Abstract
The mental health of migrants in Australia has received little research attention. With about one in four Australians having been born overseas (26%, ABS Census 1996 [1]) and the identification of mental health by both Commonwealth and State Governments as one of five priority areas, the study of mental illness in migrants is important and timely. A national survey conducted in 1997 reported that almost one in five (18%) Australians had suffered a mental disorder in the previous year [2]. The survey found that people born outside Australia had in comparison a lower prevalence of mental disorders (15.7% of those born in English speaking countries, 14.5% of those born in non-English speaking countries compared to 18.6% born in Australia) [2]. However, an analysis of the 1989–1990 National Health Survey and a survey of general practice in Victoria measuring treated prevalence in community and primary care settings showed higher rates in Italian and Greek-born migrants and lower rates in South-east Asians compared to the Australian-born [3]. In recent population surveys of Iranian migrants in Sydney and the Latin American community of NSW, the prevalence of poor mental health was 36.5% and 32.6%, respectively [4, 5]. This is almost twice the prevalence of 18–19.5% in the general population reported from recent state and national surveys [2, 6].
With migration, people may experience a greater number of stressors, but there is no consensus that migrants have higher rates of mental illness than nativeborn populations. Reviews indicate approximately equal numbers of studies reporting migrants experiencing more mental illness and reporting less than the native-born population [7, 8]. This is consistent with the theory that it is not migration alone that impacts on mental health, but the risk factors that often accompany migration combined with the stresses of migration [8]. Major risk factors include negative public attitudes, language difficulties, difficulties in gaining recognition of overseas qualifications, unemployment, and separation from family and community [7, 8]. Those at additional risk include women and individuals who are adolescent or elderly at time of migration [7, 8]. In Australia, risk factors for mental illness in the Iranian population included being a student and perceiving that migration contributed to psychological distress [4]. In the Latin American population, those who were separated, widowed or divorced, those unemployed and those dissatisfied with life in Australia were at greater risk of mental illness [5].
An important consideration in the comparison of rates of mental disorders across ethnic groups is the validity of instruments, such as the general health questionnaire (GHQ) [9], when applied across culture [10–13]. The GHQ has been shown to have reasonably good crosscultural validity when applied or adapted to ethnic groups in Australia and New Zealand such as Turkish [13] and Chinese [14] migrant populations. In the surveys of Iranian and Latin American migrants, the GHQ instruments used were translated but not validated or adapted to the specific populations.
Approximately five per cent of Australians were born in Asia and Filipinos are the third largest Asian-born migrant group, behind Vietnam and China [1]. However, in Australia there have been few population-based surveys of mental health in Asian migrant groups. Community surveys in the US report that Asian-American groups have a rate of mental illness at least equivalent to that of the white population [15, 16]. A British survey reported lower rates among Asians, but highlighted the likelihood of instrumentation limitations, including difficulty in the translation and use of mental health concepts in Asian languages [10]. In a cross-sectional study of risk factors for depression in Asian-Americans in the USA, Filipinos who were less than 30 years, female, widowed, separated or divorced, unemployed or in a part time job, of low income, low education and were Protestant were at greater risk of depression [15].
In 1996/1997 a cohort of Filipino women living in Queensland were interviewed as part of the Australian Longitudinal Study on Women's Health (ALSWH) [17]. The objective of the Filipino component of the ALSWH was to employ a multiple follow-up design to explore key themes including health service use, healthy weight and exercise, time use and social roles, life stages and key events, relationship issues, immigration experience and mental health. This paper reports on the mental health component of the questionnaires and is based on the findings of the baseline questionnaire and first follow-up of the Filipino women one year after the initial survey. Both surveys collected information on demographics, immigration experience, physical health, health service utilization and life changes and key events. The 1997/ 1998 survey included a measure of satisfaction of life in Australia and the GHQ-28 [9] as a measure of mental health. Qualitative in-depth interviews expanding on these topics were nested within the two quantitative surveys. The specific aims of this mental health component were to: (i) estimate the prevalence of psychological distress in the Queensland Filipina population based on the GHQ-28, and (ii) investigate potential predictors of mental distress including demographic, immigration-related, physical health, satisfaction with life and life-change variables. We use the term Filipina to refer to women from the Philippines now living in Australia.
Methods
Sample
According to the 1996 census of population and housing, 9111 Filipinas live in Queensland [1]. Sampling small and dispersed migrant groups such as this is difficult [4] and additionally, many Filipinas in Australia have surnames that are indistinguishable from the general population because they are married to Anglo-Australians. Random sampling based on surnames characteristic of ethnicity was not possible. As a result, the primary sample was drawn from Filipino organization membership lists and supplemented by snowball (or chain referral) sampling, with a focus on three of the highest density populations for Filipinas in Queensland: Brisbane, Mt Isa and Far North Queensland. This combination of sampling methods is particularly suited to such populations because of their strong social networks and membership in organizations. A reference group of 16 Filipinas nominated by the Filipino Community Coordinating Council of Queensland, an umbrella organization with 80 member organizations, and representing Filipinas throughout Queensland met regularly with researchers to provide community input in guiding the research.
