This study was conducted and reported in accordance with the Preferred Reporting Project (PRISMA) guidelines for systematic reviews and meta-analyses (Page et al., 2021). The study protocol was registered (CRD42025641482) in the International prospective register of systematic reviews (PROSPERO).

To comprehensively retrieve relevant information, databases such as CNKI, Wanfang Data, VIP, CBM, PubMed, Web of Science, and Cochrane Library were systematically searched. The objective was to identify published randomized controlled trials (RCTs) on the application of QYYYG in treating hypertensive vascular damage. The retrieval time range extended from the inception of each database to 15 April 2025. We combine the large language model to carry out AI-assisted retrieval. Adopt the retrieval keywords obtained by individual retrieval, and ask the DeepSeek-V3 (https://www.deepseek.com/) according to different databases. It is required to synthesize the recall rate and precision rate, generate the specific logical relationship and retrieval formula, and use the recommended retrieval formula for database retrieval. Finally, the retrieved documents are imported into Zotero document management software for management and analysis. For more details on the retrieval strategy, please refer to the Supplementary Material.

Studies with duplicate publication or duplicate research data; 2) Studies with incomplete research data and lacking indicators related to vascular damage; 3) Non-randomized controlled trial studies such as reviews, clinical experiences, and animal experiments; 4) Studies whose subjects include patients with grade 3 hypertension, secondary hypertension, or those with severe target organ functional damage, or those complicated with severe diseases of other systems.

The Cochrane Risk of Bias Assessment Tool was employed to assess the quality of the studies across seven dimensions: random sequence generation, allocation concealment, blinding of participants and research staff, blinding of outcome assessment, completeness of outcome data, selective reporting, and other potential sources of bias. Two researchers independently and blindly carried out the evaluation and cross-checked the results with each other. In the event of a disagreement, the matter will be referred to a third researcher for discussion and resolution. Subsequently, the final bias proportion diagram and quality assessment diagram will be generated. In addition, GRADEpro is used to evaluate the evidence quality of different indicators.

In accordance with the pre-determined retrieval strategy, two researchers independently and in a blinded manner utilized Zotero software to construct a relevant database for the purposes of studies screening and data extraction. Firstly, through browsing the titles and abstracts, exclude the studies that are evidently irrelevant to QYYYG or are duplicates. Secondly, read the full text meticulously and decide on the included studies based on the inclusion and exclusion criteria. Thirdly, after completing the above steps, cross check the inclusion results and reasons provided by the two researchers, and use Kappa statistic to evaluate consistency. In the event of any disagreement, the issue will be referred to a third researcher for discussion and resolution. Finally, Excel spreadsheets were used for data extraction. The data extraction items included: study title, first author’s name, publication year, sample size, mean age, gender distribution, intervention strategies, treatment duration, outcome measures, among others.

RevMan 5.3 and TSA 0.9Beta software were used to perform Meta-analysis and trial sequential analysis on the collected research data. The Q-test and I²-test were employed to assess the heterogeneity. If the P-value is greater than 0.1 and the I² value is less than 50%, it suggests that the heterogeneity is low, and the fixed-effect model should be applied for the analysis. Conversely, if the heterogeneity is high, the random-effect model will be utilized for the analysis. Subgroup analysis will be conducted based on the overall characteristics of the study to identify the sources of heterogeneity. Sensitivity analysis will be conducted using STATA 12.0 software. Meanwhile, Begg’s test and Egger’s test will be applied to check for the presence of publication bias in this study. Finally, the TSA 0.9Beta software will be used to conduct a further analysis of the studies with low heterogeneity, thereby enhancing the credibility of the results.

In accordance with the pre-formulated retrieval strategy, a total of 164 relevant studies were initially retrieved. Among them, 82 articles were identified as duplicates and thus excluded. Two researchers achieved excellent consistency in screening literature titles and abstracts (Kappa = 0.927) and good consistency in full-text screening (Kappa = 0.837). Discrepancies in literature selection were resolved through group discussions involving a third researcher, and 21 studies (Li, 2022; Dinala Jialiken, 2022; Xu et al., 2021; Tan et al., 2021; Li, 2018; Xia, 2018; Li, 2013; Xu, 2021; Mao, 2020; Yan et al., 2022; Huang, 2012; Wang, 2013; Liu et al., 2019; Chen, 2017; Du, 2017; Zhao, 2014; Ma, 2013; Fang et al., 2022a; Ning, 2017; Ding et al., 2015; Qin, 2020) were ultimately included in the analysis. The publication years of these studies ranged from 2013 to 2022. The combined sample size of all the included studies was 1,419 cases, with 728 cases in the experimental group and 691 cases in the control group. The detailed filtering process is illustrated in Figure 2.

