Epidemiology of Mood Disorders

The epidemiology of mood disorders. 

Topics covered:

  • Introduction
  • Bipolar disorders
    • Diagnostic issues
    • Prevalence
    • Comorbidity
    • Use of health services
    • Risk factors
      • Organic factors
      • Other biological factors
      • Life events
      • Childhood experiences
      • Subthreshold symptoms
  • Depressive disorders
    • Diagnostic issues
    • Prevalence
    • Risk factors
      • Genetics
      • Gender
      • Childhood experiences
      • Personality
      • Social environment
      • Physical illness
    • An integrated aetiological model
    • Comorbidity
    • Use of health services
  • References

Introduction

The Global Burden of Disease, which is a comprehensive assessment of mortality and disability from diseases and injuries in 1990 and projected to 2020, highlights the importance of mood disorders for the world. Using the measure of disability-adjusted life years, it was determined that unipolar major depression was the fourth leading cause of disease burden in the world. It was also projected that, in the year 2020, unipolar major depression would be the second leading cause of disease burden in the world. Disability-adjusted life years is based on both mortality and disability. If one looks at disability alone, then unipolar major depression was the leading cause of disability in the world in 1990, and bipolar disorder was the sixth leading cause. Across the world, 10.7 per cent of disability can be attributed to unipolar major depression and, in developed countries, unipolar major depression contributes to nearly 20 per cent of disease burden in women aged from 15 to 44 years. (1)

The mood disorders have received considerable attention in psychiatric epidemiology over the last two decades. These received particular attention in the five-site United States National Institutes of Mental Health Epidemiologic Catchment Area Study (ECA), as well as the epidemiological studies in other countries around the world that used the ECA methodology. Mood disorders also received particular attention in the National Comorbidity Survey (NCS) in the United States and in the National Psychiatric Morbidity Survey of Great Britain. Thus there is substantial data from around the world on the epidemiology of these disorders. In addition, many of the population-based twin registries, such as in Virginia, in the United States, have also paid particular interest to mood disorders and have the additional advantage of being able to consider genetic as well as environmental risk factors.

Bipolar disorders

Diagnostic issues

While classical bipolar disorder with episodes of euphoric mania interspersed with episodes of depression is one of the clearest clinical syndromes in psychiatry, the boundaries of bipolar disorder remain contested. As case definition is central to epidemiology, all the contested boundaries of bipolar disorder could influence prevalence rates and our understanding of risk factors. Some of the major boundary issues for bipolar disorder include the overlap of bipolar disorder with psychotic features, with schizoaffective disorder and schizophrenia, and the overlap of bipolar disorder with unipolar major depression when patients who present primarily with depression have brief or mild episodes of hypomania. There is also the overlap of bipolar disorder with apparent personality disorder, especially Cluster B personality disorders such as borderline and narcissistic, and the issue of when hyperthymic personality merges into bipolar disorder. (2,3)

Another important issue in determining caseness of bipolar disorder for epidemiological surveys is symptom pattern. A number ofthe diagnostic instruments for assessing bipolarity in population surveys limit the questions on mania to a type of symptom profile characterized by euphoria, grandiosity, increased energy, and decreased sleep. Whether the commonly used epidemiology interviews adequately detect those individuals who have manic episodes characterized by irritability, anger, and activation is very debatable. Furthermore, as insight is sometimes impaired in hypomania and mania, and as these are uncommon disorders, the accuracy of case detection of bipolar disorders in populations remains an issue for further research. (4)

Prevalence

Population studies such as the ECA, and its related cross-national studies, and the NCS reported that the lifetime prevalence of bipolar disorder varies from 0.3 to 1.5 per cent. The NCS data include only bipolar I data, while the ECA includes bipolar I and bipolar II disorder. (4,5) In all studies, the 6-month prevalence is not much lower than the lifetime prevalence of bipolar disorder. These findings reflect the high degree of chronicity and/or recurrence associated with bipolar disorder.

In these population studies, the mean age of onset of bipolar disorder has varied from 17 to 27 years. However, as age of onset is not normally distributed, the mean is a slightly misleading variable; in clinical samples, while the mean age of onset may be in the twenties, the most common age of onset is the late teenage years. (6)

In bipolar disorder, the prevalence in males and females is similar. This is in contrast to the consistent female excess found in major depression.

