2.1. Identification and Selection of Studies

We searched relevant RCTs from PubMed, Cochrane library, Embase, SinoMed, China National Knowledge Infrastructure (CNKI), Wanfang, and VIP databases to identify studies from their inception up to October, 2020. The searching words included “ulcerative colitis,” “UC,” “Chinese patent medicine,” “Chinese patent drug,” “Chinese proprietary medicine,” “randomized controlled trial,” “randomly,” “xileisan,” “bawei xileisan,” “yunnan baiyao,” “fufangkushen colon-coated capsule,” and “zhikang capsule”. For each database, we established a corresponding retrieval strategy with Boolean formula. PubMed search strategy was detailed in Table 1. Two reviewers independently screened citations against the following predefined selection criteria.

2.2. Population

We included participants diagnosed as mild-to-moderate active UC. There is no limitation in age, sex, nation, ethnicity, and disease stage. Children and pregnant women were excluded. Patients were excluded if they had other intestinal diseases including bacterial dysentery, amoeba colitis, schistosomiasis, intestinal tuberculosis, Crohn’s disease, and reflex enteritis.

2.3. Interventions

Chinese patent medicines with the approval batch number beginning with “Z,” approved by National Pharmaceutical Regulatory Body in China, used alone or in combination with Mesalazine (i.e., 5-aminosalicylic acid) or placebo, are eligible. No limitation was applied on drug dosage, drug formulations, and route of administration. The minimum treatment duration was at least 14 days.

2.4. Comparators

Other Chinese patent medicines, Mesalazine (i.e., 5-aminosalicylic acid (5-ASA), drugs recommended by international authorized clinical guidelines [17]), or placebo were eligible for inclusion.

2.5. Outcomes

Primary outcomes are the disappearance of symptoms including abdominal pain, diarrhea and bloody purulent stool, and Mayo score. Secondary outcomes include adverse events and recurrence rate measured in the follow-up visit from the end of treatment to two years.

2.6. Study Design

We included parallel RCTs. Conference papers were excluded if full papers were not available.

2.7. Study Selection

Two reviewers (YXS and XW) independently screened the study titles and abstracts, identifying studies that met the inclusion criteria for full-text evaluation. In studies with at least three arms, any arm not relevant to our analysis was excluded. We resolved any disagreement about the study selection through discussion, with a third author (ZLL) involved when necessary.

2.8. Data Collection Process and Data Items

The two reviewers (YXS and XW) independently extracted the information and data including study ID, publication year, interventions, outcome measures, and funding information. Any disagreement about data extraction was resolved by discussion with involvement of a third author (WBH) when necessary.

2.9. Risk of Bias Assessment

Two reviewers (YXL, YYZ) independently assessed the risk of bias for each included trial using the Cochrane’s Risk of Bias tool. We resolved any disagreements by consensus or by consulting a third review author (ZLL).

2.10. Statistical Analysis

2.11. Consistent Assessment and Publication Bias

Due to the lack of head-to-head comparisons, no node-splitting method was formed and 95% CI of inconsistence factors could not be generated, so the node-splitting model was not used for consistency test. We did not perform funnel plots to detect potential publication bias because there were no more than 10 included RCTs in each meta-analysis.

3.1. Study Selection

We identified 3222 potential articles in the initial search. A total of 1534 duplicate articles were excluded. We screened the remaining abstracts and 307 full-text articles for potentially eligible studies. Finally, 33 RCTs were included in this review (Figure 1).

3.2. Characteristics of the Selected Literature

The characteristics of the included studies are summarized in Table 2 [20–52]. A total of 33 studies, published between 2007 and 2019, involving 2862 patients with active mild-to-moderate UC, were included in this review. Among them, 55% (1209/2169) were male patients. All included studies were conducted in China. A total of 18 interventions were identified in included studies, including 3 RCTs that used Chinese patent medicine alone and 15 Chinese patent medicines used in combination with Mesalazine or placebo.

3.3. Methodological Quality

The risk of bias assessment is shown in Figure 2 and Table 3. The overall methodological quality of included studies was poor. Only one trial had high methodology quality with a low risk of bias for each item [20]. The method of random sequence generation was not explained in 42.4% (14/33) trials. Blinding was used in three trials [20–22]. None of the trials used blinding in outcome assessment. One trial registered the protocol [20]. 18.2% (6/33) trials reported funding information. No sample size estimates were reported in all trials. A portion of RCTs did not specify the inclusion and exclusion criteria for patients.

