In this study we focused on the association between acculturation and the use of health services within a population-based sample of first generation Turkish and Moroccan migrants in the Netherlands. Generally, there was only a moderate association between acculturation and health care utilisation, in that predominantly the domain communication in Dutch was related to utilisation, and GP care was not related to acculturation at all. Instead, health care utilisation was strongly related to subjective and objective measures of health care need, namely self-reported health status and the number of chronic conditions. If need factors, rather than factors like ethnic background or - as in this case - acculturation, are major determinants of health care utilisation, this is essentially a positive observation, since this can be argued to be an indicator of equity in health care access .
Some positive associations between acculturation and use of health care services were found. Among the Turkish group acculturation was generally associated with the use of mental health care services; more communication in Dutch (within one's private domain) was associated with more use of mental health care amongst Turkish men, while among women this was the case for social interaction. The direction of these associations is in line with the common hypothesis that increasing participation in the host culture is associated with higher service use. It is also in line with findings among other migrant groups in the U.S. , and - more importantly - with recent findings from a study in the Netherlands focussing on the use of mental health services .
However, among Moroccan subjects the association ran opposite while more emancipation was associated with less uptake of outpatient care (males) and more communication in Dutch was associated with less mental health care (males and females). A possible explanation for the association between more communication in Dutch and less uptake of mental health care among Moroccans may be a result of the methodology used in our study. That is, although general health status was accounted for, a specific measure of mental health was lacking. It could be that Moroccans who were better able to communicate in Dutch experienced less psychological stress and consequently a lower mental health care need. For example, in a previous study we found that lacking the skills to live/participate in the Dutch society largely related to mastery of the Dutch language, was associated with more psychological distress . If this is the case indeed, then the question of course rises why we found a comparable association with outpatient specialist care and the opposite trend among Turkish respondents.
Another explanation for the aforementioned reverse relationship among Moroccans regarding mental health care utilisation may be found in the absence or availability of social support. That is, compared to Turkish migrants, first-generation Moroccans in the Netherlands tend to have smaller social networks, which often do not extend beyond their direct families . Possibly, Moroccans with better Dutch language skills are less likely to become socially isolated, may be more likely to have alternative sources of (informal) support in case of health care need, and are consequently less likely to apply for mental health care.
Finally, an explanation of this adverse relationship, especially regarding health care from outpatient specialists, may be that higher levels of adaptation concur with a better knowledge of the Dutch health care system and the role of Dutch GPs. GPs in the Netherlands act as gatekeepers to outpatient specialist care, while in Turkey for example it is common practice to visit medical specialists directly. It might be that those who displayed better skills for living in Dutch society were more aware of this than those who did not, and were therefore more likely to remain in care in general practice. One argument against this explanation, however, lies in the observation that we found no association between acculturation and uptake of care in general practice.
The latter observation is important, because in our view it indicates and supports the notion that GP care in the Netherlands has a low threshold and is highly accessible. Recently, Uiters et al. already concluded that the gate keeping role of general practitioners in the Netherlands functions equally effectively among ethnic minority groups compared to the ethnic Dutch population . The observation that acculturation was associated with outpatient specialist and mental health care utilisation seems to suggest that if services are less accessible, or when culturally defined stigma and taboo come into play (as is the case for mental health problems), acculturation becomes more relevant as a concept in health services research.
This study has limitations. First, due to its cross-sectional design, no conclusions are allowed on the directionality of the results. For example, it is conceivable that ill health (including mental health), indicated by higher use of services, limits the ability of respondents to acculturate. Moreover, ill health may have resulted in a disproportionately negative self-evaluation regarding one's own acculturation skills. A second limitation concerns the generalisability of our results, which may have been compromised by the relatively high non-response. Selective non-response might have occurred, as people who do not use health care services may also be less willing to participate in research. However, the age/sex distribution in the responding sample was reasonable. The non-response also caused a relatively low number of respondents in each analysis. Considering the moderate range of most of the statistical associations observed, one may wonder if the associations are clinically significant. Some of the confidence intervals were extremely wide. Another source of bias might include the fact that measures were self-reported, although self-report measures have been found to be reasonably valid estimators for comparisons between migrant groups in the Netherlands .
Additionally, the acculturation instrument was strongly focused on measuring adaptation, or assimilation. As such, no statements can be made about the role of the second dimension of Berry's model of acculturation (i.e. maintenance of heritage culture and identity), while in previous studies it has been shown that this is also a very important aspect. On the other hand, measures of acculturation traditionally focus on language ability alone. Indeed, it can be derived from the results that, although the associations between acculturation and health care utilisation were limited and heterogeneous, communication in Dutch was one of the most important and central aspects of acculturation. However, other dimensions played a role of significane as well. The finding that increasing social interaction by Turkish women was associated with increasing mental health care, for example, is noteworthy in the light of marginalisation of Muslim migrant groups in Western countries . That is, if we consider increasing mental health care utilisation by Turkish women a good development, than current political and social developments, by some labelled as "Islamophobia" can be considered a threat in this respect.
Although the factor structure of the acculturation was supported by our data, reliability of the subscales was not strong. According to Nunnaly, Cronbach's alpha coefficients should be 0.70 or higher . However, lower values - such as in our study - have also been presented as adequate: by citing other studies [48, 49], Milfont and Gouveia  argue why reliabilities in the range 0.60 and 0.70 can be regarded as adequate as well, and that if samples are sizes larger than 100 (which is the case for all subgroups in our analyses) alpha coefficients greater than 0.40 are acceptable. Furthermore, the Cronbach's alpha we found corresponded with those found in another Dutch study carried out by Hosper et al.  using the same instrument (Cronbach's alpha of 0.64 is presented for the combined scale and a Cronbach's alpha of 0.84 and 0.80 on the social interaction scale).
Finally, and unfortunately, the acculturation scale was applied only in the ethnic minority groups and not among ethnic Dutch. Comparisons between ethnic Dutch, Moroccan and Turkish respondents were thus not feasible. Adjusting and applying acculturation measures for use among original inhabitants of host countries is not common use in health services research, but is a good suggestion for further studies. For example, it would have been very interesting if we could have seen how the ethnic Dutch scored on the emancipation subscale.