Filipinas involved in 24 community organizations and their social contacts, and in church groups, and Filipinas working in factories, were contacted about the study and invited to participate. These women were asked to nominate other potential respondents who were then contacted by researchers and invited to participate; these women were asked to nominate other women; and so on. A total of 487 women 16 years and older participated in the baseline survey in 1996–1997 (88% response rate) and 403 (83%) gave their consent to a follow-up interview one year later.
Instruments
The results reported here are based on information in the surveys obtained from the immigration-related items, derived from the Longitudinal Study of Immigrants to Australia [18], Mothers in a New Country study [19], the main section of the Australian Longitudinal Study on Women's Health [17], and sections on physical health, quality of life (the SF-36 [20]) and life events and life changes derived from the Holmes and Rahe Social Readjustment Rating Scale [21]. All demographic, healthrelated and immigration variables except satisfaction with life in Australia were measured at baseline. Satisfaction with life in Australia was assessed in the follow-up survey and was based on a mean score determined from responses to six questions. These questions included satisfaction with financial, educational and work opportunities; social and community life; the way children are brought up in Australia; and the general moral and ethical environment. Experience of life-events including changes in health, relationship, work and residence, death in family, decreases in finances and experience of pregnancy and birth were measured at both baseline and follow-up. Measurement of mental distress is based on information obtained from the 28-item version of the General Health Questionnaire (GHQ-28) [9] that was included in the follow-up survey. The questionnaire was pretested and reviewed by the members of the Filipina reference group.
Survey procedures
The survey questionnaires were administered in English with relevant Tagalog expressions for health conditions included. The entire questionnaire was not translated due to the high levels of fluency and literacy in English in the Filipino community [1] and because a variety of different dialects of the Filipino languages, spoken by community members, reduced the potential value of translation into any one language. Translation was available for the initial survey if necessary. The baseline survey was administered by trained interviewers face-to-face in the woman's home or at a Filipino community event. About half the follow-up surveys were administered face-to-face and the remainder over the phone, approximately 12 months after the initial questionnaire was administered. Each questionnaire took approximately half an hour to complete and was followed by an in-depth interview (described elsewhere [22]).
Data analysis
The GHQ scoring method (0-0-1-1) was used in all data analysis and a recommended threshold score of 4/5 was used for assessment of mental distress [9, 23]. The data were analysed using SAS for Windows version 8 and SAS JMP version 3.2.6.
Results
Of those 403 women consenting to follow-up, 346 responded to the second survey, resulting in a response rate of 86% of those consenting to participate and an overall follow-up rate of 71%. Reasons for non-response include participant or husband declining to participate (n = 14), unable to be contacted (n = 11), moved house and could not be located through a second contact or phone listing (n = 17), overseas or interstate (n = 10) and incomplete survey (n = 5). The 71% of women participating in the follow-up survey were older than non-participants (t-test, p < 0.01). In addition, those women participating in follow-up were more likely to have a long-term health condition (χ 2 = 4.0, df = 1, p = 0.05) and were more likely to be moderate drinkers (1 or 2 drinks per day) than abstainers or heavier drinkers (χ 2 = 9.4, df = 2, p = 0.03).
Table 1 shows a comparison between the 346 women completing the follow-up survey and the characteristics of Filipinas living in Queensland from the 1996 census [24]. Compared with the census population, a greater proportion of Filipinas in the sample were older and married or living in a relationship. Women in the sample were more likely to have a university degree and to be employed. Surprisingly, less women in the sample reported that they spoke English well compared with the census. This may reflect the different way the questions about English proficiency were worded in the two surveys. 1
Comparison between sample and Queensland Filipina sample in 1996 census
The mean age of women in the final cohort sample was 42.3 years at baseline. Most women (80%) were married or living in a relationship, half (50%) had a university degree and about half (53%) were employed in full-time or part-time work. The majority of women (72%) had children.
The proportion of women having an above threshold score (using the cut off of 4/5) on the GHQ-28 at follow-up was 23%. Table 2 shows the percentages of GHQ-28 scores above the threshold by demographic, immigration-related and health-related variables. Table 3 shows the percentages of participants with above threshold GHQ scores for lifeevent groupings assessed in both surveys.
