Introduction
Depressive disorders are common [
1]. They are associated with increased health hazards and mortality, impact on other disorders, and are predicted to be the leading cause of disease burden by 2030 [
2].
In DSM-III and DSM-IV mood disorders are defined according to a unipolar–bipolar dichotomy and major depressive disorder (MDD) is the central construct [
3,
4]. Melancholia, signified by lack of mood reactivity, psychomotor retardation and diurnal variation, as defined by Schneider [
5], is included as a specifier of MDD, but is not nosologically central. MDD is heterogeneous [
6] and the etiology is not established [
7]. Some critics argue that MDD is a watered-down entity due to the lumping together of states that have nonspecific depressive features in common [
8]. The critics have suggested that melancholia should be viewed as the central mood disorder, and that bipolar disorder, depression with psychotic features, catatonic depression, puerperal depression and abnormal bereavement are instances of melancholia [
8]. They argue that there is an overlap between these disorders in symptomatology, neuroendocrine variables and genetics [
8]. The symptomatology of melancholia is characterised by pathological mood with unrelieved gloom and apprehension that colours cognition and self-experience, resulting in preoccupation with thoughts of worthlessness, hopelessness, guilt and suicidal ideation. About 30% of melancholia patients are psychotic [
8]. Melancholia is also characterised by psychomotor change, either as retardation, agitation or reduced reactivity, and vegetative dysfunction affecting, for example, sleep, appetite and sex-drive [
8].
Given the suggestion to re-establish melancholia as the core mood disorder, the epidemiology of the DSM mood disorders dichotomised into non-melancholic and melancholic disorders is interesting. However, since 1980, most community studies of mood disorders have been restricted to MDD, dysthymic disorder and bipolar disorder.
An insufficiently understood phenomenon is the preponderance of depression among females [
9].
Substantive factors that may predispose differently to depression are gender-bound social roles, previous mental disorders, premorbid personality, and biological and genetic factors [
10]. Gender roles may influence self-affectivity, self-esteem and the development of an externalising or internalising, e.g. ruminative, coping style, affecting vulnerability. The normative gender-role relevance of a precipitant, e.g. a stressful life event, may modulate the depressogenic response. Gender differences in prior anxiety are associated with the depression gender gap [
9]. Personality dimensions, associated with depression proneness may differ. Neuroticism is associated with depression proneness and has a stronger impact in females [
11], and has been suggested to be influenced by reproductive hormones, conferring greater responsiveness to emotional stressors [
11]. Biological differences between the genders in stress reactivity and emotional regulation may also explain the gender gap in depression [
11]. Some studies support the view of a moderately higher heritability of liability to depression in females [
12]. Artefactual determinants that may explain the gender gap in depression include that females may seek help and report depressive symptoms more often, and a diagnostic bias towards women. The current concept of depression may also be more adapted to a female mode of emotional disorder presentation [
10].
In a review, the pooled overall annual incidence of MDD was 2.9 per 100 persons [
13], regardless of gender. Some population studies have reported non-differing depression incidence rates, and some cross-sectional studies have found the gender difference in depression to be absent or small [
9]. However, most incidence studies of depression have shown significantly higher overall rates in females [
14] with estimated 1.5- to 3-fold higher rates of MDD [
15]. The overall incidence rate of bipolar disorder in population studies has ranged between 0.13 and 0.53 per 100 person-years at risk, without significant gender differences [
14,
16], and in most cross-sectional population studies bipolar disorder has also been equal [
13]. In one population study, females had a higher lifetime prevalence of both DSM-IV non-melancholic and melancholic depressive disorders [
17], but in a national register study the prevalence of a first-ever ICD-10 depressive episode was about twice as high in females, but no gender differences were found in melancholic or psychotic depression [
18].
A peak age of risk for first onset of major depressive episode (including bipolar depression) is estimated to range from mid-late adolescence to the early 40 s [
19]. The mean or median ages of onset of DSM-III/III-R/IV MDD, or ICD-10 depression have ranged between approximately 20 and 35 years, without significant gender differences [
1]. The recalled mean or median ages of onset of bipolar disorder in treated or population samples have ranged from 18 to 33 years, without significant gender differences [
20]. According to an older review [
21], there had been no population study on the ages of onset of non-melancholic and melancholic depressive disorder. In a more recent study in female inpatients aged 30–60 years with DSM-IV recurrent MDD, there was a minor difference in onset; the mean age of onset in the non-melancholic cases was 34.3 years and in the melancholic 36.2 years [
22]. However, other findings indicate that younger age may be inversely related to melancholia [
23], although in a review of studies of differences between early- and late-onset depression, no evidence of differences in psychotic symptoms or psychomotor retardation was found [
24].
Although average ages of onset of depression do not differ between the genders, findings indicate that the female preponderance in depression is age specific [
25], emerges in puberty [
30], and increases with age up to middle life [
9,
10,
25]. Some studies have shown that the gender gap in depression diminishes after the age of 50–55 [
26], but others suggest that it persists [
10].
The Lundby Study is an investigation of the mental health in a total population that was monitored from 1947 to 1997 [
27]. The Lundby Study was initiated by Erik Essen-Möller (1901–1992) as an investigation of the distribution of personality traits, mental disorders and their possible forerunners in an ordinary, general and unselected population. The Lundby Study was originally meant to be a cross-sectional study [
28], but it developed into a longitudinal investigation [
27]. The original subjects recruited were everyone on the parish registers of the two adjoining rural parishes that comprised the Lundby area on 1 July 1947.
