Discussion

The present analysis indicates that residents affected by Hurricane Sandy who presented with at least mild to moderate mental health burden and were linked into care exhibited a significant decrease in anxiety symptoms from baseline to follow-up. This decrease in anxiety persisted even after adjusting for factors known to modify the risk for mental health issues including time since the hurricane, demographic characteristics (e.g., race, ethnicity, gender), and prior mental health difficulties. There is evidence that differences in the decreases in anxiety symptom scores were greater in the L2C compared to the control group. For all other outcomes, although they were not significant, perhaps due to small sample sizes, the decreases in mental health symptom scores appear to be greater among those who were linked to care compared to those not linked. This suggests that linkage to mental health care in the wake of a natural disaster may be particularly effective in decreasing anxiety burden experienced by those exposed to such disasters. In combination, however, with the univariate findings, it is clear that providing post-disaster program support for struggling community members and facilitating linkage into mental health care, is integral in long-term disaster mental health response. It is not enough to just provide support during and in the immediate aftermath in the form of crisis counseling. Mental health interventions that are longer-lasting and that can be facilitated after the initial crisis has subsided are integral to positive community mental health.

The success of PR’s linkage to care model in reducing mental health symptoms, particularly anxiety, may have been due to several factors. First, PR was specifically targeted to those who were already displaying at least mild to moderate levels of symptoms of mental health difficulties increasing the likelihood that PR participants would be willing to link into care. Second, the screening and linkage process was individually tailored in a way that may have facilitated participant empowerment. As part of the screening process, participants could see for themselves that their mental health, including anxiety, symptomology was heightened. It is likely that seeing these scores and having the results of their baseline questionnaire discussed with them individually increased participants’ insight into their difficulties and perhaps empowered them to choose to address these difficulties. ^{Reference Corrigan, Druss and Perlick39} Once they agreed to participate in L2C, the coordinator reminded participants of paperwork to complete and of appointments to make, but it was the participants themselves that had to make and attend the initial appointment. This gave participants the opportunity and support they needed to take an active role in their mental health care as well as advocate on their own behalf. It is not clear, however, as to why only the change in anxiety symptoms was significantly different between the 2 groups and not the other mental health variables. This is most likely due to issues of statistical power resulting from low sample size. However, it is possible that anxiety was particularly heightened in this community after the hurricane and therefore particularly subject to treatment effects. This is consistent with another study conducted after Hurricane Sandy in the Rockaways in which anxiety was underscored as a major mental health issue as it was found to be experienced by 52% of the sample. ^{Reference Subaiya, Stillman and Pumpalova40}

Program sustainability is an important aspect of long-term mental health interventions if they are to be successful. In addition to increasing social worker capacity at a substance abuse and mental health clinic in Far Rockaway by providing financial support for additional staff, L2C provided local community mental health providers with potential clients for their clinics and practices. Mental health resources and provisions were limited in the Rockaways even before the hurricane. ^{Reference Nir41} Scarcity of mental health providers only worsened after the hurricane as the mental health symptom burden increased as a result of the storm. ^{Reference Subaiya, Moussavi, Velasquez and Stillman42} L2C staff worked with local mental health care providers to understand what services their facilities were capable of providing and matched those providers with potential clients who were interested in linking into treatment services. L2C staff made sure that all provider information was up to date so that the participant would be able to directly link into mental health care while avoiding disconnected numbers and previous locations. Engagement by both the community and the providers indicate the potential utility in leveraging existing mental health treatment infrastructure in a community setting to address concerns among survivors. As other studies have also suggested, future research should examine the feasibility, acceptability, and cost-effectiveness of using existing mental health infrastructure as compared to building and implementing new interventions. ^{Reference Contreras, Aguilar and Eappen28} An example of an existing structure that could be leveraged to address the mental health impact of natural disasters are workplace health promotion programs. Creating and supporting collaborations between occupational and public health has been a powerful tool during the COVID-19 pandemic and will most likely continue to be, as occupational health continues to move towards a more comprehensive approach of ‘total worker health’ recognizing the numerous factors, including psychosocial factors, that impact worker health, safety, and wellbeing. ^{Reference Magnavita and Chirico43–Reference Chirico, Zaffina and Ferrari45}

In addition, awareness around post-hurricane mental health issues increased in the community through L2C’s structure, which involved partnering with community groups to gain support for the program and enter locations to hold screenings. Increased mental health awareness in the community was also promoted by having PR staff speak at community events and by having PR’s mission and screening events publicized in the local newspaper and blogs. ^46,47 Furthermore, the program may have increased the sustainability of mental health service in that community as providers were able to see more clients and become more engaged in the locality. These results may help inform mental health intervention funding for future disasters and the allocation of resources to mental health treatment.

The results should be understood within the context of the study’s limitations. First, there were differences in follow-up time between the 2 surveys, although this was partially addressed in the models by using a heterogeneous autoregressive correlation structure to account for unequal time periods and adjust for the difference in months between baseline/follow-up surveys. Although PR participants had shorter follow-up time on average, their mental health symptom scores decreased at the same or at a greater rate than the LIGHT group. Although the 2 groups used the same eligibility criteria, baseline mental health scores were still higher for the L2C group. This could have resulted in greater regression to the mean for the L2C group, but diagnostics indicated that this was not the case. Another limitation is the potential selection bias from each cohort. Potential PR participants were aware that they may be linked to mental health care so those who wanted this service, possibly due to existing mental health symptoms, were more likely to be in the cohort. Additional selection bias occurred by those who elected to complete the follow-up surveys. Moreover, the current study was limited in that it utilized a pragmatic versus a randomized controlled trial (RCT) design. RCT designs are the gold-standard for evaluating the efficacy of interventions. Data from the current study supports the feasibility, acceptability, and initial effectiveness of the L2C intervention. As such, future work is needed to evaluate the efficacy of the L2C intervention during/after a disaster although ethically and pragmatically, it does not seem feasible to implement a post-disaster RCT. In addition, the actual psychotherapeutic treatment itself likely varied by participant and provider and was most likely the reason for any symptom improvements. However, due to confidentiality, study personnel were unable to collect any detailed treatment information. Finally, small sample sizes may have hindered the ability to detect significant interactions. Although significant decreases in mental health outcomes for PTSD, depression, and stress were seen in L2C and not in the control group, the interactions between the 2 groups on these variables were not significant. It may be that the study did not have enough power to statistically detect these effects due to small sample size.

Much can be learned from the overall approach to the implementation of Project Restoration. The study team’s collaborative relationship with the Rockaways community as well as the Project Restoration’s participant-specific strategy for linking into mental health care, facilitated the program’s success. This relationship allowed for adjustment of program implementation so that the methodology was most relevant and appropriate for use in the Rockaways community specifically. For example, the input of community members and stakeholders informed every aspect of the study process from recruitment strategy to language used when linking participants into care. Also, program staff were able to cultivate strong relationships with participants over the time it took to link participants into care. In turn, they were able to tailor resources to each participant to make sure their specific barriers to mental health care were addressed. Overall, Project Restoration demonstrates the strengths and challenges to implementing an intervention and testing its effectiveness in a real-world, post-disaster context when more rigorous methodological approaches such as an RCT are not feasible or ethical.