A total of 5 English databases (the MEDLINE via PubMed, the Cochrane Library, EMBASE, the Web of Science and Ovid database) and 4 Chinese databases [China Science and Technology Journal Database (VIP), Chinese Biomedical Literature Database (CBM), Wan-fang Database, China National Knowledge Infrastructure (CNKI)] were searched for identifying studies from March 1990 to March 2022.

①Repeated publications; ②case report; ③pure theoretical research; ④The data in the literature were wrong or incomplete.

The MeSH terms of PICOS were combined to search in [Title/Abstract] by developing our search strategies sequentially. A combination of P+I, P+I+C, P+I+C+O, and P+I+C+O+S was used to search for studies. If the number of searched studies were small, we would search as P+I. The artificially screened studies according to the included and excluded criteria and the searching strategy are detailed in Supplementary File 1.

The grading recommendations assessment, development, and evaluation (GRADE) technique (13) were used to assess the evidence’s certainty following the instructions of the website¹. RCT evidence was initially classified as high quality, but it would be downgraded due to the risk of bias, inaccuracy, inconsistency, informality, and publication bias. The level of evidence was classified into four categories: “high,” “moderate,” “low,” and “very low.”

A total of 75 related articles were initially detected. After 25 duplicate studies were eliminated, 50 RCTs were included for further screening. Then 39 studies were excluded without matching the inclusion requirements, and 3 non-RCT studies were eliminated after reviewing the article in detail. Finally, 8 studies (14–21) with a total of 614 patients with AHF were incorporated for systematic review and meta-analysis. Kappa-coefficient analysis suggested that the level of agreement among the two reviewers in article selection had a high degree of consistency (Kappa = 0.805, Supplementary Table 1). Figure 1 depicted the literature screening process and results.

All 8 included RCTs were conducted in China between 2012 and 2019, the sample size ranged from 58 to 100, and the treatment duration varied from 7 days to 14 days, except in one study (21) which had no report on duration. All research interventions were Puerarin injection (PI) in combination with conventional western treatment, and the drug delivery methods of PI were intravenous injection in all the studies. In terms of the usage and dose of PI, 3 studies diluted 500 mg of Puerarin with 500 ml 5% glucose (14, 16, 17), 2 studies diluted 500 mg Puerarin with 250 ml 5% glucose (18, 21), and 3 studies diluted 200–400 mg with Puerarin with 500 ml of 5% glucose (15, 19, 20). Only 3 studies (14, 15, 17) reported that the AHF diagnostic criteria was inconsistent with acute heart failure diagnosis and treatment guide (2010 version) published by the Cardiovascular Disease Branch of Chinese Medical Association (22). None of the studies reported follow-up results. The basic characteristics of included RCTs are detailed in Table 1.

One trial (16) was rated as low risk for using random number tables to generate sequences, while the other studies (14, 15, 17–21) provided no details about the method of random sequences generation. All the included studies published complete data, and no selective outcomes were reported, so the risk of bias was considered “low.” Beyond that, no studies mentioned the information about concealing of allocation, blinding of researchers, participants, and outcome evaluators, resulting in the risk of bias regarding performance, and detection was considered “unclear.” The risk of other bias was considered “low,” since no other obvious bias was observed in all RCTs. Table 2 shows the results of the risk of bias of the included RCTs.

Five studies (14–16, 20, 21) involving 388 patients reported adverse events. As detailed in Table 3, one study (14) reported no adverse reactions in both groups and one study (20) reported the total number of adverse effects without classification. Three studies (15, 16, 21) reported a detailed number of each kind of adverse reactions in both groups. In all, it reported a total of 11.9% (23/194) adverse reactions in the PI group and 9.8% (19/194) adverse reactions in the control group. All of the adverse reactions were modest, and no significant difference in the incident rate of adverse events was observed in both groups (RR = 1.16; 95% CI, 0.66–2.05; p = 0.061, Figure 6), indicating that adjunctive use of PI was safe as a conventional medical treatment.

We assessed publication bias on the results of total effective rate, LVEF, and adverse effect, as other results had less than three studies included. We detected that there is no publication bias on the results of total effective rate, LVEF, and adverse effect (Figure 7). But because of lacking access to the information on the clinical trial registry or study protocol, it could not rule out the potential of selectively reporting existing results. The published bias result of other results are provided in Supplementary File 2.

The certainty of evidence on meta outcomes was assessed by the grading recommendations assessment, development, and evaluation (GRADE) methods, and it showed that evidentiary quality of meta results varied from “very low” to “moderate.” The rationale for the downgrade was mainly due to small sample sizes and unclear risk of bias in the selected studies, as shown in Table 4.

