Literature screening

Initially, the search strategy yielded 2,159 articles. Duplication resulted in the removal of 1,308 articles. The remaining 851 articles were filtered further and excluded according to the eligibility and exclusion criteria. After rereading the full texts, 27 studies remained for quantitative synthesis [27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53]. Figure 1 presents the details of the literature screening process.

Fig. 1

Flowchart of the literature screening process

Study characteristics

There were 25 Chinese articles and two English articles involving 11 interventions. All the articles were conducted in China. Overall, 4,131 patients (2,069 in the experimental control group and 2,062 in the control groups). Nine kinds of CPMs were enrolled: Tongxinluo capsule (TXL), Xiaoshuang granules/enteric capsule (XS), Naoxintong capsule (NXT), Xuesaitong capsule/soft capsule (XST), Jiangzhiling pill (JZL), Pushen capsule (PS), Shexiang baoxin pill (SXBX), Zhibitai (ZBT), and Dengzhan shengmai capsule (DZSM). Table 1 presents the details of ingredients of the included TCPMs. Plant names have been checked with www.theplantlist.org.

Most articles were open− label trials except for two double− blind trials. Both groups were based on CWM, with TCPM addition in the treatment group and PBO addition or blank to the control group, including CWM + TXL vs. CWM + PBO (n = 1), CWM + TXL vs. CWM (n = 6), CWM + XS vs. CWM (n = 2), CWM + NXT vs. CWM (n = 3), CWM + XST vs. CWM (n = 2), CWM + JZL vs. CWM (n = 2), CWM + PS vs. CWM (n = 3), CWM + SXBX vs. CWM (n = 3), CWM + ZBT vs. CWM (n = 3), and CWM + DZSM vs. CWM (n = 2). There were no significant differences in gender and age between the study groups with comparable baselines, and most were middle− aged or elderly. Table 2 presents the details of the included study characteristics.

Table 2 Characteristics of the studies included in this network meta− analysisRisk of bias assessment

All the included trials reported ‘randomly allocating’ participants, generating random sequences using random number tables or computer− based or lottery methods, so they were evaluated as "low risk." Two trials reported allocation concealment, evaluated as "low risk," and the other studies did not mention allocation concealment and were evaluated as "uncertain risk." One trial reported double− blind trials were evaluated as "low risk," and the other studies did not mention blinding was evaluated as "high risk" or "uncertain risk". All trials had complete data, no selective reporting or other risk bias, and were all evaluated as "low risk." Fig. 2A depicts the risk bias assessment results. Figure 2B provides the detailed and specific risk of bias assessment.

Fig. 2

Risk of bias graph of the included RCT A: the risk of bias graph; B: the risk of bias summary

OutcomesPairwise meta-analysis

We conducted eight pairwise meta− analyses comparing the effects of CWM and CWM combined with TCPM on improving the IMT, the carotid maximal plaque area, the carotid atherosclerotic plaque Course score, blood lipids, and CRP (Fig. 3). We assessed the certainty of the evidence for each outcome under the GRADE framework. The quality of the evidence for all of these comparisons was rated as low. The detailed GRADE assessment was presented in Table 3.

Fig. 3

Forest plot of Pairwise meta-analysis. A: IMT; B: carotid maximal plaque area; C: carotid atherosclerotic plaque course score; D: TC; E: TG; F: LDL; G: HDL; H: CRP; IMT carotid artery intimal- medial thickness, TC total cholesterol, TG Triglyceride, LDL low density lipoprotein, HDL high density lipoprotein, CRP C− reactive protein, AER adverse events rate

Compared to CWM, CWM combined with TCPM had a stronger effect in reducing the IMT [26 RCTs; SMD − 1.26 (95% CI − 1.59, − 0.93); p < 0.00001; I2 = 94%; low− quality of evidence] (Fig. 3A), decreasing the carotid maximal plaque area [15 RCTs; SMD − 1.27 (95% CI − 1.71, − 0.82); p < 0.00001; I2 = 94%; low− quality of evidence] (Fig. 3B), lowering the carotid atherosclerotic plaque Course score [8 RCTs; SMD − 0.72 (95% CI − 1.20, − 0.25); p < 0.00001; I2 = 91%; low− quality of evidence] (Fig. 3C), lowering the TC [20 RCTs; SMD − 1.26 (95% CI − 1.66, − 0.86); p < 0.00001; I2 = 95%; low− quality of evidence] (Fig. 3D), lowering the TG [20 RCTs; SMD 1.17 (95% CI − 1.53, − 0.81); p < 0.00001; I2 = 94%; low− quality of evidence] (Fig. 3E), lowering the LDL [20 RCTs; SMD − 1.20 (95% CI − 1.55, − 0.85); p < 0.00001; I2 = 93%; low− quality of evidence] (Fig. 3F), raising the HDL [18 RCTs; SMD 0.80 (95% CI 0.38, 1.22); p < 0.00001; I2 = 95%; low− quality of evidence] (Fig. 3G), and lowering the CRP [10 RCTs; SMD − 0.87 (95% CI − 1.11, − 0.64); p = 0.002; I2 = 66%; low− quality of evidence] (Fig. 3H). Substantial heterogeneity was observed in all results.

