Unmet need for health services among refugees and asylum seekers in Thailand, relative to the Thai population: an analytic cross sectional study, 2019

Although the Thai government has introduced policies to promote health of migrants, it is still the case that urban refugees and asylum seekers (URAS) seem to be neglected. This study aimed to explore the degree of healthcare access through the perspective of unmet need in URASs, relative to the Thai population. A cross-sectional survey, using a self-reporting questionnaire adapted from the Thai Health and Welfare Survey (HWS), was performed in late 2019, with 181 URASs attending the survey. The samples were randomly selected from the roster of the Bangkok Refugee Center. The data of the URAS survey were combined with data of the Thai population (n=2,941) from the HWS. Unmet need for health services was defined as the status of when a respondent needed healthcare in the past twelve months but failed to receive it. Bivariate analysis was conducted to explore the demographic and unmet need difference between URASs and Thais. Multivariable logistic regression and mixed-effects (ME) model were performed to determine factors associated with unmet need. more pronounced in ME model, relative to multivariable logistic regression. URASs migrating from Arab nations suffered from unmet need to a greater extent, compared with those originating from non-Arab nations. The prevalence of unmet need in URASs was drastically high, relative to the prevalence in Thais. Factors suggesting a positive relationship with unmet need included advanced age, lower education achievement, and, most evidently, being unsinured. Policy makers should consider a policy option to enrol URASs in the nationwide public insurance scheme to create health security for the entire Thai society.


Background
At present, cross-border mobility is a soaring global trend for many reasons including people searching for better economic prospects, pursuing job and education opportunities and escaping from war and political conflicts. In 2017, international cross-border populations amounted to 258 million (3. 4% of global population) [1]. Of these 258 million, 68 million were forcibly displaced people. Of the 68 million, 25 million were refugees and three million were asylum seekers [2]. The situation of refugees has gained increasing attention in the global health field in recent years, particularly since the 2011 Syrian crisis which resulted in more than six million refugees fleeing from Syria to Europe [3 ]. Asia is another region that has encountered a refugee crisis. An obvious case is the exodus of more than 700,000 Rohingya refugees from Rakhine State in Myanmar to Bangladesh, during 2015-2017 [4].
The United Nations ( UN) and the World Health Organization ( WHO) , as well as many other international development partners, have called for more concrete actions to protect refugees' rights to health and well-being. Some tangible outputs of these actions include the launch of the World Health Assembly ( WHA) Resolution 70.15, entitled ' Promoting the health of refugees and migrants' [5], the New York Declaration for Refugees and Migrants [6,7] and, recently, the Global Compact on Refugees in 2018 [8].
Thailand is one of the most popular destinations for international migrants and refugees in Southeast Asia. The majority of migrants are workers from Cambodia, Lao PDR, Myanmar and Vietnam (CLMV collectively). Some of them have entered the country unlawfully and are known as undocumented migrants. It is estimated that today, there are more than three million migrant workers living in Thailand [9].
The Thai government has implemented policies to protect the well-being of undocumented migrants for several years. One remarkable policy is the One Stop Service (OSS) registration measure for undocumented CLM migrants and their dependants [10]. Migrants who register with the OSS have their profile recorded in the civil registry and acquire a work permit, alongside undertaking nationality verification (NV). The Ministry of Public Health (MOPH) a lso instigated a nationwide public insurance policy, called the ' Health Insurance Card Scheme' ( HICS) , for these registered migrants and their dependants. The HICS benefit is comprehensive, covering inpatient (IP) care, outpatient (OP) care, high-cost care, disease prevention and health promotion [11]. almost everybody on its soil [12,13].
While undocumented migrants seem to be in the spotlight of health policies in Thailand, refugees and asylum seekers are often neglected [ 14] . All of the aforementioned policies do not include refugees and asylum seekers. The situation is more complicated among refugees and asylum seekers in urban areas compared with those in temporary sheltered areas or camps (camp refugees). This is because implementing health measures in a well-defined geographical space is relatively straightforward, and local healthcare providers are well aware of the existence of refugees in the camps. Besides, the United Nations High Commissioner for Refugees (UNHCR) and a number of international non-governmental organizations (NGOs), such as Médecins Sans Frontières and the International Rescue Committee, in coordination with public facilities along the border, have provided humanitarian assistance in the refugee camps for years [15,16].
Unlike camp refugees, urban refugees and asylum seekers (URAS) received little attention from the public health spehere in Thailand. Almost all URASs are residing in Bangkok, under the patronage of the United Nations High Commissioner for Refugees (UNHCR). So far, there are about 5,000 URASs and 97,000 camp refugees [17,18]. URASs are not covered by the HICS, nor by the public insurance schemes originally designed for Thais. Nonetheless, some private facilities or insurance companies have initiated a health insurance package for URASs, which are conditional upon affordability. Some media or local NGOs suggest that most URASs in Thailand face many hindrances in accessing health services, for instance, poverty, language difficulty, and precarious citizenship status [19,20].
Moreover, some government officials are even unaware of the existence of URASs [19]. Also, a systematic evaluation on the degree of healthcare access for URASs in Thailand is lacking.
Therefore, the objective of this study is to explore the degree of healthcare access among URASs, in comparison with the Thai population. In this regard, we use 'unmet need' for health services as an indicator to gauge the ability to access health care. The concept of unmet need originates from the reproductive health field, but during the past two decades, its application has become widespread to other fields, including population health and critical care [21][22][23].

