Current Research Status and Implication for Further Study of Real-World Data on East Asian Traditional Medicine for Heart Failure: A Scoping Review

This study used real-world data (RWD) to explore the long-term effects of East Asian traditional medicine (EATM) on heart failure (HF). A comprehensive search was conducted across five databases to identify relevant studies, which were then reviewed using the Arksey and O’Malley scoping review framework. The analysis focused on a descriptive examination of the long-term outcomes associated with EATM intervention. Methodologically, the study explored various aspects, including study subjects, interventions, applied clinical outcomes, and statistical methods. Out of 258 studies, 12 were selected. Eight studies involved patients with HF, while the others used HF as an outcome. Datasets from the National Health Insurance Research Database were used in Taiwan, while electronic medical record data were used in China and Japan. EATM interventions have been found to be associated with lower mortality and readmission rates. One study indicated that an increased dose of Fuzi, a botanical drug, or prompt use of Fuzi after diagnosis led to a decreased mortality hazard ratio. In two studies examining readmission rates, a significant increase was observed in the non-exposed group, with odds ratios of 1.28 and 1.18. Additionally, in patients with breast cancer, the subdistribution hazard ratio for the occurrence of doxorubicin-induced HF was reduced to 0.69. Although cohort studies with survival analysis were common, methodological flaws, such as issues with statistical methods and HF diagnosis, were identified. Despite these challenges, the study observed an association between EATM and improved clinical outcomes in patients with HF, emphasizing the potential of RWD studies to complement randomized controlled trials, especially for longer-term follow-ups. These results provide foundational data for future RWD research.


Introduction
Heart failure (HF) is a complex clinical syndrome resulting from structural or functional abnormalities of the heart that damage the ability of the ventricles to fill and drain blood [1].

Step 1: Checking Study Questions
The research team, comprising clinical research experts, Korean medicine specialists in circulatory diseases, and Korean medicine clinical researchers, conducted a literature search and specified the scope of the subject.The search addressed key inquiries about HF based on RWD.It provided valuable insights by examining the study design, identifying frequently used clinical outcomes, and evaluating the long-term effectiveness and safety of EATM interventions in HF.Additionally, the study scrutinized botanical drug regimens for HF treatment.It aimed to identify specific botanicals, distinguished by their scientific names, used to treat individuals with HF.Overall, these findings enhance our understanding of HF treatment in real-world clinical scenarios.

Step 2: Checking Relevant Studies
This review focused on peer-reviewed studies using RWD-based EATM interventions in patients with HF.The initial literature search was conducted in May 2022, and the final search, using the same strategy, was conducted in September 2023.The following databases were searched: MEDLINE via PubMed, EMBASE via Elsevier, Cochrane Central Register of Controlled Trials, Cumulative Index to Nursing and Allied Health Literature, and Allied and Complementary Medicine.The search strategies were discussed by clinical research experts and Korean medicine doctors, employing a combination of terms, including HF, RWD ("cohort" or "case-control" or "cross-sectional"), and EATM ("Chinese medicine" or "Kampo medicine" or "Korean medicine").The search strategy incorporated various medical subject heading terms and synonyms, and detailed search formulas are provided in Supplementary Table S1.

Step 3: Study Selection
Three authors (JP, SB, and DS) independently performed the study selection.EndNote 20 was used to remove duplicates from the searched publications that included titles and abstracts.For studies identified as potentially relevant, the full text was checked to determine inclusion.All articles were extracted in Excel and categorized as either included or excluded based on predefined criteria, with reviewers providing reasons for the exclusion.Discrepancies were resolved through discussions with other researchers.The inclusion criteria were as follows: (1) longitudinal studies using RWD, such as prospective cohort, retrospective cohort, and case-control studies; (2) studies using EATM intervention; and (3) studies focused on patients with HF or the occurrence of HF.The exclusion criteria were as follows: (1) studies not following a longitudinal design but instead using a snapshot (cross-sectional) research approach, such as assessing medical usage status in a specific year; (2) survey research and biomarker development studies; (3) interventions conducted by non-physicians; and (4) non-RWD research, including randomized controlled trials (RCTs).

