Network meta-analysis of intrathyroidal injection of combined dexamethasone with lidocaine for subacute thyroiditis Authors: Hongxia Wei, Yong Huang, Binyi Li Published: 2026 ABSTRACT Background:Intrathyroidal injection of dexamethasone (Dex) alone (monotherapy) or combined Dex with lidocaine (combination therapy) has been used for treating subacute thyroiditis (SAT). We performed this study to simultaneously determine whether intrathyroidal injection of Dex was better than oral prednisone and whether combination therapy was superior to monotherapy.Methods:We identified studies from PubMed, Embase, Cochrane library, China National Knowledge Infrastructure (CNKI), and Wanfang databases through December 31, 2021. Clinical efficacy, recurrence rate, adverse reactions, and remission time of clinical symptoms including fever, pain of thyroid and thyroid enlargement were assessed. The surface under the cumulative ranking curve (SUCRA) was estimated to determine the rank probabilities, and radar plot was to determine the optimal protocol. STATA 14 was utilized for data analysis.Results:Twenty-three studies were included. Results revealed that, compared with oral prednisone, intrathyroidal injection significantly increased clinical efficacy and shortened remission time of clinical symptoms. However, no difference between the 2 protocols of intrathyroidal injection of prednisone was detected. Radar plot suggested that monotherapy might be the optimal protocol for recurrence rate (79.9%), remission time of fever (83.6%) and pain of thyroid (92.5%), and adverse reactions (87.1%); however, combination therapy for clinical efficacy (80.0%) and remission time of thyroid enlargement (93.7%).Conclusion:Intrathyroidal injection of Dex is better than oral prednisone for treating SAT, and monotherapy should be recommended over combination therapy because combination therapy do not produce additional therapeutic benefits. 1. Introduction Subacute thyroiditis (SAT), also named as granulomatous thyroiditis (SAGT) or subacute painful thyroiditis (SAPT), is a transient and self-recoverable inflammatory thyroid disease,[1] accounting for 5% of all clinical thyroid abnormalities.[2,3] The incidence of SAT is increasing yearly, and data indicated an incidence of 4.9/100,000 persons, representing approximately 5% of all thyroid diseases[4] Specific factors contributing to the development and progression of SAT have not yet been elucidated,[5] but viral infection or autoimmune response was generally considered as the common factors,[6,7] SAT was characterized by fever, pain of the thyroid, and thyroid dysfunction;[5,8,9] however, patients will suffer from changing symptoms at different stages.[10] No etiological treatment protocols have been determined for the SAT yet, the main purpose of the primary treatment for SAT is to relieve symptoms.[11,12] The clinical practice guideline published by the American Thyroid Association (ATA) recommends anti-inflammatory agents based on the severity of symptoms; however, this guideline does not list objective criteria for identifying severe cases.[13] Furthermore, the standard recommendation was based on retrospective cohort studies, so we cannot find a basis for treatment options.[14] Additionally, corticosteroids are widely used for symptom relief because they have a quicker effect and provide pain relief more quickly than nonsteroidal anti-inflammatory drugs (NSAIDs);[15,16] however, retrospective studies have demonstrated that long-term GC use, even at low doses, is a significant independent predictor of negative effects.[16–19] Short-term use of GCs may be associated with serious adverse effects.[19] Furthermore, SAT recurrence is more likely to occur in conjunction with corticosteroid therapy.[20] Studies reported that approximately 20–30% patients suffered from relapse due to incomplete drug treatment resulted from poor medication compliance.[21,22] Therefore, more clinical studies of different therapies or novel treatments are recommended to observe the prognosis and long-term effects of SAT patients.[2] Inspired by the intrathyroidal injection of immunosuppressant for treating autoimmune thyroiditis, some Chinese practitioners used intrathyroidal injection of dexamethasone (Dex) for treating SAT and obtained promising results.[23] Li et al designed a meta-analysis to evaluate the clinical efficacy and safety of ultrasound-guided intrathyroidal injection of GCs vs routine oral administration of GCs for SAT but full text has not yet been reported.[10] Currently, 2 protocols were available for intrathyroidal injection of GCs in routine clinical practice, including intrathyroidal injection of Dex alone (monotherapy) or combined Dex with lidocaine (combination therapy), which significantly confused practitioners’ decision-making because there was no study has directly compared monotherapy with combination therapy. Therefore, we conducted the present network meta-analysis to first evaluate whether intrathyroidal injection was better than oral administration of GCs and further investigated whether intrathyroidal injection of combined Dex with lidocaine was superior to intrathyroidal injection of Dex alone. 2. Materials and methods This network meta-analysis followed recommendations proposed by the Cochrane handbook for reviewer of systematic review[24] the Preferred Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines.[25,26] Furthermore, as this study is a meta-analysis based exclusively on published literature, ethical approval and written informed consent were not required. 2.1. Search strategy Two reviewers independently searched PubMed, Embase, Cochrane Library, China National Knowledge Infrastructure (CNKI), and Wanfangdata databases for identifying eligible studies published before December 31, 2021. We used dexamethasone, subacute thyroiditis, and its analogues to develop search strategy. No restriction on language and publication status was imposed in this study. We checked the reference list of relevant reviews and eligible studies to obtain additional studies. Any discrepancies about study retrieval were resolved through discussion. The complete search strategy of three English databases is described in Table S1, Supplemental Digital Content 1. 