; this is especially true for depression in old age (75 years and more). The first manic and depressive manifestations are commonly mild, brief, or uncharacteristic, and are only diagnosed in retrospect.

Length of episodes

Most bipolar and depressive episodes are short, but a minority become chronic (lasting more than 2 years); the distribution of episode length is log normal, and therefore percentiles and not averages should be used as parameters. Using the data collected a century ago by Mendel (10) and Ziehen (11) on the natural length of episodes of mania and bipolar disorder, mainly among hospitalized patients, it is possible to compute a median length of 4 to 6 months for mania and 5 to 6 months for bipolar disorder. Wertham (12) also reports a median length of 4 to 6 months based on analysis of 2000 manic attacks. These figures do not differ from those obtained today despite a wide range of antimanic and antidepressant treatments. Among hospitalized patients episode length (median) was 4.2 months for bipolars and 5.4 months for major depressives; (13) 25 per cent of bipolar episodes lasted more than 7.3 months and 25 per cent of depressive episodes lasted more than 11 months. By contrast, in the community, where many untreated cases of depression occur, episode duration was considerably shorter; the 25th, 50th, and 75th percentiles were 4 weeks, 12 weeks, and 30 weeks respectively for 71 incident episode duration, and were 4 weeks, 8 weeks, and 16 weeks respectively for 28 recurrent episodes. (14) These subsamples were small and the data on recurrent episodes were not corrected for total number of episodes.

In 10 to 20 per cent of subjects, mood disorders take a chronic course (length over 24 months) without remission. (15) Electroconvulsive therapy can curtail depressive episodes, whereas antimanic and antidepressant drugs only alleviate the symptoms; this is why termination of treatment during an ongoing episode induces a rapid relapse.

It is important to note that the lengths of episodes of bipolar disorder and of depression have probably remained constant for over a century.


Recurrence is typical of mood disorders. It can be characterized by the number of episodes, the length of intervals (measured from remission to the onset of a new episode), and the length of cycles (measured from the beginning of one episode to the beginning of the next). Statistically, a normal distribution of cycle length can be obtained by log n transformation. In prospective studies, the length of the interval is frequently used as a parameter for survival analyses of recurrence.

In both bipolar disorder and unipolar depression the time from the first to the second episode is on average much longer than that from the second to the third. This progressive shortening of cycles and free intervals then levels off and fluctuates around a certain (but still variable) individual limit. Most published data on interval length or cycle length are methodologically flawed because they have not been corrected for the number of episodes/cycles observed, an improvement suggested by Slater (16) as early as 1938. Nonetheless, multiple episodes obviously follow each other in more rapid succession than a few episodes distributed over a lifetime. Even after taking episode numbers into account, there is a clear intra individual trend to a progressive shortening of cycle length, (17,18) dimming the prognosis for both bipolar disorder and unipolar depression. Cycle length tends to be shorter in late-onset than in early-onset mood disorders, increasing the risk of recurrence in the elderly.

Precipitating events play an important role in the onset of the first few affective episodes; thereafter recurrrence seems to become more autonomous with stressful events contributing little or nothing to the process. (19) Stressors may not only precipitate episodes but also increase a pre-existent vunerability, sensitizing the individual and thereby making him or her more vulnerable to further episodes (kindling effect (20) ). In bipolar illness there is no difference in the quality or quantity of stressors precipitating depressive and manic episodes; a legacy or the loss of a relative can induce depression or hypomania.

Counts of single-episode cases of mood disorders depend strongly on whether or not mild and brief hypomania and depression are included and on the length of observation. Single-episode bipolar disorders probably do not occur, and single-episode unipolar depression is observed in only10 to 15 per cent of cases.