Step 4: Charting Data
The pilot data extraction sheet was completed through discussions with the research team.Following several pilot tests, the extracted items included general information about the study, data source and type of RWD study, HF diagnostic criteria, statistical methods for analyzing clinical outcomes, risk factors associated with HF prognosis, co-medications, HF-related clinical presentations, and detailed information about botanical drugs.Data extraction was independently performed by three reviewers (JP, SB, and DS), who crosschecked the data from all studies.Any discrepancies among the reviewers were resolved through discussions with other researchers (JL).
in Taiwan [23, [29][30][31][32], and one in Japan [25].The National Health Insurance Research Database was used in Taiwan, whereas in the other two countries, EMR data from hospitals was used (Table 1).

Diagnosis and Patient Selection Criteria, and Statistical Analysis Methods
Regarding methodological and statistical analyses, five studies conducted survival analysis using propensity score matching [23, 24,29,31,32]; some included a dose-response analysis to explore causality [23].Independent t-tests or chi-square tests (which did not consider the time-to-event analysis) were used to compare the mean or ratio of clinical indicators [27][28][29]33,34].Logistic regression analysis was used to identify the factors influencing the occurrence of cardiac events [24,25,33,34].Regression analysis was also conducted to predict the factors influencing the use of traditional Chinese medicine [27,30,[32][33][34].One study performed multivariate logistic regression analyses and presented nomograms, receiver operating characteristic curves, and clinical decision curve analyses to construct a model for predicting survival [26].Table 2 presents the covariates used in each study.

Clinical Outcomes
The clinical outcomes used were classified into mortality, cardiovascular events (including HF occurrence), hospitalization, cardiac function, medical cost, safety, and dose-response relation (Table 3).
Although its effect on mortality remains debatable, EATM intervention in patients with HF appears to reduce mortality.In the study by Tai et al., the hazard ratio was 0.99 (95% confidence interval, 0.76−1.27),indicating no significant reduction in mortality [23].However, other studies reported a lower 5-year mortality rate in the TCM group than in the control group, with a hazard ratio of 0.24 [29] or odds ratio of 0.19 [26].In patients with myocardial infarction (MI), the odds ratio of cardiogenic death in the control group

Diagnosis and Patient Selection Criteria, and Statistical Analysis Methods
Regarding methodological and statistical analyses, five studies conducted survival analysis using propensity score matching [23, 24,29,31,32]; some included a dose-response analysis to explore causality [23].Independent t-tests or chi-square tests (which did not consider the timeto-event analysis) were used to compare the mean or ratio of clinical indicators [27][28][29]33,34].Logistic regression analysis was used to identify the factors influencing the occurrence of cardiac events [24,25,33,34].Regression analysis was also conducted to predict the factors influencing the use of traditional Chinese medicine [27,30,[32][33][34].One study performed multivariate logistic regression analyses and presented nomograms, receiver operating characteristic curves, and clinical decision curve analyses to construct a model for predicting survival [26].Table 2 presents the covariates used in each study.

