Effectiveness and Cost-Effectiveness of Mental Health Interventions Delivered by Frontline Health Care Workers in Emergency Health Services: A Systematic Review and Meta-Analysis

This systematic review is to evaluate the effectiveness and cost-effectiveness of mental health interventions delivered by frontline health care workers in disasters and public health emergencies. Six databases and trial registries were searched, and manual searches were conducted. Of the 221 studies identified, 21 were included. Meta-analyses assessed differences between the intervention and control in terms of PTSD outcomes. Eleven studies of 1802 participants were incorporated in the meta-analysis. Interventions delivered or prompted by specialist health care workers showed significant and large effects in improving PTSD-related symptoms with a SMD = 0.99 (95% CI: 0.42–1.57, p = 0.0007). Interventions delivered or prompted by frontline non-specialist health care workers showed significant but small effects in improving PTSD-related symptoms with SMD of 0.25 (95% CI: 0.11–0.39; p = 0.0007). The results showed that most mental health interventions delivered by frontline health care workers effectively supported affected people. Mental health interventions delivered by mental health care professionals are effective in reducing PTSD-related disorders in natural disasters. Future adequately powered RCTs are needed to evaluate the effectiveness of mental health interventions delivered by trained non-specialists. Economic modelling may be useful to estimate cost effectiveness in low- and middle-income countries given the difficulties of conducting studies in disaster and emergency settings.


Introduction
Natural disasters have killed approximately 60,000 people every year, on average, during the last two decades [1]. In 2021, natural disasters caused approximately 10,492 deaths and affected 101.8 million people globally, with an estimated global economic loss of $343 billion USD [2]. The World Health Organization has identified disaster-related mental health issues as among the most pressing public health issues [3]. Natural disasters (such as floods, bushfires, and earthquakes) and public emergencies (such as pandemics, wars, and civil disorders) have negative mental health implications for those who are directly or indirectly affected [4]. Due to exposure to traumatic events and chronic stress, various mental health problems, such as anxiety, depression, post-traumatic stress disorder (PTSD), and bipolar emotional disorders, occur during disasters. PTSD is the most common mental health problem following a disaster [5]. According to the American Psychiatric Association [6], when an individual lives through or witness traumatic incidents or is being threatened by death, experiencing severe injury, or sexual abuse, post-traumatic stress disorder (PTSD) may occur. Directly experiencing the traumatic incident, seeing it happen to others, hearing that it happened to a family member or close acquaintance, or indirectly being exposed due to work-related responsibilities are all forms of exposure. Examples of traumatic situations include war, physical assault, terrorist strikes, and natural catastrophes. PTSD symptoms include (a) reliving the trauma in upsetting memories, flashbacks, or frequent dreams or nightmares; (b) avoiding situations or activities that bring up the traumatic event; (c) diminished responsiveness (emotional anaesthesia or numbing); and (d) feelings of detachment and estrangement from others. Chronic physiological arousal can also cause symptoms such as an exaggerated startle (see survivor guilt).
The prevalence of mental health disorders in disaster-affected areas has been reported to be double to triple that in the general population [7]. According to a systematic review published in 2019, the prevalence of severe mental disorders in conflict settings (schizophrenia, bipolar disorder, depression, anxiety, and post-traumatic stress disorder) was 22.1%, and one out of every 11 people (9%) exposed to conflict in the previous ten years would develop a moderate or severe mental disorder [8]. A review carried out between January 2009 and March 2013 from 90 refugee camps across 15 low-and middle-income countries revealed that 77% of healthcare visits were for mental substance use disorders, psychotic disorders, and moderate and severe depression, anxiety, and PTSD [9].
In emergency situations, post-traumatic stress disorder (PTSD) is the most common mental health issue. The prevalence of PTSD among direct victims is estimated to be 30-40%, while it is 10-20% among rescue professionals and 5-10% in the general public [10]. Mental health issues can become chronic if not treated. According to research conducted in a refugee camp on the Turkish-Syrian border, 24-30% of those who were exposed to the Syrian Civil War had PTSD and depression co-morbidly, 4-6% solely had PTSD, and 19% only had depression three years later [11].
Early mental health care may reduce negative mental health effects for those affected by emergencies [12,13]. Interventions suitable for use in disaster and public emergency situations need to be brief and culturally relevant, with proven effectiveness and costeffectiveness [14]. Frontline healthcare workers must be able to provide rapid mental health support in addition to physical healthcare to affected people as first responders in emergencies [15].
Previous systematic reviews have been unable to offer clear conclusions to guide implementation or health policy on this topic because they needed to focus on all these disorders together, as opposed to one or some only, for several reasons: (1) they targeted certain clinical populations such as people suffering from depression or anxiety disorders, particularly post-traumatic stress disorder (PTSD) [16][17][18]; (2) they focused on the effects of specific psychiatric therapies such as cognitive behavioral therapy, eye movement desensitization, and reprocessing (EMDR) or therapeutic components [18][19][20][21][22]; and (3) studies did not include evidence from low-and middle-income countries (LAMICs) or underrepresented LAMICS [23,24].
The current understanding of which psychological interventions and care models can improve mental wellbeing in disasters/emergencies is incomplete, based on insufficiently robust and clear evidence [25]. This study aimed to assess the effectiveness and costeffectiveness of all types of mental health interventions delivered by frontline healthcare workers during natural disasters and in public emergency settings.