A total of 21 eligible randomized controlled studies were included, all of which were Chinese-language articles. The basic data are shown in Table 1.

The Cochrane Risk of Bias Assessment Tool was used to evaluate from the following seven aspects: 1) Random sequence generation: Five studies (Xu, 2021; Mao, 2020; Yan et al., 2022; Liu et al., 2019; Fang et al., 2022a) indicated that the random number table method was used, and a “low risk of bias” evaluation was made. Three studies (Li, 2022; Dinala Jialiken, 2022; Qin, 2020) indicated that random grouping was carried out using the R statistical software, and a “low risk of bias” evaluation was made. The remaining 13 studies only mentioned the word “random”, and the specific randomization method was not described, so an “unclear risk of bias” evaluation was made; 2) Allocation concealment: None of the 21 studies mentioned allocation concealment, so an “unclear risk of bias” evaluation was made; 3) Blinding of participants and personnel: Although no studies have mentioned it, due to the lack of development of high fidelity traditional Chinese medicine placebos, they have been evaluated as having a “high risk of bias”; 4) Blinding of outcome assessment: None of the studies mentioned it, so all studies were evaluated for “unclear risk of bias”; 5) Incomplete outcome data, 6) Selective reporting: Eight studies (Li, 2022; Dinala Jialiken, 2022; Li, 2018; Xu, 2021; Mao, 2020; Chen, 2017; Fang et al., 2022a; Qin, 2020) had cases of loss to follow-up, but the balance between groups was maintained and the reasons were similar. The outcome data of other studies were complete, and there was no selective reporting of results in all studies, so a “low risk of bias” evaluation was made for all; 7) Other bias: One study (Li, 2013) had a small sample size and an uneven proportion of male and female samples, so a “high risk of bias” evaluation was made. For the other studies, other biases were unknown, and an “unclear risk of bias” evaluation was made. The bias proportion of the included studies is shown in Figure 3A, and the quality evaluation of the included studies is shown in Figure 3B.

Among all the outcome indicators, 18 studies reported the systolic blood pressure. Begg’s test (P = 0.306) and Egger’s test (P = 0.505) were performed to evaluate publication bias. Both returned P-values exceeding 0.05, indicating no publication bias was present and confirming the robustness of the results. This is illustrated in Figure 10.

GRADEpro is used to assess the quality of evidence for different indicators. Among them, the indicators with moderate-quality evidence include: PWV (subgroup), CRP, TNF-α, and safety indicators. The indicators with low-quality evidence are: PWV, systolic blood pressure, and diastolic blood pressure. The detailed GRADE assessment results are presented in Table 2.

The Trial Sequential Analysis (TSA) was carried out using the TSA 0.9Beta software. The type I error probability was set as α = 0.05, and the statistical power was set as 80%. Taking the sample size as the required information size (RIS), and the horizontal line of Z = 1.96 as the traditional boundary value curve, the TSA was conducted on the data related to the efficacy of vascular damage.

The results of the TSA analysis of the ES value of the left common carotid artery are shown in Figure 11A. The cumulative Z value has successively crossed the traditional boundary value, the TSA boundary value curve, and the RIS line in the first three studies. The results of the TSA analysis of peripheral blood CRP are shown in Figure 11B. It also crossed the traditional boundary value and the TSA boundary value curve, and crossed the RIS line in the third study.

The TSA analysis indicates that the included sample size has reached the sample size required for the Meta-analysis. There is no need for more clinical trials, and it has been definitely concluded that the addition of QYYYG has a better efficacy in reducing the PWV-ES value of the common carotid artery and inflammatory factors.

According to the China Hypertension Survey (CHS), it is estimated that the crude prevalence rate of hypertension among adult residents in China from 2012 to 2015 was 27.9% (Wang et al., 2018), which seriously endangers public health. Modern research shows that traditional Chinese medicine has important advantages in preventing and treating hypertension in terms of enhancing efficacy and reducing side effects, multi-target effects, and long-term stability (Liu et al., 2015).