We conducted sensitivity analysis comparing pooled results from “ < 6 months of course” and “ ≥ 6 months of course” is illustrated in Fig. 3. There was no significant subgroup difference between the two groups, implying that the difference in length of course did not influence the pooled results on improving the IMT, the carotid maximal plaque area, the carotid atherosclerotic plaque Course score, blood lipids, and CRP.

Network meta− analysisIMT

A total of 27 RCTs referred to the IMT of nine types of TCPMs and 11 types of interventions, including CWM + TXL vs. CWM + PBO (n = 1), CWM + TXL vs. CWM (n = 6), CWM + XS vs. CWM (n = 2), CWM + NXT vs. CWM (n = 3), CWM + XST vs. CWM (n = 2), CWM + JZL vs. CWM (n = 2), CWM + PS vs. CWM (n = 3), CWM + SXBX vs. CWM (n = 3), CWM + ZBT vs. CWM (n = 3), and CWM + DZSM vs. CWM (n = 2) (Table 2). Figure 4A presents the network evidence plot.

Fig. 4

Network diagrams for different outcomes. A: IMT; B: carotid maximal plaque area; C: carotid atherosclerotic plaque course score; D: TC; E: TG; F: LDL; G: HDL; H: CRP; I: AER; CWM conventional western medicine, PBO placebo, TXL Tongxinluo capsule, XS Xiaoshuang granules/enteric capsule, NXT Naoxintong capsule, XST Xuesaitong capsule/soft capsule, JZL Jiangzhiling pill, PS Pushen capsule, SXBX Shexiang baoxin pill, ZBT Zhibitai, DZSM Dengzhan shengmai capsule, IMT carotid artery intimal- medial thickness, TC total cholesterol, TG Triglyceride, LDL low density lipoprotein, HDL high density lipoprotein, CRP C-reactive protein, AER adverse events rate. The width of the lines represents the proportion of the number of trials for each comparison with the total number of trials, and the size of the nodes represents the proportion of the number of randomized patients (sample sizes)

Compared to CWM, except for CWM + NXT [MD − 0.18 (95% CI − 0.39, 0.03)], CWM + XST [MD − 0.18 (95% CI − 0.43, 0.08)], CWM + PS [MD − 0.17 (95% CI: − 0.39, 0.04)] and CWM + DZSM [MD − 0.09 (95% CI − 0.34, 0.17)], other five TCPMs demonstrated a statistically significant effect in reducing the IMT. Accordingly, other interventions had no statistically significant difference. The details were shown in Table 4.

Table 4 Pairwise league table of IMT (lower− left quadrant) and carotid maximal plaque area (upper− right quadrant)

According to the SUCRA probability results (Fig. 5A), CWM + JZL was likely the best intervention for reducing the IMT. Table 5 illustrates the detailed SUCRA and ranking probability. The interventions were ranked as follows: CWM + JZL (70.6%) > CWM + SXBX (70.5%) > CWM + XS (68.6%) > CWM + TXL (57.8%) > CWM + ZBT (56.5%) > CWM + PBO (51.7%) > CWM + XST (48.0%) > CWM + NXT (46.8%) > CWM + PS (46.8%) > CWM + DZSM (27.2%) >  > CWM (5.4%).