Methods
This systematic review and associated meta-analysis were implemented according to the Cochrane guidelines on systematic reviews of observational studies and according to PRISMA-P. The PRISMA-P checklist is illustrated in Figure 1. The protocol was registered in the International Prospective Register of Systematic Reviews (CRD 42021266648).

Methods
This systematic review and associated meta-analysis were implemented according to the Cochrane guidelines on systematic reviews of observational studies and according to PRISMA-P. The PRISMA-P checklist is illustrated in Figure 1. The protocol was registered in the International Prospective Register of Systematic Reviews (CRD 42021266648).

Eligibility Criteria
Studies focusing on mental health interventions involving preparedness, response, and rehabilitation in natural disaster and emergency settings were eligible, except for qualitative studies, uncontrolled studies, conference presentations, studies of personal traumatic events, and medical emergency studies. Mental health interventions refer to general approaches to promoting awareness of mental health or measures tailored to deal with a specific mental health disorder, such as depression or anxiety. The types of interventions ranged from counselling to psychotherapy, psychoeducation, behavioral activation, and cognitive behavioral therapy (CBT). No restrictions were placed on the method used to assess mental health, with data ranging from self-report surveys to objective assessments recorded using established methodologies.

Eligibility Criteria
Studies focusing on mental health interventions involving preparedness, response, and rehabilitation in natural disaster and emergency settings were eligible, except for qualitative studies, uncontrolled studies, conference presentations, studies of personal traumatic events, and medical emergency studies. Mental health interventions refer to general approaches to promoting awareness of mental health or measures tailored to deal with a specific mental health disorder, such as depression or anxiety. The types of interventions ranged from counselling to psychotherapy, psychoeducation, behavioral activation, and cognitive behavioral therapy (CBT). No restrictions were placed on the method used to assess mental health, with data ranging from self-report surveys to objective assessments recorded using established methodologies.

Search Strategy
Six databases (Embase, PsycINFO, Epub, Global Health, APA PsycInfo, and MED-LINE) were searched from their inception to December 2021. Searches were conducted by combining combinations of the five categories of keywords (Supplementary Materials). In addition, Google Scholar and WHO databases were manually searched, and 31 additional studies were identified. There were no data limitations in this study.

Selection Process and Data Extraction
The search results were exported to EndNote 20, where duplicates were removed. The included studies were screened manually to identify other potentially relevant studies. The characteristics of the studies included are summarized in Table 1. The primary investigator (MP) reviewed the search strategy. Following the above inclusion and exclusion criteria, two reviewers (MP and JS) independently reviewed relevant studies for potential inclusion. Any differences in opinions were resolved through discussion until a consensus was reached. This process minimized the risk of bias in the decision to include or exclude studies. A PRISMA flowchart is shown in Figure 1.
Meta-analyses using random effects models were conducted in RevMan 5.4 using mean differences and 95% confidence intervals to describe between-group differences for changes in PTSD related symptoms. We also examined the effects of different types of intervention providers on the results by dividing the study participants into three groups: professional-led intervention, non-mental specialist healthcare worker intervention, and internet-based intervention.

Risk of Bias and Quality of Evidence Assessment
Two independent reviewers (MP and JS) assessed the risk of bias using the Cochrane Risk of Bias Tool (ROB2 and ROBIN-I) [57,58]. Any disagreements between the reviewers were resolved through consensus, with reference to the full text. The quality of randomized controlled studies was assessed using the Jadad scale in the following domains: measures, interventions, assignment, confounded conditions, and use of multimodal measures. The domains were confirmed as good, fair, poor, or unclear. A study was assessed as poor quality if it had at least one poor quality domain. If a study was assessed with both unclear and good quality domains, it was assessed as being of unclear quality. A funnel plot was created using RevMan 5.4.1 to show the risk of publication bias.
Quality assessment of the included 21 articles was conducted using the Jadad scale [59] and Newcastle-Ottawa Scale (NOS) [60]. The quality of the randomized controlled studies was assessed in three areas: randomization, blinding, and description of withdrawals/dropouts. The quality of the non-randomized controlled studies was assessed in four areas: selection, comparability, exposure, and outcome. The NOS consists of eight items with three subscales, and the maximum total score for these three subsets is 9. Since there is no consensus on what defines a high-quality study, we considered a study with a score of less than 7 to be of high quality [61].

