Sample

Cluster sampling was conducted, with each cluster comprising eight households. The choice of cluster size was based on the daily performance capacity of the data collection group. The statistical framework was based on the household lists available from every health department in the provinces. The ratio of sample size to the total number of households was taken as 1:1000 (13 478 households through 1681 clusters). In total, 35 014 persons (22 564 from urban areas and 12 450 from rural areas) in the age group 15 years and above were studied.

Measures

The 28-item General Health Questionnaire (GHQ–28) was used as a screening tool for the detection of mental disorders. This questionnaire was developed by Goldberg & Hillier (Reference Goldberg and Hillier1979) for screening for somatic symptoms, anxiety and insomnia, social dysfunction and severe depression. A review of studies on the validation of the GHQ–28 in different countries demonstrates its high validity and reliability as a screening tool of mental disorders in the community. This questionnaire was translated into the official language of Iran (Persian), which is comprehensible to almost every Iranian, and its validity and reliability were approved in an independent study (Reference Noorbala, Mohammad and Bagheri YazdiNoorbala et al, 1999). The best cutoff point, determined using the conventional scoring method and the minimum overall misclassification rate, was 6: that is, those scoring 6 and above were designated as possible cases of mental disorder. Sensitivity, specificity and overall misclassification rate for a GHQ–28 cut-off score of 6 were 84.7%, 93.8% and 8.2%, respectively. The reliability of the GHQ–28 was assessed on a sample of 90 participants retested 1 week after the initial referral. The estimated intraclass correlation between the test–retest scores was 0.85 (Reference Noorbala, Mohammad and Bagheri YazdiNoorbala et al, 1999). To detect psychosis, epilepsy and learning disability (‘mental retardation’), a simple semi-structured clinical interview with its limited validity and reliability was used.

Collection of data

This survey was implemented as a part of the National Health Survey in Iran. In each province, specially trained general practitioners from the provincial health centres visited the selected households and completed the GHQ–28 for the age group 15 years and above. Detection of cases of psychosis, epilepsy and learning disability was based on semi-structured clinical interviews by the general practitioners, and on available medical and paramedical records.

Statistical methods

Data relating to the survey were analysed using the Statistical Package for the Social Sciences, version 8.0 for Windows. Logistic regression modelling was used to determine the factors that affect mental disorders. Mental disorders were considered as dependent variables and gender, age, education, occupation and marital status were considered as independent variables. Using the logistic regression model, the odds ratios, their level of significance and standard deviation were calculated. Backward logistic regression (P _e=0.15 and P ^r=0.2) was also used (Reference Hosmer and LemshowHosmer & Lemshow, 1989).

Prevalence

Epidemiological surveys of mental disorders in Iran report rates varying between 11.9% and 23.8% (Reference Noorbala, Mohammad and YasamyNoorbala et al, 2001). Comparison of prevalence rates in our survey with those detected in other surveys using different methods, tools or classifications, shows that the results of surveys using the GHQ–28 are approximately compatible. Differences in methods and tools for screening and diagnosis as well as different classification systems used and age groups studied may account for the minor differences in results.

Comparing the results with Western studies, the prevalence rate obtained in this survey is higher than rates in surveys conducted by Hoeper et al (Reference Hoeper, Nycz and Cleary1979), Hodiamont et al (Reference Hodiamont, Peer and Syben1987) and Fones et al (Reference Fones, Kua and Ko1998), but lower than those obtained by Kessler et al (Reference Kessler, McGonagle and Hao1994) and Lee et al (Reference Lee, Kwak and Yamamoto1990). It is approximately similar to prevalence rates found by Stansfeld & Marmot (Reference Stansfeld and Marmot1992) and Roca et al (Reference Roca, Gili and Ferrer1999). The findings further show that the maximum rates of positive responses to questions on the GHQ–28 were for headache, distress, insomnia, sorrow and disappointment, confirming findings of other studies conducted in Iran. Anxiety and depressive symptoms were common, which is comparable with results of similar surveys in Iran and in other studies reported by Kaplan & Sadock (Reference Kaplan and Sadock2000).

