Uganda is host to nearly 1.5million conflict refugees from neighbouring countries of Rwanda, Burundi, South Sudan and Democratic Republic of Congo . These refugees exhibit poor living conditions, pre-existing mental and psychosocial challenges [38,39,40]. The inaction caused by COVID-19 preventive measures, starting from the first case identification on 21 March 2020, including restrictions on mass gatherings, public transport, entry and exit at border points, and lockdown on several social services  is envisioned to have impacted further on psychological distress in this already vulnerable population. Indeed, emerging data in vulnerable groups of refugee women and slum-dwellers suggest that COVID-19 has exacerbated their risk for stigma, all forms of violence and financial disadvantage [41, 42].While researchers have predicted the economic, psychosocial, physical, and other consequences of COVID-19 on refugees/migrants in Uganda and beyond based on previous epidemics [43,44,45,46], there is a dearth of evidence on the burden of psychological distress, social support and strength/nature of associations between these phenomena in refugee settings during the COVID-19 pandemic. The current work intends to fill this gap in the evidence using Uganda as a case study. Such data could be useful in the design of interventions to cushion psychosocial problems among refugees through the modification of social support agents such as availability and adequacy.
Study site and population
We conducted the research at 3 large refugee settlements in different regions of Uganda, hosting over 400,000 refugees:
Kisenyi, refugee settlement, an urban refugee setting in the centre of the capital city (Kampala) hosting over 70,000 refugees of mainly Somali origin. The refugees live integrated with their host.
Kyaka II refugee Settlement in the Southwestern part of Uganda, a semi-rural refugee setting hosting multinational refugees from the Democratic Republic of Congo (DRC), Burundi and Rwanda totaling approximately 124,000 refugees. The refugees live partly segregated from their host but with freedom of movement and shared services. The region can be considered as semi-rural, with a blend of rural and urban activities (e.g. farming and industrial activities)
Adjumani refugee settlement in North-West Nile Uganda, hosting about 214,000 refugees predominantly of South Sudanese nationality. The refugees live rather segregated from their host but with freedom of movement and shared services. The region is considered as rural, with agriculture as the main activity.
Cross-sectional survey data on various health and social indicators was gathered from 1014 refugees randomly selected from each of the study sites. For the current study, data on psychological distress, social support, demographic, social and behavioral indicators was of primary interest.
Participants were sampled using a two-staged cluster sampling procedure in each settlement. The first stage involved selecting clusters of zones in the main settlement using systematic random sampling with probability proportional to zone size (PPS). The second stage constituted systematic random sampling of households in selected zones. For instance, a Research Assistant picked a household at random within the zone, and then decided to visit every nth household (e.g. 3rd) moving in a specific direction (e.g. eastward) until his/her required quota (e.g. 50 households) was completed. Random numbers procedures were used to choose one adult household member (i.e. 15 years and above) from among all adults in the household to constitute the final participant. This procedure resulted in 1014 refugees, with the following distribution among the settlements: Adjumani n = 342; Kyaka 354; Kisenyi n = 318.
Thirty Research Assistants (RAs) were trained to collect data using mobile tablets, in a bid to reduce inter-individual contact and the risk of COVID-19 spread during interviews. The training oriented RAs on the purpose of the study; ethical considerations; data collection methods and tools; COVID-19 prevention, symptoms, measures and precautions; and standard operating procedures (SOPs) in fieldwork in light of COVID-19. The training also involved testing of the data collection tool among a purposively selected refugee sample of n = 30 in each of the 3 settlements, from zones neighboring but not included in the main study. Slight adjustments were made to data collection tools following this exercise.
Informed consent was received from all participants, and confidentiality was considered by inquiring of participants whether they felt safe to partake in study, emphasizing that participation was voluntary, and giving the participant liberty to choose whether he/she preferred another time and/or venue for the interview. The potential risk and benefits of the study were explained to all participants and in light of the heightened risk of COVID-19 transmission, we developed Standard Operational Procedures (SOPs) for protection of refugees as well as data collectors, guided by Safety and Security Strategy for COVID-19 of the World Health Organization (WHO) and Uganda Ministry of Health COVID-19 guidelines.
The study was approved by the Makerere University School of Public Health Institutional Review Board (MakSPH IRB) and the Uganda National Council of Science and Technology (UNCST), the two bodies governing academic research in Uganda. Additionally, the Ministry of Health (MoH), Kampala Capital City Authority (KCCA) and the Office of the Prime Minister (OPM), which is in charge of refugee affairs, gave clearance for execution of the study.
Data collection tools and study variables
A comprehensive questionnaire covering several areas of relevance to public health and COVID-19 was developed. For the current study, the following variables were of interest.
The dependent variable for the study was psychological distress, measured using Kessler’s Psychological Distress Scale (K-10) , a 10-item instrument measuring distress in terms of feelings of nervousness, hopelessness, tiredness, restlessness, fidgety, depressed mood, sadness, worthlessness, cheerlessness and loss of effort, during the past 14 days, with a 5-level response ranging from none of the time (score 1) to all of the time (score 5). A composite score for psychological distress is calculated for each participant as the sum of responses to the the 10 items. Thus, individual scores for psychological distress scale ranged from 10 to 50, with higher scores indicative of higher psychological distress. Cronbach’s alpha testing for internal consistency/reliability of K-10-Scale for the current sample was 0.91 indicating very high reliability.