• Study design, populations and samples
Both primary and secondary data collection was applied. We performed a cross-sectional survey on URASs from October to December 2019, and examined prior survey data on the Thai population through the 2019 Health Welfare Survey (HWS). HWS is a nationwide biennial survey jointly conducted by the National Statistical Office (NSO) and the International Health Policy Programme (IHPP) of the MOPH. We first contacted the Bangkok Refugee Centre (BRC), a charitable agency in collaboration with UNHCR, whose work is to support the well-being of URASs. For this study, we focused on URASs of the top-ten most common nationalities in Thailand: namely, Pakistani, Vietnamese, Cambodian, Somali, Afghan, Palestinian, Chinese, Sri Lankan, Iraqi, and Syrian, comprising 3,021 URASs in total. We then sampled 206 URASs from the pool of 3,021 URASs in the BRC roster (more details in 'Sample size calculation, sampling methods and survey design'). Among these 206 samples, 181 completed the survey questionnaire. Once the primary survey on URASs was completed, we combined the data of these 181 URASs with Thai data from HWS, focusing on those living in Bangkok (n = 2,941). The final dataset comprised 3,122 observations in total, see  All selected participants were asked to travel to BRC to complete the paper questionnaire. For those who had difficulty in travelling, a phone interview was performed instead. For a child below 15 years of age, parents or legal guardians would respond on his or her behalf. The questionnaire was translated to the respondents' own language. For those who had reading difficulty, verbal interview was performed in place of written questionnaire. On average, each respondent took approximately thirty minutes to complete the questionnaire. A focal coordinator was prepared for each nationality group. These coordinators were volunteers working with BRC. Preparatory meetings between the research team and focal coordinators was arranged prior to the survey in order to fine-tune the understanding and to assess the survey feasibility.

• Operational definitions
We set operational definitions as follows. Firstly, 'refugee' is a person who has been forced to flee his or he country because of persecution, war or violence and his o r her request for sanctuary is ratified by the UNHCR according to the 1951 Refugee Convention [26]. Secondly, asylum seeker means someone who has been forced to flee his or her country because of persecution, war or violence and his o r her request for sanctuary has yet to be processed by the UNHCR according to the 1951 Refugee Convention [26]. Lastly, unmet need refers to a status where a person reported that he or she needed health examination or treatment for any type of health issues within the past twelve months, but he or she did not receive or did not seek it. This definition is adapted from the original unmet need survey by EU-SILC [27].