Comorbidity

In the NCS, all identified bipolar I individuals suffered from at least one, and often up to three or more, comorbid disorders.The most common comorbid disorders included the full range of anxiety disorders, alcohol and drug dependence, and conduct disorder or other antisocial behaviours.

Alcohol and drug abuse and/or dependence are commonly comorbid with bipolar disorder. Old studies found that binge drinking was especially common in bipolar individuals and that this binge pattern of drinking was more associated with manic episodes than with depressive episodes. Clinical studies find that bipolar patients with a comorbid substance dependence are less compliant with prescribed mood stabilizers and have more frequent hospital readmissions.

Individuals with bipolar disorder have the full range of anxiety disorders, including phobias, panic disorder, and obsessive–compulsive disorder. In recent years, the comorbidity of bipolar disorder with panic disorder has received particular attention. In the NCS, for instance, the odds ratio for having panic disorder with major depression was 6.8, but the odds ratio for having panic disorder with bipolar disorder was 13.2. This differential pattern of comorbidity of panic disorder with bipolar disorder and major depression may assist in the search for aetiological or risk factors.

Another area of high comorbidity with bipolar disorder is that of childhood conduct disorder. One of the major issues in understanding this high rate of comorbidity is a diagnostic issue—whether childhood conduct disorder and/or childhood attention-deficit disorder are sometimes the first manifestations or precursors of bipolar disorder. Certainly, if the pattern of conduct-disorder symptoms or attention-deficit symptoms is episodic rather than consistent over time, the issue becomes whether these are not early manifestations of bipolar disorder rather than truly independent comorbid conditions.

Use of health services

In the ECA study, 39 per cent of those with bipolar I or bipolar II disorders received outpatient psychiatric treatment within 1 year and about 10 per cent would receive inpatient treatment within a 6-month period. In the NCS study, 45 per cent of those with bipolar disorder had received psychiatric treatment in the previous 12 months; although 93 per cent reported lifetime treatment for their bipolar disorder. However, both of these studies suggest that more than half the individuals with bipolar disorder are not currently in psychiatric treatment and, given the high morbidity and mortality associated with bipolar disorder, this is of major concern. (4)

Risk factors

In considering the risk factors for the development of bipolar disorder, it is useful to separate risk factors into those that are risk factors for lifetime vulnerability (for example genetic factors) and those that are risk factors for the onset of an episode of depression or mania (for example life events). Thus, in determining risk factors for lifetime vulnerability, genetic factors constitute the largest single risk factor. However, if one is considering who is vulnerable to an episode of mania over the next 6 months, genetic factors will play a relatively smaller part and predictions may best be based on other factors such as past history, childbirth, being treated for depression with antidepressant medication, and the approach of spring or summer. Genetic risk factors are discussed further in the following article. 

Organic factors

Although organic factors, such as some type of central nervous system damage, are unusual risk factors in young adults, in late-onset bipolar disorder (age of onset more than 50 years) organic disease of the central nervous system is an increasing factor for the development of mania. In younger adults, AIDS and head injury are two important aetiological factors in a limited number of cases of bipolar disorder.

Other biological factors

A range of other biological factors are particularly relevant risk factors to the onset of episodes of illness, but they may contribute a relatively small part to lifetime vulnerability. Many women have their first episode of depression or mania in the postpartum period. While a limited number of women may have manic episodes limited to the postpartum period, postpartum episodes of mania are more commonly part of a long-term bipolar disorder and these women will have episodes both precipitated by childbirth and at other times in their life. Indeed, in the postpartum period, having a history of bipolar disorder is one of the strongest risk factors for the development of a postpartum psychosis.

There is substantial evidence that seasonal patterns influence the onset of manic and depressive episodes. There are consistent findings of an excess of manic episodes in late spring and early summer. To date, however, the nature of the environmental factors that influence this late spring, early summer peak of manic episodes is less clear.

There is also substantial evidence that disruptions of normal biological rhythms may precipitate the onset of manic or depressive episodes. This has been documented in relation to international travel involving east–west or west–east travel with disruption of circadian rhythms. Disruption of circadian rhythms through shift-work or other factors, which disrupt the normal sleep cycles, may also be important triggers to the onset of episodes of mania. Indeed, this risk factor is now influencing bipolar disorder management plans by including strategies for minimizing disruptions to 24-h biological rhythms.

Life events

Adverse life events have been well documented to be precipitants of manic episodes, as well as depression. It appears that lifeevents are more likely prior to the first or second episode of mania and are less likely later in the course of illness.

