No buffer effect of perceived social support for people exposed to violence during the COVID-19 pandemic: a cross-sectional community study.

Background: The COVID-19 pandemic has represented a burden to communities worldwide. Research indicates that this burden is not equally distributed in the community, and vulnerable groups, such as violence-exposed individuals may pay a particularly high prize. Perceived social support is known to buffer against negative effects of trauma and adversity, but it is not clear whether this is the case during times of social restrictions and lockdowns. In this study, we tested if perceived social support could buffer the link between pandemic worry and psychological distress in a community sample and in the subgroup exposed to violence during the pandemic.
Methods: A stratified, presumed representative sample of the Norwegian population (N = 1,041, response rate = 39.9%) responded to a cross-sectional web survey in May 2020. Fifty-nine participants (5.7%) had been exposed to physical, sexual, and/or psychological violence during the last month.
Results: Current violence, pandemic worry, and perceived social support were independently associated with psychological distress. In the total sample, perceived social support moderated the relationship between pandemic worry and psychological distress. However, this was not found in individuals who were exposed to current violence. Conclusions: Even though high levels of perceived social support can protect against psychological distress in the face of pandemic worry in the community, it seems that this resource is not as useful for individuals exposed to current violence. Outreach health and care services are warranted to support the needs of this particular vulnerable group.

Background

The COVID-19 pandemic and the countermeasures to reduce the spread of the disease has represented a burden to communities worldwide. Researchers have aimed to assess the mental health consequences in societies and to identify risk groups and protective factors. Whereas some of the studies conducted in an early phase of the pandemic reported an alarming increase in psychological distress in the general community (Salari et al., 2020), later research with study designs of higher quality have indicated less dramatic mental health consequences of the pandemic (Prati & Mancini, 2021; Sun et al. 2021). Some researchers have proposed that the observed increase of psychological distress in the community for the most part might be attributed to vulnerable groups (Public Health England, 2021). Therefore, recent studies have pointed to the importance of identifying groups that are particularly at risk of suffering during the pandemic, such as victims of violence (Blix, Birkeland, & Thoresen, 2021; Ertan, El-Hage, Thierrée, Javelot, & Hingray, 2020). During a pandemic, people who are exposed to violence may become even more vulnerable than before the pandemic (Bradbury‐Jones & Isham, 2020).

1.1. Stressors during a pandemic – in the community and for the victims of violence

In the general community, pandemic worries, such as concerns about reduced access to health care, catching the virus and getting severely ill, losing a loved one, and infecting others, may be one of the driving force of psychological distress (Blix et al., 2021; El-Gabalawy & Sommer, 2021; Elmer, Mepham, Stadtfeld, & Capraro, 2020; Heeren, Hanseeuw, Cougnon, & Lits, 2021). In addition, the pandemic countermeasures involves stressors such as social restrictions and job loss, which can also influence psychological distress. For victims of violence, the countermeasures may have an even stronger negative impact. Countermeasures may limit access to protection, health services and ways of coping, and this may lead to increased psychological distress. Violence-exposed individuals has also been found to worry more than others about the pandemic (Blix et al., 2021).

1.2. Perceived social support

One of the most potent protective factors in the face of violence and other stressful events is perceived social support (Brewin, MacCarthy, & Furnham, 1989; Cohen & Wills, 1985; Feeney & Collins, 2015; Thoits, 2011). Perceived social support can be defined as the subjective feeling of being supported by one’s social relationships (Thoits, 2011). A prominent theoretical perspective of the significance of social support is the buffer hypothesis of social support (Cohen & Wills, 1985; Feeney & Collins, 2015). According to this hypothesis, social support exerts at least some of its effect on psychological distress through responsiveness to the needs elicited by stressful events. It has been suggested that the most important functions include instrumental, cognitive, and emotional support (Santini, Koyanagi, Tyrovolas, Mason, & Haro, 2015). Interaction with other people can facilitate problem solving, provide new thoughts and perspectives, and contribute to emotion regulation. Hence, perceived social support may also have a protecting function against psychological distress associated with pandemic worry and exposure to violence.

So far, a few studies have described perceived social support as a health-promoting factor for the general community during the pandemic (Cao et al., 2020; Fitzpatrick, Harris, & Drawve, 2020; Van der Velden, Contino, Das, Van Loon, & Bosmans, 2020; Yu et al., 2020). However, the potential buffer effect of perceived social support has not yet been investigated. Thus, it is unclear whether perceived social support mitigates the negative effects of pandemic worrying even under conditions of restricted social interaction, Furthermore, a pertinent question is whether perceived social support in a pandemic can buffer against psychological distress in vulnerable groups such as individuals exposed to violence.

