Menu
Menu
Menu
Menu
Background: The vast majority of refugees are hosted in low and middle income countries (LMICs), which are already struggling to finance and achieve universal health coverage for their own populations. While there is mounting evidence of barriers to health care access facing refugees, there is more limited evidence on equity in access to and affordability of care across refugee and host populations. The objective of this study was to examine equity in terms of health needs, service utilisation, and health care payments both within and between South Sudanese refugees and hosts communities (Ugandan nationals), in two districts of Uganda. Methods: Participants were recruited from host and refugee villages from Arua and Kiryandongo districts. Twenty host villages and 20 refugee villages were randomly selected from each district, and 30 households were sampled from each village, with a target sample size of 2400 households. The survey measured condition incidence, health care seeking and health care expenditure outcomes related to acute and chronic illness and maternal care. Equity was assessed descriptively in relation to household consumption expenditure quintiles, and using concentration indices and Kakwani indices (for expenditure outcomes). We also measured the incidence of catastrophic health expenditure- payments for healthcare and impoverishment effects of expenditure across wealth quintiles. Results: There was higher health need for acute and chronic conditions in wealthier groups, while maternal care need was greater among poorer groups for refugees and hosts. Service coverage for acute, chronic and antenatal care was similar among hosts and refugee communities. However, lower levels of delivery care access for hosts remain. Although maternal care services are now largely affordable in Uganda among the studied communities, and service access is generally pro-poor, the costs of acute and chronic care can be substantial and regressive and are largely responsible for catastrophic expenditures, with service access benefiting wealthier groups. Conclusions: Efforts are needed to enhance access among the poorest for acute and chronic care and reduce associated out-of-pocket payments and their impoverishing effects. Further research examining cost drivers and potential financing arrangements to offset these will be important.
Uganda is host to the third largest population of refugees globally, and the largest in sub-Saharan Africa, estimated at 1.5 million people [1]. Approximately 960,000 of these originate from South Sudan [28], displaced as a result of ongoing violence which broke out in December 2013 [29]. The vast majority of South Sudanese refugees live in seven settlements alongside host communities in the northwest of the country [30]. Uganda has a pluralistic health system, with government health facilities (which provide care without formal user fees) operating alongside private and not-for-profit organisations [27]. In 2018, the year of the latest available national health accounts, out of pocket payments constituted 41% of total health expenditure, while government expenditure represented just 16% [31]. Historically, refugees had access to dedicated primary healthcare facilities which were managed and funded by the United Nations High Commissioner for Refugees (UNHCR), with access to government hospital referral services. This resulted in perceived inequities between host and refugee communities in service access and tensions between these communities. However, from the early 2000s, refugee and host health services were integrated into a single health system under local government control. All health facilities in refugee settlements are now owned by the Ugandan government, and health services are run and funded by the government, with additional financial and material support provided by the UNHCR and other humanitarian actors towards health services within refugee districts [10, 32, 33]. The integration of refugee and host health services means that host populations can access the same health facilities as refugees for free [34, 35]. South Sudanese refugees in Uganda have substantial need for mental health and psychosocial support as a result of their experience of conflict and displacement [36]. There is also evidence that refugees in Uganda have greater health needs than those in host communities, with 51 percent of refugee households in Uganda defined as in need, compared to 17 percent of host households [37]. This study was conducted in two districts in Uganda: Arua, in the Northern region of the country, and Kiryandongo, in the Western region. Both districts are home to sizeable settlements of South Sudanese refugees: Rhino Camp (in Arua) has a population of 131,000, and the Kiryandongo refugee settlement has a population of 75,000 [28] . Rhino Camp is 70km from the main host town Arua, while the Kiryandongo settlement is just 5km from Bweyale town. Refugees in Rhino Camp tend to be more recent arrivals than those in Kiryandongo, who have generally been living in Uganda longer [28]. Participants were recruited from host and refugee villages from Arua and Kiryandongo districts. Every village was defined as either a host village if more than 50% of the population were Ugandan nationals, or a refugee village otherwise. A two-stage sampling approach was taken. Twenty host villages and 20 refugee villages were randomly selected from each district. Thirty households from each village were selected randomly through door-to-door household visits, and were eligible for inclusion if the household contained at least one woman of reproductive age (15-49 years). In host villages, only host households were eligible for inclusion in the survey, and in refugee villages only refugee households were eligible. In households with more than one woman of reproductive age, a pre-assigned table of random numbers was used to randomly select an interviewee, so no two women were interviewed from the same household [38]. A sample size calculation was undertaken based on estimating a prevalence of 50% for any given outcome, 30 respondents per cluster and intra-cluster correlation coefficient of 0.1 (design effect 3.9), and a 95% confidence interval and margin of error of 4%. This gave a target sample size of 1200 refugee and 1200 host women of reproductive age, or 80 villages. The survey was conducted through in-person interviews in English and Arabic, and data was entered