Childhood experiences

While there is substantial evidence that adverse childhood experiences contribute to the later development of major depression, there are relatively few data to support the idea that adversity in childhood influences the development of bipolar disorder.

Subthreshold symptoms

While there is a paucity of information from population studies as to how subsyndromal symptoms may be a risk factor for bipolar disorder, family studies have long been suggestive of the fact that individuals may manifest subsyndromal symptoms or syndromes well before the clear onset of bipolar affective disorder. Thus, there are individuals who have ‘mini-episodes' of mania or depression who are, presumably, at high risk of the subsequent developmentof bipolar disorder. Similarly, individuals in their teenage years with a pattern of mood swings and cyclothymia are probably also at increased risk of the development of bipolar disorder. Also, individuals with a hyperthymic personality, particularly if they have had past episodes of depression, are a high-risk group for the subsequent development of bipolar disorder.

Depressive disorders

Diagnostic issues

A key issue for the epidemiology of depressive disorders is defining the boundaries of major depression and dysthymia. Depressive symptoms in the community are common, and defining both the symptom count and the duration at which depressive symptoms count as part of a clinical disorder is arbitrary. Recently, Kendler and Gardner (7) examined the boundaries of major depression as defined by DSM-IV in a population-based twin sample of women. They found that, if a twin had four or fewer depressive symptoms, syndromes composed of symptoms involving no or minimal impairment, and episodes lasting less than 14 days, then the individual's co-twin was still at an increased risk of major depression. Kendler and Gardner concluded that they could find no empirical support for the DSM-IV requirement of duration for 2 weeks, five symptoms, or clinically significant impairment. These authors suggested that major depression, as articulated by DSM-IV, may be a diagnostic convention imposed on a continuum of depressive symptoms of varying severity and duration. Wainwright et al., (8) using data from the National Psychiatric Morbidity Survey of Great Britain, have also suggested that research should move beyond a binary decision of case versus non-case, and utilize a probablistic measure of psychiatric case status, replacing the arbitrary threshold with a smooth transition. This type of approach allows the benefits of syndrome diagnosis to be retained, while not falling into the dilemma of an arbitrary threshold that may lack validity.

Provided that one accepts the arbitrary definition of major depression, then determining the rates of current depressive disorders is not especially problematic. However, there are major methodological issues involved in determining whether an individual has ever had a lifetime episode of major depression. Lifetime prevalence rates vary from 4.4 per cent in the United States ECA study, to 17.1 per cent in the NCS, and to over 30 per cent in Kendler's Virginia twin sample of women. In part, subjects in the community may forget or fail to report past episodes of major depression, and the manner in which the questions are asked may importantly influence lifetime rates of depression. In the Diagnostic Interview Schedule, which was used in the ECA, respondents were asked about lifetime symptoms, a lifetime diagnosis was made, and then recency of the lifetime diagnosis was determined. More recent diagnostic interview schedules, such as the Composite International Diagnostic Interview, first ask about current depressive symptoms and then, having ‘primed' individuals about depressive symptoms, go on to enquire about past depressive episodes. Interviews that follow the schedule of ‘priming' before asking about past episodes appear to obtain considerably higher rates of lifetime major depression. Determining lifetime rates of depression with greater precision is an important task, as the vulnerability to depression conferred by risk factors such as genetic factors and childhood experiences may be wrongly estimated if lifetime rates of major depression are imprecise. For instance, when Kendler et al. (9) examined the heritability of major depression and corrected for the moderate reliability of a lifetime diagnosis of major depression, the heritability estimate increased from 40 per cent to over 70 per cent. As concluded by Kendler, major depression is not a disorder of high reliability and moderate heritability, but is a diagnosis of moderate reliability and high heritability.

