Herbal Medicine for Behavioral and Psychological Symptoms of Dementia: A Systematic Review and Meta-Analysis

Background: Dementia is a global health concern, causing serious health and socioeconomic burdens with population aging. The associated symptoms of dementia, called behavioral and psychological symptoms of dementia (BPSD), are factors contributing to the socioeconomic burden of dementia. Recently, herbal medicine (HM) has attracted attention as a potential complementary therapy for BPSD. Therefore, this systematic review was aimed at analyzing the effectiveness (or efficacy), safety, and research status of HM in BPSD management through a comprehensive review. Methods: Thirteen electronic databases were searched comprehensively. Related clinical studies published until December 28, 2020, were collected. The methodological quality was evaluated using tools such as the Cochrane Collaboration’s risk of bias tool according to the study design. The effectiveness (or efficacy) was analyzed for randomized controlled trials (RCTs) only, and when sufficient homogeneity was assured, effect estimates were presented as mean difference (MD) and risk ratio (RR), with 95% confidence interval (CIs), through a meta-analysis. Results: A total of 52 clinical studies, including 36 RCTs, were included in this review. As an adjunctive therapy, HM showed statistically significant benefits in BPSD severity assessed by the Behavior Pathology in Alzheimer’s Disease Rating Scale (combined with psychotropic drugs: MD = −3.48, 95% CI: −3.96 to −2.99; with anti-dementia drugs: MD = −2.81, 95% CI: −3.17 to −2.45) and Neuropsychiatric Inventory (with anti-dementia drugs: MD = −3.23, 95% CI: −4.06 to −2.40). Adverse events were significantly less frequent in the HM group (RR = 0.50; 95% CI: 0.28 to 0.88). However, the methodological quality of the RCTs included in this systematic review was not optimal overall. Conclusion: According to the findings of this review, HM may be associated with additional benefits in BPSD treatment, particularly when used as an adjunct to conventional medications, including psychotropic and anti-dementia drugs. However, considering the methodological quality of the included RCTs, this clinical evidence is not robust. Nevertheless, dementia is a global health concern, and considering the limitations of conventional psychotropic drugs for BPSD, a major cause of the disease burden, HM appears to be a promising complementary therapy that warrants further research.


INTRODUCTION
Dementia is a global health concern, causing serious health and socioeconomic burdens with population aging. A study comparing its prevalence and costs between 2010 and 2015 calculated overall annual trends and predicted that the worldwide costs of dementia in 2030 would reach approximately US $2 trillion (Wimo et al., 2017). The clinical manifestation of dementia can be classified into cognitive decline, i.e., core symptoms, and associated symptoms called behavioral and psychological symptoms of dementia (BPSD) (Ohno et al., 2019). BPSD is a term that encompasses various behavioral problems and psychological symptoms that may occur in patients with dementia and is related to the poor patient prognosis, burden of caregivers, and risk of institutionalization, consequently contributing to the socioeconomic burden of dementia (Cerejeira et al., 2012). BPSD is present in most patients with dementia, particularly hyperactivity, apathy, depression, and anxiety, with moderate or higher incidence (van der Linde et al., 2016).
Although pharmacological approaches, including psychotropic drugs, are frequently used to manage BPSD in clinical settings (Ozaki et al., 2017), the results are occasionally unsatisfactory, and drugs such as antipsychotics, benzodiazepines, and Z-drugs are associated with adverse events (AEs), such as increased risk of falls and all-cause mortality (Landi et al., 2005;Ralph and Espinet, 2018). Moreover, patients with dementia are mostly elderly, and the use of several psychotropic drugs in the population is considered a "potentially inappropriate medication," which discourages the use of psychotropic drugs for BPSD (By the 2019 American Geri, 2019). Therefore, more effective and safe treatments for BPSD management are necessary.
East Asian traditional medicine (EATM) is a medical system that has been established in Asian countries for a long time, and some countries, such as Korea, Japan, China, and Taiwan, use it in their national medical systems (Park et al., 2012). As an EATM modality, herbal medicine (HM) is considered to be a management strategy for dementia, particularly BPSD. For example, an HM called Yokukansan is effective against the positive symptoms of BPSD (Matsuda et al., 2013). However, other types of HMs can also be considered in the management of BPSD, highlighting the need for a comprehensive review of the various HMs that can be used in BPSD (Howes et al., 2017). Therefore, this systematic review was aimed at analyzing the effectiveness (or efficacy), safety, and research status of HM in BPSD management through a comprehensive review.