Clinical Outcomes
The clinical outcomes used were classified into mortality, cardiovascular events (including HF occurrence), hospitalization, cardiac function, medical cost, safety, and doseresponse relation (Table 3).
Although its effect on mortality remains debatable, EATM intervention in patients with HF appears to reduce mortality.In the study by Tai et al., the hazard ratio was 0.99 (95% confidence interval, 0.76-1.27),indicating no significant reduction in mortality [23].However, other studies reported a lower 5-year mortality rate in the TCM group than in the control group, with a hazard ratio of 0.24 [29] or odds ratio of 0.19 [26].In patients with myocardial infarction (MI), the odds ratio of cardiogenic death in the control group ranged from 1.34 to 2.64 [33,34].Regarding hospital visits, the overall TCM use was associated with a reduction in readmission rates.Guan et al. showed that the readmission rate due to HF was significantly lower in a TCM user group [24].Another study comparing TCM users of Fuzi (a medicinal botanical drug) versus TCM users without Fuzi found no significant differences in admission rates to intensive care units and hospitalization rates due to HF, MI, and stroke [23].However, the study noted a trend indicating a lower mortality rate associated with the prompt use of Fuzi after the diagnosis of HF and an increase in Fuzi dosage [23].Among individuals undergoing doxorubicin chemotherapy for breast cancer, TCM use was associated with a lower incidence of CHF [31].Another study found a lower incidence of HF when botanical drugs were used for over 180 days in patients with hypertension [32].Among patients with MI, the TCM group had a lower incidence of acute HF [33,34].Regarding cardiac function, the TCM group showed a greater improvement in NYHA classification and LVEF [24].In another study, TCM use had no significant effect on LVDD and cardiac index but was associated with improvement in LVEF, CO, every cardiac output, and 6-min walking test scores [28].Regarding medical costs, expenses for outpatient treatment or hospitalization within one year after HF diagnosis were significantly lower in a TCM user group [29].No significant adverse events were reported in the available literature (Table 4) [23,27,28].

