The influence of age on the female/male ratio of treated incidence rates in depression
Poor data exist on the influence of psychosocial variables on the female/male ratio of depression because of the small number of cases and the resulting limited numbers of variables available for investigation. For this investigation a large number of first admitted depressed patients (N = 2599) was available which offered the unique opportunity to calculate age specific sex ratios for different marital and employment status categories.
Age and sex specific population based depression rates were calculated for first ever admissions for single year intervals. Moving averages with interpolated corrections for marginal values in the age distribution were employed.
For the total group the female/male ratio of depression showed an inverted U-shape over the life-cycle. This pattern was influenced by the group of married persons, which showed a sex-ratio of 3:1 between the age of 30–50, but ratios of around 1:1 at younger and older ages. For not married persons the female/male ratio was already around 2:1 at the age of 18 and rose to 2.5:1 in mid-life and declined to 1 at around 55. The almost parallel decline of depression rates in employed men and women resulted in a female/male ratio of about 2:1 from age 18 to age 50 and became 1 after the age of 60. The female/male ratio among the not employed was about 1, in mid-life it became negative.
Our analyses show that the gender-gap in first admitted depressed patients is age dependent and that psychosocial factors modify the sex ratio.
KeywordsEmployment Status Married Person Epidemiologic Catchment Area Psychosocial Mechanism Involutional Melancholia
In reviewing published studies on age specific sex differences in depression, Jorm  concluded that the broad claim that depression is more common in women than in men is an oversimplification. He found that an inverted U-shape of the age curve of the female/male depression ratios fit the data best, reflecting the finding that the difference between male and female rates are rather small in young age, increase towards mid-life and decrease thereafter.
More recent studies confirm this finding. Anthony and Petronis  have shown in an ECA-Follow-up study that the incidence rate of depression rises towards midlife in women but not in men. In Australia, Wilhelm et al.  found an increasing female preponderance of life time prevalence in depression rates when following-up a cohort of 170 teacher trainees from age 23 (when the female/male ratio was about 1:1) to age 38, when it became 1.5:1. Jorm's observation of a narrowing down of the gender gap after mid-life is supported by the accumulating evidence that – contrary to common belief- depression in women becomes less frequent after the menopause [4, 5, 6]. Also, in the US National Comorbidity Survey (NCS)  which provides age specific 12 months prevalence rates in five year categories up to the age of 55, an increasing gender gap was found towards mid-life with a decrease thereafter. In the earlier Epidemiologic Catchment Area Study (ECA) hardly any gender difference was observed after the age of 55 . Bebbington et al  in a representative survey also confirmed, that the female preponderance in depression is evident only before the age of 55 due to an absolute fall in female prevalences thereafter. Der and Bebbington  found a decreasing gender gap in first admission depressives after the age of 55 (55–64 = 1.3; 65–74 = 1.2; 75 += 1.0) and Meltzer et al.  reported for the British National Survey that after the age of 55 the sex ratio changes to a male preponderance. Beyond the ages of 65 and 75 the situation seems to be less clear. While some investigators did not observe any gender differences among the very old [12, 13], recently Prince et al.  found a significant excess of women in a survey of 21.724 subjects from 11 countries aged 65 and over.
At first sight, age seems to offer an intriguing possibility to narrow down the number of explanations for the gender gap. Since hormonal activity is clearly related to age, one interpretation of the rise and fall of the female/male ratio over the life cycle could be that female sex hormones might contribute to the higher female rates. However, social roles -especially marital and employment status – are also changing over the life cycle, as Mirowsky  has pointed out. Therefore only the combined analysis of age and marital and employment status could shed some light on the mechanisms which influence the rise and fall of the gender gap with advancing age.
In a preceding paper analysing first ever in-patient admission rates for depression, we have shown that the generally reported female to male ratio of 2:1 can be higher in specific marital and employment status categories, but also much smaller, that it may disappear or even become reverse . The latter pattern was observed in the employed widowed persons, suggesting that age might be a further intervening variable. While marital and employment status probably exert their effect through psychosocial mechanisms, the interpretation of the impact of age is more complex, since it may influence the risk of depression both by biological and psychosocial mechanisms.
Poor data exist on the combined influence of age, marital or employment status on the female/male ratio of depression. The reason is that most epidemiological studies identify only relatively small absolute numbers of cases and are therefore limited in the number of variables they can study simultaneously. Usually they have to compromise between the breadth of age categories and the number of additional variables studied. The study by Anthony and Petronis , while taking many variables into consideration, suffers from a small sample (N = 193) of depressed patients, and conclusions, especially about the interaction of the many variables must remain inconclusive. In contrast, for the present investigation a large (population based) number of first admitted depressed patients (N = 2599) was available which offered the unique opportunity to calculate age specific sex ratios for different marital and employment status categories.