1.3. Perceived social support in victims of violence

Studies in ordinary (non-pandemic) settings indicate that perceived social support seems to reduce the negative effects of violence (Coker et al., 2002; Schumm, Briggs‐Phillips, & Hobfoll, 2006; Tirone et al., 2021). The protective effect of perceived social support may be particularly strong for individuals who experience a high level of strain during the pandemic, such as those exposed to current violence. As they need social support more, they may also benefit more from it.

Unfortunately, exposure to violence is related to lower levels of perceived social support (Dias et al., 2019; McCaw, Golding, Farley, & Minkoff, 2007). We might speculate that people exposed to violence might have weaker social networks and fewer social relationships that are less available or usable under pandemic-induced social restrictions. For example, it might be more difficult to maintain weaker social bonds via digital communication (particularly for individuals living with a partner who uses violence) when the frequency of physical contact is reduced. Previous research has shown that trauma-exposed individuals may suffer from social exclusion (Kaniasty & Norris, 2008). When social restrictions allow individuals to have only a few physical contacts, victims of violence may not be selected as preferred companions. In line with this proposal, qualitative studies indicate that access to otherwise available sources of perceived social support was reduced during the pandemic (Fawole, Okedare, & Reed, 2021; Mahapatro, Prasad, & Singh, 2021). Therefore, it is also possible that the buffer effect of perceived social support might be weaker for those exposed to current violence, who are less protected against the negative effects of both the current violence and pandemic worry. It is unclear whether the buffer effect of perceived social support during a pandemic with social restrictions is stronger or weaker for individuals exposed to violence.

1.4. The present study

In this study, we aim to 1) determine whether perceived social support moderated the association between pandemic worry and psychological distress in a presumed representative community sample; 2) explore whether perceived social support moderated the association between exposure to current violence and psychological distress; and 3) investigate whether the potential moderator effect of perceived social support differed across individuals exposed to current violence and those not exposed to current violence. Based on the previous literature on the general buffer effect of social support, we hypothesize that perceived social support will be associated with a weaker relationship between pandemic worry and psychological distress. However, the literature did not provide us with sufficient background to pose strong hypotheses about the buffer effects of perceived social support in individuals exposed to current violence, so for this part, we have a more explorative approach.

Methods

Participants and procedure

We conducted a cross-sectional web survey in Norway between 19 and 26 May 2020. At the time, the COVID-19 situation was described as ‘under control’ in Norwegian society, and the government had recently started easing the countermeasures after approximately two months of lockdown. Norway is a high-income country with a well-functioning welfare system and low pre-pandemic rates of unemployment. The hospitals have so far not been overloaded with patients during the COVID-19 pandemic. Vaccines were not yet provided. This lockdown included restrictions on physical social contacts (maximum five outside one’s household, and given a physical distance of two metres), but there were no restrictions on social contact between romantic partners who did not live together. Other countermeasures were school closure, closed or limited health care services, closed restaurants/bars, and people were encouraged to work from home if possible. Shelters were open, but reported reductions in the number of requests during the lockdown (Bergman, Bjørnholt, & Helseth, 2021). Family service clinics were closed the first few weeks before allowing phone consultations. In April 2020, approximately 11% of the Norwegian work force were unemployed, an increase from approximately 2% in February 2020.

A data collection agency (Kantar/Gallup) collected data from a probability-sampled panel. Participants are not self-recruited, but have been invited by Kantar/Gallup, in order to construct a nationally representative sample. The panel of approximately 46,000 participants is considered representative of Norwegians with access to the internet, which constitutes about 97% of the total Norwegian population. Individuals were invited until the pre-specified sample size (1,000) had been achieved. This sample size was chosen to obtain a high level of congruence between the distribution of the demographics in the sample and the population (in terms of age, gender, and living area) and was recommended by the COSMO study (Betsch, Wieler, & Habersaat, 2020). Sampling and weighting were performed based on official statistics from Statistics Norway.