Our study also suggested that PI treatment in conjunction with conventional medical treatment was superior in increasing the total effective rate, improving heart function of the valve of LVEF, LVEDD, and IVRT, and improving the left ventricular diastolic function of the valve of Peak A, Peak E and stroke volume, in terms of comparing with conventional medical treatment alone. In addition, subgroup analysis was performed on the primary outcomes of total effective rate and LVEF. Interestingly, the results showed that adjunctive use of PI could both improve the LVEF value and increase the total effective rate regardless of the dose (200–400 mg/day or 500 mg/day). As previously reported, PI showed a satisfied clinical efficacy in the treatment of cardiovascular diseases that PI was more effective than using conventional western medical alone in the treatment of unstable angina pectoris (9). It is also more effective and relatively safe in the clinic for treating acute ischemic stroke and diabetic peripheral neuropathy (25, 26).

When we further explored the association between PI and favorable results in patients with AHF, it was proposed that PI had the effect of alleviating impaired heart function and inhibiting the levels of myocardial injury and inflammatory markers (27), as it was found that inflammation and heart function impaired in AHF resulted in neutral effects or worsening of clinical outcomes (28, 29). In addition, patients with AHF presented with similar congestion symptoms, which could lead to HF decompensation, which occurred owing to both fluid accumulation and redistribution, and further progress in the deterioration of AHF, thus decongestive therapy and diuretic drugs were recommended for AHF (1). PI was found that it could expand the coronary artery to promote coronary blood flow (10) and improve microcirculation to alleviate congestion symptoms (11), which might be the mechanism that PI could alleviate the AHF symptoms. Furthermore, clinical trials pointed out that a higher heart rate was a strong predictor of 1-year mortality of AHF, and reductions in coronary blood flow and myocardial oxygen consumption may be beneficial for AHF treatment (30, 31). Song et al reported that PI had the effect of decreasing heart rate and reducing myocardial oxygen consumption (32), which may also be the potential mechanism that PI had favorable results in patients with AHF.

Regarding clinical safety, a total of 9.8% (19/194) adverse reactions occurred in the control groups while 11.9% (23/194) in the PI group, including nausea, hypotension, vomiting, headache, and slow heart rate. As 5 (62.5%) studies (14–16, 20, 21) reported the adverse effects and moderate evidence for safety assessment, we preliminary put forward the argument that combination therapy of PI was safe in treating AHF. But since the record for risk of bias assessment of included RCTs was “unclear,” it implied that there is still a need for further eligible and critical clinical trials to validate the safety of PI.

The findings of meta-results were consistent with previously published research (33). To assess the credible clinical evidence of our results, evaluation of bias risk and evidence’s confidence were performed. It showed that all the included studies lack details in selection bias, blinding performance, and blinding outcome assessment (Table 2), which may result in the overstated effect of outcomes and reported bias in selected results. In addition, GRADE evaluation indicated that the confidence of the evidence was graded, which varied from very low to moderate quality for evidences (Table 4), and risk of bias, inconsistency, imprecision, and publication bias were mainly responsible for the downgrading of evidence because of the quality of included RCTs, thus larger RCTs with improved methodological quality in future are expected to further update the results of this systematic review.

The adjunctive efficacy and safety of PI regarding curative effect among patients with AHF were for the first time systematically reviewed and evaluated in this study. At present, AHF treatment still lacks specific and effective medicine, leading to a relatively high recurrence rate, hospitalization rate, and mortality rate. We found that integrated use with PI could improve heart function, increased total effective rate, and was safe as the conventional western medication, which could be chosen by physicians when patients with AHF faced with unexpected treatment effects. The methodology of the present study was designed to a high standard according to the methodological quality of systematic reviews-2 (AMSTAR 2) by identifying relevant literature comprehensively, developing evaluation plans, and strict implementation, which could improve the accuracy and clinical applicability of the results of this study (34).

Although the results were encouraging, restrictions were unavoidably present in this study. Due to the small number and low to moderate quality of included studies, strictly designed trials according to the Consolidated Standards of Reporting Trials (CONSORT) statement also need to be further performed to verify the efficacy of PI as an adjunctive therapy for AHF. Duration included 7 days and 14 days, and the dose of PI included 500 mg/day and 200–400 mg/day, but we only did perform subgroup analysis of dose on the total effective rate due to small number of studies. Besides, the control group involves different conventional medical treatments, which potentially led to heterogeneity between the studies. Although there was no restriction on language when screening literature, the final included studies were all performed in China, which may lead to potential selection bias in the research. In addition, a few have data available for each outcome, for instance, the number of studies included in the meta-analysis of LVEDD, IVRT, Peak A, Peak E, and SV was 2-3/8 (25–37.5%), which limited the credibility of the above results. Thus, much more caution should be taken about the results until further trials in different populations and high-quality designed studies were performed to strengthen and update the results of the present meta-results.