Gender distribution

The study found higher prevalence rates of mental disorder in women than in men (25.9% v. 14.9%), compatible with results of other surveys in Iran and those conducted in other countries. Gender and marital roles can be considered as possible explanations for the higher rates. The majority of women are bound to their social roles as housewives; even when women work outside the home, they still have the burden of housework. Hence, the latter group should be more subject to strains and stress; however, our research showed that working only within the home has a more serious impact on psychiatric morbidity. The fact that women in Iran are more at risk of mental disorders than is the case in Western cultures may be due to the robust effect of biological factors or to social inconveniences experienced more by women than by men. However, the type of such social problems may differ between cultures. The study demonstrates higher rates in rural populations than in city dwellers (21.3% v. 20.9%), but the difference is not statistically significant.

Age distribution

The study revealed a significant correlation between age and the occurrence of mental disorders. Prevalence rates increase with age, supporting the results of Lee et al (Reference Lee, Kwak and Yamamoto1990) and Hodiamont et al (Reference Hodiamont, Peer and Syben1987). This may be explained by reduction in physical vigour and the greater vulnerability of older people to stress as well as mental and physical diseases. This finding is not compatible with those of similar surveys conducted in Iran showing higher rates in people aged up to 45 years compared with those aged 45 and above (Reference Noorbala, Mohammad and YasamyNoorbala et al, 2001).

Stress

This study supports the results of earlier studies showing higher rates of mental disorders among illiterate and semi-literate groups. Sociocultural constraints in such groups posing limits to their coping styles in the face of stress may be considered as one of the main factors. Confirming the results of other epidemiological studies in Iran (Reference Noorbala, Mohammad and Bagheri YazdiNoorbala et al, 1998), our survey's findings demonstrate higher prevalence rates of mental disorder among the married: these may be due to economic and social stress factors such as financial matters, family management and child care. The study's findings of higher rates of mental disorder among housewives and unemployed men also reported by Hodiamont et al (Reference Hodiamont, Peer and Syben1987), Stansfeld & Marmot (Reference Stansfeld and Marmot1992), Murthy & Burns (Reference Murthy and Burns1992), Bahar et al (Reference Bahar, Henderson and Mackinnon1992) and Noorbala et al (Reference Noorbala, Mohammad and Bagheri Yazdi1998) may be explained as the overall outcome of insufficient income, the stress of unemployment, limited social relations and monotonous lifestyle. There is also a possibility that mental disorders have contributed to unemployment.

Implications for health care

Our finding that about a fifth of the population surveyed in the age group 15 years and above suffers from mental disorders suggests that 10–12 million persons in Iran require mental health care. Taking into consideration the present number of Iranian psychiatrists (735) and the present number of Iranian mental hospital beds (7850; Reference Yasamy, Shahmohammadi and Bagheri YazdiYasamy et al, 2001), the need to provide appropriate staff and facilities to render mental health care is more evident than ever.

CLINICAL IMPLICATIONS

1. ▪ The prevalence of mental disorders in Iran is comparable with most other countries.

2. ▪ Mental disorders are more prevalent in women than in men; unemployment and being divorced or widowed are other main correlates of morbidity.

3. ▪ Iranian mental health facilities should be extended to provide for the estimated 10–12 million persons suffering from psychiatric illness.

LIMITATIONS

1. ▪ This survey was conducted using the 28-item General Health Questionnaire, which is not a structured clinical interview, and was unable to assess the prevalence of different disorders.

2. ▪ Frequency of psychosis, epilepsy and learning disability was measured by general physicians using semistructured clinical interviews with limited validity and reliability.

3. ▪ The sample does not include children and adolescents, who constitute a major part of the Iranian population.