Of the 1.26 million new asylum claims made in European Union (EU) countries at the end of 2015, 176,000 were made by Afghans, ranking just second to claims made by Syrians [1
]. The migration of Afghans to the EU has increased dramatically in recent years due to socio-economic challenges and a prolonged conflict in Afghanistan, in addition to deteriorating economic conditions and policy shifts toward existing (undocumented) Afghan refugees in Iran and Pakistan [2
] that result in resettling to other regions by these migrants. Turkey continues to be a major transit hub for Afghans seeking to reach EU countries [3
], but also a place of permanent residence for some, as evidenced by thousands of asylum claims lodged by Afghans in recent years [4
However, Turkish asylum and settlement laws grant protection only to refugees from Europe and to individuals of Turkish descent, respectively [5
], which marks the presence of Afghans in Turkey as irregular. It appears that Turkey has recently recognized the importance of creating a coherent migration management framework. Adopted in 2013 and implemented in 2014 by the Turkish Parliament, the Law 6458 on Foreigners and International Protection (Yabancılar ve Uluslararası Koruma Kanunu
) is intended as a step toward managing both legal and non-formal migration to Turkey [6
]. Under this Law’s Article 96, the mutual “adaptation” of immigrants and society is supposed to be facilitated through courses and information campaigns, depending on available funds and stakeholders’ recommendations.
Nevertheless, until this law applauded by the United Nations (UN) and EU is fully shared with all stakeholders and enforced, the predicament of the status quo remains, with all its adverse effects on migrant health.
Post-migration stressors (e.g., asylum difficulties or administrative difficulties in the application process, employment-related problems, etc.) associated with their insecure status as irregular migrants in Turkey have led to high rates of psychological distress, as demonstrated recently by the authors of this study [7
]. The situation of Afghans in Turkey, however, remains understudied, and the significance of further investigating their health needs rests on several factors, including: (1) their continual migration to Turkey; (2) health needs that may transcend psychological help for traumatic symptoms to include non-communicable diseases such as hypertension and chronic pain commonly observed in refugees [8
]; and, (3) the fact that unmet health needs are likely exacerbated by difficulties in accessing health services due to their legal status.
Despite these needs, we found no studies of access to and use of health services among irregular/undocumented migrants or asylum-seekers in Turkey. Previous work has focused on migrants in the EU where access to health services for irregular migrants may be relatively poor partly due to linguistic and cultural barriers [9
], the fact that few countries grant access to care beyond emergency services, and because migrants often experience fears of being reported to police and immigration authorities by health care staff [10
]. A later review confirms these access barriers among asylum-seekers, and additionally highlights the inability to pay for services, discrimination on the part of health care staff, and difficulties in navigating health services (improved through social/community supports) as access barriers [11
In contrast, Hadgkiss and Renzaho cite various studies showing that asylum seekers’ utilization of general practitioners/primary care services and nurses is high (ranging from 55% to 73% report using such services), with prescription of medication being a common outcome of such encounters. Specifically, Gerritsen et al. [12
] showed that Afghans in the Netherlands report higher contacts with a general practitioner (GP) and more medication (analgesics, sleeping pills) use than a Somali comparison group, and, that female gender, older age, and perceived (poor) general health status predict health service use. As opposed to the Netherlands where health services are more easily accessible for irregular migrants, regardless of their nationality [10
], in Turkey irregular or undocumented migrants are not entitled to any public health services. Instead, only Syrians, who are under “temporary protection” status, are legally covered by Turkey’s public health care system [13
]. Asylum-seekers from other countries of origin (international protection applicants) do not automatically qualify for coverage, and as a general rule can only utilize health services in the province where they are registered and required to reside. Registration is processed with provincial DGMM Directorates (Directorate General of Migration Management), which under normal circumstances ensures international protection applicants free-of-charge access to primary (e.g., health centers, tuberculosis dispensaries), secondary (state hospitals), and tertiary (research and training hospitals) health services. Provincial DGMMs provide international protection applicants a Foreigners Identification Number (FIN), which healthcare providers use for intake and processing purposes; therefore, with the exception of emergency health services that can be spontaneously accessed at any time, without a FIN irregular migrants cannot access the full range of services [14