The course of the mood disorders of the overwhelming majority of patients, whether psychiatric inpatients, outpatients, or general practitioner patients, is recurrent; even milder depressive episodes tend to be recurrent. (21)

The pattern of recurrence is irregular. Compared with normal subjects, the daily mood ratings of bipolar subjects were characterized as a low-dimensional chaotic process and true cyclicity was not apparent in the power spectra of either group (normal subjects or bipolars). (22)

Over a lifetime bipolar patients experience twice as many episodes as unipolar depressives, a difference which is not explained by the manifestation of manic episodes in addition to depression. The total number of episodes observed depends on the length of observation. In a 22- to 26-year follow-up study, bipolar patients experienced a median of 10 episodes, whereas depressive patients experienced four episodes. (13) On average, 0.44 episodes per year were observed in bipolar patients and 0.30 in depressive patients.

A follow-up study of manic subjects applying survival analysis demonstrated relapse and recurrence rates of 20 per cent by 6 months, 48 per cent by 1 year, and 81 per cent by 5 years; patients with mixed/cycling features had a 10 per cent higher recurrence rate than pure manics. In a recent large representative Danish record study (N= 20 350 first admissions) unipolar depressives had strikingly lower recurrence rates (hospitalizations) than bipolars, both correlating with the number of previous episodes. (18) The authors concluded: ‘The course of severe unipolar and bipolar disorder seems to be progressive in nature despite the effect of treatment and irrespective of gender, age and type of disorder'.


Incomplete remission

Remission after affective episodes is frequently incomplete. Residual symptoms are common in psychiatric and general practitioner patients and in cases identified in community studies. Comparatively little is currently known about the residual symptoms of mania, although they do exist. Residual symptoms of major depression, defined by a score of 8 or more on the 17-item Hamilton Depression Scale, were found in 32 per cent of 60 patients 12 to 15 months after remission. (23) Residual symptoms represent a strong risk factor for further recurrence; a survival analysis by Paykel et al. (23) found a threefold higher risk of recurrence (76 per cent) in patients with residual symptoms than in those without (25 per cent). Chronic residual symptoms are those typical of depression: mood, anxiety, genital symptoms, (24) insomnia, headaches, neurasthenic complaints, reduced libido, and gastrointestinal symptoms. They are frequently treated by long-term antidepressant medication, which should not be withdrawn until the patient has had 4 months completely free of symptoms.

Long-term outcome

The long-term outcome of mood disorders is usually unfavourable. Five comparable long-term follow-up studies (10 years or more) are considered. All re-examined depressive patients, most if not all of whom had originally been admitted to psychiatric hospitals. After 10 years,roughly one-third had been readmitted, and this figure rose to about two-thirds after 20 years. A poor outcome, as operationally defined by Lee and Murray, (25) was observed in 11 to 25 per cent of patients. The outcome was in no way more favourable in the two studies that included psychiatric outpatients. 

Bipolar disorder has a poorer outcome than depression. After a follow-up of 22 to 26 years, definitive recovery (at least 5 years with good social adaptation) was found in 25 per cent of 186 depressive subjects, whereas the figure was 16 per cent of the 220 bipolar patients; a chronic course lasting at least 2 years without remission was present in 11.8 per cent of the depressive subjects compared with 14.1 per cent of the bipolars. (29)

Modern treatment may have changed the outcome of mood disorders by reducing chronicity and rehospitalization. A recent 10-year prospective study of 131 hospitalized bipolar patients (30) found that only 4 per cent had developed chronicity, although most patients still had recurrences. The frequency of episodes did not increase with time, so no further kindling effect was observed. Comorbidity with alcoholism, a factor known to correlate with poorer outcome, was found in 30 per cent of the patients. Coryell et al. (7) found that, over 5 years, bipolar II disorder had a better outcome than bipolar I disorder in terms of rehospitalizations; furthermore, the two groups tended to remain diagnostically stable. Episode frequency was comparable in the two groups.


