BACKGROUND/RATIONALE:
The COVID-19 pandemic has created sweeping changes in daily life across the world; yet, given the newness of this lethal virus, limited research has been conducted on its impact on well-being and mental health. Mandatory quarantines can have both immediate and long-term effects on social engagement and isolation. In particular, support, dependence, and trust in social networks may be disrupted and eroded, which, in turn, may impact Veteran well-being and suicide risk. Suicide rates are already alarming in Veterans, with 17 veterans dying per day. Although social distancing is importantly designed to flatten the trajectory of COVID-19, such practices could have other unintentional negative outcomes on Veterans' social networks, and overall well-being.

OBJECTIVE(S):
This project will assess experiences of Veterans during the COVID-19 pandemic, changes in their social networks over time due to social distancing via a social network analysis, and modifiable treatment factors to inform intervention development.

METHODS:
Potential participants were identified using VA Corporate Data Warehouse (CDW). Veterans with PTSD, mood disorders, or other psychiatric disorders were oversampled at a rate of 50%, as Veterans with psychiatric disorders are at higher risk for suicide than Veterans without psychiatric disorders. A new COVID-19 CDW database came online prior to the start of recruitment. This database was used to oversample for Veterans with a COVID-19 diagnosis at the rate of 50%. Recruitment letters were sent to Veterans nationwide and invite participants to complete a one-time, 90-minute survey via a survey on their mobile device, tablet, or computer. Measures included the Ambiguous Intentions Hostility Questionnaire (Combs et al., 2007) to assess hostile attributional bias, a form of social cognition, the posttraumatic stress disorder (PTSD) checklist - 5 (Weathers et al., 2013) to assess PTSD symptoms, the Patient Health Questionnaire - 9 to assess depressive symptoms (Kroenke et al., 2001), and the Beck Scale for Suicidal Ideation to assess thoughts and behaviors related to suicide (Beck et al., 1991). Measures were chosen due to their strong psychometric properties. All participation occurred online, allowing for the study to assess a national sample of Veterans. Overall, 293 Veterans were enrolled, and 233 completed the survey. A personal-level social network analysis was conducted (Acosta et al., 2018, 2015; Moore et al., 2008; Varda et al., 2009). A composite adverse social connectedness measure was created based on a combination of social network measures, perceptions of social support relationships, and perceived overall level of social support. The first component measures the lack of coordination within the support network. A higher value here indicates less coordination within the support network. The second component is a weighted average of the relationships in the support network based on the respondents' perceptions of trust and dependency on these relationships. A higher value indicates that more of the relationships in the network are "lower quality." "Lower quality" indicates there is high dependency on this support but their trust in that support is low. Thus, a respondent with a higher weighted network value has a support network made up of more low-quality relationships. A respondent with a lower weighted network value would have a support network made up of supports that they have high trust in but are not highly dependent upon. The third component is a perception of their overall level of support, "to what degree are you getting the support you need from the people and resources around you?" A higher score indicates that the respondent perceives a lower level of support from their network. From these three components a composite measure of adverse social connectedness was created. The three components are 1) degree of care coordination within their support network 2) adverse social connectedness within their support network and 3) perceived overall level of social support. The maximum adverse social connectedness score for an individual is 3, with higher values indicating a higher level of adverse social connectedness.

FINDINGS/RESULTS:
Of the 233 study participants, 66.5% (n = 155) had a diagnosis of COVID-19. Of these individuals, 69.7% (n = 108) individuals had a mild case, such that their symptoms were able to be treated at home, 22.6% (n = 35) required hospitalization, and 7.7% (n = 12) were hospitalized and required ventilation. The average for depressive symptoms via the PHQ-9 was 9.40 (SD = 8.37), indicating mild symptoms, close to the cut-off of moderate. The average for posttraumatic stress disorder symptoms was 22.91 (SD = 22.51) on the PTSD Checklist - 5. This is below the cut off of 31, yet elevated. Differences in depressive symptoms for individuals with a COVID-19 diagnosis in comparison to individuals who did not have a COVID-19 diagnosis were trend significant (individuals with no COVID-19 diagnosis: M = 7.93, SD = 8.59 [mild symptoms]; individuals with a COVID-19 diagnosis: M = 10.06 [moderate symptoms], SD = 8.18, p = .085). There were no differences between groups in terms of PTSD symptoms. Of the sample 6.87% (n = 16) experienced suicidal ideation in the past week as per the Beck Scale for Suicidal Ideation (BSSI). Scores on the BSSI were an average of 1.52. Of the sample, 4.30% (n = 10) individuals indicated they had made a prior suicide attempt on the BSSI.