This paper aims to explore the extent to which migrant Afghans residing in Istanbul, Turkey use a variety of health services. These services include encounters with primary care physicians (PCPs), outpatient (OP) specialists, and the use of prescription as well as nonprescription medications—used as proxies for health services use, as shown in a recent large-scale epidemiological study of EU migrants [15
]. This paper also aims to identify the factors associated with the use of health services, as guided by Andersen’s Behavioral Model of Health Services Use or the “Andersen Model” [16
Health services utilization, according to the Andersen model, is a function of three components: (1) predisposing factors, which may include demographic characteristics (e.g., age, gender); (2) enabling factors or conditions that make health service resources available to an individual (e.g., income, access to a regular source of care); and (3) perceived (e.g., quality of life, symptoms) as well as evaluated (e.g., diagnoses) need factors, which are described as the most immediate reason for health services use to take place. This model has been widely applied to studies of migrant populations, notably by Laban et al. [17
], who used Andersen’s model with a sample of Iraqi asylum-seekers in the Netherlands where it was hypothesized that a special set of need variables, including those reflecting post-migration living difficulties (e.g., long asylum application procedures) would predict service use.
Results showed that long asylum procedures did not predict preventive service use, but did predict mental health service use and prescription drug use only, possibly due to the stress associated with long asylum processes. Moreover, this study cites that other need variables, including low perceived quality of general health and functional disabilities, were the most important predictors of use. We tested similar relationships here, and hypothesized that health service use would be predicted by various predisposing and enabling factors, and that the effects of (perceived) need factors, including health and mental health measures and a range of post-migration living difficulties, would the most important predictors of use.
2. Materials and Methods
2.1. Participants and Procedures
The inclusion criteria for this cross-sectional study was limited to migrant adults over the age of 18 who are of Afghan ancestry and currently reside in Istanbul. The ethical merits of this study were reviewed and approved by the first author’s affiliated university Institutional Review Board (IRB), with approvals also obtained locally through Galatasaray University in Istanbul (Ethical approval code: 5150291). An IRB-approved information sheet explaining participants’ rights and the survey were used for soliciting participants using a combination of convenience, snowball, and street-intercept sampling techniques in Istanbul’s Zeytinburnu district over a 10-day recruitment period in September 2015. The first author of this study along with a research assistant from a local university in Istanbul, both fluent in Dari, explained the survey to potential participants as an effort to document their health and life experiences. Then participants completed the survey on their own on sidewalks, in local cafés, public venues such as parks, in local shops where they worked, and in their residences (a few participants with literacy challenges were interviewed).
Surveys were completed in Dari, which is a language commonly spoken among Afghans. With the exception of the Afghan Symptom Checklist (ASCL), in which a Dari version was already available (courtesy of Ken Miller), all other survey items and scales, including the SF-8 and Post-Migration Living Difficulties (PMLDs) checklist went through a rigorous translation-back translation process with assistance from members of the Afghan refugee community in Southern California.
2.2.1. Predisposing Factors
Surveys assessed information on age, gender, marital status (1 = currently married, 0 = unmarried (includes never married, divorced/separated, widowed)), and educational attainment (1 = post-secondary and beyond, 0 = secondary school and lower).
2.2.2. Enabling Factors
Surveys included questions on length of residence in Turkey (1 = ≥1 year, 0 = <1 year) while a question on family presence in Turkey provided indication of social capital. Socio-economic (SES) variables included questions on employment status (1 = employed, 0 = unemployed, including retired and disabled) and income stability (using a binary “yes/no” response assessing whether respondents are able to meet monthly financial needs with current income, 1 = stable, 0 = unstable). We also asked participants whether they had access to a regular source of health care using a binary “yes/no” response choice (1 = access, 0 = no access); however, this question did not refer to a specific source of health care.