Fig. 5

Surface under the cumulative ranking curve (SUCRA) plots for different outcomes. The vertical axis represents cumulative probabilities and the horizontal axis represents rank. A: IMT; B: carotid maximal plaque area; C: carotid atherosclerotic plaque course score; D: TC; E: TG; F: LDL; G: HDL; H: CRP; I: AER; CWM conventional western medicine, PBO placebo; TXL Tongxinluo capsule, XS Xiaoshuang granules/enteric capsule, NXT Naoxintong capsule, XST Xuesaitong capsule/soft capsule, JZL Jiangzhiling pill, PS Pushen capsule, SXBX Shexiang baoxin pill, ZBT Zhibitai, DZSM Dengzhan shengmai capsule, IMT carotid artery intimal-medial thickness, TC total cholesterol, TG Triglyceride, LDL low density lipoprotein, HDL high density lipoprotein, CRP C− reactive protein

Table 5 Pairwise league table of TC (lower− left quadrant) and carotid atherosclerotic plaque course score (upper− right quadrant)Carotid maximal plaque area

A total of 16 RCTs referred to the carotid maximal plaque area of eight types of TCPMs and 10 types of interventions, including CWM + TXL vs. CWM + PBO (n = 1), CWM + TXL vs. CWM (n = 4), CWM + NXT vs. CWM (n = 1), CWM + XST vs. CWM (n = 1), CWM + JZL vs. CWM (n = 2), CWM + PS vs. CWM (n = 2), CWM + SXBX vs. CWM (n = 2), CWM + ZBT vs. CWM (n = 2), and CWM + DZSM vs. CWM (n = 1) (Table 2). Figure 4B presents the network evidence plot. All interventions had no statistically significant difference. The details were shown in Table 4.

According to the SUCRA probability results (Fig. 5B), CWM + SXBX was the most likely the best intervention for reducing the carotid maximal plaque area. Table 8 presents the detailed SUCRA and ranking probability. The ranking of interventions was as follows: CWM + SXBX (83.0%) > CWM + JZL (82.7%) > CWM + XST (53.1%) > CWM + ZBT (52.0%) > CWM + TXL (48.4%) > CWM + NXT (45.3%) > CWM + DZSM (44.7%) > CWM + PS (35.0%) > CWM + PBO (31.1%) > CWM (24.8%).

Carotid atherosclerotic plaque course score

Eight RCTs referred to the carotid atherosclerotic plaque Course score of six types of TCPMs and seven types of interventions, including CWM + TXL vs. CWM (n = 2), CWM + XST vs. CWM (n = 1), CWM + PS vs. CWM (n = 1), CWM + SXBX vs. CWM (n = 1), CWM + ZBT vs. CWM (n = 2), and CWM + DZSM vs. CWM (n = 1). (Table 2). Figure 4C presents the network evidence plot. All interventions had no statistically significant differences. The details were shown in Table 5.

According to the SUCRA probability results (Fig. 5C), CWM + XSBX was the most likely the best intervention for lowering the carotid atherosclerotic plaque Course score. Table 8 depicts the detailed SUCRA and ranking probability. The interventions were ranked as follows: CWM + SXBX (92.5%) > CWM + TXL (85.9%) > CWM + ZBT (61.0%) > CWM + PS (55.0%) > CWM (23.2%) > CWM + XST (22.7%) > CWM + DZSM (9.7%).

TC

A total of 21 RCTs referred to the TC of nine types of TCPMs and 11 types of interventions, including CWM + TXL vs. CWM + PBO (n = 1), CWM + TXL vs. CWM (n = 4), CWM + XS vs. CWM (n = 1), CWM + NXT vs. CWM (n = 1), CWM + XST vs. CWM (n = 1), CWM + JZL vs. CWM (n = 2), CWM + PS vs. CWM (n = 3), CWM + SXBX vs. CWM (n = 3), CWM + ZBT vs. CWM (n = 3), and CWM + DZSM vs. CWM (n = 2). (Table 2). Figure 4D presents the network evidence plot.

CWM + TXL [MD − 0.58 (95% CI − 1.14, − 0.03)], CWM + SXBX [MD − 1.33 (95% CI − 1.95, − 0.70)], and CWM + ZBT [MD − 0.69 (95% CI − 1.32, − 0.07)] had a statistically significant effect on lowering TC compared to CWM. CWM + SXBX [MD − 1.30 (95% CI − 2.57, − 0.03)] had a statistically significant effect on lowering TC compared to CWM + XST. Accordingly, other interventions had no statistically significant differences. The details were shown in Table 5.