Discussion
In this study, a scoping review method was used to investigate the RWD research status of EATM interventions for HF.Among the 12 selected studies, 8 were conducted in patients with HF [23-30], 2 used HF as the primary outcome [31,32], and 2 used HF as the secondary outcome [33,34].All interventions were based on traditional botanical drugs, including TCM and Kampo.Mortality was the most frequently used outcome variable [23][24][25][26][27]29,31,33,34].Overall, the use of EATM was associated with lower mortality and readmission rates.Furthermore, TCM use in patients with other conditions (such as breast cancer, hypertension, and MI) was associated with a lower incidence of HF.Among the 12 studies, 10 confirmed the use of botanical drugs for treatment, with Panax ginseng C. A. Meyer being the most used.The incidence of adverse events was not higher in the TCM group.
Regarding cost-effectiveness, one year after the occurrence of HF, the TCM group exhibited lower medical costs and shorter hospital stays.However, five years after the onset of HF, the TCM group experienced higher costs, accompanied by an increased frequency of outpatient visits.Additionally, traditional medicine is commonly used concurrently with standard treatments, making it costlier than groups receiving only standard treatments.Therefore, economic evaluations should consider not only direct medical costs but also direct non-medical costs, quality-adjusted life years, and other relevant factors [35,36].The criteria for economic evaluation may vary by country, emphasizing the need for additional country-specific research.
Regarding the herbs prescribed for treatment, Panax ginseng C. A. Meyer (Renshen), Astragalus membranaceus Bunge (Huangqi), and Salvia miltiorrhiza Bunge (Danshen) were predominantly used.Panax ginseng C. A. Meyer (Renshen), known for its ability to tonify qi and yang, has diverse pharmacological effects, regulates blood glucose and cholesterol levels, and lowers blood pressure, thereby influencing the risk factors associated with cardiovascular diseases [37,38].Similar to Panax ginseng C. A. Meyer, Astragalus membranaceus Bunge (Huangqi), which tonifies qi, is reported to have anti-inflammatory, antioxidant, vascular protective, and diuretic effects [39][40][41].In China, the combination of Panax ginseng C. A. Meyer (Renshen) and Astragalus membranaceus Bunge (Huangqi) is commonly used for CHF because of the synergistic effects of the combination of the two herbs [40].Salvia miltiorrhiza Bunge (Danshen), known for its blood-activating and stasis-resolving characteristics, is reported to exert cardiac protective effects attributed to its antioxidative, anti-inflammatory, and antiapoptotic properties [42].Guo et al. conducted data mining to report the prescription rules for herbs commonly used for preserving ejection fraction in HF [43].All three of these herbs were included, highlighting the primary functions of herbal medicine as qi-replenishing, yang-warming, blood-activating, and diuresis-inducing.This suggests that the appropriateness of considering these three herbs for HF aligns with previous findings and our research findings.Despite the positive therapeutic effects of botanical drugs on cardiovascular health, six studies in the analysis included interventions with concurrent standard treatments.Hence, it is crucial to consider the potential interaction between botanical drugs and standard treatments, rather than attributing effects solely to botanical drugs.
In a previous RCT, the traditional botanical drug Qili Qiangxin capsule reduced N-terminal-pro-brain natriuretic peptide levels, although all-cause mortality was not explored owing to the relatively short duration of the RCT [44].In contrast, our study employed a retrospective cohort design with a long-term follow-up, allowing for the analysis of all-cause mortality [23, 25,26,29,31].We observed differences in the treatment effects between RWDs and RCTs.Regarding readmission, an RCT by Li et al. on the effect of Qili Qiangxin capsules on HF reported a readmission rate of 3.28% in the treatment group (8 of 244 patients) versus 6.48% in the control group (16 of 247 patients) [44].However, in an RWD study by Guan et al. who investigated the effect of Shenmai injection on CHF, the readmission rate was 32.37% in the treatment group (146 out of 451 patients) versus 38.93% in the control group (232 of 596 patients) [24].Although participant baseline characteristics, study settings, and interventions differed slightly, the effect size of TCM treatment varied considerably between RWD and RCT design studies, despite measuring the same outcome variable.Similar differences between RWD and RCT results have been observed in other studies [45][46][47].RWD studies analyze larger sample sizes over longer periods, catering to a more diverse population, and can yield different effect estimates even when the outcome variable is the same as in existing RCTs.Generalizing and predicting which research design will provide a more significant effect on the intervention between RCT and RWD is challenging.However, unlike RCTs, observational studies cannot establish causality and can only demonstrate an association.Therefore, both research designs should be used complementarily.Some problems have not been described in detail regarding study subjects, and previous RCTs have been criticized for not providing detailed information on HF diagnoses [48].A review related to the existing RCTs treating HF has highlighted the ambiguity in distinguishing between HF with reduced (LVEF ≤ 40%) and preserved (LVEF ≥ 50%) ejection fractions [19], which have different treatment strategies and prognoses [49].Furthermore, the evaluation of cardiac function using the NYHA class or ejection fraction has not been widely used.These indices can significantly influence prognosis and treatment response [49].Consequently, follow-up studies should explicitly present patient characteristics by providing a detailed diagnosis of individuals with HF and an assessment of cardiac function.
The selection of the exposed and non-exposed groups lacked the appropriate utilization of physical, radiological, and biochemical indices, leading to limitations in assessing comparability between the groups.This limitation might be inherent in claims data, and there could be systematic differences between users and non-users of EATM.Although statistical adjustments were implemented in the analysis stage, such as regression, efforts to minimize bias in subject group selection are crucial.Another issue pertains to the statistical analysis methods.To analyze time-to-event data, survival analysis, such as Cox proportional hazard regression, should be applied.However, several studies used mean comparison (t-test) or frequency comparison (chi-square test) instead, representing a methodological error [28,29,33,34].Statistical analysis methods suitable for the long-term follow-up of RWD research should be considered.Additionally, study design considerations, such as setting the first diagnosis time as the index date rather than the date of the first administration of the treatment group, should be adopted to avoid immortal bias [50,51].In addition, as in the study by Tai et al. included in this review, dose-response analysis should be actively applied to explore causality [23].
To overcome the limitations observed in existing observational studies on HF, the following recommendations are proposed.First, regarding health insurance data, the inadequacy of baseline characteristic information that can be used as covariates is noted, while EMR data from hospitals lacks long-term clinical outcome data.Therefore, there is a need to integrate these two data sources in future studies.Additionally, synchronizing pattern identification should be considered as it influences the treatment strategy in RWD studies [52].Second, conducting dose-response analyses is essential to strengthen the evidence of association.Tai et al.'s study employed the restricted cubic spline method in evaluating dose or treatment initiation time [23].Whereas previous research categorized participants into quartiles, adopting an analysis method that maximizes the advantages of continuous variable data is crucial.Third, it is emphasized that existing HF clinical trials predominantly used surrogate outcomes such as echocardiography, symptoms, and blood tests.However, outcomes such as cardiogenic death or HF readmission, which could provide insights into long-term prognosis, should be actively incorporated.Lastly, quantitative synthesis faces challenges due to the different outcomes reported in each study.A study on the core outcome set for a clinical trial of TCM for HF is currently underway [53].However, extending this set is crucial when using RWD.Accumulating evidence with common outcomes offers the potential for future quantitative synthesis in observational studies using RWD.
To our knowledge, this study has the advantage of being the first literature review on RWD-based EATM interventions for HF.This study not only explored the effectiveness, but also presented the applied design, regimen, study subjects, and statistical methods.In addition, the outcomes and covariates used are described in detail to provide fundamental data to those who would conduct RWD studies using EATM for HF in the future.However, because of the inherent limitations of the observational study, this research also has fundamental constraints.First, the study's findings only suggest an association and do not confirm the efficacy of EATM.Whether the effect is from the botanical drug alone or synergistic with standard treatment is unclear; thus, it requires supplementation through experimental studies.Additionally, the included studies lacked precise dosages or administration details, highlighting the need for more accurate descriptions in further research.Second, analyzing safety or compliance was challenging owing to data limitations.To overcome these limitations, future studies should incorporate health insurance data and hospital EMRs to establish diverse measures for evaluating safety and compliance, including laboratory tests and confirmation of medication adherence.Third, the study's limitations include its applicability beyond East Asia, as it focuses on traditional medicine for HF in East Asian populations.A comprehensive review is needed to determine if the findings apply universally or have geographic specificity.Fourth, the search strategy, encompassing "Chinese medicine", "Kampo medicine", and "Korean medicine", aimed to investigate overall traditional medicine status.However, the extracted studies only broadly assessed the exposure to traditional medicine or included interventions based on herbs such as injections and decoctions, omitting diverse EATM interventions such as acupuncture and moxibustion.This is considered a limitation of RWD studies because, in retrospective studies based on health insurance data or hospital EMRs, the exact treatment procedure cannot be accurately determined.That the treatment was not mentioned in the paper does not imply non-administration [54,55].Thus, a more comprehensive evaluation of various East Asian traditional medicine interventions is needed through registries, prospective cohorts, and other forward-looking research.