2.2.3. Need Factors (Perceived)
Psychological Distress Symptoms
The frequency of distress symptoms was measured using a culturally-grounded measure of mental health, the Afghan Symptom Checklist or ASCL [20
]. The ASCL consisted of 23 items familiar to both western psychiatry and Afghan society (e.g., “asabi”/high stressed, difficulty concentrating, sadness). Each question asked the participant to think back on the previous two weeks. Scores ranged from 23 to 115 derived from response choices ranging from “1” (“never”) to “5” (“everyday”). The ASCL demonstrated excellent reliability in this sample (Cronbach’s α
Post-Migration Living Difficulties (PMLDs)
The PMLD checklist [21
] was used for measuring the severity of post-migration problems commonly encountered by asylum-seekers within the past 12 months. Responses were rated on a five-point Likert-type scale ranging from “0” (“no problem at all”) to “4” (“a very serious problem”). In a previous paper based on this sample [6
], the authors identified five subscales through principal components analysis (PCA) (their respective Cronbach’s alphas ranging from 0.719 to 0.891). These five subscales were labeled (1) “conditions of extreme precarity” (nine-items: fears of being sent home, loneliness, poverty, etc.); (2) “asylum difficulties” (four-items: delays in processing asylum applications); (3) “employment-related problems” (three-items: e.g., no permission to work); (4) “access to medical and social services” (four-items: e.g., little help from charities), and (5) “marginalization and family-related stressors” (four-items: e.g., worries about family back home, communication difficulties/language barriers).
2.2.4. Health Service Use (Dependent Variable)
Informed by previous studies examining service use among Afghans and other ethnicities residing in the EU [12
], we asked participants whether they accessed any of the following health services within specified time-frames with a dichotomous “yes/no” response choice. Services included encounters with PCPs (past two months), and outpatient (OP) specialists (past two months), as well as the use of prescription (past 14 days) and nonprescription medicines (past 14 days).
2.3. Data Analysis
SPSS, version 23.0 (SPSS Inc., Chicago, IL, USA) was used for all data analysis. First, we replaced missing scores for scales with less than 10% missing items by using the average of completed scale items. Bivariate tests were used to examine associations between the covariates (predisposing, enabling, and need factors) and the health service use outcomes (PCPs, OP specialists, prescribed medicines, un-prescribed medicines), where chi-square tests of independence and independent samples t-tests were applied as appropriate. We subsequently entered variables into multivariate logistic regressions to identify factors predictive of health service use. For a consistent set of parsimonious multivariate models, variables were selected based on their conceptual relevance, as informed by previous research, and their statistical significance (p < 0.05) at the bivariate level. For each predictive model, in the first step we entered predisposing factors (gender); next, we entered enabling factors (length of residence in Turkey, family presence in Turkey, income, access to a regular provider), followed by perceived need factors (PCS and MCS indices, ASCL scores, and scores for PMLD subscales). Multivariate logistic regression analysis computed odds ratios (ORs) with 95% confidence intervals and p-values of the association between each covariate and use of health services. Additionally, multiple regressions were carried out to screen for multicollinearity through an examination of tolerance statistics for predictor variables in each model. Tolerance was found to be greater than 0.1 for all variables, indicating that multicollinearity was not a problem. Linearity of the logit was ascertained by ensuring that p-values for Hosmer and Lemeshow tests for each model exceeded 0.05, which we confirmed. Statistical significance was considered at p < 0.05; however, given the sample size, relationships approaching significance at p < 0.10 are reported as well.
