Victims

We first compared the two victim groups: respondents victimised between March 2018 and March 2019 (group 1^victims) and respondents victimised between March 2020 and March 2021 (group 2^victims). For group 1^victims, we linked data on mental health and lack of support assessed at T ₁ (before victimisation between T ₁ and T ₂) and for group 2^victims we linked the same data assessed at T ₃ (before victimisation between T ₃ and T ₄).

Preliminary analyses showed that 105 respondents were victimised between T ₁ and T ₂, as well as between T ₃ and T ₄, prohibiting an independent comparison between both victim groups. To solve this problem, we randomly divided this subgroup into two subgroups. One-half (n = 52) was added to the group of respondents who participated at T ₁ and T ₂ (group 1^total, n = 4717) and the other half (n = 53) to respondents who participated at T ₃ and T ₄ (group 2^total, n = 4875). In this way, the proportion of group 1^victims and group 2^victims is similar (56% versus 44%) to the original proportion of people with VAT cases in between T ₁ and T ₂, and in between T ₃ and T ₄.

Next, we weighted the data of group 1^total and group 2^total using 16 exclusive demographic profiles among the total adult Dutch population (2018 and 2020 respectively), based on the data of Statistics Netherlands (see: https://opendata.cbs.nl/#/CBS/en/). These profiles were constructed using the variables gender (male, female), age (18–34, 35–49, 50–64, 65 years and older) and marital status (married and unmarried). After this procedure and weighting, we finally selected the victims of group 1 (n = 421) and the (non-overlapping) victims of group 2 (n = 319).

Non-victims

To examine differences in mental health and social support among non-victims before and after the outbreak, we assembled a second data-set. For this data-set, we first selected respondents who participated at T ₂, T ₃ and T ₄ (regardless of being victimised, n = 4020). After this selection, we weighted the data using the 16 demographic profiles of the Dutch adult population (2019). We finally selected respondents who were not victimised by VATs in the period T ₁ to T ₄ (n ^non-victims = 3245).

All results are based on weighted data (except Cronbach's alpha).

Experiences with VATs

At each survey, respondents were asked to indicate for each of the 21 listed potentially traumatic events, if they were exposed to this specific event in the past 12 months (1, no; 2,  yes) including the following VAT events:

1. (a) physical violence (sexual violence/sexual abuse (not online); online sexual violence/sexual abuse; robbery; physical violence, but not by own partner; physical violence by own partner);

2. (b) accidents (traffic accidents, disasters, fire, medical errors); and

3. (c) threats (serious threat; without the use of physical violence (not online); online serious threat, without use of physical violence).

The list also included events like (sudden) death of a (significant) other, (online) theft and serious illness categorised here as other stressful life events (SLEs). Participants were finally given the opportunity to report events in the previous 12 months that were not listed, which were recoded afterwards into existing or new categories.^{Reference van der Velden, Komproe, Contino, de Bruijne, Kleber and Das1}

Lastly, they were asked when the (most drastic or stressful) event occurred (coded as 1 for 1 week ago to 8 for 7–12 months ago), and asked to rate the level of stress during the event (coded 1 for none to 5 for very much). To reduce the number of cells with low cell counts, both time since the event and stress during the event were dichotomised (time: 1, up to 2 months ago (1, 2, 3, 4 and 5); 2, 3–12 months ago (6, 7 and 8); stress: 1, not very stressful (1, 2 and 3); 2,  very stressful (4 and 5)).

Anxiety and depression symptoms

In all surveys, anxiety and depression symptoms were assessed using the five-item Mental Health Inventory: this scale covers symptoms during the past month on six-point Likert scales, such as ‘This past month I felt very anxious’ and ‘I felt depressed and gloomy’ (0, never to 5, continuously).^{Reference Ware and Sherbourne11,Reference Means-Christensen, Arnau, Tonidandel, Bramson and Meagher12} After recoding the negative formulated items, the total scores were computed and multiplied by four (range 0 to 100), where lower scores indicate higher anxiety and depression symptom levels (all Cronbach's alpha of group 1^victims, group 2^victims and non-victims ≥0.85). A cut-off score of ≤59 was used to identify respondents with moderate–severe anxiety and depression symptom levels.^{Reference Driessen13}

Post-traumatic stress symptoms

To examine PTSD symptomatology at T ₂ and T ₄, victims were asked to fill in the eight-item version of PTSD Checklist for DSM-5 (PCL-5)^{Reference Price, Szafranski, van Stolk-Cooke and Gros14,Reference Weathers, Reyes and Elhai15} with the VAT event (reported in the same survey) in mind. When respondents reported two or more events, they were asked to focus on the most impactful or stressful VAT event when filling in the PCL-5. This event was presented on the screen. Examples of items are ‘Feeling very upset when something reminded you of the stressful experience’ and ‘Loss of interest in activities that you used to enjoy’. The items of the PCL-5 have a five-point Likert scale (0, not at all to 4, extremely) and examine symptoms across the four symptom clusters of PTSD according to DSM-5 and focus on symptoms in the past month (range 0 to 32). Higher scores reflect higher PTSD-symptom levels (Cronbach's alpha of group 1^victims and group 2^victims ≥ 0. 92). To identify victims with probable PTSD, a cut-off of ≥13 was applied.^{Reference Pereira-Lima, Loureiro, Bolsoni, Apolinario da Silva and Osório16}

General mental health problems and physical problems

The brief screening Problems and Help Inventarisation List^{Reference van der Velden, Das, Contino and van der Knaap17} was administered at all surveys to examine the prevalence of various problems such as general mental health problems, financial problems and physical problems. All respondents were asked to indicate for each of the problem types listed whether they experience these problems or not (1, yes; 2,  no). For the present study, data was extracted from the items ‘psychological problems’ (as a proxy for general mental health problems), and ‘physical problems’.