MATERIALS AND METHODS
We registered the protocol of this systematic review in the OSF registries (URL: https://osf.io/3u8ch) and International Prospective Register of Systematic Reviews (URL: https://www. crd.york.ac.uk/prospero/display_record.php?ID CRD42020211000) before beginning the study. The study protocol was as previously described (Kwon et al., 2021). No amendments were made to the information provided in the protocol. We report the systematic review in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis 2020 checklist (Page et al., 2021). and Wanfang Data on December 28, 2020. Articles published from the inception of the database to the search date were screened. We also identified additional eligible articles through reviews of relevant literature reference lists and trial registries, such as clinicaltrials.gov, and consultation with experts in this area to include additional gray literature. The detailed search strategies are described in Supplementary Material S1.

Eligibility Criteria
We included all types of original clinical studies, including randomized controlled clinical trials (RCTs), nonrandomized controlled clinical trials (CCTs), and before-after studies without restrictions on the publication language or publication status. Studies involving patients with any type of dementia in long-term care facilities, community, or specialized geriatric assessments and psychiatric units were included. Although there were no restrictions on the sex, age, or race of the participants, studies that did not provide diagnostic criteria or a validated assessment tool for inclusion and studies on patients with drug allergies or other serious illnesses, such as cancer, liver disease, or kidney disease, were excluded. We included studies involving oral HM based on EATM theories as a monotherapy or adjunctive therapies to psychotropic drugs, with or without routine care for dementia as treatment interventions. Although there were no restrictions on the dosage form of HM, we excluded studies that did not list the composition of HM, except for patent drugs. For the control intervention, we included studies involving wait-list, placebo, or psychotropic drugs, with or without routine care for dementia, such as antidementia drugs.
The primary outcome was the severity of BPSD symptoms, such as scores of the Behavior Pathology in Alzheimer's Disease Rating Scale (BEHAVE-AD) (Sclan et al., 1996), Neuropsychiatric Inventory (NPI) (Cummings et al., 1994), and Brief Psychiatric Rating Scale (BPRS) (Overall and Gorham, 1962). The secondary outcomes included 1) total effective rate (TER) for BPSD symptoms; 2) activities of daily living (ADLs) of patients, such as the Barthel Index (Mahoney and Barthel, 1965) and the Functional Independence Measure (Linacre et al., 1994), as well as instrumental ADL (IADL), such as the Activities of Daily Living Prevention Instrument (Galasko et al., 2006); 3) quality of life (QoL) of patients, such as the Alzheimer Disease Related Quality of Life (Kasper et al., 2009); 4) caregiver burden of caregivers, such as the Caregiver Burden Inventory (Novak and Guest, 1989); 5) QoL of caregivers, such as the Short Form 36 Health Survey (Ware and Sherbourne, 1992); 6) placement in a long-term care facility from home; and 7) safety data, such as incidence of AEs.

Study Selection
All documents retrieved from the databases and other sources were imported into EndNote X8 (Clarivate Analytics, Philadelphia, United States). Using "Find Duplicates" function in EndNote X8 and manual searching, duplicate documents were excluded, and two researchers (CY Kwon and B Lee) independently reviewed the possibility of inclusion by reviewing the titles and abstracts. For the first included documents, the final documents to be included were determined through a review of full texts. Disagreements between the two researchers in the study selection process were resolved through consensus.

Data Extraction
Two researchers (CY Kwon and B Lee) independently extracted the data from the included studies using a pre-defined form in Excel 2016 (Microsoft, Redmond, WA, United States). The extracted information included the first author's name, publication year, country, sample size and dropout, details of participants, treatment and control intervention, duration of intervention, main outcome measures and results after treatment ended, AEs, and information to assess the risk of bias (RoB). When the data in each included study were insufficient, we contacted the corresponding authors of the original studies via e-mail. Disagreements between the researchers in the data extraction process were resolved through consensus.