• Questionnaire and determinants of interest
The questionnaire for the URAS survey was adapted from the HWS questionnaire. Two rounds of consultative meetings between the research team, health system academics and BRC staff were arranged to ensure content validity and to make sure that the participants clearly understood the We also conducted mixed-effects (ME) logistic regression, having done multivariable logistic regression at a prior stage. This time, the ME model took the nationalities of the participants into account. We categorised nationalities into three main clusters: Thai, non-Arab Asian, and Arab Asian.
The results were presented in terms of crude and adjusted odds ratios (OR) with 95% confidence interval (CI). Inverse probability weighting was applied when assessing statistical significance in order to take the survey design into account.

• Subgroup analysis
Subgroup analysis was exercised by limiting the analysis on URASs. We then broke down the degree of unmet need by nationalities and types of URASs (urban refugee versus asylum seeker). The analysis was performed in the same fashion as the full-sample analysis.

• Subgroup analysis
Subgroup analysis found that there was no significant difference in the unmet need in urban refugees, relative to asylum seekers (P-value = 0.523 for OP care and 0.549 for IP care), Figure 3.

• Result discussion
To our knowledge, this piece of work is among the first few studies in Asia that quantitatively investigate the degree of healthcare access through the perspective of unmet need among URASs.
From a macro-perspective, the demographic data showed that most URASs were relatively younger, had lower educational backgrounds, and were living in economically deprived households. With this finding, it is not an exaggeration to state that URASs are one of the most vulnerable groups in Thailand. The evidence points to about one fifth to one quarter of URASs faced unmet need for health services while the prevalence of unmet need in the Thai population was very small.
Determinants that potentially contributed to unmet need included increasing age, low education achievement, and, most prominently, the lack of health insurance. This finding coincides with the same discovery in some other foreign studies. Wang et al suggested that higher education background was negatively associated with unmet need for supportive care among Chinese women [29]. Hailemariam and Haddis also flagged that low levels of education resulted in increasing degrees of unmet need for family planning in the Ethiopian population [30]. Bhattathiry and Ethirajan reported that unmet need for family planning decreased as age advanced [31]. This finding contradicts our discovery, which found that people with advanced age were more likely to have unmet need than those in lower age groups. Some of the explanations for this phenomenon is, gigantic. This finding also corresponds with the fact that the majority of URASs pointed towards financial difficulties to afford the treatment cost as the most important concern. In other words, URASs are at huge risk of impoverishment at any time when they seek treatment, and it means that Thailand has not yet achieved UHC for everybody on its soil as intended [32]. Since the concept of UHC covers not only the provision of essential quality health services, but also the prevention of impoverishment from healthcare spending, the issue of URASs accessing health care has a strong policy implication. Thailand is committed to the Sustainable Development Goals (SDG), including SDG target 3.8, which focuses on UHC [33]; therefore policies to enroll URASs in a public health insurance scheme should be seriously considered. In addition, leaving URASs uninsured potentially results in low access to essential healthcare, and this may undermine the health security of society as a whole. Experiences from other countries that offer health insurance for URASs, such as Iran and Malaysia, are of great value and warrant further exploration [34,35].
As Thailand is not a party to the 1951 Refugee Convention [36], the Thai government is not obliged to guarantee any health measures for urban refugees, compared with asylum seekers whose application for refugee status is still in process. The subgroup analysis reflected this fact, showing no significant difference in the unmet need for healthcare in urban refugees, relative to asylum seekers.
Despite not being a primary objective of the study, the varying degree of unmet need among diverse national groups was thought-provoking. This was evidenced by the fact that the adjusted OR in the ME model, which had already considered the clustering effect of nationalities on unmet need, greatly expanded, comapred with the ratio in the multivariable logistic regression, which assumed no correlation between observations. The descriptive subgroup analysis also showed that Cambodian and Vietnamese URASs suffered least from unmet need, compared with other nationals.
A possible explanation is that URASs from Southeast Asia nations may have lifestyle and beliefs close to Thais; and that Thai society is already acquainted with migrants travelling from neighbouring countries (especially from CLMV nations). In contrast, URASs from Arab nations (for instance, Iraqis, Palestinians and Syrians) presented with a relatively large degree of unmet need.
As the Arab people are the minority in Bangkok, they possibly need a huge adaption to incorporate the Arab way of life to the Asia culture. This picture alludes to the concept of acculturation proposed by a great deal of prior research [37][38][39]. That is, refugees who can assimilate or integrate themselves into a new culture tend to have better health outcomes, compared with the poorly adjusted ones [37][38][39].