Coping self-efficacy

To examine VAT-related coping self-efficacy among victims at T ₂ and T ₄, the seven-item Coping Self-Efficacy (CSE-7)^{Reference Bosmans, Komproe, Van der Knaap, Benight and van der Velden18} measure was administered with items such as ‘dealing with frightening images or dreams about the event’ and ‘being optimistic since the event’. The CSE-7 was administered in the same way as the PCL-5. For each item, respondents rated their perceived self-efficacy level on a seven-point Likert scale (1, I am completely incapable of to 7, I am perfectly capable of; Cronbach's alpha of group 1^victims and group 2 ^victims ≥0.93). Higher scores reflect higher coping self-efficacy levels (range 7–49). A cut-off of ≤28, corresponding to (about) 10% lowest scores, was used to identify victims with relatively low versus high coping self-efficacy levels.

Social acknowledgement

To assess event-related social acknowledgement among victims at T ₂ and T ₄, the five-item general disapproval scale of the Social Acknowledgment Questionnaire (SAQ)^{Reference Maercker and Müller19} was administered, in the same way as the PCL-5 and CSE-7. The items have five-point answer categories (1, totally disagree to 5, totally agree). Examples of items are ‘There is not enough sympathy for what happened to me’ and ‘Most people cannot understand what I went through’ (both Cronbach's alpha of group 1^victims and group 2^victims ≥0.85). Higher scores reflect lower social acknowledgement levels (range 5–25). A cut-off of ≥17, corresponding to (about) the 10% highest scores, was used to identify respondents with a high lack of social acknowledgement.

Lack of emotional support

In all surveys, lack of emotional support in response to problems was examined among all respondents with the eight-item subscale lack of emotional support of the Social Support List-Discrepancy (SSL-D).^{Reference Bridges, Sanderman and van Sonderen20,Reference van Sonderen21} Examples of items are ‘comfort you’ and ‘help you to clarify your problems’. The SSL-D items apply four-point Likert scales (1, I miss this, I would like it to happen more often to 4, It happens too often). Low scores reflect a higher lack of emotional support levels (range 8–32) (Cronbach's alpha of group 1^victims, group 2^victims and non-victims ≥0.88). A cut-off of ≤16, corresponding to (about) the 10% lowest scores, was used to identify respondents with a high lack of emotional support.

Post-event differences between both victim groups

To examine differences between both victim groups for the prevalence of post-event (for group 1^victims at T ₂, and for group 2^victims at T ₄) anxiety and depression symptoms, post-traumatic stress symptoms, general mental health problems, coping self-efficacy, lack of emotional support and social acknowledgement, multivariate logistic regression analyses were conducted. To test differences between both victim groups in a rigorous way and to rule out alternative explanations for differences between the two groups as much as possible, we controlled for characteristics that differed between both groups not only at a P < 0.05 level but also at a P ≤ 0.10 level (see Table 1), for example gender, number of different VAT events in the past 12 months and general mental health problems in the year before.

Table 1 Characteristic victims before and after the coronavirus disease 2019 (COVID-19) outbreak^a

PTE, potentially traumatic event; SLE, stressful life event; VAT, exposure to violence, accidents and/or serious threat.

a. As a result of weighting, the numbers (n) may differ slightly.

b. Low, primary school, intermediate secondary education (USA: junior high school); medium, higher secondary education/preparatory university education (USA: senior high school); intermediate, vocational education (USA: junior college); high, higher vocational education, (USA: college, university), according to education level categories of Statistics Netherlands (CBS).

c. Respondents who were confronted with more than one type of VAT event, were asked to have the most stressful VAT event in the past 12 months in mind when answering the question. Of all 740 VAT victims, 33 answered this question with a theft-related event in mind. They were omitted from these analyses (n ^{victims group 1} = 407, n ^{victims group 2} = 300).

d. Three types of VATs were distinguished: violence, accidents and serious threats.

To examine differences in mean scores for post-event anxiety and depression symptoms, post-traumatic stress symptoms, coping self-efficacy, lack of emotional support and social acknowledgement between both victim groups, mixed-effects models were conducted with the same control variables as in the logistic regression analyses.

Differences in incidence (change) between both victim groups

Differences in the incidence of anxiety and depression symptoms, general mental health problems and lack of social support were conducted with similar logistic regression analyses (the incidence of other variables could not be calculated). The incidence was calculated by dividing the number of respondents with post-event symptoms, problems and lack of support among those without pre-event symptoms, problems and lack of support, by the total number of the victim group.

Differences after and before the outbreak among non-victims

To examine the extent to which the prevalence of anxiety and depression symptoms, general mental health problems and lack of social support among non-victims changed after the outbreak (T ₄) compared with before the outbreak (T ₂), repeated measures multivariate logistic regression analyses were performed. In the analyses, we controlled for T ₂- and T ₄-related demographics and reported other potentially traumatic or SLEs. Mixed-effects models were conducted to examine differences in mean scores of symptoms, problems, and support at T ₂ and T ₄ on the same (time-dependent) variables.

All analyses were performed with IBM SPSS version 26.