RoB Assessment
To assess the RoB of the included RCTs, we used Cochrane Collaboration's RoB tool comprising domains of random sequence generation, allocation concealment, blinding of participants, personnel, and outcome assessors, completeness of outcome data, selective reporting, and other biases. In particular, we assessed other bias items based on the statistical baseline imbalance between the treatment and control groups, such as the participant's mean age, sex, disease period, or disease severity. Each domain was assessed as "low risk," "unclear risk," or "high risk" (Higgins, 2011), and the evaluation results are presented as a figure using Review Manager software, version 5.4 (Cochrane, London, United Kingdom). For included CCTs, before-after studies, and case reports, we used the Risk Of Bias In Non-randomized Studies of Interventions tool (Sterne et al., 2016), The Quality Assessment Tool for Before-After (Pre-Post) Studies With No Control Group National Heart, Lung, and Blood Institute (NHLBI) (2013), and the Quality Assessment Tool for Case Series Studies National Heart, Lung, and Blood Institute (NHLBI) (2013), respectively. Two researchers (CY Kwon and B Lee) independently assessed the RoB of the included studies, and discrepancies were resolved through consensus.

Data Synthesis and Analysis
A descriptive analysis of the findings, including the demographic characteristics of the participants, details of the interventions, and outcomes, were conducted for all included studies. If there were two or more studies using the same type of treatment and control interventions, with the same outcome measures among our primary and secondary outcomes, a meta-analysis was conducted using Review Manager software (version 5.4; Cochrane, London, United Kingdom). For continuous and binary outcomes, the mean difference (MD) and risk ratio (RR) were calculated with 95% confidence interval (CI). We assessed heterogeneity using both the χ 2 test and the I 2 statistic, and I 2 values greater than 50 and 75% were interpreted as substantial and considerable heterogeneity, respectively. We pooled the meta-analyzed results using a random-effects model if the included studies had significant heterogeneity (I 2 > 50%) and a fixed-effect model if the heterogeneity was insignificant or if less than five studies were included in the meta-analysis because of lack of precision in the estimate of the between-study variance (Guyatt et al., 2002;Balshem et al., 2011). We planned subgroup analyses according to the severity of dementia, type of dementia, severity of BPSD, and treatment duration, if necessary data were available. The Mini-Mental State Examination score was used to classify the severity of dementia of the participants, with scores of 20-24, 13-20, and 12 or less regarded as mild, moderate, and severe, respectively. Additionally, we conducted a sensitivity analysis to identify the robustness of the results of the metaanalysis by excluding 1) studies with high RoB and 2) outliers that are numerically distant from the rest of the data. If more than ten studies were included in each meta-analysis, we planned to assess the publication bias using a funnel plot.

RoB in Studies
For RCTs, a total of 19 Furukawa et al., 2017;Han, 2018;Lin et al., 2016;Mizukami et al., 2009;Monji et al., 2009;Pan et al., 2014;Pu et al., 2014;Shi et al., 2020;Teranishi et al., 2013;Zhang et al., 2015a;Zhang et al., 2018; and two studies Zhang et al., 2015a) were evaluated as having a low RoB in the corresponding domain, mentioning the appropriate random sequence generation method and allocation concealment, respectively. Each of four Shi et al., 2020;Teranishi et al., 2013;Zhang et al., 2015a) and five studies Furukawa et al., 2017;Shi et al., 2020;Teranishi et al., 2013;Zhang et al., 2015a) were evaluated with low risk of performance or detection bias by appropriately performing blinding of participants, personnel, or outcome assessors. In one study , the number of dropouts in each group was not described in detail, and the lack of outcomes related to BPSD was evaluated as a high risk of attrition and reporting bias. Two studies Motohashi, 2006) were evaluated as having an unclear risk of other bias because there was no information on the homogeneity of baseline clinical characteristics between the two groups ( Figure 2). In one CCT (Xu, 2018), the treatment or control group or outcome measures were not properly specified. In 12 before-after studies Xu et al., 2007;Shinno et al., 2008;Hayashi et al., 2010;Kawanabe et al., 2010;Iwasaki et al., 2012;Nagata et al., 2012;Yang et al., 2012;Sumiyoshi et al., 2013;Ohsawa et al., 2017;Manabe, 2020), study questions, eligibility criteria for the study population, interventions, and outcome measures were clearly stated in most studies. However, blinding of outcome assessors was not reported in all studies, and only two studies Manabe, 2020) provided individual-level data (Supplementary Material S5).