In a previous study, it has been shown that for depression, broadly defined, the female incidence rate is higher than the male [
29]. In the present study, we aimed to investigate whether any gender differences have emerged during the 50-year follow-up in terms of the overall first incidence rates, average ages of first onset, or incidence rate by age of first onset patterns in different groupings of disorders, with depressive features focusing on: (1) severity of depression, (2) DSM-IV disorder subtype, and (3) non-melancholic and melancholic depression.
Materials and methods
Study area, population and case identification
The Lundby area surrounds a village in the south of Sweden. The study had intakes on 1 July 1947 (
N = 2550) and 1957 (
N = 1013), when all inhabitants including newcomers were recruited (Table
1). At inception, the subjects (
N = 3563) were between 0 and 95 years old (median 31 years). No new subjects have been added since 1957. The subjects were followed up in field surveys, regardless of residence, in 1957, 1972 and 1997.
Table 1
Sources of information for case finding and attrition rate by year of field study, and alive and deceased subjects in the total Lundby population (N = 3563)
1947 | 2550 | 2520 | 13 | 17 | 0 | 0 |
1957 | 3563a
| 3260 | 31 | 19 | 233 | 20 |
1972 | 3310 | 2777 | 46 | 4 | 481 | 2 |
1997 | 2827 | 1559 | 82 | 156 | 1018 | 12 |
Considerable societal changes took place during the follow-up in the rural area, including development into semi-rural/suburban character, a shift from farming to industry and service professions, and an increasing location of places of employment outside the area [
27]. More than half of the population moved out from the Lundby area during the study.
The field studies generated data from semi-structured interviews, informants (e.g. relatives and nursing staff), registers and case records. Psychiatrists conducted and evaluated all four field investigations. The interviews, which contained itemised checklists of observed behaviours and subjective reports, retained the same form throughout the study. Key points in the semistructured interviews were physical and mental health, contacts with medical services, primary care, psychiatric care, somatic and mental illnesses, complaints, medication, smoking habits, alcohol and substance use and women’s health. The social situation was investigated with questions about satisfaction with life, work, the emotional climate in the family and the relationship with partner. The itemised checklists focused on observable behaviours (affective and vegetative reactions) such as tension, gloominess, torpidity and sensitivity, and questions about personality traits and habitual dispositions/symptoms (mostly related to affect) such as “Are you of a nervous disposition?”, “Do you cry easily?”, “Do you get tired easily?”, “Are you easily hurt?” and “Do you feel unjustly treated?”. Discussions were also held, and the psychiatrists wrote down their impressions. Registers included nearby hospital archives, regional registers and a national patient register. Case records included in- and outpatient records from general practitioners, somatic and psychiatric clinics.
Data permitted evaluation in 99% of the study subjects between 1947 and 1972, and 94% between 1972 and 1997 (Table
1). Attrition was lower in males (5%) compared to females (7%) during the period 1972–1997 with drop-outs more common in age groups under 50 years, varying between 6–8% in males and 7–13% in females [
27].
Diagnostic assessment
Lundby depression consists of ‘depression proper’ and ‘depression plus other symptoms’. ‘Depression proper’ includes lowered mood, feelings of low vitality, lowered enjoyment of life, lack of initiative, reduced activity, inhibition, retardation, sleep disturbances, loss of appetite and weight, anxiety and fear, reduced self-esteem, guilt feelings, and diurnal variation. Depression with the addendum ‘plus other symptoms’ refers to states that are also essentially depressive, but that, in addition to depressive symptoms, also display other symptoms, e.g. somatisation or delusions, which are consistent with depression.
Lundby depression (proper and plus) lumps together mild, reactive and atypical depressive states with profound endogenous depression with inhibition, retardation and psychosis. However, in conjunction with the diagnostic assessment, the severity was classified. Every episode with depression was scored in terms of the degree of impairment that the depression was judged to cause, in accordance with Leighton et al. [
30]. A mild degree of impairment, which means that daily work is usually possible but with lowered achievement, roughly corresponds to a GAF score between 70 and 61; medium impairment is roughly 60–51; severe impairment, which practically always involves a marked reduction in functional capacity or a total inability to work, is 50–31; very severe impairment, which would include, for example, depression with retardation or delusions, is 30–1 [
4,
27].
After the 1997 field investigation, consensus DSM-IV diagnoses, including mood disorder, depressive disorder, MDD, dysthymic disorder, depressive disorder not otherwise specified (NOS), bipolar depression, other mood disorder (mood disorder due to a general medical condition and substance-induced mood disorder) and adjustment disorder with depressed mood, were assessed alongside Lundby diagnoses for the period 1972–1997. The impairment degree medium was chosen as threshold for caseness. Subsequently, all first episodes with Lundby depression (proper and plus) with medium, severe or very severe impairment 1947–1972 were diagnosed according to the DSM-IV, using all available information. To study depression by severity, the threshold for caseness of Lundby depression was moved from medium to severe and very severe, respectively. To study non-melancholic and melancholic depression, groups were constructed with the aid of the melancholia concept according to Taylor and Fink [
8]. Melancholic depression included MDD with melancholic and/or psychotic features and/or catatonic features, bipolar depression, puerperal depression and abnormal bereavement.
Statistical procedures
To study the first incidence rate and age at first onset of depression, a risk sample was defined by excluding from the total population subjects who, already before intake in the study, had suffered a depression or fallen ill with schizophrenia or dementia. Incidence rates for first episodes (IR) were obtained as the number of first occurrences of a disorder in subjects aged 15 years or more divided by the total number of person-years under risk for that disorder. Female/male differences of IR and mean age of onset were tested by constructing 95% confidence intervals (CI) for the female/male IR ratios and age of onset differences, respectively [
31].