The data collection agency approached 2,612 individuals stratified on gender, age, education, and area of residence. In total, 39.9% (N = 1,041) completed the survey, 55.8% (N = 1,457) did not respond, 2.7% (N = 71) started the survey but did not complete it, 1.6% (N = 41) clicked on the link to participate but did not confirm agreement with the terms of the study, and 0.1% (N = 2) withdrew from the study. The study participants did not differ from non-responders in gender, education or living situation (living alone vs with someone), but the sample was highly skewed towards older individuals, with a mean age of 54.1 in responders and 43.3 in non-responders (Blix et al., 2021). This means that caution should be taken when interpreting results for the youngest age group. The Norwegian Regional Committee for Medical and Health Research Ethics approved the study (registration number 133,226/2020).

2.2. Measures

Current violence was measured by a set of questions asking whether participants, during the last month, had been exposed to interpersonal violence according to WHO’s definition, which includes physical, sexual, psychological violence as well as deprivation/neglect (WHO, 2002). As this study included adult participants, we measured psychological, physical, and sexual violence.

Psychological violence was measured by a slightly adapted single question from the Stressful Life Events Screening Questionnaire (Goodman, Corcoran, Turner, Yuan, & Green, 1998); ‘Have you, during the last month, been repeatedly ridiculed, put down, ignored, or told that they were no good?’ Mild and severe physical violence was measured by two single questions, each collapsing four and six questions, respectively, we have previously used in our national study on rape and violence in Norway (Thoresen, Myhre, Wentzel-Larsen, Aakvaag, & Hjemdal, 2015): ‘Have you, during the last month, been slapped, pinched, pulled, or shook violently?’ and ‘Have you, during the last month, been hit with a fist or hard object, kicked, strangled, beat, threatened with a weapon, or physically attacked in other ways?’. These were derived and culturally adapted from national studies in the United States (Kilpatrick, Edmunds, & Seymour, 1992; Kilpatrick, Resnick, Baber, Guille, & Gros, 2011; Kilpatrick et al., 2003). Sexual violence was measured with one single catch-all question: “Have you, during the last month, been exposed to any form of sexual assault or violation. We created a dichotomous variable where an answer of ‘yes’ to any of these questions would qualify the individual as ‘exposed to current violence.’

Psychological distress in the last two weeks was measured by the abbreviated 5-item version of the Hopkins Symptom Checklist–25 (HSCL-25;Derogatis, Lipman, Rickels, Uhlenhuth, & Covi, 1974). The five items measures the most common symptoms of depression and anxiety: ‘Feeling hopeless about the future’; ‘feeling blue’; ‘worrying too much about things’; ‘feeling fearful’; and ‘feeling tense or worked up’. Participants responded on a scale from 0 (not bothered) to 3 (bothered a great deal). This abbreviated version has shown good psychometric properties and has previously been found to correlate highly (r = 0.92) with the HSCL–25 in a general population sample (Tambs & Moum, 1993). The item scores were averaged. We used a cut-off value of >2, which a previous study has shown to have the best combination of specificity, sensitivity, and predictive values (Strand, Dalgard, Tambs, & Rognerud, 2003). In the present study, the Cronbach’s alpha value for the 5-item HSCL was .91. In another study based on the same data set, we found that individuals exposed to current violence reported significantly higher levels of psychological distress compared to individuals not exposed to current violence (Blix et al., 2021).

Perceived social support was measured by four items from the Crisis Support Scale (Joseph, Andrews, Williams, & Yule, 1992). The Crisis Support Scale (Joseph, Williams, & Yule, 1992) was designed to measure post-disaster perceived social support. The original scale comprise seven questions: Four questions on perceived social support, one question about negative social support (‘feeling let-down’), one question about contact with other disaster survivors, and one question about overall satisfaction with support. The authors recommend analysing the negative support item separately, as negative support (or negative responses) is conceived as a separate phenomenon. The question about contact with other disaster survivors and the overall satisfaction questions were omitted from this study, as they do not point to specific ingredients of perceived social support. Hence, we used the four items tapping into perceived social support: (‘someone willing to listen,’ ‘able to talk about thoughts and feelings,’ ‘sympathy and support from others,’ and ‘practical help’). Previous research supports the psychometric properties and the validity of the scale (Elklit, Pedersen, & Jind, 2001). Participants responded on a scale from 1 (never) to 7 (always), and the item scores were averaged. The Cronbach’s alpha in the present study was .85.

