Health-related quality of life in refugee youth and the mediating role of mental distress and post-migration stressors

Syrians were the largest group of forced migrants to enter Norway at the time of the study, and 5553 were registered in the age groups recruited in 2018 [25]. The inclusion criterion for age was 12–24 years, as they have the right to attend secondary and upper secondary schools. Using strategic sampling to recruit recently resettled youth, 40 schools with introductory classes for newly arrived immigrants were contacted. 23 schools located in nine different regions of Norway agreed to participate. Reasons for not participating were no response to request (nine schools), no Syrian students (six schools) or already participating in other studies (two schools). The participating schools had between one and 23 Syrian students attending. Three students declined to participate due to exam preparations or language difficulties. The recruitment period lasted from May to December 2018 and a total of 160 youth from Syria were included in the final study sample.

The present study utilised a cross-sectional, questionnaire-based design. Questionnaires were administered in both Arabic and Norwegian. The teachers distributed written information about the study in Arabic and Norwegian to the students in advance, and consent forms to parents with children under the age of 16. The youth consenting to participate completed the questionnaire whilst at school with a researcher present to answer questions.

All measures had validated Arabic language versions available and permission to use these were sought from all copyright holders. Measures for PTE and Post-migration stressors were translated from Swedish to Norwegian and reviewed to detect and remove any discrepancies in meaning. A cultural expert reviewed the questionnaire for cultural appropriateness and comprehensibility, and a pilot was conducted in a refugee learning centre.

Our conceptual model was based on previous theory and included pre-migration PTEs as the main predictor, post-migration stressors and mental distress (1. general mental distress and 2. PTSD) as potential mediators and HRQoL as the outcome (see Fig. 1).

The mediating effect was examined by a regression-based approach in the SPSS PROCESS 3.4 macro (model 6), with 5000 bootstrap samples in the procedure suggested by Hayes [30]. This serial mediation procedure enables the isolation of each mediator’s indirect effect as well as the indirect effect passing through both mediators in a series. Considering that the mediators could act in reverse order to the conceptual model, this was tested in separate models. An indirect effect is assumed to be significant at an alpha level of 0.05 if its 95% confidence interval (CI) does not include zero. Gender and age were controlled for as covariates. Initially, residence time was also controlled for, but as this was not significant in the final models, it was removed to increase power of the estimates. Missing values from KIDSCREEN were replaced according to the KIDSCREEN manual using calculated estimates based on Rasch analyses [38], whilst responses missing two or less items in HSCL-10 or CRIES-8 were replaced with individual means. Responses missing more items were categorised as below cut-off values (CRIES-8 and HSCL-10) when assessing the prevalence as not to inflate frequency of cases but were not included in further analyses. A total of 19 participants opted for a shortened questionnaire due to language difficulties. This shortened questionnaire did not contain HSCL-10, CRIES-8 or post-migration stressors. General descriptive statistics, t-tests and correlation analyses were used to explore the variables.

The sample included 160 youth from Syria (37.5% female, mean age 18 years) (Table 1). All participants attended school full time and were located in either local secondary or upper secondary schools or adult learning centres. The majority were living with their parents (75%) and of the ones that were not, mostly were > 19 years of age (M_age = 20.1 years, range 16–24 years), lived alone and were mostly male (89%). The average time since they left Syria was 5.3 years, and the mean residence time in Norway was 2 years. Most had Arabic as their mother tongue (76.3%).

The means and correlation coefficients are presented in Table 2. The participants (n = 151) reported a mean of 4.5 PTEs (SD = 2.64), the most prevalent being witnessing war (68%), feeling your life was in danger (59%) and seeing someone die (55%). A total of 88% reported at least one event, and 61% four or more events. Youth older than 18 years had experienced significantly more PTEs (19–24 years: t(151) = -3.60, p < 0.001) and boys more so than girls (Boys: t(151) = -3.43, p = 0.001). Scores for post-migration stressors were an average of 13.56 (SD = 7.56), and dichotomising scores into experiencing each stressor rarely (never, rarely, sometimes) or often (often or very often) showed that participants (n = 127) reported an average of 2.2 post-migration stressors (SD = 2.0) they experienced often. The most common were missing family (33%), economic concerns (29%), missing previous life (28%) and language problems (26%). Least common was feeling unsafe (8%) and perceived discrimination (6%). There were no gender differences, but again the youth over 18 years reported more frequent stressors (t(123) = -2.19, p = 0.031). The mean score for general mental distress (HSCL-10) (n = 132) was 1.77 (SD = 0.61) and 36% had scores above the recommended threshold (≥ 1.85) considering an indication of mental distress [33]. Almost half the participants (48%) had scores above the recommended threshold for PTSD (≥ 17) [36]. The mean score for global HRQoL (44.63, SD = 10.0) was slightly lower than norm data (48.51 SD = 9.28) and 26% reported low HRQoL [38]. All the main study variables were negatively correlated to the main outcome with moderate-to-high correlation coefficients (Small = 0.1 to 0.3, medium = 0.3 to 0.5, large = 0.5 to 1). The negative correlation between PTEs and HRQoL (r = -0.318, p < 0.001) supports the first hypothesis. No correlations between the independent (PTEs) and the mediator variables (Post-migration stressors, mental distress/PTSD) exceeded the 0.80 multicollinearity threshold suggested by Field [39].