DSM-IV allows major depression to be further subclassified into subtypes, such as melancholia and atypical and psychotic depression. Most of the traditional epidemiology studies have tended to ignore the issue of subtyping major depression. Recently, however, the issue of the atypical depression subtype has received particular attention in the study of the Virginia twins and in the NCS. In both these studies, latent class analysis suggests that atypical depression is a distinct subtype with several distinctive features, such as higher rates of parental alcohol- and drug-use disorders, higher interpersonal dependency, and higher rates of conduct disorder. If risk factors for atypical depression are, in part, distinct from risk factors for other subtypes of major depression, then for epidemiology to contribute to an understanding of aetiology it will be important to undertake further work on depressive subtypes. (10,11)

Prevalence

In the ECA, the 6-month prevalence of major depression across five sites was 2.2 per 100. (12) In Edmonton, Canada, the 6-month prevalence rate was 3.2 per 100 (13) and, in Christchurch, New Zealand, the rate was 5.3 per 100. (14) In the NCS, the 1-month prevalence of a major depressive episode was 6.1 per 100. (15) In the National Psychiatric Morbidity Surveys of Great Britain, the 1-week rate of a depressive episode was 2.1 per 100. (16) Together, these studies would suggest that the current rate of major depression is in the realm of 2 to 5 per cent.

The estimates of the lifetime rate of major depression are much more variable. The lowest rate reported is 4.4 per 100 from the ECA study, while, in the study of Virginia twins, the lifetime rate of major depression is over 30 per cent. It is reasonable to believe that the true lifetime rate of major depression is probably in the realm of 10 to 20 per 100, and so caution should be exercised in expressing lifetime rates of depression with undue precision.

These rates of major depression may also be lower if the rate of bipolar disorder is higher. Isolated clinical studies have found that 1 in 2, not 1 in 10, individuals presenting with depressive disorders have features of bipolar spectrum disorders. If these figures are correct, then this would presumably lower the rates of major depression, but would correspondingly increase the rates of bipolar disorders.

Over the past decade, one of the controversial findings in the epidemiology of major depression has been whether the rates of depression are increasing, and whether it is occurring at a younger age. The initial suggestions by Klerman and Weissman (17) provoked a variety of studies. Despite methodological concerns about the reliability of lifetime major depression, studies across countries have reasonably consistently documented an increasing rate of major depression with an earlier age of onset. (18) As mood disorders are the single largest risk factor for suicide, it is also of note that, in most Western countries, the rate of suicide, especially in young adults, has increased considerably over the past 20 years.

Risk factors

Genetics

There is now substantial evidence that genetic factors are of major importance as risk factors for vulnerability to major depression. While traditional estimates have put the heritability at about 40 per cent, when Kendler et al. (9) allowed for the moderate reliability of the diagnosis of major depression, the heritability estimate increased to 70 per cent. Of greater interest is that the genes for major depression do not appear to be unique for depression, but overlap with the genes for anxiety and the genes for neuroticism. (19,20) Wilhelm et al. (21) have suggested that the greater prevalence of depression in women is due to the strong association of anxiety and neuroticism with depression, and that the higher rates of anxiety and neuroticism in women lead to higher rates of depression.

Gender

One of the most consistent findings in the epidemiology of major depression is that the ratio of women to men is approximately 2:1. This increased rate of major depression in women arises during puberty, as in childhood there is a slightly higher prevalence of depression in boys than girls. The timing of this transition in rates by gender is related to biological puberty rather than just to age.

Childhood experiences

Early theorizing suggested that the loss of a parent in childhood increased the later risk for major depression; although many studies have examined this issue, they have inconsistently found it to be a risk factor for adult depression. (22) However, studies that examine the nature of child–parent attachment using a measure such as the Parental Bonding Instrument have consistently found that a lack of parental care is associated with increased rates of depression. (23) More recently, childhood sexual abuse has been established as a risk factor for adult major depression.

However, cumulative childhood disadvantage almost certainly poses a greater risk to later depression than any single childhood variable in isolation. Thus, if studies only look at single childhood risk factors, they may miss the full impact of global childhood adversity. The converse of childhood risk factors is childhood resilience, and it is probable that one good relationship with an adult and high intelligence in the child may, in part, protect from other adversities.

Personality

There has been a long history of interest in the likelihood that people with certain personality traits are more vulnerable to depression than others. It is likely that those individuals who are unduly anxious, impulsive, and obsessional may have increased rates of later major depression. The best data exists for neuroticism, which emerges as a clear risk factor for the later development of depressive and anxiety disorders. However, as already mentioned, the same genes seem to contribute to the development of neuroticism and to later anxiety and depressive disorders.

Social environment

There has been considerable interest in the role of marital status as a risk factor for major depression. For men, it appears clear that married men have the lowest rate of depression, while separated or divorced men have the highest rates of major depression. In women, the association is slightly less clear, but in the ECA study the same findings applied for women as for men. Understanding the nature of the association between marital status and rates of depression is more problematic. If personality is a risk factor for depression, then the same traits could interfere with the ability to marry or to stay married. There is little doubt that depression sometimes contributes to marital maladjustment and separation or divorce. Finally, the stresses associated with divorce and separation could increase the likelihood of an episode of depression occurring.