on tablets using SurveyCTO platform. Interviews lasted approximately 45 minutes and were conducted by a team of 25 trained enumerators in July-August 2019. Data were collected on individual and household characteristics, healthcare need, care seeking and costs of care as outlined below. We examined equity across illness incidence as a measure of health need, health care seeking and health care expenditure outcomes. We included three measures of healthcare need, for acute sickness, chronic sickness and maternal care. Households were defined as having experienced acute sickness if anyone in the household had a short-term illness in the two weeks preceding the survey, and chronic sickness if anyone in the household had a long-lasting illness in the preceding month. Acute sickness was defined as an illness or injury that occurs suddenly with a rapid onset, and tends to resolve quickly on their own or with medical treatment, or is so severe and fast acting that a patient does not survive. Chronic sickness was defined as an illness which has a slow progression that builds over time, and tends to be a long lasting problem requiring multiple visits to a health facility. Women were defined as having need for maternal care if they reported having delivered a baby in Uganda in the preceding year. Care seeking for households with acute or chronic sickness was defined as the sick household member having attended a public health centre, hospital or private clinic (excluding traditional healers, herbalists and community health workers) during the recall period (two weeks for acute sickness and one month for chronic sickness) . The two measures of cares seeking for maternal care were four or more antenatal care (ANC) visits, and facility-based delivery. Women were defined as having had four or more ANC visits if they reported seeking ANC at least four times during their pregnancy, regardless of timing or facility type. They were defined as having a facility-based delivery if the baby was delivered at a health centre, hospital or private clinic. Annual healthcare expenditures were estimated for acute and chronic care. Annual acute and chronic care costs were defined as the total expenditure on medicines, tests and consultation fees for any acute or chronic sickness in the household in the preceding two weeks (for acute care) or one month (chronic care), multiplied by 26 (acute care) or 12 (chronic care) to estimate annual expenditures. Maternal care costs were defined as the total expenditure on all reported components of ANC received before a delivery in Uganda in the last year and all costs of the delivery (including fees, and medicines). We estimated equity in relation to household wealth, measured as reported monthly household consumption expenditure, and adjusted per adult equivalent. Consumption expenditure was considered a more reliable measure of ability to pay than income, which is often under-reported in developing countries [39]. We generated two measures of financial protection to monitor progress towards UHC, catastrophic health expenditure and impoverishment by health expenditure. Catastrophic expenditure was defined for categories of care (acute, chronic, and maternal) and all care at two thresholds, 10% and 25% of household expenditure. Households were defined as poor if their daily consumption expenditure per adult equivalent was below the international poverty line of 1.9USD per day (2011 PPP prices). The number of adult equivalents in the household was calculated using the formula adult equivalents = (children × 0.33 + adults)0.9. As household size was reported categorically, households reporting 5 or 6 members were given a nominal size of 5.5, and households reporting 9+ members a nominal size of 9. As we did not have numbers of children and adults in the household separately, the number of household members who were adults was estimated indirectly from other questions about the respondent’s circumstances. We did this by summing the respondent herself, and any adults the respondent reported living with either as husbands, non-husband heads of household, or other women of reproductive age, as the number of adults. All other members of the household were assumed to be children. Households were defined as being impoverished by health expenditure if their adult equivalent consumption expenditure exceeded $1.90 per day, but subtracting their total annualised health expenditure from their total annualised expenditure pushed them below the international poverty line. We generated quintiles of household consumption expenditure by ranking households based on expenditure per adult equivalent and dividing them into five equally sized groups from richest to poorest. We first described the characteristics of the overall sample, as well as separately for refugees and hosts, using descriptive statistics. We compared these to the same statistics drawn from the 2016 Demographic and Health Survey for the national Ugandan population [40]. We analysed equity in relation to condition incidence, service utilisation and expenditures across household wealth quintiles, measured by their reported monthly household consumption expenditure per adult equivalent. We estimated concentration indices to assess whether the distribution of outcomes was pro-rich (positive index value) or pro-poor (negative index value) [39] using the conindex command in Stata. The Gini index was used to estimate wealth inequality, and the Kakwani index, defined as the Gini index subtracted from the concentration index (CI), was used to estimate the progressivity of health expenditures [41]. The Gini index can vary between 0 and 1, with 0 indicating perfect wealth equality (the population having exactly the same wealth) and 1 maximum inequality. The Kakwani index can vary between -1 and 1, with negative values indicating the poor contribute a higher share of their income, or a regressive financing system, 0 indicating perfectly proportional expenditure and positive values indicating progressive contributions, whereby the richest pay a higher proportion of their income. Dominance tests were carried out to ascertain whether the concentration indices were significantly pro-rich or pro-poor, and whether the Kakwani index was significantly progressive or regressive. We finally estimated the incidence of catastrophic and impoverishing health expenditure outcomes across the whole sample, and comparing refugee and host communities.
N/A