3.4. Results of Individual Studies and Synthesis of Results

3.4.2. Results of the Network Meta-Analysis

The networks for comparison established in this study are presented in Figure 3. The detailed results are summarized in Table 5. All the PSFR value of operation time ranged from 1.00 to 1.05, indicating complete convergence, good iterative effects, and stable results of the model. In terms of adverse events, Xilei Powder was safer than the remaining 12 interventions, including Aconitum Lizhong pill plus Mesalazine, Xilei powder plus Mesalazine, Bupi Yichang pill plus Mesalazine, Zhikang capsule plus Mesalazine, ChangYanNing capsule plus Mesalazine, Mesalazine, Shengmai injection plus Mesalazine, Kangfuxin lotion plus Mesalazine, compound Huangbai liquid plus Mesalazine, Yunnan white drug-powder plus Mesalazine, Danshen freeze-dried powder plus Mesalazine, and Danshen injection plus Mesalazine. Mesalazine-combined treatments, Kangfuxin lotion plus Mesalazine and Danshen freeze-dried powder plus Mesalazine, were superior to Mesalazine alone in reducing recurrence rate, while for disappearance of tenesmus, the combined treatment Kangfuxin lotion plus Mesalazine was less effective than Mesalazine and Zhikang capsule plus Mesalazine. No statistically significant difference was found in other outcomes.

Table 5. Adverse events, recurrence rate, disappearance of mucopurulent bloody stool/abdominal pain/diarrhea/tenesmus, and Mayo score based on network meta-analysis.

(a) Adverse events
XLSplusM
0.00 (0.00, 0.02)	FZLZplusM
0.00 (0.00, 0.02)	0.84 (0.04, 20.86)	XLS
0.00 (0.00, 0.10)	5.53 (0.33, 133.79)	6.67 (0.31, 213.75)	BPYCplusM
0.00 (0.00, 0.02)	1.31 (0.07, 22.05)	1.48 (0.06, 33.09)	0.23 (0.01, 3.90)	ZKplusM
0.00 (0.00, 0.02)	1.37 (0.13, 15.57)	1.59 (0.11, 25.97)	0.26 (0.02, 2.64)	1.08 (0.09, 12.93)	CYNplusM
0.00 (0.00, 0.02)	1.42 (0.19, 11.04)	1.63 (0.16, 21.02)	0.26 (0.02, 1.84)	1.12 (0.15, 9.04)	1.03 (0.28, 4.01)	M
0.00 (0.00, 0.03)	1.78 (0.13, 26.89)	2.00 (0.11, 43.38)	0.32 (0.02, 4.61)	1.36 (0.10, 21.44)	1.32 (0.14, 12.25)	1.24 (0.21, 7.51)	SMIplusM
0.00 (0.00, 0.02)	1.11 (0.13, 9.98)	1.35 (0.11, 18.09)	0.21 (0.02, 1.73)	0.91 (0.10, 8.17)	0.84 (0.18, 3.88)	0.80 (0.35, 1.79)	0.65 (0.09, 4.39)	KFXplusM
0.00 (0.00, 0.01)	0.89 (0.07, 11.73)	1.00 (0.06, 19.30)	0.16 (0.01, 2.02)	0.66 (0.05, 9.93)	0.63 (0.08, 5.21)	0.62 (0.12, 3.13)	0.48 (0.04, 5.38)	0.77 (0.12, 4.85)	FFHBplusM
0.00 (0.00, 0.07)	3.59 (0.17, 76.39)	3.98 (0.17, 125.54)	0.64 (0.02, 13.17)	2.69 (0.14, 65.55)	2.60 (0.20, 38.32)	2.48 (0.27, 26.88)	2.04 (0.12, 36.46)	3.05 (0.30, 38.01)	4.18 (0.26, 71.82)	YNBYplusM
0.00 (0.00, 0.02)	1.26 (0.13, 14.65)	1.52 (0.11, 24.46)	0.24 (0.02, 2.44)	1.03 (0.10, 11.71)	0.95 (0.16, 5.77)	0.92 (0.27, 3.16)	0.73 (0.09, 6.22)	1.12 (0.27, 5.01)	1.49 (0.20, 11.78)	0.38 (0.03, 4.50)	DSIplusM
0.00 (0.00, 0.01)	0.78 (0.06, 11.28)	0.90 (0.05, 17.74)	0.14 (0.01, 1.94)	0.61 (0.05, 9.39)	0.57 (0.07, 5.10)	0.56 (0.10, 3.07)	0.44 (0.04, 5.17)	0.68 (0.11, 4.74)	0.92 (0.09, 9.72)	0.22 (0.01, 3.68)	0.60 (0.07, 4.93)	DSFplusM
(b) Recurrence rate
DSFplusM
0.28 (0.02, 4.13)	DSIplusM
0.26 (0.02, 3.25)	0.87 (0.07, 11.72)	GCZXplusM
0.59 (0.08, 5.26)	1.96 (0.26, 19.28)	2.23 (0.33, 20.99)	KFXplusM
0.13 (0.02, 0.78)	0.45 (0.07, 3.03)	0.51 (0.09, 3.32)	0.24 (0.07, 0.57)	M
0.58 (0.03, 16.00)	2.07 (0.11, 62.18)	2.17 (0.12, 50.57)	1.05 (0.08, 14.13)	4.61 (0.44, 56.19)	XLSplusM
(c) Disappearance of mucopurulent bloody stool
YNBYplusM
2.10 (0.32, 16.03)	M
0.79 (0.05, 13.00)	0.38 (0.05, 2.82)	ZKplusM
0.45 (0.02, 8.86)	0.22 (0.02, 1.85)	0.57 (0.03, 11.39)	XLSplusM
1.91 (0.12, 34.79)	0.90 (0.12, 7.23)	2.43 (0.14, 44.25)	4.36 (0.22, 90.77)	XLS
(d) Disappearance of abdominal pain
M
0.30 (0.03, 3.05)	ZKplusM
0.16 (0.01, 1.86)	0.51 (0.01, 16.91)	XLSplusM
0.40 (0.04, 4.63)	1.37 (0.05, 36.72)	2.68 (0.08, 97.04)	XLSplusM
(e) Disappearance of diarrhea
KFXplusM
7.40 (0.52, 117.45)	M
2.21 (0.07, 83.33)	0.31 (0.03, 3.26)	ZKplusM
1.85 (0.05, 76.22)	0.26 (0.02, 2.94)	0.85 (0.02, 22.98)	XLSplusM
6.31 (0.18, 268.20)	0.85 (0.08, 9.65)	2.75 (0.10, 79.92)	3.30 (0.11, 114.24)	XLS
(f) Disappearance of tenesmus
ZKplusM
0.47 (0.01, 27.37)	M
0.00 (0.00, 0.02)	0.00 (0.00, 0.02)	KFXplusM
(g) Mayo score
DSIplusM
−1.36 (−4.42, 1.75)	JCNplusM
−1.59 (−3.73, 0.54)	−0.22 (−2.46, 2.00)	M