Mortality expressed by the standardized mortality ratio (SMR) is elevated in subjects suffering from mood disorders, irrespective of the sample selection (community or psychiatric inpatients). An SMR of between 1.37 and 2.49 has been found among mood disorder subjects; in the normal population it is 1.0. Suicide is the main, but not the sole, cause of this elevated mortality. Other over-represented causes of death among patients with mood disorders are accidents, cardiovascular disorders, cerebrovascular disorders, respiratory infections, thyroid disorders, and secondary substance abuse/dependence. (31)

A meta-analysis (32) of 58 studies covering 2257 cases of suicide gave the following SMRs for suicide mortality compared with suicides in the general population (SMR = 1.0): bipolar disorder, 15.05, major depressive disorder/major depressive episode, 20.35; dysthymia, 12.12; depression not otherwise specified, 16.10. Taking all diagnostic categories together, suicide among mood-disorder patients was 13.65 times more frequent than in the general population.

The long-term follow-up studies of severely depressed patients gave suicide rates between zero and 13 per cent. The frequently quoted rate of 12 to 19 per cent (33) may only be valid for hospitalized patient samples, which by definition include many suicidal patients. In community and outpatient samples, suicide accounted for only a small percentage of deaths. Many suicides can probably be prevented by administering long-term antidepressant medication (34) and lithium. (35,36)


Predicting course remains an uncertain art. In general, studies of predictors should be prospective and the results replicated, prerequisites which are rarely fulfilled. Most of the reported predictors tend to be hypothetical in value.

The previous course of mood disorders predicts future course, recurrence (number of previous episodes) predicts further recurrence, and the length of episodes and residual symptoms predict recurrence. Acute onset and few attacks predict a better outcome, whereas poor functioning over years predicts a worse outcome.

Age does not seem to have a great effect on outcome, but depression in the elderly, although milder, is often more chronic.

Neuroticism as a measure of emotional and vegetative lability tends to remain stable in depressives, if the effect of ageing istaken into account. High-baseline neuroticism scores predict a poorer outcome in patients, primiparae with postnatal depression, and subjects in the community.

Neuroticism is a risk factor for recurrence perhaps because it includes many items measuring depressive symptoms (and subclinical depression is known to predict incidence and recurrence of depression). The Symptom Check List 90R (37) is also a predictor for recurrence; it correlates closely with neuroticism and includes both emotional and vegetative lability.

Low self-esteem was not found to be a predictor of the first onset of depression but rather a symptom that waxes and wanes withdepression itself. (38) Lack of self-confidence predicted the recurrence of depression in a prospective study. (39)

Neurotic depression has a poorer prognosis than endogenous depression and a much poorer prognosis than retarded depression. However, on the whole the subtyping of depression (melancholic, endogenous, simple, delusional) has little predictive value.

Comorbidity of alcoholism and personality disorders with bipolar disorder and depression make for a poorer prognosis and outcome of both.

Course of other subtypes of mood disorders

Minor depression

In the community, measures of the recurrence of minor depression are similar to those for major depression, a fact confirmed by survival analyses. (40)

A diagnostic change from minor to major depression, or the reverse, is frequent during the course of mood disorders. Primary minor depression, like depressive symptoms in general, is a significant risk factor for major depression. It may represent a residual state of major depression and a risk factor for recurrence. Among the elderly, in which age group only about one-third of cases of major depression fully recover, minor depression is common as a residual state of major depression.

Both minor and major depression should be seriously considered as a target for preventive intervention and treatment. (40)

Seasonal affective disorder

Many patients experience affective episodes mainly in the spring or in autumn to winter; others suffer from both types of seasonal recurrence. Seasonal affective disorder (SAD) remained seasonal in 70 per cent of 43 cases followed up over 2 to 5 years, but the rate of diagnostic stability of SAD was limited to 26 per cent because 44 per cent of the subjects later developed seasonal subthreshold (subsyndromal) depression and another 20 per cent recovered. (41) In three other studies SAD had a higher diagnostic stability (between 38 per cent and 57 per cent).

Thus the diagnostic stability of SAD is clearly weaker than the seasonality of depressive symptoms and syndromes.