Data Synthesis
Narrative synthesis was used to report the main characteristics, methods, and findings of the included studies. In addition, a meta-analysis was conducted on studies with primary outcomes centered on PTSD reduction. A mean difference combined with a random-effects model was used to synthesize continuous data. The mean difference, 95% confidence interval, and p-value were provided with an associated I 2 measure of heterogeneity. If I 2 > 80%, this result was explored using the following subgroups to explain the heterogeneity: age, frailty, intervention type, sex, study size, and study-level quality assessment.

Study Characteristics
A total of 151 reports were identified from the electronic database searches after removing duplicate records. After screening titles and abstracts for relevance, 58 full texts were assessed to yield the 21 studies included in the review. The remaining 21 studies were quality-assessed and included in the final synthesis. Of the included studies, 16 were randomized controlled trials, three were controlled studies [27,34,71], and two were cost effectiveness analyses [42,54]. See Table 1 for details.
All included studies were of good quality. See Tables 2 and 3 for details.  Table 3. Quality Assessment of non-randomised studies using the Newcastle-Ottawa Scale (NOS) assessment tool [60].
Regarding intervention type, six were psychotherapies, including three using narrative exposure therapy (NET) and one each using eye movement desensitization and reprocessing (EMDR), cognitive behavioral technique (CBT), and interpersonal psychotherapy (IPT). The duration of psychological intervention for affected people varied from 1 h behavior treatment [14] to two to three-hour psychotherapy [37] to five weekly sessions of psychosocial intervention [44,55] and 12 h psychotherapy [36]. The length of mental health training intervention for providers (practitioners) ranged from one day [34,53] to 6 h training [51] to one or two weeks [38,44] to a maximum period of 40 h, as, for example, in Problem Management Plus (PM+) [54].
In terms of sample size, 14.3% of the studies had small samples (n < 30), most of the studies (66.7%) were between 31 and 600, and 19.0% were above 600.

Effectiveness of the Interventions Psychotherapies
We found seven studies that investigated this type of intervention-they found it was largely effective in all studies [11,14,31,[35][36][37]44]. Psychotherapy and narrative exposure therapy significantly reduced PTSD symptoms in adult refugees living in the Kilis Refugee Camp at the Turkish-Syrian border [11], as well as in Rwandan widows and orphans intervened by psychologists [37], when compared with the control groups. Interpersonal psychotherapy, administered by mental health professionals, was found to reduce depression in people who continued to suffer mental health impacts from the 2008 Wenchuan earthquake in Sichuan, China [36]. Narrative exposure treatment provided by psychologists had a significant effect on lowering PTSD, depression, and anxiety as well as improving post-traumatic growth and perceived social support in Sichuan earthquake survivors [31,35].

Psychoeducation or Trainings
We found eight studies that investigated these types of intervention-they found it was largely effective in eight studies [29,38,38,40,46,51,53,54]. Counselling delivered by community mental health workers in primary health clinics in northern Iraq significantly reduced depression, dysfunction, and anxiety in adults [40]. A brief group psychological intervention administered by non-specialists working with local community health professionals had a significant effect on anxiety, depression, and dysfunction in women aged 18-60 years living in rural villages in Pakistan [38,54]. Psychological first aid training provided by trained mental health nurses significantly improved self-efficacy and perceived preparedness among nurses in Turkey [51], significantly enhanced their knowledge and understanding of appropriate psychosocial responses, and improved skills in primary health care workers in Sierra Leone [53]. When compared with control groups, two web-based psychoeducation interventions had significantly fewer PTSD symptoms [38] and depression [29]. The brief mental-health-integrated disaster preparedness training conducted by trained lay mental health workers in Haiti was effective [46].

Psychosocial Support
We found two studies investigating psychosocial support. One found that it was ineffective in reducing PTSD symptoms, depression symptoms, or improving hope among adolescents delivered by trained school instructors [49], while another study found that the control group was more effective in reducing PTSD symptoms compared with the intervention group which provided support by community non-mental health professionals [55].