HM as a Monotherapy
Compared to psychotropic drugs, the severity of BPSD symptoms evaluated by BEHAVE-AD was not significantly different between the two groups (MD −0.36; 95% CI: −1.09 to 0.36), regardless of the treatment duration. However, as a result of subgroup analysis according to the severity and type of dementia, in the case of severe dementia and vascular dementia, BEHAVE-AD was significantly reduced in the HM group (MD −2.34, 95% CI: −4.15 to −0.53). The Hamilton depression rating scale (HAMD) was significantly improved in the HM group (MD −2.86; 95% CI: −3.85 to −1.87), but there was no significant difference between the two groups in TER calculated based on BPSD symptoms (RR 1.05, 95% CI: 0.95-1.16). To evaluate the severity of BPSD, NPI , NPI-nursing home , and the Cornell scale for depression in dementia (CSDD)  were evaluated in one each, and there was no significant difference between the two groups in all studies. In addition, AE occurred significantly less in the HM group (RR 0.40, 95% CI: 0.25-0.64), regardless of the treatment duration (Table 2). However, when the sensitivity analysis was performed on studies with low risk of performance and detection bias, there was no significant difference in the meta-analysis of AE between the two groups (RR 0.83; 95% CI: 0.27-2.61). Overall, severe AEs related to HM were rarely reported, but hypokalemia in HMs containing licorice, such as Yokukansan, required attention (Supplementary Material S6). When comparing HM and placebo, in one study , there was no significant difference between the two groups in the NPI-questionnaire. However, in another study  that measured the severity of BPSD symptoms, the HM group showed significant improvement in hallucinations, activity disturbances, aggressiveness, and anxieties and phobias of BEHAVE-AD, and in delusions, hallucinations, agitation, aberrant motor behavior, and sleep disturbance of NPI (p < 0.05), although there were no significant differences between the two groups in other domains. In a study comparing the group administered HM and donepezil placebo with the group administered with HM placebo and donepezil placebo , it was reported that the change in the NPI total score after treatment was significantly greater in the group administered with HM and donepezil placebo (p < 0.05). Finally, a study that evaluated TER based on psychiatric symptoms reported that TER of the HM group was significantly higher than that of the placebo group . There was no significant difference in the incidence of AE between the two groups (RR 2.67; 95% CI: 0.93-7.65) ( Table 2). Even when the sensitivity analysis was performed only with studies with a low risk of performance and detection bias, this result was not affected.
When comparing HM and anti-dementia drugs, one study (Zhou, 2018) showed that the Pittsburgh Sleep Quality Index (PSQI) was significantly reduced in the HM group (p < 0.05). In one study comparing HM and no treatment , the severity of BPSD symptoms measured by NPI after 4 weeks of treatment was significantly improved in the HM group Frontiers in Pharmacology | www.frontiersin.org July 2021 | Volume 12 | Article 713287 (p < 0.05), but the Barthel index or IADL was not significantly different between the two groups. In one study comparing HM and hydergine , there was no significant difference in the HAMD score after treatment.

HM as an Adjunctive Therapy
When HM was additionally used for psychotropic drugs, the BEHAVE-AD (MD −3.48, 95% CI: −3.96 to −2.99) and TER calculated based on BPSD symptoms (RR 1.16, 95% CI: 1.05-1.29) significantly improved, compared with psychotropic drugs alone. In addition, the incidence of AE was also significantly lower in the HM group (RR 0.71; 95% CI: 0.50-0.99; Table 2). When HM was additionally used, one study Gu et al. (2018) reported that the BPRS score improved significantly (p < 0.05), and another study Zhang et al. (2018) reported that QoL and the severity of BPSD evaluated by CSDD, Hamilton Anxiety Rating Scale, and HAMD significantly improved, and the frequency of side effects evaluated by treatment emergent symptom scale was significantly reduced (p < 0.05, all). When HM was additionally used as an anti-dementia drug, the severity of BPSD symptoms measured by BEHAVE-AD (MD −2.81, 95% CI: −3.17 to −2.45), NPI (MD −3.23, 95% CI: −4.06 to −2.40), and HAMD (MD −4.92, 95% CI: −5.48 to −4.37) significantly improved, compared with anti-dementia drugs alone. The TER calculated based on BPSD symptoms (RR 1.29, 95% CI: 1.13-1.47) and Barthel index (MD 3.42, 95% CI: 2.67-4.16) also significantly improved in the HM group. The severity of BPSD symptoms was evaluated using PSQI(30) (p < 0.05) and NPI-questionnaire  (p < 0.01) in each study, and both showed significantly improved results in the HM group. In one study , the degree of depression and burden of caregivers after treatment were reported through a self-rating depression scale and a Zarit burden interview, respectively, but there was no significant difference between the two groups. The frequency of AE was also significantly lower in the HM group (RR 0.50, 95% CI: 0.28-0.88), although there was no consistent result according to the subgroups ( Table 2,Supplementary Material S6).
Furthermore, when HM was additionally used, there was no significant difference in the CSDD score compared with the oxiracetam alone . However, HAMD significantly improved (p < 0.05) when HM was additionally used to clopidogrel in one study .