GHQ-28 scores by demographic, immigration-related, health-related variables
General Health Questionnaire (GHQ-28) scores by experience of life events variables
Of the demographic variables examined, only marital status was significantly associated with an above-threshold GHQ-28 score (p < 0.01). Single women were significantly more likely to have an above-threshold score. Women who were married or living in a relationship were more likely to have an above-threshold score than women who were widowed, separated or divorced. A greater proportion of women who were employed or studying had above-threshold scores than women who were unemployed outside the home. This finding, although only a trend (p = 0.08), was in the opposite direction to that expected.
None of the immigration-related variables measured at the time of the baseline survey were significantly associated with GHQ-28 scores. Those women who were less satisfied with life in Australia at follow-up were significantly more likely to have an above-threshold score (p < 0.001). Taken separately, all six components of satisfaction with life in Australia were associated with GHQ scores (p < 0.05). Respondents were least satisfied with work opportunities, with 43% of the total sample reporting that they were not satisfied or only a little satisfied with opportunities to work in a desirable job for which they were qualified. Twenty-nine per cent and 20%, respectively, said they were not satisfied or only a little satisfied with financial and educational opportunities. Self-rating of general health as poor, fair or good at baseline was associated with an above threshold GHQ-28 score at follow-up (p < 0.01).
A high proportion of women (39%) experienced a major decrease in finance in the 12 months prior to the follow-up survey. Thirty-two per cent of women experienced a death of a family member or close friend, 28% experienced a deterioration in either their own health or the health of a family member or close friend, 29% had a change of employment, 18% changed where they lived, 17% had a change in relationship, and 8% experienced a pregnancy or birth.
The experience of each category of life events, except a pregnancy or birth in the 12 months prior to the follow-up survey, was significantly associated with an above-threshold GHQ-28 score (see Table 3). Those women experiencing a deterioration in health, change in relationship, residence and work, a death of a family member or close friend or a major decrease in finance were more likely to have an above-threshold score than women not experiencing these events. These same life-event experiences measured at baseline were associated with GHQ-28 score in the expected direction but were not statistically significant. Not surprisingly, several of these life-event factors were significantly correlated. Examination of univariate relationships between life-event variables yields two configurations of correlated factors. Change in finance, work, relationship and residence forms one group of correlated variables and change in health and death of a family member or close friend forms a second configuration (Pearson χ 2, p < 0.05). This is not unexpected as major changes in work, residence, relationship and financial situation may be all caused by a common event and any one of these life events may lead to experience of the others. Similarly, change in health of a family member or close friend may result in or result from a death. Each pair of life-event factors measured at baseline and follow-up was significantly correlated (e.g. change in work in previous 12 months at baseline was related to change of work in previous 12 months at follow-up).
A logistic regression analysis was carried out to identify ‘independent’ predictors of above-threshold GHQ scores. All statistically significant variables were included in the model. Age and employment status were initially included in the model because of their low (although not statistically significant) p-values in the univariate analysis. Because of their small contribution to the model, they were excluded from the final analysis. Inclusion of baseline measures of life events (all highly correlated with the measurement of the same events at follow-up) decreased the fit of the logistic model. The final model included five statistically significant variables – marital status, satisfaction with life in Australia, change in financial situation, change in relationship and change in health in past 12 months.
Taking into account other variables in the model, those women who were single were more likely to have an above-threshold score than those women who were separated, widowed or divorced (odds ratio (OR) = 6.54, 95% confidence interval (CI) = 1.18–35.3). Those who were married were more likely to have an above-threshold score than those who were separated, widowed or divorced, although the confidence interval includes 1 (OR = 1.68, 95% CI = 0.57–5.35). Women whose satisfaction with life in Australia was rated as low were more than three times as likely to have an above-threshold GHQ score compared with women who were rated as having high satisfaction with life in Australia (OR = 3.33, 95% CI = 1.79–6.37).
Women who experienced a change in financial situation in the 12 months between the two surveys were more than twice as likely to have scored above the threshold than women not experiencing a change in finance (OR = 2.38, 95% CI = 1.31–4.37). Similarly women experiencing a change in relationship during this time were more than twice as likely to have an above-threshold GHQ score compared to women not experiencing a change in relationship (OR = 2.33, 95% CI = 1.11–4.87). Similarly women experiencing a change in health between surveys were more likely to have an abovethreshold GHQ score than women not experiencing a change in health (OR = 1.92, 95% CI = 1.02–3.61).
The remaining life event variables associated with above-threshold GHQ scores in univariate analysis (change in work and residence and death of family member or close friend) were not significant factors in the logistic model because of their associations with other life event variables. Change in finance mediated the effect of change in work on mental distress. That is, change in work is related both to change in finance and to mental distress, but the relationship between change in work and mental distress disappears if change in finance is entered as a covariate. That indicates that change in work appears to increase mental distress through decreases in finance. Change in residence and change in relationship tend to occur together. Similarly, death of family member or close friend occurs together with deterioration of health. The factors included in the final model (change in relationship and deterioration of health) add to the model in excess of the two changes (in residence and death) that occur with them.