Rapid cycling mood disorder

Rapid cycling is found almost exclusively in bipolar disorder. It is defined by the presence of four or more affective episodesin 12 months, and is more frequent in females and in the bipolar II subtype. Rapid cycling does not appear to represent a chronic endstage form of bipolar disorder; it is often a transient non-familial manifestation of bipolar disorder. (42) A prospective follow-up study conducted over approximately 3 years showed diagnostic stability in about half the cases and the other half had fewer than four further episodes per year; in a control group 10 per cent of the non-rapid-cycling patients converted to rapid cycling. (43)

Rapid cycling does occur in unipolar depression, but is usually a transient and unstable phenomenon. Bipolar rapid cycling takes a very autonomous course and is very hard to treat.

Consequences for treatment

After the initial split into bipolar disorder and unipolar depression, the last 30 years have seen the progressive subdivision of mood disorders into further subgroups: bipolar I, bipolar II, rapid cycling bipolar disorder, minor depression, dysthymia, SAD, and so on. In the process the characteristics of the course of a disorder have been incorporated into the classification, which means that course and prognosis are no longer independent of the diagnostic definition.

The distinction between bipolar disorder and unipolar depression is fundamental. These disorders differ markedly in their course and outcome, with bipolar disorder having an earlier onset, higher episode frequency, slightly shorter episode length, and poorer outcome (fewer full recoveries, slightly more chronicity), but, unexpectedly, probably fewer suicides. All bipolar disorders and most unipolar depressive disorders are recurrent, with a minority having a really good prognosis without residual symptoms and further recurrences. Bipolar II disorders probably have a slightly better prognosis than bipolar I disorders.

Therapeutic decisions on the length of acute treatment will depend on the length of the individual's previous episodes and on the average episode length observed in follow-up studies. The length of affective episodes has probably not changed in 100 years. Antidepressants cannot shorten the episodes but can minimize the symptoms. Treatment should be maintained for the duration of episodes, which are frequently masked; otherwise relapses must be expected.

The choice of a long-term prophylactic medication also has to take into consideration the previous individual course of the disorder plus the general scientific knowledge about course and prognosis, and to keep in mind the increased mortality, especially the high suicide mortality associated with mood disorders. (44) Recurrence is also a feature in mild cases, but in contrast with severe cases the suicide mortality is probably low. Decisions about long-term medication also have to take into account whether there are residual symptoms during the intervals; such symptoms are a strong risk factor for further recurrence. Over the lifetime, each new recurrence is associated with a new risk of suicide. Any cessation of treatment has to take these risks into account. So far there are no positive recommendations for the cessation of prophylactic treatment.


1.Kraepelin, E. (1899). Psychiatrie. Ein Lehrbuch für Studierende und Ärzte(6th edn). Barth, Leipzig.

2.Falret, J.P. (1851). De la folie circulaire ou forme de maladie mentale caractérisée par l'alternative régulière de la manie etde la mélancholie. Bulletin de l'Académie de Médecine (Paris), 6, 382–400.

3.Kahlbaum, K. (1863). Die Gruppirung der psychischen Krankheiten und die Eintheilung der Seelenstörungen. A.W. Kafemann, Danzig.

4.Angst, J. (1966). Zur Ätiologie und Nosologie endogener depressiver Psychosen. Monographien aus dem Gesamtgebiete der Neurologie und Psychiatrie. Springer-Verlag, Berlin.

5.Perris, C. (1966). A study of bipolar (manic–depressive) and unipolar recurrent depressive psychoses. Acta Psychiatrica Sandinavica, 194(Supplement), 1–189.

6.Winokur, G. and Clayton, P.J. (1967). Family history studies. I. Two types of affective disorders separated according to genetic and clinical factors. In Recent advances in biological psychiatry, Vol. 9 (ed. J. Wortis), pp. 35–50. Plenum Press, New York.