Among a community-based sample in Istanbul, this study aimed to explore the extent to which first generation Afghan migrants use a variety of health services, and secondly to investigate factors associated with service use, as guided by the Andersen Model. Overall, our data suggests that the frequency of health service use among Afghan migrants residing in Turkey is quite low when compared to rates observed among asylum-seekers in the EU [11
], which provides the closest comparison possible, given the sparse literature on (irregular and asylum-seeking) migrants’ access to health services in Turkey. In this study, we found that encounters with OP specialists (20%) were lowest, PCP use slightly higher, and medication use, namely nonprescription medications (37%), was highest. The low usage rates could be a result of our sample being predominantly composed of young men, who have been shown to use preventive health services less frequently than women or older men [12
]. Moreover, women’s higher use of preventive services observed here may be due to the greater need for PCPs that provide support and consultations for various types of health screenings and prenatal care.
As expected, family presence in Turkey, income, and having access to a regular source of care positively correlate with use of all services other than medication (prescription and nonprescription) use at the bivariate level, and to some degree in multivariate analyses after adjusting for need factors. While income may certainly provide individuals with the means of covering out-of-pocket health care costs in Turkey, having access to a regular source (a gatekeeper), and family facilitates linkages to more specialized OP services, which concurs with previous findings [23
]. Also, other enabling factors could be operating here, unaccounted for in our survey, such as where Afghans reside in Turkey (Zeytinburnu—a large resettlement hub in Istanbul) and the assistance from fellow Afghans familiar with navigating local health clinics. Future research is needed to identify and characterize these and other access-enabling channels that Afghans depend on.
In sum, having personal and social resources are so critical to the use of health services, that these factors may override the influence of physical and mental health problems observed in our sample. It is important to note that despite the low HR-QoL indices and high psychological distress levels observed here, these perceived illness factors had little to no influence on service use as hypothesized, contrasting with previous research [17
]. The reason for this unexpected result is not entirely clear; however, it may indicate an unmet health need further exacerbated by their precarious situations and structural barriers that restrict Afghans in need of care from accessing professional health services. This hypothesis is supported by the greater reliance on nonprescription medications (e.g., over-the-counter analgesics) for physical health problems.
Moreover, our data suggests that linguistic and discrimination-related factors deter the use of health services, as demonstrated in the inverse bivariate relationship between PMLD-5 stressors and use of PCPs and OP specialists. This aligns with a recent review of asylum-seekers in the EU [11
] indicating that factors such as discrimination on the part of health care provider and difficulty navigating health care systems serve as barriers to health care use. However, the effect of PMLD-5 did not persist in multivariate analyses, and strikingly PMLD-2 emerged as a significant predictor of service use (except for nonprescription medications). The positive association between PMLD-2 and service use is noteworthy because this implies that individuals facing asylum difficulties are meeting with government or charity agencies who are connecting them to services. Thus, asylum difficulties, though stressful in this population [7
], may be reflecting a kind of social capital and would be better classified as an enabling factor for health care utilization, or a mixture of need and enabling, which future research may examine.
A strength of our study was that it is among the first papers to examine health care utilization patterns among irregular migrants in Turkey. Nevertheless, the study has several important limitations. The small and non-random sample limits generalizability, and female participants were underrepresented; however, this may reflect actual migration trends among Afghans which favors males. Nonetheless, future research with Afghans and other irregular migrant populations in Turkey should place strong emphasis on pregnant and childrearing women, a critical period for both child and mother. Second, we assessed health services use based on self-report data rather than using medical records. Also, we did not account for use of emergency department (ED) services in our analysis, a major portal by which irregular migrants seek health care services. Future research ought to consider this either as an outcome variable or as an enabling factor given that ED encounters may result in referrals to other services. Lastly, the cross-sectional design does not allow for cause-effect relationships to be established. We suggest future longitudinal research to further explore health care services use among this population, which may change for some who build stronger social networks, learn to better navigate the health system, and gain some degree of socio-economic stability needed to pay for out-of-pocket costs for services.