The first analysis with post-migration stressors and general mental distress as mediators is presented in Fig. 2.

Supporting our second hypothesis of post-migration stressors as an independent mediator, we found that PTEs influence quality of life indirectly through post-migration stressors and that this indirect effect was significant (indirect effect (a₁ and b₁): -0.32, 95% CI (-0.66, -0.03)). Hypothesis three was not supported, as general mental distress did not act as an independent mediator (indirect effect (a₂ and b₂): -0.27, 95% CI (-0.94, 0.67)). However, as the relationship between the mediators (path d₂₁) was positive and significant, there was support for a serial mediation model (hypothesis 4) (indirect effect (a₁ d₂₁ b₂): -0.19, 95% CI (-0.42, -0.05)). The model including these variables accounted for 39% of the variance in HRQoL scores (F(5,111) = 13.97, p < 0.001, R² = 0.386).

Repeating the same analysis, but replacing general mental distress with PTSD, reiterated the support for the second hypothesis of post-migration stressors as an independent mediator (indirect effect (a₁ and b₁): -0.46, 95% CI (-0.89, -0.14)), see Fig. 3. Contrary to general mental distress, PTEs were significantly associated with PTSD, even after controlling for post-migration stressors (path a₂). There was also a significant association between the two mediators. In fact, PTEs and post-migration stressors contributed to 41% of the variance in PTSD (F(5,109) = 19.15, p < 0.001, R² = 0.412). However, when controlling for PTEs, there was no significant association between PTSD and quality of life (path b₂), and hypotheses 3 and 4 were therefore not supported in this model. Considering that the direction of influence is unknown, we reversed the order of the mediators in a new analysis (see Fig. 4). This reversal resulted in the confirmation of hypothesis 4 also for PTSD with a small, but significant indirect effect via serial mediation (indirect effect (a₁ d₂₁ b₂): -0.18, 95% CI (-0.35, -0.05)). In this reversed model, PTEs are associated with increased PTSD, which in turn increases the frequency of experienced post-migration stressors, which negatively affects quality of life. The total model accounted for 25% of the variance in HRQoL scores (F(5,108) = 7.27, p < 0.001, R² = 0.251).

Lastly, we explored the pathway between post-migration stressors and HRQoL further by analysing types of stressors and dimensions of HRQoL as well as the global HRQoL used in the mediation models (see Table 3).

Economic concerns had the highest means and showed significant correlation with almost all dimensions. The second last item, perceived discrimination, also had significant correlation across several outcomes despite having lower means. Two of the items with higher means (missing family and my previous life) did not affect the outcomes greatly. Global HRQoL and psychological wellbeing were affected by several of the same stressors, while few stressors were significant for physical wellbeing. Worrying about being able to stay in Norway was uniquely associated with the dimension for Parents/Autonomy.

The aim of this study was to explore how experiences from war and forced migration influence quality of life after recent resettlement. Based on the “Ecological model of refugee distress”, we explored models including both post-migrations stressors and mental distress as mediators. We found that PTEs reduced quality of life and that this was mediated by post-migration stressors alone or in sequence with mental distress, but not by mental distress alone. The participants had high levels of reported PTEs (61%), general mental distress (36%) and PTSD (48%), akin to other studies of resettled Syrian children and adolescents [4, 5, 40‐42]. The HRQoL was moderately good but lower than population norms [38]. The level of post-migration stressors was slightly lower compared to Syrian adults [43] or other resettlement contexts such as refugee camps [16], low-income host countries [15] and among asylum seekers [17, 44].