The main independent variable for this study was social support, with the aim to assess its association with psychological distress, and whether such associations differ between refugees in rural, semi-rural and urban settlements. Social support was measured using a modified version of the Interview Schedule for Social Integration (ISSI) , which assesses social support in terms of the Availability and Adequacy of Social Interaction and Social Attachment.
Availability of Social Interaction (AVSI) was assessed using six items inquiring of participants to indicate if they have anyone/persons: with whom they have common interest, meet and talk to regularly, can speak with openly, can borrow things from and can turn to when in trouble. This was coded as 1 if the answer was in affirmative and zero if the response was”No”. A composite score was formed ranging between 0 and 6 to represent, with higher scores indicative of higher availability. The participants were in addition requested to rate the Adequacy of these person/persons by inquiring if they desired more (coded as 1), less (coded as 1) or no change (coded as zero). Desiring “more” or “less” in a specific item was considered “inadequate”, while desiring neither more nor less was considered “adequate”. Thus, Adequacy of Social Interaction (ADSI) was rated on a total scale ranging between 0 and 6, with higher scores indicative of lower adequacy. Cronbach’s alpha testing for internal consistency/reliability of Availability and Adequacy of Social Interaction respectively for the current sample was 0.71 and 0.81 respectively, indicative of good reliability.
Availability of Social Attachment (AVSA) was assessed based on six items inquiring of participants to indicate using a “Yes” (coded as 1) or “No” (coded as 0) response regarding whether there is someone special: from whom they derive support, they feel close to, they share happy moments, they can embrace for comfort, who appreciates what they do, and with whom they can share inner thoughts. For social attachment, composite individual scores are calculated as the sum of responses to each item. Thus, scores for social attachment range between 0–6, with higher scores representing higher availability. Adequacy of Social Attachment (ADSA) was assessed by inquiries to participants on whether they desired more (coded 1), less (coded 1) or no change (coded 0) regarding the mentioned attachments. Thus, scores for Adequacy of Social Attachment ranged between 0 and 6, with higher scores indicative of lower adequacy. Cronbach’s alpha testing for internal consistency/reliability of Availability and Adequacy of social attachment respectively for the current sample was 0.55 and 0.87 respectively, indicating low and high reliability respectively.
Other independent variables included in the study were:
Demographic and Social characteristics i..e. refugee settlement (Rural, Semi-rural, Urban), nationality (South Sudanese, Congolese, Somali, Rwandese, Burundian), gender, age, marital status, religion, income (earnings per week), employment status and education (highest level achieved).
Behavioral characteristics assessed by indicators including: alcohol use which were assessed by asking participants if they took alcohol regularly (with “Yes/No” response), smoking assessed by inquiring of participants if they currently smoke (with “Yes/No” response)and physical activity assessed by inquiring of participants how often they engaged in exercise in a week (with response options “never”, “once”, “2–3 times” and “4 or more times”). As these variables are from previous studies generally known to be associated both with social support and psychological distress, it is prudent to adjust for them in the main analyses to control for possible confounding.
COVID-19 symptoms were assessed by asking participants if they had currently or within the past 14 days experienced/exhibited symptoms of coughing, sneezing, running nose, sore throat, difficulty in breathing, loss of taste, and loss of smell. The number of symptoms was calculated per individual and used in the analyses to represent COVID-19 risk. This variable therefore ranged from 0 to 8, with higher scores indicative of higher risk of COVID-19 transmission.
Detailed categorization of all variables are shown in Table 1.
Cronbach’s Alpha coefficients were calculated to assess for reliability (internal consistency) of the dependent variable and the main independent variables i.e. Kessler’s Psychological Distress Scale and ISSI sub-scales) in the current sample. To compare the burden of psychological distress between rural, semi-rural and urban refugee populations, Analysis of Variance (ANOVA) was used, and post hoc tests with Bonferroni correction applied to account for multiple pairwise comparisons of means between the three sub-populations. Similarly, to examine differences in availability and adequacy of social support between refugees in rural, semi-rural and urban settlements (ANOVA) were used, with post hoc tests according to Bonferroni method applied. To assess for bivariate associations between psychological distress and sex, independent sample t-tests was used. To assess for bivariate associations between psychological distress and settlement, nationality, occupation and religion ANOVA (contrasting 3 or more means) was used respectively, with post hoc corrections according to Bonferroni method. To assess for bivariate associations between psychological distress and age, education, smoking, alcohol use, exercise and social support indicators respectively, Pearson’s Correlations tests was used respectively.
To assess the independent association between psychological distress and social support while controlling potential confounder, all independent variables exhibiting statistical significance in the bivariate tests were entered in Multivariable Linear Regressions (MLR). Assumptions for Multiple Linear Regression (MLR) i.e. linearity, homoscedasticity of variance and multicollinearity were met. Deviations were noted regarding the normality assumption though a preference not to transform the data was adopted due to several reasons as discussed under study limitations.
Ordinal and continuous independent variables that were significant in the bivariate analyses were entered in the regression in their original form, while nominal variables were transformed to dummy variables prior to entry in the regression. Regressions were run for the entire sample to assess associations between psychological distress and social support among refugees in general, while controlling for potential confounders. Additionally, regressions analyses stratified by settlement were run to compare the association between psychological distress and social support between settlements (i.e. rural, semi-rural and urban) while controlling for potential confounders. The same variables entered in the un-stratified analyses were included in the stratified analyses except for settlement (the stratification variable), which was excluded in the stratified analyses.
SPSS version 22 was used for all analyses and a statistical significance of p < 0.05 assumed for all tests.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.