2 MRC Social and Community Psychiatric Unit, The Institute of Psychiatry. London, UK ..... psychiatric morbidity, in females the association was sig nificant in the ...
Soc Psychiatry P~ychiatr Epidemio! (19SJ2) 27: 62-68 © Springer-Verlag 1992
Social Psychiatry and Psychiatric Epidemiology
Sex differences in mental illness: a community study of the influence of physical health and sociodemographic factors J. L. Vazquez.Rarquero 1, J. F. Diez Manrique', J.Muiiozl, J. Menendez Arango\ L. Gaite\ S. Herrera", and G. J. Der2 1
Social Psychiatry Research Unit of Cantabria, University Hospital "Marques de Valdccilla", University of Cantabria, Santander, Spain MRC Social and Community Psychiatric Unit, The Institute of Psychiatry. London, UK
Accepted: October 22, 1991
Summary. This paper examines sex differences in psy chiatric morbidity, using data from a community sample. The PSE-CATEGO-ID system was used to evaluate psy chopathology. Six sociodemographic factors and physical illness were taken as independent variables. Females ex hibit both a significantly higher psychiatric morbidity than males and a predominance of syndromes closely related to depression (SD; OD syndromes) and anxiety (GA; SA; TE; IT syndromes). Logistic modelling analysis, carried out separately for each sex, yielded different models. Psy chiatric illness in men was best predicted by physical ill ness, unemployment and the interaction between the two. In contrast, physical illness emerged, in women, as the only factor exerting significant effects on psychiatric mor bidity.
In the last few decades there has been growing evidence of the existence of significant differences between the sexes in the rates of specific mental disorders. This evidence derives from studies of psychiatric in-patient populations (Cochrane and Stopes-Roe 1981), of general practice pa tients (Shepherd et a1. 1966; Briscoe 1982; Williams et a1. 1986), of homogeneous "non-treated" samples (Jenkins 19K5), and also from community studies. In this last group two main methodological strategies of exploring psycho pathology have been developed in the last few years. The first centered on the use, by psychiatrists, of the PSE CATEGO-ID system (Wing et a1.1974), the other on the administration of the NIMH Diagnostic Interview Sched ule by lay interviewers (Robins et aJ. 1981; Regier ct al, 1988). As they present marked differences in the concepts of mental illness they use and in the mechanisms adopted for exploring psychopathology we will compare our re sults mainly with studies which, like the present one, apply the PSE-CATEGO-ID system (Bebbington et al. 1981; Henderson et a1. 1981; Mavreas et al. 1986;Lehtinen et al. 1990). The female excess of mental disorders reported in the different studies appears to be due mainly to neurosis and manic-depressive psychosis, while in men the tendency isto
produce higher rates of personality disorders, alcoholism and anti-social conduct (Dohrenwend and Dohrenwend 1976: Vazquez-Barquero et al. 1987; Regier et al. 1988). This sex specific pattern of psychopathology has to be taken into account when considering the findings obtained with an instrument which, like the PSE-CATEGO-ID, does not cover the disorders that are more common in men. Differences in psychiatric morbidity between males and females can also be shown by examining the distribu tion of psychiatric symptoms and syndromes. Jenkins (1985), using the GHQ and the CIS found more somatic symptoms, more fatigue, irritability and depression in women, while in men there was a greater tendency to re port lack of concentration. Similarly Mowbray (1972), in an in-patient study using the Hamilton Rating Scale, dem onstrated that females had significantly higher scores for somatic anxiety, general somatic symptoms and psychic anxiety. Different community studies using the PSE-9 have in vestigated syndrome profiles. Orley and Wing (1979) in their study in Uganda found no differences in the syn drome profiles of women and men. Henderson et al. (1981) found that women in Canberra reported more de pression, anxiety and irritability than men. Mavrcas and Bebbington (1988) in their comparison of the syndrome profiles obtained in the two independent surveys of Cam berwell and Athens showed that in both communities women tend to score higher in all syndromes and that the differences were especially marked in the "depression", "general anxiety", "situational anxiety", "tension", and "worrying" syndromes. In the last few years a variety of social factors have been examined in an attempt to explain the excess of certain psychiatric disorders in women and their utilization of health resources. These include, among others, maternal loss in childhood, unemployment, presence of children under the age of 14 at home, and low social and educa tional levels (Brown and Harris 1978; Wan and Parry 1982;Suttees et a1.1983;Bebbington et a1. 1984;Vazquez Barquero et a1. 19K7). Similarly the relevance of physical complaints to the higher female rates of mental illness and
Tahle I. Contact, response, and refusals in the two stages of the sur vey
Too m lo be interviewed
mately 9250 inhabitants. There is a mixture of dairy farm ing, fishing, light industry and tourist activities. This com munity was chosen as representative, within the region of Cantabria, of the non-urban coastal zone, where approxi mately 30 % of the total population of Cantab ria lives.