Cost-Effectiveness of the Interventions
Only two studies [42,54] assessed the cost-effectiveness of psychological skills and behavior technique interventions in disaster settings. A simulation model found that stepped care case-finding with referral to cognitive behavioral therapy for US hurricane survivors was more cost-effective than referral to coping skills alone at an acceptable additional cost per disability-adjusted life year [42]. Problem management interventions using problem-solving and behavioral techniques delivered by trained non-specialist health workers to distressed primary care patients were found to be more expensive but more effective than treatment as usual [54]. They concluded that the intervention was likely to be more cost-effective, although threshold willingness to pay values were problematic in the LAMIC context. Table 4 presents an overview of the results of the RCTs and controlled studies.

Meta-Analysis
A meta-analysis of eleven PTSD studies was performed. PTSD-focused psychotherapy interventions delivered by professionals were compared with a waitlist control group [11,14,31,35,37] or a usual care control group [36] during the synthesis process. PTSD-focused mental health interventions conducted by non-professionals were compared with waitlist controls [27,46] or a comparison group [44]. PTSD-focused online mental health interventions were compared with no intervention [38] or usual care conditions [29]). The last time-point of data collection was used as the effect size. In those studies that used waitlist control groups, the pre-intervention data of the waitlist group were compared with the post-intervention data of the intervention group. Nine different measures were used to assess PTSD symptoms. See Table 5. Potential publication bias was assessed by visual examination of the funnel plots. See Figure 2.
(2019) [46] community members, drawn from three disaster-affected communities 37 grated dis-aster preparedness Waitlist control 6 regression coeffi-cients Subgroup: Effect of Mental Health Intervention on PTSD-Related Symptoms Potential publication bias was assessed by visual examination of the funnel plots. See Figure 2.

Comparison of Mental-Health-Professional-Delivered Mental Health Interventions versus Control Group
Six studies with 313 participants compared interventions delivered by specialist mental health professionals, including psychotherapists, psychiatrists, and clinical psychologists or social and human service providers who provide support to improve an individual's mental health or treat mental disorders. The average age of the participants was 23.0 years, and 58.0% were female. A mean difference fitted with a random-effects model was used to synthesize the continuous data. The random effect model was to estimate the impact of a person's intrinsic and immeasurable qualities. Five studies found that mental healthcare provided by specialists was more successful in reducing PTSD-related symptoms compared with controls. One study found that mental health interventions were ineffective at reducing PTSD [11].
After pooling, we found that interventions delivered by mental health specialists had a clear effect on PTSD reduction, with a standardized MD = 0.99 (95% CI: 0.42-1.57, I 2 = 80, p = 0.0007). Thus, compared with the control group, mental health interventions administered by health professionals were beneficial. We fitted professional-delivered intervention subgroups to this finding and were unable to explain the heterogeneity. See Figure 3.
PTSD reduction with professional-delivered mental health interventions compared with PTSD reduction with control group. Effect sizes (Std diff in means) were computed for control designs. 0.20 = small, 0.50 = medium, and 0.80 = large [62].

Comparison of Non-Specialists' Mental Health Interventions versus PTSD Reduction without Mental Health Interventions
Three studies with 831 participants compared these outcomes. After pooling, we found that the SMD was 0.25 (95% CI: 0.11-0.39; I 2 = 0%, p = 0.0007). The effectiveness of non-specialist mental health interventions was shown to be small compared with waitlist controls. One of the three included studies found mental-health-integrated disaster preparedness intervention to be more effective in reducing PTSD-related symptoms [46], while the other two found no significant effect of trauma-focused cognitive behavior therapy and a school-based intervention programme which combined psychoeducation and cognitive behavioral techniques in improving the symptoms of post-trauma compared with the control groups [27,44]. See Figure 3.
After pooling, we found that interventions delivered by mental health specialists had a clear effect on PTSD reduction, with a standardized MD = 0.99 (95% CI: 0.42-1.57, I 2 = 80, p = 0.0007). Thus, compared with the control group, mental health interventions administered by health professionals were beneficial. We fitted professional-delivered intervention subgroups to this finding and were unable to explain the heterogeneity. See Figure 3. Figure 3. PTSD reduction with professional-delivered mental health interventions compared with PTSD reduction with control group [11,14,27,29,31,[35][36][37][38]44,46]. PTSD reduction with professional-delivered mental health interventions compared with PTSD reduction with control group. Effect sizes (Std diff in means) were computed for control designs. 0.20 = small, 0.50 = medium, and 0.80 = large [62].