According to the SUCRA probability results (Fig. 5D), CWM + XSBX was the most likely the best intervention for lowering TC. Table 8 indicates the detailed SUCRA and ranking probability. The 11 types of interventions were ranked as follows: CWM + SXBX (95.6%) > CWM + XS (73.6%) > CWM + ZBT (63.8%) > CWM + JZL (57.1%) > CWM + TXL (57.0%) > CWM + PS (56.3%) > CWM + PBO (46.9%) > CWM + NXT (37.9%) > CWM + DZSM (24.6%) > CWM + XST (23.2%) > CWM (14.0%).

TG

A total of 21 RCTs referred to the TG of nine types of TCPMs and 11 types of interventions, including CWM + TXL vs. CWM + PBO (n = 1), CWM + TXL vs. CWM (n = 4), CWM + XS vs. CWM (n = 1), CWM + NXT vs. CWM (n = 1), CWM + XST vs. CWM (n = 1), CWM + JZL vs. CWM (n = 2), CWM + PS vs. CWM (n = 3), CWM + SXBX vs. CWM (n = 3), CWM + ZBT vs. CWM (n = 3), and CWM + DZSM vs. CWM (n = 2) (Table 2). Figure 4E presents the network evidence plot.

CWM + NXT [MD − 0.76 (95% CI − 1.35, − 0.17)], CWM + JZL [MD − 0.52 (95% CI − 0.94, − 0.10)] and CWM + SXBX [MD − 0.59 (95% CI − 0.95, − 0.23)] had a statistically significant effect on lowering TG compared to CWM. Consequently, other interventions had no statistically significant differences. The details were shown in Table 6.

Table 6 Pairwise league table of LDL (lower− left quadrant) and TG (upper− right quadrant)

According to the SUCRA probability results (Fig. 5E), CWM + NXT was the most likely the best intervention for lowering the TG. Table 8 presents the detailed SUCRA and ranking probability. The interventions were ranked as follows: CWM + NXT (90.1%) > CWM + SXBX (81.1%) > CWM + JZL (72.7%) > CWM + XS (66.1%) > CWM + PS (52.5%) > CWM + TXL (47.0%) > CWM + PBO (44.7%) > CWM + ZBT (41.6%) > CWM + DZSM (22.2%) > CWM + XST (21.9%) > CWM (10.0%).

LDL

A total of 21 RCTs referred to the LDL of nine types of TCPMs and 11 types of interventions, including CWM + TXL vs. CWM + PBO (n = 1), CWM + TXL vs. CWM (n = 4), CWM + XS vs. CWM (n = 1), CWM + NXT vs. CWM (n = 1), CWM + XST vs. CWM (n = 1), CWM + JZL vs. CWM (n = 2), CWM + PS vs. CWM (n = 3), CWM + SXBX vs. CWM (n = 3), CWM + ZBT vs. CWM (n = 3), and CWM + DZSM vs. CWM (n = 2). (Table 2). Figure 4F presents the network evidence plot.

CWM + TXL [MD − 0.43 (95% CI − 0.84, − 0.02)], CWM + JZL [MD − 0.63 (95% CI − 1.22, − 0.05)], CWM + SXBX [MD − 0.96 (95% CI − 1.44, − 0.48)], and CWM + ZBT [MD − 0.56 (95% CI − 1.04, − 0.09)] has a statistically significant effect on lowering LDL compared to CWM. CWM + SXBX [MD − 0.86 (95% CI − 1.60, − 0.11)] had a statistically significant effect on lowering LDL compared to CWM + DZSM. Therefore, other interventions had no statistically significant difference. The details were shown in Table 6.

According to the SUCRA probability results (Fig. 5F), CWM + SXBX was the most likely the best intervention for lowering the LDL. Table 8 depicts the detailed SUCRA and ranking probability. The interventions were ranked as follows: CWM + SXBX (92.6%) > CWM + XS (76.9%) > CWM + JZL (69.9%) > CWM + ZBT (64.4%) > CWM + TXL (53.5%) > CWM + PS (49.1%) > CWM + PBO (47.0%) > CWM + NXT (35.9%) > CWM + XST (24.9%) > CWM + DZSM (23.5%) > CWM (12.3%).

HDL

A total of 19 RCTs referred to the HDL of eight types of TCPMs and 10 types of interventions, including CWM + TXL vs. CWM + PBO (n = 1), CWM + TXL vs. CWM (n = 4), CWM + XS vs. CWM (n = 1), CWM + NXT vs. CWM (n = 1), CWM + JZL vs. CWM (n = 2), CWM + PS vs. CWM (n = 3), CWM + SXBX vs. CWM (n = 2), CWM + ZBT vs. CWM (n = 3), and CWM + DZSM vs. CWM (n = 2). (Table 2). Figure 4G presents the network evidence plot.