Pandemic worry was measured with six questions related to the COVID-19 pandemic, indicating whether the participants worried about ‘losing someone they loved,’ ‘becoming seriously ill from the virus,’ ‘infecting others,’ ‘the health system being overloaded,’ ‘not being able to visit people who depended on them,’ and ‘a new outbreak of COVID-19.’ The questions were translated and adapted to the Norwegian context from the COSMO study (Betsch et al., 2020), and participants responded on a scale from 1 (not worried) to 7 (very worried). Cronbach’s alpha was .85. In another study based on the same data set, we found that individuals exposed to current violence reported significantly higher levels of pandemic worry compared to individuals not exposed to current violence (Blix et al., 2021).

2.3. Analyses

Of the 1,041 participants, 882 responded to all the items used in the study (15% of the participants did not respond to at least one item). Average scores were computed if the participant had responded to at least half of the items of the scale (according to the ‘half-item rule’). As a result, 1,027 of the 1,041 participants were included in the analyses.

We performed an ANOVA to compare levels of perceived social support across participants exposed to current violence and participants not exposed to current violence. We conducted linear regression analyses with psychological distress as the outcome and exposure to current violence, pandemic worry, perceived social support, the two-way and three-way interaction terms, and background factors (gender, age, and level of education) as predictors. All predictors were centred before the regression analyses were conducted. We tested the significance of the simple slopes of the interactions following the recommendations of Dawson and Richter (2006). To adjust for potential sources of bias, we also performed analyses weighted by age, education, and area of residence. All data analyses were performed with IBM SPSS Statistics version 26.

Results

Of the 1,041 participants, 49.0% (n = 510) were females. The age range was 18–89 years, with a mean of 54.1 (SD = 15.9), and 35.9% (n = 374) had college/university education. The majority of the sample was living together with someone – 67% lived with their spouse with or without children (n = 694), 4% lived only with their children (n = 45), 3% lived with their parents (n = 32), and 4% (n = 39 = lived in collectives or in other arrangement. A minority of 22% (n = 231) reported living alone. Fifty-nine participants (5.7%) had been exposed to violence during the last month. Of these, 44 had been exposed to psychological violence, 23 had been exposed to physical violence, and 17 had been exposed to sexual violence. While 43 had been exposed to one type of violence, nine had been exposed to two types of violence, and seven had been exposed to all three types of violence during the last month. People in all types of living situations were represented among those who reported violence. Participants exposed to current violence had lower levels of perceived social support (M = 3.06, SD = 0.93) than participants not exposed to current violence (M = 3.70, SD = 0.78), F(1) = 37.79, p < .001.

A linear regression analysis (Table 1; Model 1) showed that when mutually adjusted and adjusted for background variables, current violence, pandemic worry, and perceived social support were independently associated with psychological distress. The total model explained a significant proportion of the variance in psychological distress, R = .34, F(3, 1,008) = 128.56, p < .001. The significant two-way interaction effect between pandemic worry and perceived social support on psychological distress (Table 1; Model 2) indicates that the association between pandemic worry and psychological distress was weaker for participants with high levels of perceived social support than for participants with low levels of perceived social support. Thus, we found evidence of a general buffer effect of perceived social support during the pandemic.

Table 1.

Associations between exposure to violence (Viol), pandemic worry (worr), perceived social support (soc), interaction terms, and psychological distress (n = 1041)

	Model 1				Model 2				Model 3
	B	95% CI	p	β	B	95% CI	p	β	B	95% CI	p	β
Violence	.44	.24 to .63	<.001	.16	.45	.16 to .70	<.001	.16	.47	.19 to .74	<.001	.16
Worry	.22	.19 to .26	<.001	.35	.22	.19 to .26	<.001	.34	.22	.18 to .25	<.001	.33
Social support	−.21	−.27 to −.15	<.001	−.25	−.20	−.26 to −.15	<.001	−.24	−.20	−.25 to −.14	<.001	−.23
Viol X worr					.03	−.14 to .21	.682	.02	−.10	−.21 to .18	.890	−.10
Viol X soc					.19	−.11 to .54	.033	.07	.23	−.07 to .60	.011	.08
Worr X soc					−.10	−.15 to −.06	<.001	−.14	−.09	−.13 to −.05	<.001	−.12
Viol X worr X soc									.04	.002 to .12	.018	.07
Adj. R2	.34				.36				.37

All models adjusted for background variables (sex, age, and university/college education).