7.Coryell, W., Keller, M.B., Endicott, J., Andreasen, N.C., Clayton, P.J., and Hirschfeld, R.M.A. (1989) Bipolar II illness: course and outcome over a five-year period. Psychological Medicine, 19, 129–41.

8.Stephens, J.J. and McHugh, P.R. (1991). Characteristics and long-term follow-up of patients hospitalized for mood disorders in the Phipps Clinic. Journal of Nervous and Mental Diseases, 179, 64–73.

9.Bland, R.C., Newman, S.C., and Orn, H. (1988). Age of onset of psychiatric disorders. Acta Psychiatrica Scandinavica, 338(Supplement), 43–9.

10.Mendel, E. (1881). Die Manie. Urban and Schwarzenberg, Vienna.

11.Ziehen, T. (1896). Die Erkennung und Behandlung der Melancholie in der Praxis. Karl Marhold, Halle.

12.Wertham, F.I. (1929). A group of benign chronic psychoses: prolonged manic excitements. With a statistical study of age, duration and frequency in 2000 manic attacks. American Journal of Psychiatry, 9, 17–28.

13.Angst, J. and Preisig, M. (1995) Course of a clinical cohort of unipolar, bipolar and schizoaffective patients. Results of a prospective study from 1959 to 1985. Schweizer Archiv für Psychiatrie und Neurologie, 146, 5–16.

14.Eaton, W.W., Anthony, J.C., Gallo, J., et al.(1997) Natural history of Diagnostic Interview Schedule DSM-IV major depression. The Baltimore Epidemiologic Catchment Area follow-up. Archives of General Psychiatry, 54, 993–9.

15.Angst, J. (1988). Clinical course of affective disorders. In Depressive illness: prediction of course and outcome(ed. T. Helgason and R.J. Daly), pp. 1–48. Springer-Verlag, Berlin.

16.Slater, E. (1938) Zur Periodik des manisch-depressiven Irreseins. Zeitschrift für die Gesamte Neurologie und Psychiatrie, 162, 784–801.

17.Angst, J. (1981). Course of affective disorders. In Handbook of biological psychiatry. Part IV: Brain mechanisms and abnormal behavior-chemistry(ed. H.M. van Praag), pp. 225–42. Marcel Dekker, New York.

18.Kessing, L.V., Andersen, P.K., Mortensen, P.B., and Bolwig, T.G. (1998) Recurrence in affective disorder. I. Case register study. British Journal of Psychiatry, 172, 23–8.

19.Goodwin, F.K. and Jamison, K.R. (1990). Manic–depressive illness. Oxford University Press.

20.Post, R.M., Rubinow, D.R., and Ballenger, J.C. (1984). Conditioning, sensitization, and kindling: implications for the course of affective illness. In Neurobiology of mood disorders(ed. R.M. Post and J.C. Ballenger), pp. 432–66. Williams and Wilkins, Baltimore, MD.

21.Angst, J. (1990). Depression and anxiety: a review of studies in the community and primary health care. In Psychological disorders in general medical settings(ed. J. Costa e Silva, Y. Lecrubier, and U. Wittchen), pp. 60–8. Hogrefe and Huber, Toronto.

22.Gottschalk, A., Bauer, M.S., and Whybrow, P.C. (1995). Evidence of chaotic mood variation in bipolar disorder. Archives of General Psychiatry, 52, 947–59.

23.Paykel, E.S., Ramana, R., Cooper, Z., Hayhurst, H., Kerr, J., and Barocka, A. (1995). Residual symptoms after partial remission: an important outcome in depression. Psychological Medicine, 25, 1171–80.

24.Paykel, E.S. (1998). Remission and residual symptomatology in major depression. Psychopathology, 31, 5–14.

25.Lee, A.S. and Murray, R.M. (1988). The long-term outcome of Maudsley depressives. British Journal of Psychiatry, 153, 741–51.