Publication Bias
Since there was no meta-analysis that included more than ten studies, we could not assess the publication bias using a funnel plot.
FIGURE 2 | Risk of bias for all included studies. Low, unclear, and high risk, respectively, are represented with the following symbols: "+", "?", and "-". However, considering the design of the study, these studies were not included as evidence for analyzing the effectiveness or efficacy of HM for BPSD but reviewed in terms of the current status of research in this field. More information can be found in Supplementary Material S3, S6.

Summary of Evidence
In this systematic review, the most comprehensive review and meta-analysis to date was conducted to analyze the effectiveness (or efficacy), safety, and research status of HM for BPSD. According to the meta-analysis, HM did not show statistically significant differences from psychotropic drugs in the effectiveness of BPSD evaluated as BEHAVE-AD or TER or in the subgroup analysis of dementia severity, dementia type, or treatment duration. However, a few studies reported that HM showed statistically significant improvement in patients with Alzheimer's disease compared to psychotropic drugs in HAMD. In addition, HM appeared to be safer compared to psychotropic drugs in terms of the incidence of AEs. Comparisons between HM and placebo in four studies did not have homogeneous outcomes in the meta-analysis. Of the four studies, three showed significant differences between the HM and placebo groups in improving BPSD symptoms but not in the one remaining study. The incidence of AE was not significantly different between the groups, and these results did not change according to the subgroup analysis on dementia severity, dementia type, or treatment duration. When HM was compared to anti-dementia drugs or no treatment, there were statistically significant benefits in improving sleep quality assessed by PSQI or BPSD symptoms assessed by NPI in a study, respectively. When HM was used as an adjunctive therapy, it showed the most consistent benefit. When HM was used in combination with psychotropic or anti-dementia drugs, there were statistically significant benefits in BEHAVE-AD, NPI, HAMD, TER based on BPSD symptoms, incidence of AEs, and Barthel index, compared to monotherapy with psychotropic or antidementia drugs. The methodological quality of the RCTs included in this systematic review was not optimal overall. In particular, allocation concealment and blinding domains were evaluated as unclear in most studies.

Clinical Implications
HM is an EATM modality that has long been used in Asian countries for health improvement and disease treatment. The results of this systematic review and meta-analysis provided limited evidence that HM may be associated with additional benefits in BPSD treatment, particularly when used as an adjunct to conventional medications, including psychotropic and antidementia drugs. Although the clinical evidence supporting the effectiveness (or efficacy) and safety of HM for BSPD is insufficient, this topic has clinical relevance considering that many elderly patients already use prescription drugs and HM in combination (de Souza Silva et al., 2014;Agbabiaka et al., 2017). Additionally, the use of HMs is not limited to EATM. Herbs used in other traditions, such as Ginkgo biloba, Withania somnifera, Panax ginseng, and Curcuma longa, and some phytochemicals have also shown promising results in the treatment of dementia (Alzobaidi et al., 2021). Currently, studies are focused on drug delivery, such as improving the potential anti-dementia effect of HM by using a targeted nanocarrier system (Moradi et al., 2020;Singh et al., 2021). Any current or future studies that explore the therapeutic potential of HM for dementia should be encouraged, as these can provide valuable insight in the field.
Although not within the scope of this review, HM is also used to delay cognitive decline, a core symptom of dementia, and its mechanisms are being studied to be related to mechanisms such as anti-inflammatory, antioxidative, and antiapoptotic activity (Tewari et al., 2018). However, the underlying mechanism of HM for the core and associated symptoms of dementia is yet unclear, and it may be related to some challenges including non-uniform chemical composition, non-standardized ratio of herb ingredients, and its multi-component and multi-target mechanism (Zhou et al., 2019). Moreover, there are safety issues associated with HM, such as lack of safety monitoring and potential interactions with conventional pharmaceuticals (Ekor, 2014). Fortunately, for some standardized HMs, such as Yokukansan, underlying therapeutic mechanisms for dementia (Takeyoshi et al., 2016), potential interactions with conventional medications (Soraoka et al., 2016) and safety issues have been documented . Similarly, a database of some potential herb-drug interactions relevant to the management of cognitive impairment has been recently developed. It provides the pharmacological interactions of 170 bio-actives with 10 commonly-used drugs (Auxtero et al., 2021). However, other heterogeneous HMs are also used in clinical practice and their safety profiles need further clarification. Spontaneous reporting systems and active pharmacovigilance for the use of HMs should be encouraged with stringent oversight by a national-level regulatory body to ensure patient safety and satisfaction (Zhang et al., 2015b).
In summary, in order for HM to be seamlessly integrated into the conventional medical system in the management of dementia, particularly to treat BPSD, the use of standardized HM with wellmanaged quality should be encouraged, and the underlying mechanisms and possible interactions with conventional pharmaceuticals should be further investigated. It should also be used by health care professionals in clinics or hospital-based settings for meticulous effectiveness and safety monitoring.