Up to mid 1974 all inhabitants of the city of Vienna requiring psychiatric in-patient treatment were admitted to one central hospital, the Department of Psychiatry of the University of Vienna. On January 1 1971, a computerised case record system for in-patients (which is still in operation today)  was set up, which allows to register – in addition to demographic data – a number of clinical variables in a structured way. Up to mid 1974, after a thorough diagnostic check-up (applying ICD-8 criteria) and acute treatment, most patients were referred to a large mental hospital after some days. Since admission policy was strictly regulated according to administrative boundaries and since there existed no other facilities for psychiatric in-patient treatment in or near Vienna, this computerised case record system offers the unique opportunity to analyse a large representative population of in-patient admissions. In the second half of 1974, in-patient admissions were no longer centralised at the Department of Psychiatry. Therefore, our analysis covers the 42 months between 1st January 1971 and 30th June 1974.
Only patients fulfilling ICD-8 diagnostic criteria for 300.4 (neurotic depression) or 296.0/296.2 (involutional melancholia, manic-depressive psychosis-depressed type) and having been admitted for the first time ever to the Department of Psychiatry were included in the present analysis. In ICD-8 the term "manic-depressive" was used for both unipolar and bipolar affective disorders, but ICD 296.2 exclusively refers to unipolar depression. (ICD-8 criteria for depression are largely similar to ICD-9 criteria which are still in use today in Austria). The population census of 1971 (1,614.841 inhabitants) allowed to calculate age and sex specific rates for first ever admissions per 100.000 of the Viennese population for single age intervals. The age range studied was confined to the five decades between 18 and 67 years, since absolute numbers were small below and above these limits. Thus, our study does not relate to the very old. Marital and employment status for admitted patients were recorded in the same way as in the population census. It was therefore possible to calculate age and sex specific population based depression rates also for these two variables. For the present analysis "employed" covers both, full-time and part-time employment; "not employed" is a residual category including those currently not on a payroll, i.e. unemployed, but also students, houseworkers and retired persons.
The large number of patients even allowed the use of single year age intervals for representing the results as graphs for age specific female and male rates as well as for the female/male sex ratios.
For this purpose the method of "moving averages" with interpolated corrections for marginal values in the age distribution was employed. The moving average was calculated by averaging two points on either side of a target value. Thus, the length of the moving average was five. This smoothing technique minimises the effect of irregular or random variations. 95 % confidence limits are also shown in the graphs, in order to be able to observe deviations from the null hypothesis of no differences between the female and the male rates (i.e. from a ratio of 1). Conclusions are drawn from inspecting the course of the specific curves and the confidence limits over the life time and from employing one-tailed t-tests for comparisons between the age groups of 18–34 ("young") and of 35–51 ("middle aged") (increase of the female/male ratio?) and of 35–51 ("middle aged") and 52–67 ("old") (decline of the female/male ratio?).
Sex and age
Marital status, sex and age
With few exceptions, married persons had lower rates than not married persons. These exceptions included very young men (up to the age of 25) and women beyond the age of 50 years where no or only minor marital status differences were found.
It can be concluded so far, that, if marital status is considered, the inverted U-shape of the age specific sex ratio found in the total population was more pronounced among married persons and much less so or nearly non-existent among not married persons.
Employment status, sex and age
The findings of our detailed analysis are not easy to interpret, first of all, because other variables changing over the life cycle (like hormonal status, child care, social conditions) were not assessed. The fourfold interaction between sex, age, marital and employment status could not be analysed neither, because numbers would have become too small. Finally, marital status and employment status may act as factors influencing referral. Concerning the latter problem, it has to be stressed though, that we exclusively analysed first ever admissions from a strictly defined catchment area and that at the time period for which data were collected the psychiatric out-patient care system was underdeveloped in Vienna. If nothing else, our data can show what might happen to the sex ratios of incidence rates if marital and employment status were considered together with age.
With these caveats in mind, we can first confirm the finding by Jorm  of an inverted U-shape of the female/male ratios of depression rates over the life cycle. While being around 1.5:1 both in young adulthood and around the age of 60, the female/male ratio increased towards mid-life, with a peak at 45 to 50 and declined thereafter. Thus, our data confirmed the results of other studies quoted in the introduction of the present paper [4, 5, 6, 9], showing that depression in women becomes less frequent after the menopause. Eaton et al.  in an incidence study starting from ECA data also found an inverted U-shape for female depression rates.