• Significant results are in bold.

3.5. Rank Probabilities

The probability rankings are shown in Table 6 and Figure 4. The rank of each treatment is shown on the histogram, which indicates the probability of being ranked in that position. Lower rank indicates a better effect. Rank 1 is best, and Rank N is worst.

4.1. Main Finding

To our knowledge, this is the first SR and NMA evaluating the relative effectiveness and safety of Chinese patent medicine for mild-to-moderate active UC. In the included 33 RCTs, Chinese patent medicines were frequently combined with Mesalazine in the treatment of UC. The findings from network meta-analyses indicate that using Xilei powder plus Mesalazine generates statistically less incidence of adverse events than that of other interventions. Combined treatments including Danshen freeze-dried powder plus Mesalazine and Kangfuxin lotion plus Mesalazine were superior in reducing recurrence rate, and Kangfuxin lotion plus Mesalazine and Zhikang capsule plus Mesalazine were superior in the disappearance of tenesmus. Danshen injection plus Mesalazine leads to statistical improvements in Mayo score.

The overall quality of the included RCTs in this NMA is low. Most studies were judged as unclear risk of performance bias, detection bias, and reporting bias, because of inadequate information about allocation concealment, blinding, and protocol registration. In addition, the disease phase and severity were not clearly reported, which limited the number of studies to be included. These methodological defects of the included studies may affect the authenticity and reliability of our findings.

A previous NMA indicated that there is no statistical difference between the Chinese herbal injection combined with Mesalazine and Mesalazine alone groups in the treatment of mild-to-moderate UC for composite outcome total effective rate [14]. Our findings about the effects of Chinese patent medicines in combination with Mesalazine on safety and recurrence rate are consistent with the previous findings. In contrast, we found that Danshen injection plus Mesalazine was superior to Mesalazine in improving the Mayo score. The different findings might be due to the use of different outcomes. Instead of the “total effective rate,” which is a composite, dichotomous variable, with no specific definition/criteria for each level, we use the specific Mayo score as an outcome.

In this study, the network plot is star shaped and does not form a typical network structure, which is consistent with the findings of two previous studies [13, 14]. There is a lack of direct comparison among the interventions included in this study. As a result, it is impossible to detect the inconsistency between included studies. Our findings are consistent with that of two previous NMA. Meanwhile, the network plot does not form a typical closed loop, reducing the certainty of our analyses. The results need to be interpreted with caution.

4.2. Strength and Limitation

This is the first systematic review and NMA evaluating the relative safety and effectiveness of Chinese patent medicines for mild-to-moderate active UC. Given the lack of head-to-head studies comparing Chinese patent medicines in this population, the rank probability of this work could inform clinical practice.

However, there were several limitations to this study. First, the sample sizes of the included studies were insufficient to draw definitive conclusions with respect to certain Chinese patent medicines. Second, the inclusion and exclusion criteria of this study are limited to the clinical trials of Chinese patent medicine combined with Mesalazine compared with Mesalazine and placebo, and the selection of outcome indicators also limits the inclusion of more RCTs.