Strengths and Limitations
This systematic review comprehensively reviewed the studies published to date on this issue and summarized the clinical evidence supporting the effectiveness and safety of HM in the management of BPSD. Considering the limitations of psychotropic drugs in the management of BPSD, particularly in the elderly, and many elderly patients already use HM, this topic has great clinical relevance. Our study highlights the limited evidence of HM for BPSD management and discusses the future directions necessary for HM to be integrated into conventional dementia care systems as an adjuvant therapy.
The findings of this systematic review should be interpreted with careful consideration of some limitations (Wimo et al., 2017). Although this review collected clinical evidence of HM for BPSD as the most comprehensive, the number of studies included in each meta-analysis was less than six because the studies included were heterogeneous. In particular, some standardized HMs, such as Yokukansan, existed, but most studies used HMs of heterogeneous composition. Although EATM is a medicine system that emphasizes holistic and individualized approaches (Fung and Linn, 2015), the use of standardized HM is emphasized in order to establish an effective HM use strategy for BPSD treatment and to confirm its expected effectiveness and safety. In addition, in order to accumulate robust clinical evidence of HM for BPSD management, the design of dementia severity, dementia type, BPSD severity, and treatment duration of subjects should be homogeneous (Ohno et al., 2019). In the protocol of this review (Kwon et al., 2021), the subgroup analysis was planned according to the severity of baseline BPSD of participants, but this subgroup analysis was not possible because of the heterogeneity of the indicators. However, since psychotropic drugs, such as antipsychotics, are generally more recommended for severe BPSD compared to safe non-pharmacological therapy (Masopust et al., 2018), finding other safe alternatives, including HM, in patients with severe BPSD, is necessary (Cerejeira et al., 2012). Since only a few studies were included in each meta-analysis, evaluation of publication bias through funnel plots was not possible. However, most studies included in the analysis were conducted and reported in China, which suggests a potential publication bias in the results. Although HM is mainly used in Asian countries as an EATM modality, rigorous clinical trials conducted in Taiwan and Korea, in addition to China and Japan, are encouraged to address this issue (van der Linde et al., 2016). None of the included studies reported the results of economic value related to HM for BPSD. Dementia causes a huge socioeconomic burden worldwide, and BPSD is a major contributing factor (Cerejeira et al., 2012). Therefore, effective alternatives to BPSD in the future require cost-effectiveness, effectiveness, and safety. Considering that the cost-effectiveness of HM is being studied for other clinical topics, such as chronic low back pain (Sung et al., 2019), further clinical research on HM for BPSD should encompass economic evaluation.

CONCLUSION
According to the findings of this review, HM may be associated with additional benefits in BPSD treatment, particularly when used as an adjunct to conventional medications, including psychotropic and anti-dementia drugs. However, considering the methodological quality of the included RCTs, this clinical evidence is not robust. In addition, the heterogeneity of HMs used in each study encourages the use of standardized HMs in the future. Nevertheless, dementia is a global health concern, and considering the limitations of conventional psychotropic drugs for BPSD, a major cause of the disease burden, HM appears to be a promising complementary therapy that warrants further research.

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.