Overall, 37 participants were missing data for the social network analysis. These participants were removed from the social network analyses. On average, Veterans named approximately 3 people in their support networks. This increased slightly post-March 2020 (pre-March 2020: M = 3.20 [SD = 3.13], post-March 2020 M = 3.26 [SD = 3.18]), although this change was not statistically significant. Overall, social connectedness worsened during the pandemic (social connectedness pre-pandemic: M = 1.84 [SD = .35], post-pandemic: M = 1.91 [SD = .35], t(1, 195) = -3.52, p = .001). In particular, dependency in their support network members increased post-March 2020, which was statistically significant (dependency pre-pandemic: M = 3.09 [SD = .75], dependency during the pandemic: M = 3.22 [SD = .74], t(1, 195) = -2.52, p < .05). There were no significant differences in trust from prior to the pandemic to during the pandemic. Overall findings indicate that the quality of Veterans' social relationships worsened during the pandemic.
A path analysis was conducted to examine relationships between modifiable treatment factors (i.e., social cognition - hostile attributional bias, psychological inflexibility, and expressed emotion), the social network analysis, PTSD and depressive symptoms, and suicidal ideation. Due to high overlap, PTSD and depressive symptoms were placed into a single factor. Expressed emotion was not related to the social network score at the bivariate level, and was removed from the analyses. Psychological inflexibility was highly related to PTSD and depressive symptoms at a bivariate level, and was removed from the analyses. Analyses began by starting with a saturated model, in which all paths between all variables were specified. The primary path was social cognition - hostile attributional bias, to the social network analysis, to a PTSD/depressive symptoms factor, to current suicidal ideation. Two non-significant paths were excluded: one between social cognition - hostile attributional bias and suicidal ideation, and one between the social network analysis and suicidal ideation.
Fit indices were used to determine model fit. Chi-square test with a non-significant p value (greater than .05) indicates model fit. In the current model, 2 = .440, p = .80, representing adequate model fit. Other fit indices were examined, including the comparative fit index (CFI), the Tucker-Lewis Index (TLI), root mean squared residual (RMSEA). A model with CFI and TLI greater than .95 indicates good fit, and in the current model TLI = 1.086 and CFI = 1.000. RMSEA below .05 indicates good model fit, and in the current model, RMSEA = .000. Thus, overall the model indicates strong fit.
For the primary path, social cognition - hostile attributional bias was associated with the social network analysis, = .16, p = .03, such that greater hostile attributional bias indicated poorer quality relationships. Social cognition - hostile attributional bias was also directly associated with PTSD and depressive symptoms, = .44, p < .001. The social network analysis was associated with PTSD and depressive symptoms, = .26, p < .001, such that poorer quality relationships were associated with higher symptoms. PTSD and depressive symptoms were associated with current suicidal ideation, = .44, p < .001. In terms of indirect effects, there was an indirect effect of social cognition - hostile attributional bias on PTSD and depressive symptoms, = .06 through social the social network analysis. There was an indirect effect of social cognition - hostile attributional bias on current suicidal ideation through both the social network analysis and PTSD symptoms, = .02. There was an indirect effect of the social network on suicidal ideation through PTSD and depressive symptoms, = .03.
The majority of the sample felt that they had at least some degree of social support. This support often came from family members or the VA. Approximately 3.00% (n = 7) Veterans indicated that they did not have a support network to turn to when they needed help since March 2020. Even those with support networks often find that they are not addressing their most pressing needs. The majority of Veterans that indicated they had pressing needs did not have a person or organization in their network that helped them with those needs. In fact, 87.12% (n = 203) of Veterans with a health/healthcare need and 87.12% (n = 203) of those with a mental/behavioral health need did not have someone in their network to help with those needs. They were also likely to lack support for their food (95.70%, n = 223), transportation (97.42%, n = 227), work (95.27%, n = 222), legal (97.85%, n = 227), money (77.30%, n = 180), and emergency needs (81.54%, n = 190). The majority of Veterans (56.65%, n = 132) did not have people or organizations in their identified support networks that they perceived to be coordinating their care or helping them with needs (e.g., finances, transportation, etc.).

IMPACT:
Results indicate that the quality of relationships worsened over the course of the pandemic. A form of social cognition, hostile attributional bias, was directly related to poor quality relationships. This is important, as hostile attributional bias is modifiable through treatment (Horan et al., 2011). Poor quality relationships assessed via the social network analysis were also associated with current PTSD and depressive symptoms, which were then associated with thoughts of suicide. As observed in prior work (DeBeer et al., 2014), social support is a critical buffering factor in the relationship between PTSD and depressive symptoms and suicidal thoughts. Overall, results indicate that hostile attributional bias is associated with low quality relationships, and these low quality relationships increase PTSD/depressive symptoms, thereby increasing suicide risk. Thus, interventions that target hostile attributional bias, such as social cognitive interventions may be important in improving hostile attributional bias, thereby improving relationships, and potentially PTSD and depressive symptoms, and thoughts of suicide. However, social cognitive treatments have primarily been tested in individuals with schizophrenia, and have not yet been tested in those with PTSD or depressive symptoms, or individuals at high risk for suicide.

Additionally, results revealed that many Veterans do not have people in their lives they can rely on to help them coordinate their needs related to social determinants of health. This is particularly salient within the context of the pandemic, when needs have significantly increased. However, many Veterans did rely upon the VA to assist with coordination of needs, indicating the VA may play a critical role in assisting Veterans in getting their needs met. Ensuring that coordination and communication between providers, Veterans, and their family members continues to be important as the pandemic moves into a recovery phase. Assisting Veterans in finding the resources that they need to address their current pressing needs, including health/healthcare, mental and behavioral health and support, and emergency needs are particularly critical within the context of the pandemic. In addition, inquiring about pressing social determinant of health needs will be essential to providing Veterans the best possible treatment plans and care. For example, if a Veteran does not have transportation or the money needed for out-of-pocket expenses, care plans cannot be carried out to their full potential. Helping Veterans connect to resources related to these social determinant of health needs could be particularly valuable as the country emerges from the pandemic.

PUBLICATIONS:

None at this time.