QYYYG are composed of six kinds of medicinal herbs, namely, Herba Bidentis Bipinnatae, Scrophulariae Radix, Alismatis Rhizoma, Polygoni Multiflori Radix, Corni Fructus, Cyathulae Radix (Fan et al., 2025). With three tonifying and three purging herbs, it can achieve the effects of tonifying the healthy qi without leaving the pathogenic factors and purging the pathogenic factors without harming the healthy qi (Zou et al., 2017). The whole formula combines the functions of attacking and tonifying, harmonizes yin and yang, makes the blood vessels unobstructed, dissipates the heat-toxin, and nourishes the burned blood vessels. Current clinical trials have demonstrated that QYYYG exhibits particularly prominent therapeutic effects in elderly populations. Specifically, QYYYG not only reduces the coefficient of blood pressure variability across all time periods in elderly patients (Ni et al., 2016) but also ameliorates early renal damage and renal fibrosis (Fang et al., 2022a). Additionally, it can delay the aging process by enhancing energy metabolism, mitigating DNA oxidative damage, and maintaining the balance of inflammatory factors (Fang et al., 2025). These unique effects in the elderly population are closely associated with age-related physiological changes, including decreased vascular elasticity, increased arterial stiffness, neurohumoral regulatory imbalance, and the accumulation of multiple risk factors.

Clinical pharmacological studies have demonstrated that QYYYG not only effectively controls blood pressure but also ameliorates target organ damage in hypertension. Specifically, QYYYG blocks the activation of the NLRP3 inflammasome by enhancing Nrf2 signaling and inhibiting myocardial inflammation, oxidative stress, and pyroptosis, thereby effectively improving cardiac remodeling in hypertensive patients (Xu et al., 2025). Additionally, QYYYG exerts protective effects against hypertensive renal injury through multiple mechanisms, including regulation of the TRPC6-CaMKKβ-AMPK-mTOR pathway (Ma et al., 2023), modulation of the HIF-1α/PKM2 positive feedback loop (Qian et al., 2021), and epigenetic mechanisms associated with nicotinamide N-methyltransferase (NNMT) expression (Zhang et al., 2020).

High-quality randomized controlled trials (RCTs) are essential for clinical medical research. However, due to the inherent characteristics of traditional Chinese medicine (TCM), insufficient methodological awareness among researchers, considerations of implementation costs, and clinical ethical requirements, current RCTs on TCM for hypertension are deficient in both scale and quality. Notably, deficiencies in blinding are particularly prominent, manifesting in inadequate allocation concealment, incomplete blinding of participants and researchers, and lack of blinding in outcome assessment. To address these limitations, future efforts should focus on: 1) Developing high-fidelity TCM placebos (e.g., bionic technology to simulate appearance and odor); 2) Optimizing outcome measures (increasing objective indicators); 3) Exploring “adaptive designs” (retaining moderate individualization while maintaining standardization). These strategies may progressively improve the methodological rigor of TCM clinical trials.

The Meta-analysis shows that the combination of conventional treatment methods and QYYYG in the treatment of hypertensive vascular damage has obvious value in improving vascular elasticity, reducing inflammatory response, and lowering blood pressure. Moreover, it does not increase the related adverse drug reactions and has good clinical application safety. In this study, the results of the subgroup analysis of PWV in the left and right common carotid arteries indicate that QYYYG have different degrees of influence on the vascular elasticity of hypertensive patients of different age groups. This conclusion suggests that when studying the treatment of hypertensive vascular damage, more attention should be paid to the baseline level of the average age of patients.

The application of trial sequential analysis overcomes the limitations of traditional Meta-analysis, reduces false positive results caused by random errors, and estimates the RIS value to provide a termination criterion for clinical trials (Xia et al., 2013). After analysis and correction, there is conclusive evidence that the addition of QYYYG can reduce the PWV-ES and peripheral blood CRP. There is no need to include more studies to verify this conclusion.

The advantages of this study are as follows. Firstly, by centering on the renowned prescriptions that are extensively applied in clinical practice, a meta-analysis was conducted on controversial issues from the perspective of vascular damage. Secondly, the integrated application of meta-analysis and sequential analysis furnishes the study with the statistical significance threshold and the ineffectiveness threshold of the intervention effect. Thirdly, subgroup analysis revealed the impact of age and other factors on the efficacy of the drug. Fourthly, the heterogeneity of the included indicators is basically low, and the conclusion is credible.

This study still has some limitations. On the one hand, all the retrieved literature are in Chinese, and there are no English articles, which may lead to bias. On the other hand, the included studies carry a risk of bias, which reduces the credibility of the evidence. Notably, due to limitations in the current state of TCM-related research, none of the trials provided information on blinding procedures. Additionally, most existing studies on QYYYG have used endpoint values, while neglecting the change values compared with baseline. Change values can better reflect therapeutic effects and minimize biases caused by baseline heterogeneity. Future researchers should pay attention to these issues.