The two interpretations which come immediately to one's mind are biological on the one hand (female rates are increasing during the hormonally active period of women and decrease thereafter) and psychosocial on the other hand (life stress is increasing for women with middle age and decreases thereafter), which might mutually influence each other. No decision, however, is possible between these interpretations without considering other variables.
When analysing the additional influence of marital status it becomes clear that it was marriage which largely accounted for the inverted U-shaped curve. In married persons the female/male ratios climbed early (at around 30) to the 3:1 level, stayed there until the age of fifty and declined sharply thereafter. In other words: among the married the inverted U was broad, while in the total population it was rather a V than an U.
Thus, while among the married, rates were usually lower than among the not married, the gender gap was larger at a much earlier age than in the total population. This favours a psychosocial interpretation (marriage being more stressful for women than for men), but still, a biological interpretation is not ruled out, since married women may rather have children -and child bearing is related to hormonal changes (there was no possibility in our analyses to find out how frequently post-partum depression occurred).
Among the not married (taken as a group, including the never married, the divorced and the widowed) the inverted U-shape curve was found only in a rudimentary form, with a decline after the age of 40 (from around 2.8:1 to 1:1 at the age of 50 and later). It is difficult to interpret this curve, also if considering that the subgroup of the never married corresponded to the total group of the not married (for divorced and widowed persons the analysis could not be carried out due to the small numbers of cases in these subgroups if divided into single age groups.)
Employment status also changed the general pattern of an inverted U-shape found for the total population to a substantial degree. Among the employed the female/male ratio was constant (below 2:1) from the age of 18 to 50, and dropped thereafter to 1:1. The inverted U-shape virtually disappeared in this analysis. It seems that up to the age of fifty the raised risk for employed women was not age-dependent on age; the female preponderance tended to disappear beyond the age of 50. The not employed showed a pattern which was the opposite to that observed in the total population – there is a U-curve – but not an inverted one, but an upright U, i.e. in midlife not employed men had a clearly higher risk than not employed women.
The employment findings clearly favour a psychosocial explanation of the variations of the female/male depression rates over the life cycle as opposed to a biological one. If men were not employed in midlife, the risk of being admitted for a first ever depression episode was much higher than that for women. Data support the hypothesis that unemployment increases the risk for psychological disturbances- especially for affective disorders – in both men and women [2, 19, 20, 21, 22]. The reverse sex ratio found in the not employed group of our investigation may reflect differences in the demands and social role expectations connected to the specific social roles of men and women. Unemployment may undermine menus ability to fulfil their role obligations. For married men unemployment has additional implications for their family status and relationships. In this context, it has to be stressed that women's and men's labour market position and role in the family have undergone substantial changes since the 1970s and the impact of psychosocial mechanisms on both women and men might have changed during the last decades.
Our analyses show, that the female/male ratio in first ever admitted depressed patients was age dependent and that there is a complicated interaction with marital and employment status. As the presented data are cross-sectional, the age group differences could as well rather reflect cohort than ageing effects. While we cannot be sure whether these findings are relevant for the aetiology of depression or only reflect referral peculiarities, they suggest that incidence studies should be carried out in unselected community populations considering sex, age marital and employment status simultaneously. Only in doing so we can narrow down the number of possible psychosocial and biological explanations of the gender gap in depression.
- 8.Robins LN, Regier DA: Psychiatric Disorders in America: The Epidemiologic Catchment Area Study,. Free Press, New York,. 1991Google Scholar
- 11.Meltzer H, Gill B, Petticrew M, Hinds K: The prevalence of psychiatric morbidity among adults living in private households. OPCS Survey of Psychiatric Morbidity in Great Britain. Report 1. HMSO, London,. 1995Google Scholar
- 16.Gutiérrez-Lobos K, Wölfl G, Scherer M, Schmidl-Mohl B: The gender gap in depression reconsidered: The influence of marital and employment status on the female/male ratio of treated incidence rates. Soc Psychiatry and Psychiatric Epidemiology. 2000, 35: 202-210. 10.1007/s001270050229.CrossRefGoogle Scholar
- 17.Katschnig H, Poustka F: Zur Sozialpsychologie des Dokumentationsverhaltens im Krankenhausbetrieb. In: Dokumentation und Information im Dienste der Gesundheitspflege. Edited by: Nacke O, Wagner G. 1976, Stuttgart-New York, Schattauer, 67-80.Google Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-244X/2/3/prepub
This article is published under license to BioMed Central Ltd. This is an Open Access article: verbatim copying and redistribution of this article are permitted in all media for any purpose, provided this notice is preserved along with the article's original URL.