The purpose of the analyses presented in this paper was to determine gender differences in health and health care utilisation within and between various ethnic groups in the Netherlands. This information might be helpful to develop policy to focus on the health status and accessibility of the health care system of specific groups according to gender and ethnicity.
In general women showed poorer health than men; the largest differences were found for the Turkish respondents, followed by Moroccans, and Surinamese. Furthermore, women from Morocco and the Netherlands Antilles more often contacted a general practitioner than men with this background. Women from Turkey were more hospitalised than Turkish men. Women from Morocco more often contacted ambulatory mental health care than men with this background, and women with an indigenous background more often used over the counter medication than men with an indigenous background.
It turned out to be difficult to obtain estimates of gender differences for the different ethnic groups in their countries of origin. A study of service utilization in Curaçao, Netherlands Antilles, showed that women were more likely to consult a general practitioner or specialist than men, which is in agreement with our findings (although the specialist consultation was borderline significant).  Furthermore, nationwide surveys of the Ministry of Health of Turkey indicate that the health status of women is poorer (self perceived health status and chronic diseases) than that of men and that there is a need to study the underlying psycho-social causes of this situation. They furthermore show that women are hospitalized more than men. Both findings are in agreement with the findings in the current study.
In the literature only four studies were found which also focused on gender differences in health in various ethnic groups. One study examined inequalities in the self-reported health of men and women from white and minority ethnic groups in the UK. The results showed higher morbidity (i.e. worse general health status) for women from the Black Caribbean and Indian populations, but not for women from the white (indigenous), Pakistani and Bangladeshi populations. Furthermore, the analyses showed substantially poorer health among all minority ethnic groups (men and women) compared to white men. Another study assessed the association between gender, race/ethnicity (white, moreno, mulatto, black), and social class and prevalence of depressive disorders in an urban sample in Brazil. The study showed that there was no female:male difference in depression among Whites, and that the highest ORs for gender difference were found in the moreno and black ethnic group (adjusted for i.e. social class). A recently conducted study assessed differences in men's and women's self-rated health, functional limitations and life-threatening medical conditions across five major US racial/ethnic populations. The results showed that the magnitude of gender differences varies considerably by racial/ethnic group, health outcome, and comparison category. When compared to white men, Non-Hispanic blacks (men and women), Mexican women, Puerto Ricans (men and women) and Cuban women are more likely than white men to report fair/poor general health status when adjusted for demographic and socio-economic factors. Finally, a study examining gender differences in health care utilization among older Americans found that gender differences in medical use vary according to the type of services used: women are less likely to use hospitalization and outpatient surgery but are more likely to use physician and home health services than men.  These differences are largely consistent in direction and magnitude across racial/ethnic groups (white, African American, Hispanic).
The results of the current study regarding gender differences within ethnic groups are in agreement with the study conducted in the UK, showing that the general health status of women is worse compared to that of men in some groups (e.g. the Moroccan, Turkish and Surinamese groups) and that the gender differences are small or absent in other groups (e.g. the indigenous group and the Antillean group, respectively). Regarding health care utilization, the results differ somewhat from the study conducted among older Americans, as gender differences in health care use vary among the different ethnic groups, although differences are in general small and there is no clear trend. Based on the results of the current study we can not conclude that gender differences in health and health care utilization among the ethnic groups clearly vary due to cultural differences, for example regarding the position of women in the society. Other studies should be conducted to explain the gender differences between the ethnic groups, by focussing on cultural, but also on e.g. biological and socio-economic factors.
Strengths and limitations
The non-response rate in the migrant groups was higher than in the indigenous population. This was mainly due to difficulties in reaching the sampled persons, a well-known problem in population-based studies among migrants.[10, 18] It is not clear whether the results of the study were consequently biased. However, no differences in age, gender, socio-economic position and general health status (obtained from the census data) between respondents and non-respondents were found.
There is the possibility of bias from using self-reported data. However, general health status has been extensively used in surveys all over the world and the outcome measures on acute and chronic conditions are included in the surveys of Statistics Netherlands,. although evidence on their cross-cultural validity is limited. Reporting of health and use of health care might differ between women and men, although women are not per se more willing to report symptoms than men. Self-reported data turned out to be a useful method for providing a valid estimation of ethnic differences in health care utilisation.[14, 34] In order to restrict bias as much as possible the comprehensibility and acceptability of the questionnaire was tested in a pilot. Furthermore, interviews were held in the respondent's own language if necessary.
Given the large sample size of the indigenous population, some of the gender differences in health and health care utilisation are statistically significant (e.g. level of education, general health status, contact with a general practitioner, physiotherapist and ambulatory mental health care), although some were small and not socially or biologically relevant. On the other hand, the numbers of respondents in the migrant groups are relatively small, yielding broad confidence intervals for the ORs (e.g. for general health status and chronic conditions in Antilleans, ambulatory mental health care in Surinamese). Therefore, P <.05 was used for the multivariate analyses, despite multiple testing. This was also justified because of the explorative nature of the study.
Gender and ethnic differences in health and health care have shown to be related to age and socio-economic position.[23, 24] Consequently, the multivariate models also contained these variables. However, one could argue whether controlling for education and insurance type is sufficient to cover socio-economic position. Especially for the few more educated migrants, level of education might not be a good indicator, as issues such as being Muslim might make them less employable and therefore poorer compared to their indigenous equivalents, hence leading to worse health outcomes.
The strengths of the current study include the relatively large sample sizes of the migrant groups compared to other studies on health and health care utilisation among these groups conducted in the Netherlands,[10, 19] and the use of several health outcomes and particularly the use health care utilisation rates compared to other studies on gender and ethnicity. [23, 24]
Furthermore, a recent study that addresses the importance of the need to integrate a gender perspective into epidemiological studies on migration and health states that information on both ethnic background and sex, together with socio-economic status is not usually available in most health information systems.  All these variables are included in the survey data on which the current study is based.