• Methodological discussion
Methodology-wise, this study bears some strengths and weaknesses. Regarding strengths, the study employed a systemic approach for data sampling, and we recruited participants from a household level, even though there were no physical visits to the participants' households. Another strength of the study is the use of Thai respondents' data as a comparator. We would not have a clear view on the extent of unmet need for health services in URASs had the comparator (HWS data) been missing.
However, there remain some weaknesses. Firstly, as the nationalities of URASs are vastly diverse, we could not guarantee a perfect translation of the questionnaire. This problem would rarely occur in the HWS questionnaire as Thai is the only formal language for Thai citizens. Nonetheless, we tried to minimize the language barriers by arranging a training workshop for the survey volunteers to achieve mutual understanding between the volunteers and the research team. These volunteers mostly worked with BRC and some of them were also URASs.
Secondly, since the unmet need question inquired about a history of healthcare access in the past twelve months, a recall bias was inevitable. This problem might not severely undermine the validity of the analysis as the bias could be present in both the URAS survey and the HWS. However, the bias might be more pronounced in the URAS survey compared with the HWS because of the difference in survey practice. In the URAS survey, when people with travelling difficulties were recruited, we asked a surrogate respondent to answer the questionnaire on their behalf. In contrast, the HWS surveyors always visited the participants at their households, resulting in a lower reliance on surrogate respondents in comparison with the survey on URASs.
Thirdly, as mentioned earlier, we could not perform physical visits to participants' households.
Accordingly, some key household information that necessitates direct observation, such as household infrastructure and owner's equity, was missing. Such information serves as the main ingredient for estimating household prosperity through the indicator called 'asset index' [40]. The lack of this indicator, in combination with a fair amount of missing data on household economy, might explain why the economic wealth of URASs did not exhibit a statistically significant relationship with unmet need, although the direction of effect implied that the less affluent participants tended to face greater odds of unmet need, compared with the well-off group. The original HWS questionnaire contains questions about household properties, and the surveyors were able to use the answers from these questions to estimate asset index. However, we dropped such questions in the URAS survey as we decided not to visit URASs' household to avoid a risk of disclosing their residential address.
Fourthly, though the URAS survey and HWS followed the same set of questions, the timeline for conducting both surveys and human resources used were different. Therefore a direct comparison between URASs and Thais should take into account this limitation.
Lastly, the people of interest in this study were those presenting on the BRC roster only, not all URASs in Bangkok., We totally missed URASs in non-household settings, such as shelters or detention centres. This definitely limits the generalisability power of our study. To expand the academic richness in this field, further studies on other types of refugees are strongly recommended.

Conclusion
Overall, URASs had lower educational attainment and faced more severe financial hardship than instead. All respondents were assured that their participation was voluntary and they had the right to withdraw from the survey at any time. All individual information was strictly kept confidential and would not be reported to the wider public.

• Consent to publish
Not applicable

• Availability of data and materials
The raw data used by this study jointly belonged to BRC and IHPP. The analysed data are however available from the authors upon reasonable request.

• Competing interests
The authors declare no conflict of interest.

• Funding
This study received funding support from the Health Systems Research Institute, Thailand.
• Author contributions

• Acknowledgement
We are immensely grateful for the support from BRC, IHPP and UNHCR during the survey process.
The advice from Ms Bongkot Napaumporn and Dr Herve Isambert is hugely appreciated.