We also identified a significant two-way interaction between exposure to current violence and perceived social support. For participants with high levels of perceived social support, the relationship between exposure to current violence and psychological distress was stronger than for participants with low levels of perceived social support. This means that we did not find evidence of a buffer effect of perceived social support on the potential negative effects of current violence. On the contrary, the findings indicate a potentiating effect – that in individuals with high levels of perceived social support, exposure to current violence was more strongly associated with psychological distress than in individuals with low levels of perceived social support. No interaction effect between exposure to current violence and pandemic worry on psychological distress was found. Adding the three two-way interaction terms to the model contributed significantly to explaining the variance in psychological distress, R change = .018, F(1, 1,005) = 9.58, p < .001.

The three-way interaction effect between violence, pandemic worry, and perceived social support on psychological distress was significant (Table 1; Model 3). Adding the three-way interaction term to the model contributed significantly to explaining the variance in psychological distress, R change = .004, F(1, 1,004) = 5.62, p = .018. Figure 1 depicts the simple slopes for all the variables. Among the participants not exposed to current violence, we found a negative association between perceived social support and psychological distress. This was particularly salient for participants with high levels of pandemic worry (simple slope = −.31, t = −9.70, p < .001) but was also evident in participants with low levels of pandemic worry (simple slope = −.12, t = −3.76, p < .001).

Figure 1.

Three-way interaction plot for associations between exposure to current violence, perceived social support, and pandemic worry on psychological distress. High/low perceived social support and pandemic worry represent one standard deviation over and under the mean

In contrast, for participants exposed to current violence, we found no significant associations between perceived social support and psychological distress, neither in participants with high levels of pandemic worry (simple slope = −.04, t = −.40, p = .691) nor low levels of pandemic worry (simple slope = .07, t = 0.76, p = .446). We performed the same analyses weighted by gender, age, education, and area of residence. These yielded highly similar results (not displayed).

Discussion

In this study, we aimed to determine whether perceived social support can have a protective function during a pandemic, both for the population and for individuals exposed to current violence. Our results showed that 1) in general, perceived social support seemed to protect against psychological distress for individuals who worried about the pandemic; 2) we found no evidence indicating that perceived social support protected against the psychological distress associated with exposure to current violence during the pandemic; and 3)) we found no evidence indicating that perceived social support protected against the psychological distress associated with pandemic worry, for individuals exposed to current violence.

That perceived social support was negatively associated with psychological distress is in line with previous studies conducted during the COVID-19 pandemic (Cao et al., 2020; Fitzpatrick et al., 2020; Van der Velden et al., 2020; Yu et al., 2020). However, these studies could not provide evidence that the psychological distress observed was connected to the pandemic. Our findings indicate that perceived social support may act as a buffer against the negative impact of pandemic-related worries. This suggests that perceived social support is particularly beneficial for people who worry excessively. In line with this, our results suggest that during stressors such as pandemics, perceived social support can help manage worry. Similar results have been found for ruminative thoughts (Birkeland, Blix, & Thoresen, 2020; Marroquín & Nolen-Hoeksema, 2015) and negative appraisals (Khoury, Atkinson, Bennett, Jack, & Gonzalez, 2021). Thus, access to social support may play a role in coping with maladaptive thoughts, such as excessive worry about the future.

Importantly, perceived social support seems not to have a similar buffering function for victims of current violence. This result is in contrast to previous reviews of (non-pandemic) studies that have shown that perceived social support mitigates the negative consequences of violence (Coker et al., 2002; Tirone et al., 2021). During a pandemic in which social restrictions are in place, people are more isolated at home, and those exposed to violence may be isolated together with the perpetrator of the violent acts. In addition, their need for help may go unnoticed.

Previous studies suggest that exposure to violence is associated with lower perceived social support (Dias et al., 2019; McCaw et al., 2007), and that perceived social support may deteriorate over time in people exposed to trauma (Thoresen, Birkeland, Arnberg, Wentzel-Larsen, & Blix, 2019). If the social support network is low on resources, or the individual perceives these resources to be limited, barriers to seeking support may arise, such as a reluctance to overburden friends and family due to the feeling that they have enough to cope with already or that they would not understand. Such social support barriers have been found to have strong associations with psychological distress in previous trauma samples (Smith, Felix, Benight, & Jones, 2017; Thoresen, Jensen, Wentzel-Larsen, & Dyb, 2014). A weak existing social network may be difficult to maintain and access digitally and may not be able to provide a buffering function during the pandemic.