Conclusions
This study investigated mortality, readmission rates, cardiac function, and medical expenses as the clinical endpoints of EATM treatment for HF.The review results showed

Figure 1 .
Figure 1.PRISMA flow chart: Selection of studies for scoping review.

Figure 1 .
Figure 1.PRISMA flow chart: Selection of studies for scoping review.

Table 1 .
Characteristics of the included studies with key questions and real-world data sources.

Table 2 .
Diagnosis and patient selection criteria, and statistical analysis methods with considered covariates.

Table 3 .
Types of clinical outcomes used in included studies.

Table 3 .
Cont.The outcome used in each study is denoted by an 'O'.6 MWT, 6-min walking test; A *, dosage of Fuzi, and their association with mortality and cardiovascular events; AHF, acute heart failure; AMI, acute myocardial infarction; B *, commonly used TCM and a predictive factor for the utilization of TCM; CHF, chronic heart failure; DM, diabetes mellitus; HF, heart failure; HTN, hypertension; ICU, intensive care unit; LVEF, left ventricular ejection fraction; LVDD, left ventricular diastolic dysfunction; MI, myocardial infarction; NYHA, New York Heart Association; NT-proBNP, N-terminal pro-brain natriuretic peptide.

Table 4 .
Effectiveness of East Asian traditional medicine intervention in included studies.

Table 5 .
Components of prescribed botanical drugs (herbal medicines) in the included studies.
The prescribed herb in each study is denoted by an 'O'.A * Liriope Tuber.

Table 6 .
Botanical drugs (herbal medicines) used in the included studies.