In the classic and influential work of George Brown on working-class women, (24) having three or more children, a lack of paid employment, and the lack of a confidant were risk factors for the development of an episode of depression. Subsequent studies have inconsistently replicated the risk factors of having children or lack of paid employment, but have supported the finding that the lack of a confidant increases the risk of depression.

It is well established that adverse life events, particularly those characterized by loss, increases the risk of an episode of major depression. The increased vulnerability to an episode appears to last for a period of 2 to 3 months following such an event. (25,26)

Early thinking about depression suggested that there would be those depressions that occurred for largely biological reasons and those precipitated by adverse life events; however, it is now clear that such a dichotomous view is incorrect. Kendler et al. (26) showed that there is a significant genotype by environment interaction in the prediction of onset of major depression. They proposed that genetic factors influence the risk of onset of depression, in part, by increasing the sensitivity of individuals to the depression-inducing effects of stressful life events.

Physical illness

Having a chronic or severe physical illness is associated with an increased risk for depression. The mechanisms behind this increased risk may vary depending upon the physical disorder. In disorders such as Parkinson's disease, it is possible that there are shared neurotransmitter abnormalities between Parkinson's disease and depression. In post-stroke depression, there is good evidence that the location of the lesion, at least in part, contributes to the rate of depression, which suggests a neuroanatomical/neurotransmitter connection between the physical illness and the likelihood of depression. For non-central nervous system disorders, such as acute myocardial infarction, diabetes, and cancers, the mechanism for this association is less clear. However, at least in the case of patients with cancer, most do not suffer from major depression and, if they do, the key risk factors are family history and a past history of depression. This suggests that the stress associated with a serious or chronic physical illness may act by bringing out an individual's lifetime vulnerability to depression.

An integrated aetiological model

The ultimate purpose of studying risk factors for depressive disorders is to contribute to the development of an integrated aetiological model. The most promising research in this area has been performed by Kendler and colleagues on women from the Virginia Twin Register. (27) In this study, 680 female–female twin pairs of known zygosity were assessed on three occasions at longer than 1-year intervals. Nine predictor variables (genetic factors, parental warmth, childhood parental loss, lifetime traumas, neuroticism, social support, past depressive episodes, recent difficulties, and recent stressful life events) were examined to see how they contributed prospectively to the development of an episode of major depression over the next 12 months. In considering the results from this study, it is important to bear in mind the limitations of this landmark study—including the facts that the sample was limited to women and that variables such as marital status, the presence of young children in the home, and a history of non-affective psychiatric disorders were not included in the analyses. The model was also predicting the onset of an episode over 12 months and not predicting lifetime episode, when a different combination of variables may have been found. However, despite these caveats, Kendler and colleagues developed a model that predicted 50 per cent of the variance in the liability to develop major depression. The four strongest predictors in order of liability to depression were as follows:

  • stressful life events
  • genetic factors
  • previous history of major depression
  • neuroticism.

It is of note that some of the risk factors exerted these effects directly, while other effects were largely indirect. Thus, 60 per cent of the effect of genetic factors on liability to depression was direct, but the remaining 40 per cent was indirect and largely mediated by past episodes of depression, stressful life events, and neuroticism. Variables such as perceived parental warmth had no direct effect on liability to develop an episode of major depression, but did impact upon neuroticism, a history of a past depressive episode, recent difficulties, and lifetime traumas.

The most comparable prospective studies looking at risk factors for the development of major depression have been undertaken during the postpartum period, which is a time of increased risk of depression. In this special case, the most consistent risk factors are family history and a past history of depression, and there is lesser support for a lack of social support, neuroticism, and complications during childbirth.

As one of the key tasks of epidemiology is to contribute to an understanding of aetiology, models that integrate risk factors are important strategies for further research. They provide clinicians with predictive power, and can also guide intervention studies to prevent the onset of episodes of depression.