Our analyses suggest that negative experiences from forced migration reduce quality of life after resettlement, as proposed in our first hypothesis. As proposed in the “Ecological model of refugee distress”, post-migration stressors independently and fully mediated the relationship between PTEs and HRQoL, supporting its relevance and as a mediator also for HRQoL. This reflects earlier findings of post-migration stressors as a mediator to mental distress [7, 10, 15‐18] but is contrary to the results of a recent meta-analysis that found no mediating effect of post-migration stressors on wellbeing indicators [7]. This could be due to only a handful of studies being relevant and the use of different measures in a variety of contexts, but is also an indication of the need to replicate our results in larger studies. Although numerous studies have found that mental distress predict quality of life [27, 28], few have investigated this as a mediator between trauma and quality of life as we proposed in our third hypothesis. One study of adult Ethiopians in refugee camps found that trauma from displacement had both a direct effect upon HRQoL and an indirect effect through mental distress [29]. However, in our study mental distress did not act as an independent mediator, and hypothesis three was not supported. The different results could be due to differences in age, contexts and levels of distress. Conversely, we found that mental distress acted as a mediator in conjunction with post-migration stressors, supporting our fourth hypothesis, and reaffirming the importance of including both factors when assessing HRQoL in refugees.

Longitudinal studies on refugee youth suggest differential impact of pre- and post-migration factors throughout the resettlement process, as traumatic experiences before arrival predict short-term reactions and stressors in exile better predict psychological problems a decade after arrival [45]. In the present study, the impact of pre- and post-factors also varied with type of mental distress, as post-migration stressors seemed to be more relevant for general mental distress and explained more variance in HRQoL, whilst PTEs seemed to have more relevance for PTSD and explained less of the variance. These differences have also been found in other studies [46] and suggest that discussions on pre- and post-migration influence should not be an either/or debate, but relate to their differential impact throughout the settlement process. Type of mental distress is also pertinent to consider in addition to levels above thresholds, as there was no significant association between PTSD and HRQoL after controlling for earlier experiences (PTEs), despite almost half of the participants scoring above the thresholds.

A cumulative effect of post-migration stressors was supported by our analyses, but further exploration also revealed differences between types of stressors and dimensions of HRQoL. Global HRQoL and the dimension of psychological wellbeing were most affected by post-migration stressors and physical wellbeing the least, suggesting that other factors (e.g. somatic symptoms) could be relevant for the latter. Some stressors were relevant across several dimensions, notably economic concerns and perceived discrimination, which are repeatedly found as detrimental to health and wellbeing [3, 7, 47]. In contrast, two of the most commonly reported experiences (missing family or previous life) had low correlation, possibly indicating less severe emotional distress [17]. These results reflect the “ecological model of refugee distress” where factors at multiple levels influence health and wellbeing. They also inspire hope as the most influential stressors are malleable factors. However, it also highlights the responsibility of host nations for protecting against or alleviating unnecessary pressure on an already vulnerable population as the most influential types of stressors seem to be embedded in social structures.

As mentioned earlier, Miller and Rasmussen [12] suggested that post-migration stressors could deplete coping mechanisms, affecting the capacity to recover from trauma. Other theories, such as the “Stress sensitivity theory” propose that past trauma activates an overreaction to ongoing demands thus decreasing the tolerance for stressors perceived as manageable by others [48]. These could be explanations as to why increasing numbers of PTEs were associated with higher frequencies of post-migration stressors in our study. Emotion dysregulation has been found to mediate mental distress in refugees [4, 49]. Also, psychological processes such as emotion regulation, memory and executive functions are susceptible to the influence of trauma or elevated stress during development with long-lasting effects, as this in turn affects the capacity to regulate future stress responses [50]. The proposed model might also be reversed in a “Stress generation model” [51], as tested in Fig. 4. Mental distress would then make people behave and react in ways that create more stressful situations, for example, by avoiding social interaction or making impulsive financial choices. The general burden of mental distress could also act as a worry or stressor in itself and is in fact included in some post-migration stressor scales [52]. The interaction between distress and stressors suggests a bidirectional relationship where both processes occur, also shown in other studies [53, 54], creating a more complex and transactional model [51]. These proposed processes could inform current treatments for mental distress or general psychosocial interventions in schools. For example, the models could help explain why some experience reduced symptom burden after treatment for PTSD but not increased quality of life, and vice versa. Implications for a therapeutic approach would be the importance of addressing currents stressors, not only symptoms, as failure to do so may limit the effectiveness of treatment [10] and also to include holistic treatment goals such as quality of life. Lastly, broader and structural interventions, including all refugee youth, may alleviate or prevent post-migration stressors and mental distress and should be implemented by host nations.

Although psychological problems are frequent in refugee children, the extents are reduced over time in settlement [13, 45] and wellbeing remains high [23]. This could be a sign of natural recovery processes or resilience, where youth develop new resources such as cultural competence, language skills or networks. Resettlement can therefore be stressful, but also involve personal growth and resilience [55], and as such it supports Antonovsky’s criticism of early stress theories for assuming that stressors were inherently negative and the importance of focussing on health resources [56]. Social support and positive coping styles are suggested as protective factors buffering mental distress in refugee children and youth [3, 4] and are also associated with increased quality of life [18, 27]. Our models did not investigate any protective factors, and further studies should include such measures.