Not known (no information available)
Selection and study of the sample
In the first stage, a random sample of 1223 persons aged 17 years and over (581 males and 642 females), stratified by sex and age, was drawn from the electoral register. All members of the sample were interviewed in their homes using a variety of medical and social questionnaires. The GHQ-60 (Goldberg and Williams 1988) in its Spanish ver sion (Vazquez-Barquero et al. 1986), was used to assess the mental health ofthe first stage sample. The G HQ scor ing method was used and the "non-case" "case" cut off score of 11/12 applied (Goldberg and Williams 1988). The criteria for the selection of subjects to be included in the second stage were: i) all persons who in the first stage scored 12 or more in the GHQ-60; ii) an approxi mately equal number of persons who scored below 12; this second group was selected at random from two different batches of below GIIQ scores. The suhjects selected in this stage were interviewed at their homes using the ninth edition of the PSE (Wing et al. 1974). The PSE data were analyzed using the CATEGO program and the Index of Definition Il). The ID allocates each subject to one of eight levels of definition of symptomatology. LevelS rep resents the threshold category, and levels 6, 7, and 8 are definite cases. Levels 5 and above are conventionally re ported as "cases". CATEGO is then used to allocate a single diagnostic category to all threshold or above sub jects. It does this via 8 stages, in which the original clinical items are progressively combined. The second stage generates syndromes) which are themselves of a descrip tive nature. Table 1 shows that three hundred and forty five (28.2 % ) individuals were replaced for different reasons at the first stage; of these 98 (8 % ) openly refused to be inter viewed. This refusal rate increases to 13 % if we add the "never available" group. The majority of the remaining replacements were due to census deficiencies. The num ber of subjects replaced at the second stage are, as we see, very low. Thus, in general, our replacement figures are comparable to those of other community surveys (Vaz quez-Barquero et a1.1987).
Moved definitively out of the area
Temporarily away out ofihe area
Never available or difficult to contact Refusals Data incomplete from first stage Total replaced
medical consultation has been frequently demonstrated (Eastwood 1975; Howith 1977; Goldberg and Huxley, 1980), We have shown, in this respect, not only that the fe male predominance of mental illness is independently in fluenced by social factors and by the presence of physical illness, but also that the interaction between physical and mental disease is modified by sociodemographic variables (Vazquez-Barquero et a1. 1988). Despite all the evidence suggesting that sex differences in psychiatric morbidity may be accounted for by a com plex interaction of social factors and physical illness, com parative analysis of men and women is very rare (Briscoe 1982; Jenkins 1985; Zimmermann-Tansella et a1. 1990). The aim of this study is, therefore, to investigate the char acteristics of sex differences in psychiatric morbidity as defined by the PSE-CATEGO-ID system. To do this, data from The Community Survey of Cantabria was used (Vaz quez-Barquero el al. 1987), thus avoiding the selection bias derived from the process of establishing medical con tact,
Design The design selected was a two-stage cross-sectional anal ysis of a random sample of the population from a rural area of Cantabria. A detailed description of the design and of the characteristics of the sample has been given in previous publications (Vazquez-Barquero et al. 1986, 1987), but for the purpose of this study this can be sum marized as follow:
Area The area studied eonsists of three adjoining counties, on the eastern side of the bay of Santander. With an approxi mate area of 105.4 Km2 they have a population of approxi
Variables used in the analysis Psychiatric illness. The presence of psychiatric illness was established in the second stage sample with the PSE CATEGO-lD system. Sociodemographic factors. For the present analysis six sociodemographic variables are used. Age was divided into 17-44 and 45 + age groups. Marital status was classi fied into married and not married. Employment status was classified as employed or unernployed; by unemployed we meant the lack of paid working activity either through un employment, retirement or for medical reasons. This lat
Table 2. Sociodernographic characteristics of the second stage Characteristics