26.Kiloh, L.G., Andrews, G., and Neilson, M. (1988). The long-term outcome of depressive illness. British Journal of Psychiatry, 153, 752–7.

27.Thornicroft, G. and Sartorius, N. (1993). The course and outcome of depression in different cultures: 10-year follow-up of the WHO Collaborative Study on the Assessment of Depressive Disorders. Psychological Medicine, 23, 1023–32.

28.Surtees, P.G. and Barkley, C. (1994). Future imperfect: the long-term outcome of depression. British Journal of Psychiatry, 164, 327–41.

29.Angst, J. and Preisig, M. (1995). Outcome of a clinical cohort of unipolar, bipolar and schizoaffective patients. Results of a prospective study from 1959 to 1985. Schweizer Archiv für Psychiatrie und Neurologie, 146, 17–23.

30.Tohen, M., Waternaux, C. and Tsuang, M.T. (1990). Outcome in mania. A 4-year prospective follow-up of 75 patients utilizing survival analysis. Archives of General Psychiatry, 47, 1106–11.

31.Angst, F., Sellaro, R., Stassen, H.H., and Angst, J. Mortality of patients with mood disorders: follow-up over 34 to 38 years. In preparation.

32.Harris, E.C. and Barraclough, B. (1997). Suicide as an outcome for mental disorders. A meta-analysis. British Journal of Psychiatry, 170, 205–28.

33.Guze, S.B. and Robins, E. (1970). Suicide and primary affective disorder. British Journal of Psychiatry, 117, 437–8.

34.Angst, F., Sellaro, R., and Angst, J. (1998) The natural history of bipolar disorder. Presented at the Summer Meeting of the British Association for Psychopharmacology, July 1998, Cambridge. Journal of Psychopharmacology, 12(Supplement A to No. 3), A1.

35.Coppen, A., Standish-Barry, H., Bailey, J., Houston, G., Silcocks, P., and Hermon, C. (1991). Does lithium reduce the mortalityof recurrent mood disorders? Journal of Affective Disorders, 23, 1–7.

36.Müller-Oerlinghausen, B., Ahrens, B., Grof, E., et al.(1992). The effect of long-term lithium treatment on the mortality of patients with manic–depressive and schizoaffective illness. Acta Psychiatrica Scandinavica, 86, 218–22.

37.Derogatis, L.R. (1977). SCL-90. Administration, scoring and procedures manual I for the R (revised) version and other instruments of the psychopathology rating scales series. Johns Hopkins School of Medicine, Baltimore, MD.

38.Ernst, C., Schmid, G.B., and Angst, J. (1992). The Zurich Study: XVI. Early antecedents of depression. A longitudinal prospective study on incidence in young adults. European Archives of Psychiatry and Clinical Neuroscience, 242, 142–51.

39.Surtees, P.G. and Wainwright, N.W.J. (1996). Fragile states of the mind: neuroticism, vulnerability and the long-term outcome of depression. British Journal of Psychiatry, 169, 338–47.

40.Kessler, R.C., Zhao, S., Blazer, D.G., and Swartz, M. (1997). Prevalence, correlates, and course of minor depression and major depression in the National Comorbidity Survey. Journal of Affective Disorders, 45, 19–30.

41.Graw, P., Gisin, B., and Wirz-Justice, A. (1997). Follow-up study of seasonal affective disorder in Switzerland. Psychopathology, 30, 208–14.

42.Coryell, W., Endicott, J., and Keller, M. (1992) Rapidly cycling affective disorder: demographics, family history, and course. Archives of General Psychiatry, 49, 126–31.

43.Bauer, M.S., Calabrese, J., Dunner, D.L., et al.(1994) Multisite data reanalysis of the validity of rapid cycling as a course modifier for bipolar disorder in DSM-IV. American Journal of Psychiatry, 151, 506–15.

44.Schou, M. (1998). The effect of prophylactic lithium treatment on mortality and suicidal behavior: a review for clinicians. Journal of Affective Disorders, 50, 253–9.