Discussion
Overall, 23% of the sample had above-threshold GHQ-28 scores. This figure is higher than the prevalence of 18% in the 1997 National Survey of Mental Disorders [2] and the 19.5% reported in a sample of South Australians using the same instrument (GHQ-28) [6]. Although the level of above-threshold GHQ scores is somewhat lower than the 36.5% and 32.6% reported in two community surveys of Latin Americans and Iranian immigrants to Australia [4, 5], it supports the findings of community-based surveys in the US that mental illness is at least as prevalent in some Asian immigrant groups as in the general population. This is in contrast to lower rates of treated prevalence of mental disorder reported for those born in South-east Asia compared to Australian-born people [3]. As several researchers both in Australia and overseas have argued, treatment rates in Asianborn populations are likely to be an underestimate due to stigma and the tendency to somatize [3, 11, 12].
The unique experience of Filipinas, many migrating to Australia as brides of Australian men [1], needs to be considered in generalizing the prevalence rates to other Asian groups. For example, like-ethnic community support has been shown to have a positive impact on migrant mental health [25, 26]. However, a limitation of this study was that the measurement of social networks and support were not included in the surveys because of concerns of the community reference group about the women's sensitivities about directly questioning support from husbands and family.
In the general Australian population, full-time employment has been shown to be a protective factor for mental health [2, 6] and in a number of other migrant studies in Australia and overseas, dissatisfaction with employment had been shown to be associated with mental distress [5, 7]. In our study, a greater proportion of employed women had above-threshold GHQ-28 scores, although the finding was a trend. In the study of Filipino-Americans, those who were unemployed were at greatest risk of mental distress, but those who were housewives or retired had the lowest rates [15]. The rate among housewives is consistent with the current study. Very few of our women reported that they were not employed. Although the numbers are small, they indicate that being a student is associated with above-threshold GHQ-28 scores. In many cases working can be a source of conflict in their marriage to Australian or European husbands and is often experienced as an additional role on top of the role as housewife and mother [22, 27]. Filipinas who report being a student may be combining studies with full-time work in addition to housework and caring for children. Furthermore, many employed women report they are underemployed largely because of the lack of recognition of overseas qualifications [22]. In univariate analysis, Filipinas who said they were dissatisfied with opportunities to work in a job for which they were qualified were more likely to have an abovethreshold GHQ-28 score.
Overseas evidence suggests that immigrants who are disappointed with opportunities in the host country have higher levels of depression [28]. Consistent with the study in the Latin American population in the Hunter Valley, overall satisfaction with life in Australia was significantly associated with GHQ score. However, as this was measured at the time the GHQ-28 was administered it is difficult to separate cause and effect. That is, women with distress may be more likely to report they are dissatisfied with life in Australia.
Changes in financial situation, relationship and health in the year between surveys was associated with mental distress. This finding itself is not surprising as each is a stressful life event. What is striking however, is the frequency with which these events are reported to occur in this population. For example, more than one-third of women reported that they had experienced a major change in financial situation within 12 months. In previous qualitative research in this cohort, financial pressures have been shown to be a major concern for Filipinas [27]. In the data set, we lack information about the context of financial change. It is not known whether the financial pressures are in fact single major events or if participants are responding positively to the major financial change questions because of their experience of daily financial hassles. A substantial proportion of women in the study (29%) also said they were not satisfied with the financial opportunities in Australia and this may indicate more constant financial stress than a change in circumstance. In a study of Asian immigrants in the US, 37% reported that they experienced daily financial hassles, that is, worry that the total family income would not be enough to meet the family's expenses and bills [29].
A limitation of our study is that the life event variables associated with above-threshold GHQ scores were measured at the same time as the GHQ and change in work, relationship and residence, decrease in financial situation and deterioration of health can all be outcomes of psychiatric morbidity. So, as for the measurement of satisfaction with life in Australia, the direction of the causal association between experience of life events and distress cannot be determined. GHQ scores were only measured at follow-up and cannot therefore be adjusted for pathology at baseline. Therefore, the contribution of life events and satisfaction with life in Australia in the 12 months between surveys to mental distress beyond pre-existing levels is not known.
Footnotes
Acknowledgements
This research was funded by the Commonwealth Department of Health and Aged Care. We thank the Filipino Community Co-ordinating Council of Queensland, affiliated organizations and members of our reference group. We also thank Jenny Phillips, Alla Ryboy and Anne Marie Benedicto.