Comparison of PTSD-Related Symptoms with Web Mental Health Interventions versus PTSD-Related Symptoms without Mental Health Interventions
Two studies with 658 participants evaluated web-based interventions. After pooling, the SMD was 0.19 (95% CI: 0.14-0.35; I 2 = 0%, p = 0.01), suggesting that mental health interventions delivered via the Internet were only marginally effective. One study found that mental health interventions were more effective in reducing PTSD [38], while another study found that mental health interventions were not effective [29]. The latter had a limited sample size of 37 [29] compared with those that identified an effect, which had a median of 188 participants. Consequently, a lack of significant effect was associated with a lack of power. See Figure 3.

Discussion
This is the first systematic review to comprehensively explore both effectiveness and cost-effectiveness of the full range of mental health interventions delivered by frontline health care workers in disaster and emergency contexts. We found evidence of a strong and significant association between mental health interventions provided by frontline specialist health care workers and reduced PTSD symptoms in survivors of disasters and health emergencies. We identified a lack of health economic evidence needed to support decisionmaking and public investment in enhancing mental health skills training in disasters and public health emergencies. These findings support scaling up the timely and effective mental health interventions to enhance mental health capability of responders in disasters and emergencies and better support people in need.
This systematic review identified 21 studies, including 17 randomized controlled trials and four non-randomized studies, evaluating the effectiveness and cost-effectiveness of mental health interventions delivered by frontline healthcare workers. Only two of these studies assessed the cost-effectiveness of the interventions. The vast majority of RCTs (15/17, 88%) found an improvement in mental health outcomes. Most of these studies examined the primary outcome of PTSD reduction (11/17, 65%). Owing to the heterogeneity of the identified studies, a meta-analysis could not be performed. However, it was possible to implement a meta-analysis of 11 studies on interventions targeted at post-disaster traumatic symptoms and PTSD. This meta-analysis found a strong and significant association between mental health interventions provided by frontline specialist health care workers and reduced PTSD symptoms in survivors of disasters and health emergencies but only small or marginal effects for mental health interventions delivered by non-specialists or web-based, respectively.
There is a need for more studies on mental health interventions for various groups with specific requirements, such as young people, women, the disabled, and the elderly [27,38,38,44]. The results of studies focused on school aged adolescents and children were mixed. School-based mental health interventions targeting children and their educational environment significantly improved overall daily functioning but not trauma-related symptoms [27,44]. The latter could be explained by the fact that it used problem-solving as a comparison instead of no intervention [44]. However, trained teachers' psychosocial support has no significant effect on children's PTSD or depression symptoms [49], implying that closer collaboration between mental health professionals and teachers, as well as more specific training, is required for school-based interventions to be both feasible and sustainable [27].
Low intensity and easy-to-implement psychological interventions were also more acceptable for survivors following disasters, as well as more likely to be delivered by trained non-mental health frontline workers, and significantly reduced distress, PTSD symptoms, and functional impairment [55]. Such "low-intensity intervention" or support by community-based providers with less mental health expertise has had a big impact on providing accessible mental health support in resource-poor settings in humanitarian circumstances [35,37,38,40,46,71]. Furthermore, because hiring local trainers/supervisors can reduce costs, local mental health capacity-building is recommended. For disaster response and recovery, multi-sector collaboration and coordination of psychological techniques as well as community-based efforts have been advocated [71]. Additional resources should also be assigned to improve preparedness and response capabilities [46,71].
This review included only publications written in English. For the data synthesis, the narrative methodology for whole sampling may duplicate evidence and risk subjective analysis of studies. Despite these limitations, narrative reviews offer a breadth of literature coverage and flexibility to deal with evolving knowledge and concepts.

Conclusions
This systematic review synthesised randomized controlled trials and cost-effectiveness studies as key evidence for assessing psychological and mental health interventions delivered by frontline healthcare workers in disaster contexts. Mental health interventions delivered by frontline healthcare workers may be time-and cost-effective for lowering psychological distress in natural disaster settings [44,46]. This review demonstrates that brief and low-intensity interventions, such as psychoeducation and social support programs, as well as skills training, are viable options in disasters.
More controlled trials with adequately powered sample sizes and longer-term followups are needed. Given the inherent difficulty in conducting RCTs among disaster survivors in LAMICs, economic modelling may be particularly useful to estimate cost effectiveness at the population level.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/ijerph192315847/s1. Author Contributions: All authors have contributed sufficiently to the manuscript to be included as authors. Conception and design of the study: M.P., J.S. and T.X.; review of quantitative data analysis: B.C. and P.C.; drafting the manuscript: M.P. and J.S. All authors have read and agreed to the published version of the manuscript.

Abbreviations
The following abbreviations are used in this manuscript: