Effectiveness of Resilience Interventions on Psychosocial Outcomes for Persons With Neurocognitive Disorders: A Systematic Review and Meta-Analysis

Background: Neurocognitive disorders, such as mild cognitive impairment (MCI), dementia, and Alzheimer's disease, not only harm people's cognitive function but also lead to negative emotions, poor quality of life (QOL), and unsatisfactory level of well-being. Resilience can be defined as a dynamic and amendable process, which maintains or improves life satisfaction and quick recovery from own dilemma. However, no meta-analysis of randomized controlled trials (RCTs) has thus far examined the effectiveness of resilience interventions among persons with neurocognitive disorders, and the results of RCTs were inconsistent. This systematic review aimed to assess the effectiveness of resilience interventions on psychosocial outcomes among persons with neurocognitive disorders. Methods: Nine electronic Chinese and English databases (the Cochrane Library, PsycINFO, Web of Science, PubMed, Medline, Eric, JSTOR, CNKI, and WANGFANG) were searched through April 2021. Only RCTs were included, and the quality of the included studies was assessed by the Cochrane “Risk of Bias” tool. Meta-analysis was carried out on psychosocial outcomes, and heterogeneity was investigated by subgroup and sensitivity analysis. RevMan 5.4 was used for meta-analysis. Results: Fourteen RCT studies were identified, representing a total of 2,442 participants with neurocognitive disorders. The risk of bias was high or unclear for most included studies in the domains of allocation concealment, blinding participants, and interventionists. Meta-analysis showed that heterogeneity was low or moderate. There were significant differences in favor of resilience interventions compared with control on the outcome of QOL, using the Quality of Life-Alzheimer Disease scale (QOL-AD) [I2 = 36%, standardized mean difference (SMD) = 0.14, 95% CI (0.02, 0.26), p = 0.02], and no significant differences on depression, using the Cornell Scale for Depression in Dementia (CSDD) [I2 = 41%, SMD = −0.14, 95% CI (−0.34, 0.05), p = 0.16], and neuropsychiatric symptoms using the Neuropsychiatric Inventory Questionnaire (NPI-Q) [I2 = 62%, SMD = −0.10, 95% CI (−0.37, −0.16), p ≤ 0.46]. Conclusions: Resilience interventions had a significant benefit on QOL but no significant benefit on depression and neuropsychiatric behavioral symptoms. More evidence is needed to answer questions about how to implement resilience interventions and how to evaluate their effectiveness.


INTRODUCTION
Mild cognitive impairment (MCI), dementia, or Alzheimer's disease (following abbreviation neurocognitive disorders) are chronic progressive syndrome. During the phase transition process, divergent sections of the brain are affected, and a persons' capability of adaptation to the disease and environment gradually decreases (1). Neurocognitive disorders not only impair the person's memory, orientation, thinking, cognitive functioning, and language (2), but also trigger emotions and psychological symptoms, such as depression and anxiety (3,4). In addition, the impact of neurocognitive disorders is longlasting; it is extremely exacting or nearly impossible to be cured completely (5,6). With these detrimental outcomes, neurocognitive disorders further lead to an unsatisfactory level of well-being and quality of life (QOL), such as physical function, and financial instability of individuals and families; these results also undermine people's ability to fulfill family, social, and professional roles (7). Neurocognitive disorders currently affected tens of millions of people all over the world and caused enormous medical and economic burdens. For example, there were 50 million people with dementia worldwide (2), and the number of people with dementia worldwide is projected to be 152 million in 2050 (8). Dementia contributes significantly to the global burden of disease, costing an estimated $818 billion annually (9), expected to reach $2 trillion by 2030 (8,(10)(11)(12). Thus, neurocognitive disorders are regarded as one of the greatest social, health, and economic challenges of the twentyfirst century (8,(11)(12)(13). It is progressively crucial to develop strategies that facilitate and help persons with neurocognitive disorders to maintain independence, well-functioning, and high QOL in the long run.
There were various approaches to coping with the challenges affiliated with neurocognitive disorders. Resilience-centered interventions can be seen as one important approach to adapting to stress and reducing the adverse impact of the stressors (7). Luthar and Cicchetti (14) defined resilience as "a dynamic and amendable process, " in which people use resources to acclimate to adversity (15). Meanwhile, Kunzler et al. (16) highlighted that resilience-centered interventions could be seen as the process of maintenance, withstanding, overcoming, adjustment, adaptation, posttraumatic growth, stress-related growth, rebound from a stressor, or rapid readjustment. As mentioned above, resilience can be defined as a process: after experiencing acute (short-run) or chronic (long-run) issues on health and stress, an individual can actively adapt, withstand, overcome, adjust, cope, and grow to maintain and improve his or her QOL with the support from multifaceted resources on individual and social levels.
The literature illustrated that protective factors of resilience were diverse, such as self-care, adherence to treatment programs, patient perceptions of pain and disease, adherence to physical activities, self-empowerment, health-related QOL, self-efficacy improvement, stress, depression, and anxiety reduction, optimism in viewpoints, and recovery acceleration (17)(18)(19)(20)(21). However, the process of resilience interventions focuses on reinforcing personal characteristics and exterior assets in response to a severe challenge to build an inclusive environment with multifaceted psychosocial supports (22). Similarly, other literature also suggested that resilience framework should include individual, family, community, and social components (23,24). For example, Harris (24) stressed strengthening personal attributes in the process of resilience interventions, which could include self-acceptance of a person with neurocognitive disorders with the shifts in self, nurturing the individual's remaining competence, a positive perspective on diseases and dilemmas, and recognition of numerous means in which someone with dementia can contribute meaningly to their friends, family, and/or the community. Casey et al. (22) suggested five domains to implement resilience interventions: having a "fighting spirit" and personal control, maintaining solid family relationships, maintaining ties to communities, increasing awareness, addressing negative attitudes through dementia education, and engaging in physical activity. Kunzler et al. (16) suggested that resilience should include supportive doctors, linkages to helpful community groups and events, and sympathetic and supportive social surroundings. Overall, the literature mentioned above indicated that in the resilience process, multifaceted interventions should be taken, involving interactions between individuals and the external resources.
The outcomes of resilience-centered interventions are psychosocial, such as improving QOL, restoring normal performance, maintaining mental health, improving adaptability (25), better adjustment (26), enhancing mental well-being (27), reducing care dependence, good social relations, positive self-image (28), reducing burden or stress (27,29), enhancing intent or meaning of life, and obtaining self-esteem, positive emotions, self-efficacy, boldness, active coping, optimism, social support, adaptation, and cognitive flexibility (including positive reassessment and acceptance) (16). However, one question arises regarding how to measure resilience-centered intervention outcomes in different resilience approaches. Windle et al. (30) studied tools for resilience interventions and concluded that the conceptual and theoretical adequacy of the scales was questionable, with no existing "gold standard" of resilience measures. Whelan et al. (15) indicated that key sets of outcomes for resilience in neurocognitive disorders have not been identified. For example, Ghanei Gheshlagh et al. (7) used three scales, which are Resilience Scale-25 (RS-25), Connor-Davidson Resilience Scale-10 (CDRISC-10), and Connor-Davidson Resilience Scale-25 (CDRISC-25), to assess the effectiveness of the resilience process for people with chronic physical diseases. Saint-Bryant et al. (26) used the Cornell Scale for Depression in Dementia (CSDD), Quality of Life Alzheimer's Disease scale (QOL-AD), and the Index of Relocation Adjustment Scale (IRA) to measure depression, QOL, and adjustment conditions as the outcomes of resilience process for older adults with dementia.
Another question is that the effect of resilience interventions is inconclusive. Thus far, there are only three systematic reviews (31-33) related to resilience interventions or outcome measures. Although Li et al. (31) claimed to contain resilience training, it was a mere strength training different from resilience. They also assessed cognitive outcomes, such as executive cognitive ability, global cognitive function, memory, and attention. Findings indicated positive effects on the executive cognitive capability and overall cognitive function, a weak-positive effect on memory, and no significance in attention. In the review of Carrion et al. (32), the resilience interventions focused on cognition-oriented caregiving approaches. The included 47 randomized controlled trials (RCTs) did not conduct a meta-analysis, and the results were inconclusive. In the review of Regan and Varanelli (33), the resilience interventions used modified cognitive behavior therapy (CBT) and problem-solving approach. They assessed three outcomes, including depression, anxiety, and adjustment in older adults with mild cognitive impairment and early dementia. The included seven RCTs and eight pre-post studies indicated positive effects in reducing depression in older persons with early dementia. However, Regan and Varanelli's (33) review did not conduct a meta-analysis, so it was unable to draw a clear conclusion about the intervention effect because of the divergent methods of the included studies. Additionally, the narrative review (15) identified five resilience interventions in three empirical studies of six papers, including Peer Support Network Services, Dementia Advisors, Memory Makers, Visual Arts Enrichment Activities, and Early-Stage and Beyond Community Activities. However, this narrative review included empirical studies, and the effectiveness of the resilience interventions could not be determined due to the study design.
Overall, there exist several dissimilarities regarding methodology and quality of studies among previous systematic reviews, which leads to inconsistent results regarding the effectiveness of resilience interventions. Currently, there is no systematic review that both includes RCT studies and conducts a meta-analysis. Therefore, this review aimed to identify RCT resilience interventions among persons with neurocognitive disorders to assess the effectiveness and provide further detailed evidence. This may contribute to enhancing existing resilience interventions and to facilitating the future development of such programs.

Participants
Participants were people of all ages with neurocognitive disorders, including dementia, mild cognitive impairment (MCI), and Alzheimer's disease. Participants' formal diagnoses on types and severity of those neurocognitive disorders were based on corresponding scales, including the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) (34); International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10) (35); or other comparable diagnostic criteria. We included people living in diverse settings, such as the community, hospitals, and nursing homes. We did not use a criterion for age so as not to exclude studies in which some participants were below 60 years old. If a mixed sample of participants (e.g., people with dementia and their caregivers) were found and the data of persons with dementia were reported separately or were collected by contacting the author, these studies also were included.

Interventions
Any intervention that promotes a person's state of adaptation and adjustment with the help of personal attributes and external assets, regardless of content, duration, setting, or mode of delivery, was included. Resilience, for example, can include active coping (e.g., planning, problem solving), self-efficacy, optimism or positive attributional style, cognitive flexibility (e.g., positive reassessment and acceptance of negative emotions and conditions), religiosity and spirituality (e.g., frequent religious visits), positive emotions or positive affect, hardiness, self-esteem, intent or meaning of life, sense of coherency (internal), locus of control, coping flexibility, hope, humor, altruism (16), physical strength training, formal or informal care, social connection, and community or other external resource support.
Studies were excluded if they involved animal trials and non-psychological or non-social interventions of resilience, such as pharmacological interventions (e.g., treatment with antidepressants).

Comparators
Comparators included no treatment, treatment as usual (TAU) (e.g., routine medication and usual social activities), and waitlist control. If the control group adopted active control, such as music, physical, and cognitive-behavioral, rather than no treatment, TAU, or wait-list control, the literature was excluded. For studies with two or more controls, our meta-analysis was conducted only using the control group of no treatment, TAU, or wait-list control.

Outcome Measures
We defined outcomes as assessments of psychosocial adaptation. For these outcomes, QOL was a primary outcome, and others were secondary outcomes, such as social relations, positive self-image, self-efficacy, hardiness, anxiety, and depression. We accepted all psychosocial assessment tools used in the included studies. Outcomes were assessed before the treatment, upon completion of the treatment, and follow-up evaluations to assess long-term effects. We considered measures self-assessed and scored by observers or clinicians.
Studies were excluded if the studies contained nonpsychosocial outcomes of resilience, such as brain structure, immediate memory, attention and calculation, deferred memory, time orientation, location orientation, language, visual space, or the geographical environment. This ensured that the review focused on the psychosocial outcomes of resilience interventions. The absence of the outcome values was an exclusion criterion for this review: if the values of mean and standard deviation (SD) were not reported in the description of outcome, mean and SD cannot be obtained by contacting the authors, or mean and SD cannot be calculated by the review manager software or calculator provided by Cochran, the original study was deleted (Review manager software or calculator: https://training. cochrane.org/online-learning/core-software-cochrane-reviews/ revman/revman-5-download; https://training.cochrane.org/ resource/revman-calculator).

Types of Studies
Our review intended to include both published and unpublished RCTs in Chinese or English language. We also took into account cluster RCTs.

Electronic Searches
Nine electronic databases (the Cochrane Library, PsycINFO, Web of Science, PubMed, Medline, Eric, JSTOR, CNKI, and WANGFANG) were searched through April 2021. Gray literature was also searched from ProQuest Dissertations & Theses Database (PQDT) and DUXIU and reviewed. Authors of relevant conference abstracts were reached out for possible information sharing. The search terms used were the following: (a) dementia or Alzheimer or cognitive loss: MCI or dementia or Alzheimer or ADRD or "cognitive impai * " or "cognitive loss" or "cognitive decline"; (b) resilience or resiliency: resilien * OR adjust * OR adapt * OR "post-traumatic growth" OR "post-traumatic growth" OR "stress-related growth" OR withstand * OR overcom * OR resist * OR recover * OR thriv * OR adapt * OR adjust * OR bounc * back; and (c) RCT or random * . We used "subject OR title OR abstract OR keywords OR topic" to search. Search strategy was (MCI or dementia or Alzheimer or "cognitive impai * " or "cognitive loss" or "cognitive decline") AND (resilien * OR adjust * OR adapt * OR "post-traumatic growth" OR "posttraumatic growth" OR "stress-related growth" OR withstand * OR overcom * OR resist * OR recover * OR thriv * OR adapt * OR adjust * OR bounc * back) AND (RCT or random * ).

Assessment of Risk of Bias in Included Studies
We employed the Cochrane "Risk of Bias" tool (36) to identify any risk of bias with a judgment of low risk, high risk, or unclear risk of bias for each study of the following areas: (1) selection bias, (2) random sequence generation, (3) allocation concealment, (4) blinding of participants and personnel, (5) blinding of outcome assessment, (6) incomplete outcome data, and (7) selective reporting.

Studies Screening
Three reviewers (YW, YZH, and WCH) screened articles according to inclusion and exclusion criteria. Three reviewers independently reviewed the studies' title and abstract, then screened the full paper, and independently evaluated methodological quality. Any uncertainties concerning suitability were discussed at weekly group meetings with all reviewers.

Data Extraction
Three reviewers independently extracted data using a predesigned form from the included studies. The following data were extracted: (1) basic study information, namely, authors, reference, and country/region; (2) participant characteristics, namely, illness/condition, total number, and number in each group, age, gender, and race/ethnicity; (3) intervention characteristics, namely, intervention content, individual or group format, in-person or virtual, setting, length (e.g., number of weeks), number of sessions, duration per session, and control; (4) intervention assessment information, namely, time point (e.g., pretest, posttest, follow-up), measures, outcomes with screenshots (including the mean, standard deviation, and number of participants in each group at each time point), and outcome raters (e.g., patients, caregivers, and staff); and (5) information on bias risk assessment (see Assessment of risk of bias in included studies). After comparing results, any uncertainties that could not be solved were discussed in weekly meetings with all reviewers.

Data Analysis
RevMan 5.4 was used for meta-analysis. Meta-analysis was performed if outcomes were measured by the same scales in at least two studies. Heterogeneity was assessed using an I 2 statistic. To interpret heterogeneity, reviewers followed Cochrane guidance: 0-40% as not important, 30-60% as moderate heterogeneity, 50-90% as substantial heterogeneity, and 75-100% as considerable heterogeneity (37). A random-effects model was used if I 2 statistics reports the value of 50% or above. A fixedeffects model was used if I 2 statistics were lower than 50%. If one study used more than one instrument to measure the same outcome variable, the team employed the more commonly used instrument for the analysis.
Subgroup analyses were conducted with the following characteristics if applicable: outcome instrument, disease/conditions, country/region, rater, and follow-up. If the heterogeneity showed moderate or high, we performed subgroup and sensitivity analysis. Publication biases were assessed by a funnel plot if the number of studies used for meta-analysis was more than 10.

Figure 1 shows the Preferred Reporting Items for Systematic
Reviews and Meta-analyses (PRISMA) flow chart of the study review and selection process. A total of 2,890 studies were searched from electronic searches. After deduplication, we considered a total of 2,885 studies. The remaining studies were screened at the title and abstract level based on the pre-established inclusion and exclusion criteria, depending on the study type, population, intervention, control, and outcome. Forty-five full papers were then reviewed, from which 31 were excluded utilizing the same criteria. Fourteen RCT studies satisfied all the inclusion criteria. Thus, 14 RCT studies representing a total of 2,442 participants with neurocognitive disorders were included in the systematic review and meta-analysis.

Included Study Characteristics
Location Among 14 studies, 7 were conducted in the UK, 2 in the USA, 1 in Germany, 1 in Denmark, 1 in France, 1 in Netherlands, and 1 in Norway ( Table 1). FIGURE 1 | Flowchart of the study selection process.

Participants
Participants in most studies (n = 10) were persons with dementia (26, 28, 29, 38, 41-44, 46, 47), one as persons with MCI (39), one as persons with MCI and early dementia (27), one as persons with dementia or cognitive impairment (45), and one was Alzheimer's (40). The number of participants ranged from 19 to 726, and seven studies had more than 100 participants. Participants were recruited from communities (n = 6) and nursing homes/residential care facilities (n = 8).

Interventions
We identified six resilience approaches based on content descriptions, including integrated approaches (n = 5), exercise regimen or activity programs (n = 4), psychological interventional technique (n = 2), a psychiatric intervention (n = 1), disease/case management (n = 1), and cognitive stimulation therapy (n = 1). Among five studies with integrated approaches, one was individualized person-centered care provided by standard procedures to diminish turmoil in care home residents with dementia (41), one was multimodal nondrug therapy on dementia's symptom and care need (38), one was multidimensional home-based care coordination provided by an interdisciplinary team to maximize independence for persons with MCI living home (39), one was a staff-led intervention that comprises four mandatory modules and one optional module to facilitate the adaptation of seniors with dementia after placement into residential care (26), and one was a psychosocial intervention including multifaceted and semi-tailored counseling, education, and support (40). Four studies provided exercise regimens or activity programs (28,(43)(44)(45), such as adapted Tai Chi and cognition-action program, walking, aerobic, and strength exercise training. Two mental interventional techniques used mindfulness (29) and reminiscence (46). One study of recovery-orientated psychiatric intervention packages included prediagnostic wellbeing assessment and counseling, diagnostic consultation with written feedback, and postdiagnostic support (27). One disease management employed an internet-based care management software system for care planning and coordination (42). One individual cognitive stimulation therapy intervened and assessed adaptive functioning and QOL of participants with dementia (47). All interventions were conducted in groups of persons older than 40 years old.

Effects of Interventions
Overall, the effects of resilience interventions were diverse in various outcomes. In terms of social behavior, one study indicated significant differences in favor of resilience interventions compared with controls (38). Meanwhile, no significant differences were found in neuropsychiatric symptoms (39-41, 43, 44), adjustment (26), stress (29), and anxiety (29) in favor of resilience interventions compared with controls. Besides, the results of other outcomes' assessments were inconsistent. Specifically, regarding QOL, seven studies showed no statistical significance in favor of resilience interventions compared with controls (26, 27, 40-42, 44, 46), while two showed a significant effect (29,47). One showed a significant enhancement in selfreported QOL but no significant improvement in proxy-rated QOL (39). One showed that activities of daily living (ADL) training had positively affected overall QOL, but no benefits were observed for exercise on QOL (28). As for depression, four included studies showed that the resilience interventions were not statistically different compared with controls (27,29,39,40), while one showed statistically significant differences (45). In terms of well-being, one study showed no significance in favor of resilience interventions compared with controls (26), while the other study showed statistical significance (27). About the effects on agitation, one showed no significance in favor of resilience interventions compared with controls (41), while one showed statistical significance (45).

Risk of Bias
Risks of bias are summarized in Figure 2. The main flaws for risks of bias across 14 studies were in allocation concealment and blinding participants and interventionists. Regarding random sequence generation, 13 studies were judged to be at low risk, which used computer-generated random numbers, a block randomization method, a custom Excel program, or a webbased system.
For allocation concealment, five studies were rated as low risk, which used sealed envelopes, allocation numbers by a blind assigner, or emphasis on allocation concealment. Five studies were judged to be high risk, which reported no concealment in the allocation process. The remaining four studies did not report information on allocation concealment and were rated as unclear.
Regarding blinding participants and interventionists, 3 studies were rated as low risk, 10 studies as high risk, and 1 study as unclear. Meanwhile, most studies (n = 13) were judged as low risk for blinding outcome assessment, by using personnel not included in the intervention process, and one study was judged as high risk due to non-blinding outcome assessment.
All studies were judged to be at low risk for incomplete outcome data because the dropout rate was low (<30%) during the intervention, and they explained the numbers and reasons for dropout and the data analysis methods of dealing with missing values. Lastly, all studies were judged as low risk for selective reporting.

Meta-Analysis
Seven studies reported data on quality of life, assessed by QOL-AD and were pooled for a meta-analysis using a fixed-effects model. Results illustrated that there were significant standardized mean differences in favor of resilience interventions compared with controls for QOL [SMD = 0.14, 95% CI (0.02, 0.32), p = 0.02] (Figure 3).

Subgroup Analyses
We further performed subgroup analyses with results shown in

Assessment of Sensitivity
The heterogeneity of the seven included studies was moderate (I 2 = 36, chi 2 = 9.41), which suggested that heterogeneity might not be important as explained in Cochrane guidance. The sensitivity analysis showed no heterogeneity (I 2 = 0% chi 2 = 4.55) after a small sample study (n = 40) (47) was deleted (see Figure 5).

Analysis of Publication
Only seven RCTs were included, so the funnel plot was not made, but publication bias may exist.

Meta-Analysis
Five studies reported data on depression assessed by the CSDD and were pooled for a meta-analysis using a fixed-effects model. Results demonstrated that there were no significant standardized mean differences in favor of resilience interventions compared with controls for depression [SMD = −0.14, 95% CI (−0.34, 0.05), p = 0.16] (Figure 6). Figure 7 shows that no significant standardized mean differences in favor of resilience interventions compared with controls for CSDD were found in subgroup analyses. The heterogeneity ranged from 0 to 83%. Specifically, the subgroup result of participants with dementia showed no heterogeneity (I 2 = 0%). Other subgroup results showed moderate or high heterogeneity (I 2 = 53%−83%).

Assessment of Sensitivity
The heterogeneity of the five included studies was moderate (I 2 = 41 chi 2 = 6.77). The sensitivity analysis showed that the heterogeneity decreased (I 2 = 0% chi 2 = 1.26) after a small sample study (n = 19) (26) was deleted (see Figure 8).

Analysis of Publication Bias
Only five RCTs were included, so the funnel plot was not made, but publication bias may exist.

Meta-Analysis
Two studies reported data on neuropsychiatric symptoms assessed by NPI-Q and were pooled for a meta-analysis using the random-effects model. Results revealed that there were no significant standardized mean differences in favor of resilience interventions compared with controls for neuropsychiatric symptoms [SMD = −0.10, 95% CI (−0.37, −0.16), p = 0.46] (Figure 9). Meanwhile, there tended to be substantial heterogeneity in the two included studies (I 2 = 62%, chi 2 = 2.62).

Analysis of Publication Bias
Only two RCTs were included, so the funnel plot was not made, but publication bias may exist.