Comorbidity

One of the important contributions of epidemiology to the study of mood disorders over the past 15 years has been the recognition of the extent to which depression and other psychiatric disorders are often comorbid. In both the ECA study and NCS, over two-thirds of all individuals identified as having an episode of major depression also met the criteria for one or more other psychiatric disorders. Not surprisingly, the most common comorbid disorders are anxiety disorders and substance-abuse disorders. In the NCS, the anxiety disorders with the highest odds ratios indicating comorbidity were generalized anxiety disorder, panic disorder, and post-traumatic stress disorder. It is also important to note that for most anxiety disorders, with the exception of panic disorder, the anxiety disorder usually predates the onset of the depressive disorder. (28) This is of considerable importance, as the risk factors for pure major depression differed from the risk factors for comorbid major depression. Furthermore, the cohort effects of increasing rates of major depression were largely attributable to increasing rates of comorbid major depression, rather than to increasing rates of pure major depression. These results raise important issues for prevention, as it may well be that targeting young people with anxiety disorders could be a major step to the prevention of the development of later major depressive disorders.

The second key area of comorbidity with major depression is with alcohol dependence. Data from the Virginia Twin Register suggest that part of this comorbidity is due to shared genetic factors, although there is also a smaller common environmental risk factor to both disorders.

Another area of considerable comorbidity with major depression are the personality disorders. The comorbidity between major depression and these disorders is receiving considerable attention in clinical samples but, to date, there are only limited data in epidemiological samples on the importance of these patterns of comorbidity.

Use of health services

One of the major challenges for psychiatry presented by epidemiological studies of depression has been the consistent finding that the majority of cases of depression in the community are neither recognized, diagnosed, nor treated. In the ECA study, it was found that 65 to 70 per cent of people with depression had visited a health professional in the last 6 months, but only 15 to 20 per cent had had a visit for a mental health reason and only about 10 per cent had seen a mental health specialist. Ormel et al. (29) found that patients with depression who present with largely somatic rather than psychological symptoms are extremely unlikely to be recognized by general practitioners. Similar findings have also been reported in England. (30) Even if major depression is recognized in the primary care setting, it is often not adequately treated. Even when major depression is treated in primary care, there is evidence to suggest that the outcome is worse if treated by the primary physician than when treated by a mental health specialist. (31)

The findings on the non-recognition and low levels of treatment in primary care have prompted national efforts in the United States (Depression Awareness Recognition and Treatment) and the United Kingdom (with a joint project between the College of Psychiatrists and the College of General Practitioners) to improve the recognition and treatment of depression. Given the importance of depression as the leading risk factor for suicide, many countries interested in lowering suicide rates are targeting the recognition and treatment of depression in the community. The Global Burden of Disease study is further evidence of the impact of mood disorders on our populations.

Top

References

1.Murray, C.J.L. and Lopez, A.D. (1996). The global burden of disease and global health statistics. Harvard University Press, Boston, MA.

2.Akiskal, H.S. (1996). The prevalent clinical spectrum of bipolar disorders beyond DSM-IV. Journal of Clinical Psychopharmacology, 16, 4s–14s.

3.Blacker, D. and Tsuang, M.T. (1992). Contested boundaries of bipolar disorder and the limits of categorical diagnosis in psychiatry. American Journal of Psychiatry, 149, 1473–83.

4.Kessler, R.C., Rubinow, D.R., Holmes, C., Abelson, J.M., and Zhao, S. (1997). The epidemiology of DSM-III-R bipolar I disorder in a general population survey. Psychological Medicine, 27, 1079–89.

5.Weissman, M.M., Bland, R.C., Canino, G.J., et al.(1996). Cross national epidemiology of major depression and bipolar disorder. Journal of the American Medical Association, 276, 293–9.

6.Joyce, P.R. (1984). Age of onset in bipolar affective disorder and misdiagnosis as schizophrenia. Psychological Medicine, 14, 145–9.

7.Kendler, K.S. and Gardner, C.O. (1998). Boundaries of major depression: an evaluation of DSM-IV criteria. American Journal of Psychiatry, 155, 172–7.

8.Wainwright, N.W.J., Surtees, P.G., and Gilks, W.R. (1997). Diagnostic boundaries, reasoning and depressive disorder, I. Development of a probabilistic morbidity model for public health psychiatry. Psychological Medicine, 27, 835–45.

9.Kendler, K.S., Neale, M.C., Kessler, R.C., Heath, A.C., and Eaves, L.J. (1993). The lifetime history of major depression in women. Reliability of diagnosis and heritability. Archives of General Psychiatry, 50, 863–70.