A few recent qualitative studies have shed light on interpersonal processes in victims of violence during the COVID-19 pandemic. For example, despite some reports that violence intensified during this period (Lyons & Brewer, 2021), it may be more difficult for the victim to access the support of others when the perpetrator is present in the household, or others may not be able to help them due to social restrictions (Fawole et al., 2021; Mahapatro et al., 2021). In line with this, other (non-pandemic) studies indicate that at high levels of victimization, the protective function of perceived social support seems to break down (Beeble, Bybee, Sullivan, & Adams, 2009; Scarpa, Haden, & Hurley, 2006). Some problems, such as living in an unsafe environment with exposure to current violence during a lockdown, may require more support than the available social network can provide. Such situations need to be solved in more tangible ways (e.g. by moving the victim to a safe place), and institutional support from, for example, shelters and health services may be necessary.

This study has some strengths and limitations. We used stratified probability sampling, which means that every individual in the population had a chance of being selected for the study and that subgroups in terms of gender, age, education, and area of residence were properly represented among those we approached. Self-selection to this particular study may have influenced our results. We were able to assess representativeness by comparing the demographic characteristics of those were approached and those who responded. Analyses weighted by demographic characteristics provided similar results to those of the main analysis, but it is uncertain whether our results are valid for young adults. Among the limitations is that this is a cross-sectional study based on self-report. Our sample size was too small to investigate relations within each type of violence, and results should be confirmed in studies with greater sample size and in other contexts. We did not have information on relationship status or details on types of violence, which mean that we could not examine associations between these and perceived social support.

The cross-sectional study represents a snapshot of the situation in a specific phase of the pandemic and in a specific country, and we were not able to study the trajectories of countermeasures and psychological distress. Norway is a country with a strong economy and a well-functioning and accessible healthcare system. The context may influence levels of pandemic-related worry, current violence, perceived social support, and the relationships between these factors. Physical violence, psychological violence, and sexual abuse occur in many variants, and in this study, we were not able to include extensive exploration of these phenomena. Regarding physical violence, we used behaviourally specific questions, but collapsed different forms of physical violence into one question about mild physical violence and one question about more severe physical violence. Sexual abuse was measured with one simple catch-all question. It may have made it difficult for the respondent to decide if an event they had experienced would fall into this category or not (Thoresen & Øverlien, 2009). In addition, the one item measuring psychological violence, derived from the Stressful Life Event Screening Questionnaire (Goodman et al., 1998) and, is perhaps more subjective in nature, compared to the more behaviourally specific questions about physical abuse.

In conclusion, our results indicate that perceived social support during the pandemic moderated the association between excessive pandemic worry and psychological distress. However, for individuals exposed to current violence, perceived social support did not seem to act as a buffer against either the violence or pandemic worry. Our study is the first to indicate that perceived social support has a protective function also during pandemic countermeasures, but that this does not necessarily apply for individuals exposed to violence.

In the next few decades, new pandemics are likely to occur (Madhav et al., 2017). To be prepared for this, we need to learn from the present and plan for targeted interventions that will reduce the potential negative consequences of pandemics and pandemic-induced restrictions on psychological distress in the community. Facilitating social support from existing social networks or by supplementing these networks would likely lessen psychological distress, particularly in people with low perceived social support. During a pandemic, providing opportunities for people to spend time together and maintaining or increasing the availability of services that provide social support may be powerful public health interventions.

Importantly, whereas social support did buffer against psychological distress for the majority of our sample, this was not the case for victims of violence. Taking measures to protect personal safety are acceptable and necessary exceptions to the social distancing policy, and shelters need to be open at all times. Health agents and policy makers should make sure that information that it is acceptable and possible to reach out to shelters and other health care services is widely disseminated to the public. In addition to being open and available, it may be necessary for care and health services to follow an active outreach strategy by contacting people who might be at risk of exposure to violence, and check if they have any current unmet health care needs. Additionally, crisis management plans for pandemics should include plans for protecting the mental health of vulnerable groups such as individuals exposed to violence.

Questions remain regarding which types of social support are helpful for whom in what situations, and how to create interventions that increase perceived social support. As worries and emotions may be transient, studies that measure these frequently (e.g. by using an experience sampling methodology) may provide more specific knowledge that can be used to design interventions to improve social relationships.