DISCUSSION
This systematic review and meta-analysis examined the effectiveness of resilience interventions among persons with neurocognitive disorders. A total of 14 RCT studies representing 2,442 participants were identified that fulfilled the inclusion criteria of this review. The risk of bias was either high or unclear for most studies in allocation concealment, blinding participants, and interventionists domains. Meta-analyses were conducted for a primary outcome of QOL and secondary outcomes of depression and neuropsychiatric symptoms. Our results indicated that resilience interventions had a significant positive effect on persons with neurocognitive disorders in enhancing QOL but might not be beneficial in decreasing depression and neuropsychiatric symptoms. Meanwhile, many other psychosocial outcomes were measured less frequently.
Our review identified target groups of neurocognitive disorders based on various conditions including symptoms and level of severities: mild or moderate dementia, cognitive impairment or MCI, and Alzheimer's disease. Similarly, specific target groups in the review of Regan and Varanelli (33) included mild cognitive impairment and early dementia. Inconsistent with our study, some other reviews (31, 32) focused on mixed target groups, including both healthy older adults and older adults at risk of dementia, MCI, and Alzheimer's disease, focusing on the early prevention and intervention of neurocognitive disorders. This might also be one reason why it was impossible to judge the effect of interventions due to sample heterogeneity.
Meanwhile, our review included RCTs and conducted metaanalyses to assess the effect of resilience interventions. Consistent with our study, two previous systematic reviews also contained RCTs, which included 12 RCTs (31) and 47 RCTs (32), respectively. However, they applied a narrative approach to synthesize the findings without conducting a meta-analysis. Inconsistent with our study, the review of Regan et al. (33) included other study designs, such as pre-post studies, besides RCTs, which might be one of the reasons why the findings   were inconclusive. However, the quality of RCTs included in our review is not very high. Only 5 of the 14 studies were judged low risk in allocation concealment, and only 3 of the 14 studies were judged low risk in blinding of participants and interventionists domains. Although it is very difficult to implement allocation concealment and blinding of participants and interventionists domains in real-world RCT research, it is strongly recommended that more rigorous RCT research should be carried out in the future, with special attention to allocation concealment and blinding of participants and interventionists domains.
In addition, our review identified integrated resilience approaches. In contrast, the reviews of Regan and Varanelli (33) focused on psychotherapeutic approaches; the review of Li et al. (31) paid more attention to resistance training, strength, and exercise programs; and the review of Carrion et al. (32) solely focused on cognitive therapy. A broader range of resilience approaches was considered in our review, which rendered interventions diverse. Among them, integrated resilience approaches that were used by most included studies involved multiple components, such as caregiver's support, social connection, and resource support. Results indicated the possible advantages of the multiple-component interventions and multidisciplinary teamwork in active coping with complex symptoms and stress of persons with neurocognitive disorders. Thus, we call for more resilience research using integrated approaches for persons with dementia to better understand the effectiveness of integrated approaches and how to appropriately adopt and implement them.
Furthermore, our review focused on psychosocial outcomes (e.g., QOL, ADL, mental health, coping ability, adaption, adjustment). However, social outcomes were measured less frequently. The meta-analysis of psychological outcomes was conducted, so it is impossible to judge the effect of resilience interventions on social outcomes, such as improving social connection, social well-being, and resource support. Similarly, Regan and Varanelli (33) assessed depression, anxiety, and adjustment. Carrion et al. (32) rated depression. In contrast, Li et al. (31) largely focused on the effect of resistance training on cognitive function, such as executive cognitive ability, global cognitive function, attention, and memory. Since there is no "gold standard" for measuring the outcomes of resilience interventions in persons with neurocognitive disorders, we recommend that the measurement instruments can be further developed and validated to measure more effectively.
Additionally, regarding the effectiveness of resilience interventions, the previous systematic reviews' findings indicated positive effects (31,33) or inconclusive (32) on different outcomes. However, due to the different design of the included original studies or the lack of meta-analysis of RCTs in these reviews, the statistical significance of the effect of resilience  interventions could not be judged, and the level of evidence was not high. Therefore, these conclusions of intervention effects needed to be drawn cautiously. One synthesis of systematic reviews also indicated that due to the heterogeneity of the included studies, there was no sufficient evidence to determine whether resilience interventions may promote psychosocial outcomes (48). Our meta-analysis of the RCTs confirmed that resilience interventions had significant benefit to persons with neurocognitive disorders in enhancing QOL but might not be beneficial in decreasing depression and neuropsychiatric symptoms. It should be pointed out that several aspects of the original study limited the generalizability of our results: distinct approaches, diverse measurement tools and raters, and divergent settings and locations, and different data collection points during interventions and in f/u assessments. Therefore, further research is needed to address the development, implementation, and application of resilience interventions and conduct more rigorous and higher-quality RCT trials among persons with neurocognitive disorders.
This study has several limitations. First, due to the research team's language capacity, we only included English and Chinese literature, thus excluding potential useful information written in other languages. Second, although to some extent there was an accord on resilience as a dynamic process (14) and leading to psychosocial outcome (49, 50), and our review also identified various resilience interventions assessed by psychosocial outcomes, there was still no consensus about the definition of resilience and proper outcome measures. Thus, there may be some resilience studies that our study did not identify, and there could be other, equally valid, ways to define resilience that we did not consider. Last but not least, this study might be limited by the selected databases. Although the investigators included the most widely used English and Chinese databases, it remains possible that some works, particularly unpublished studies conducted in other countries, were not located and examined.

CONCLUSION
The study findings indicated significant benefits of resilience interventions on QOL but no significant benefits of resilience interventions on depression and neuropsychiatric behavioral symptoms among persons with neurocognitive disorders.
There is an ongoing need for additional evidence to support the effectiveness of resilience interventions, how to further improve resilience interventions, how to implement them, and how to evaluate the effectiveness in persons with neurocognitive disorders. In addition, there is a need to strengthen methodological quality to assess and determine the effects of resilience interventions.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.