10.Sullivan, P.F., Kessler, R.C., and Kendler, K.S. (1998). Latent class analysis of lifetime depressive symptoms in the National Comorbidity Survey. American Journal of Psychiatry, 155, 1399–406.

11.Kendler, K.S., Eaves, L.J., Walters, E.E., Neale, M.C., Heath, A.C., and Kessler, R.C. (1996). The identification and validation of distinct depressive syndromes in a population-based sample of female twins. Archives of General Psychiatry, 53, 391–9.

12.Weissman, M.M., Leaf, P.J., Tischler, G.L., et al.(1988). Affective disorders in five United States communities. Psychological Medicine, 18, 141–53.

13.Bland, R.C., Orn, H., and Newman, S.C. (1988). Epidemiology of psychiatric disorders in Edmonton. Acta Psychiatrica Scandinavica, Supplement, 1–80.

14.Oakley-Browne, M.A., Joyce, P.R., Wells, J.E., Bushnell, J.A., and Hornblow, A.R. (1989). Christchurch Psychiatric EpidemiologyStudy, Part 2. Six month and other period prevalences for specific psychiatric disorders. Australian and New Zealand Journal of Psychiatry, 23, 327–40.

15.Blazer, D.G., Kessler, R.C., McGonagle, K.A., and Swartz, M.S. (1994). The prevalence and distribution of major depression in anational community sample: the National Comorbidity Survey. American Journal of Psychiatry, 151, 979–86.

16.Jenkins, R., Lewis, G., Bebbington, P., et al. (1997). The National Psychiatric Morbidity Surveys of Great Britain—initial findings from the household survey. Psychological Medicine, 27, 775–89.

17.Klerman, G.L. and Weissman, M.M. (1989). Increasing rates of depression. Journal of the American Medical Association, 261, 2229–35.

18.Cross National Collaborative Group (1992). The changing rate of major depression. Cross national comparisons. Journal of the American Medical Association, 268, 3098–105.

19.Kendler, K.S., Neale, M.C., Kessler, R.C., Heath, A.C., and Eaves, L.J. (1992). Major depression and generalized anxiety disorder. Same genes (partly) different environment. Archives of General Psychiatry, 49, 716–22.

20.Andrews, G., Stewart, G., Allen, R., and Henderson, A.S. (1990). The genetics of six neurotic disorders: a twin study. Journal of Affective Disorders, 19, 23–9.

21.Wilhelm, K., Parker, G., and Hadzi-Pavlovic, D. (1997). Fifteen years on: evolving ideas in researching sex differences in depression. Psychological Medicine, 27, 875–83.

22.Tennant, C. (1988). Parental loss in childhood. Its effect in adult life. Archives of General Psychiatry, 45, 1045–50.

23.Parker, G. (1983). Parental ‘affectionless control' as an antecedent to adult depression. A risk factor delineated. Archives of General Psychiatry, 40, 956–60.

24.Brown, G.W. and Harris, T. (1978). Social origins of depression: a study of psychiatric disorders in women. Tavistock Publications, London.

25.Paykel, E.S. (1978). Contribution of life events to causation of psychiatric illness. Psychological Medicine, 8, 245–53.

26.Kendler, K.S., Kessler, R.C., Walters, E.E., et al.(1995). Stressful life events, genetic liability and onset of an episode of major depression in women. American Journal of Psychiatry, 152, 833–42.

27.Kendler, K.S., Kessler, R.C., Neale, M.C., Heath, A.C., and Eaves, L.J. (1993). The prediction of major depression in women: toward an integrated etiological model. American Journal of Psychiatry, 150, 1139–48.

28.Kessler, R.C., Nelson, C.B., McGonagle, K.A., Liu, J., Swartz, M., and Blazer, D.G. (1996). Comorbidity of DSM-III-R major depressive disorder in the general population: results from the US National Comorbidity Survey. British Journal of Psychiatry, 168, 17–30.

29.Ormel, J., Koeter, M.W.J., Van den Brink, W., and van de Willige, G. (1991). Recognition, management and course of anxiety and depression in general practice. Archives of General Psychiatry, 48, 700–6.

30.Gater, R. and Goldberg, D. (1991). Pathways to psychiatric care in South Manchester. British Journal of Psychiatry, 159, 90–6.

31.Schulberg, H., Block, H.R., Madonia, M.J., et al.(1996). Treating major depression in primary care practice, eight-month clinical outcomes. Archives of General Psychiatry, 53, 913–19