CWM + TXL [MD 0.34 (95% CI: 0.05, 0.64)] had a statistically significant effect on raising HDL compared to CWM. Thus, no statistically significant difference existed between the other interventions. The details were shown in Table 7.

Table 7 Pairwise league table of HDL (lower-left quadrant) and CRP (upper-right quadrant)

According to the SUCRA probability results (Fig. 5G), CWM + XS was the most likely the best intervention for improving HDL. Table 8 illustrates the detailed SUCRA and ranking probability. The interventions were ranked as follows: CWM + XS (86.1%) > CWM + JZL (72.9%) > CWM + TXL (72.9%) > CWM + PBO (62.4%) > CWM + PS (45.6%) > CWM + NXT (45.2%) > CWM + DZSM (43.1%) > CWM + ZBT (28.6%) > CWM + SXBX (26.8%) > CWM (16.4%).

Table 8 Surface under the cumulative ranking curve and ranking probability of different Chinese patent medicines on each outcomeCRP

A total of 11 RCTs referred to the CRP of five types of TCPMs and seven types of interventions, including CWM + TXL vs. CWM + PBO (n = 1), CWM + TXL vs. CWM (n = 5), CWM + NXT vs. CWM (n = 2), CWM + PS vs. CWM (n = 2), CWM + SXBX vs. CWM (n = 1), and CWM + ZBT vs. CWM (n = 1). (Table 2). Figure 4H presents the network evidence plot. All interventions had no statistically significant difference. The details were shown in Table 7.

According to the SUCRA probability results (Fig. 5H), CWM + ZBT was the most likely the best intervention for lowering the CRP. Table 8 presents the detailed SUCRA and ranking probability. The interventions were ranked as follows: CWM + ZBT (71.3%) > CWM + PS (67.0%) > CWM + NXT (64.9%) > CWM + TXL (52.3%) > CWM + SXBX (45.7%) > CWM + PBO (42.8%) > CWM (6.0%).

Safety

A total of 18 RCTs reported the number of the AER of eight types of TCPMs and 10 types of interventions, including CWM + TXL vs. CWM + PBO (n = 1), CWM + TXL vs. CWM (n = 5), CWM + XS vs. CWM (n = 1), CWM + XST vs. CWM (n = 1), CWM + JZL vs. CWM (n = 2), CWM + PS vs. CWM (n = 2), CWM + SXBX vs. CWM (n = 2), CWM + ZBT vs. CWM (n = 2), and CWM + DZSM vs. CWM (n = 2) (Table 2). Figure 4I presents the network evidence plot.

Four studies reported no adverse reactions in the experimental and control groups, while the remaining 14 studies reported 204 cases of adverse reactions. Adverse events included gastrointestinal reactions, such as nausea, discomfort, indigestion, abdominal distension, pain, and diarrhea. Autonomic nervous dysfunction symptoms had dizziness, headache, rash, myalgia, mild hepatic or renal insufficiency, bleeding, and delayed PT. However, most resolved spontaneously without special treatment. The detailed list of adverse reactions was shown in Table 9.

Table 9 Occurrence of adverse reactionsInconsistency test

No closed loops were found in the NMA due to the lack of direct comparison of TCPMs. The inconsistency test could not be carried out. Hence, the results were analyzed using a consistency model.

Publication bias

IMT is the leading indicator for publishing the results of the evaluation applications. The comparison− adjusted funnel plots were plotted to test the publication bias of IMT. When the points in the funnel chart are symmetrical based on the position of the centerline, presenting that there is no publication bias. Figure 6 depicts that the points in the funnel chart are asymmetrical along the center line, indicating the potential presence of publication bias favoring CWM + TCPMs in reducing IMT, as compared to CWM and CWM + PBO.

Fig. 6

Funnel plot of IMT. CWM conventional western medicine, PBO placebo, TXL Tongxinluo capsule, XS Xiaoshuang granules/enteric capsule, NXT Naoxintong capsule, XST Xuesaitong capsule/soft capsule, JZL Jiangzhiling pill, PS Pushen capsule, SXBX Shexiang baoxin pill, ZBT Zhibitai, Dengzhan shengmai capsule