2.2. Study selection Two independent reviewers checked the eligibility of each study according to the following criteria: 2.2.1. Population Patients confirmed with SAT using recognized diagnostic criteria, such as the 2016 American Thyroid Association Guidelines for Diagnosis and Management of Hyperthyroidism and Other Causes of Thyrotoxicosis[13] or the Guideline for the Management of Thyroid Disease in China.[27] 2.2.2. Intervention/comparison Intrathyroidal injection of Dex alone or combined with lidocaine vs oral prednisone, or versus each other, in randomized controlled studies. 2.2.3. Outcomes At least one of clinical efficacy, recurrence rate, adverse reactions, or remission time of clinical symptoms, including fever, pain of thyroid, and thyroid enlargement. 2.2.4. Exclusion Duplicate reports with less data; studies with essential data unavailable after contacting the corresponding author; sample size < 20 per group. 2.3. Data extraction Two independent reviewers extracted the following data from eligible studies: the first author, country, publication year, sample size and the number of patients assigned in each group, mean age of patients, details of comparisons, treatment duration and outcomes, and the information of methodological quality. Any discrepancies about data extraction were resolved through discussion. 2.4. Quality assessment Two reviewers used the Cochrane risk of bias tool to independently evaluate study quality from 7 aspects.[28] Each item was rated as low risk if the method was correctly used, rated as unclear risk if the method was not clearly described, and rated as high risk if the method was incorrectly used. Any discrepancies about quality assessment were resolved through discussion. 2.5. Outcomes In this network meta-analysis, we defined clinical efficacy, recurrence rate, and the risk of adverse reactions as the primary outcomes. We also defined remission time of clinical symptoms, including fever, pain of thyroid, and thyroid enlargement, as the secondary outcomes. 2.6. Statistical analysis We performed this Bayesian network meta-analysis by using STATA 14.0 (StataCorp LP, College Station, Texas, USA).[29] We did not check global or local inconsistency because no closed loop was available for all outcomes.[30,31]. A random-effects model was selected to calculate the relative efficacy of available treatment protocols.[32] Mean difference (MD) with corresponding 95% credible interval (CrI) was utilized to express the pooled results of remission time clinical symptoms; however, odds ratio (OR) with 95% CrI for clinical efficacy, recurrence rate, and adverse reactions. We also estimated the surface under the cumulative ranking (SUCRA) value to determine the ranking probabilities of treatment protocols for all outcomes.[33,34] A comparison-adjusted funnel plot was created to examine the possibility of the presence of publication bias.[35] Network plot of evidence, forest plot of relative efficacy, SUCRA plot, and funnel plot were created using Stata 14.0 with commands developed by Chaimani et al,[36] and a radar plot was created using Microsoft® Excel 2010. 3. Results 3.1. Search results A total of 192 relevant studies were identified from initial literature retrieval, and 53 duplicate studies were deleted. Then, 50 studies were selected through title and abstract selection. After checking the full text, 27 ineligible studies were excluded according to the following reasons: unrelated to topic (n = 11), insufficient data (n = 3), duplicate studies (n = 4), ineligible controls (n = 4), and non-RCT (n = 4). The relevant information could not be further extracted from 18 articles, which were then excluded, and 23 articles[37–59] were finally included in this network meta-analysis. We designed Figure 1 to depict the process of study selection. PRISMA flow-diagram of study selection. CNKI, China National Knowledge Infrastructure. 3.2. Basic information of eligible studies There were 23 studies that met our inclusion criteria. The basic information of eligible studies is summarized in Table 1. All studies were published in China, and the sample size of individual studies ranged from 40 to 800, with an accumulated number of 2689. All studies confirmed SAT using the Guideline for the Management of Thyroid Disease in China except for 2 studies, which confirmed SAT based on Japanese Guidelines for the diagnosis and management of thyroid disease and their utility[60] and American Thyroid Association Guidelines for Diagnosis and Management of Hyperthyroidism and Other Causes of Thyrotoxicosis.[13] Among these eligible studies, 1343 patients were treated with oral prednisone in the active control group, 949 patients were treated with monotherapy, and 397 patients were treated with combination therapy. Transitivity assessment confirmed the comparability of all studies based on diagnostic criteria, publication year, sample size, mean age, disease duration, and treatment duration (Table S2, Supplemental Digital Content 2). Sixteen studies reported the data of clinical efficacy, 7 studies reported the data of recurrence rate, eleven studies reported the data of adverse actions, fourteen studies reported the data of remission time of fever, and sixteen studies reported the data of remission time of pain of thyroid and thyroid enlargement. Network maps of all comparisons for all outcomes are displayed in Figure 2. Table 1Basic characteristics of included studies (n = 23).StudyResearch regimeControl regimeDiagnostic criteriaSample sizeAge, yearsDisease durationTreatment durationGao et al[37]injection of 5mg Dexoral prednisoneCCPG40 vs 45(43.35 ± 10.27) vs (44.72 ± 12.51)(16.31 ± 5.13) vs (15.68 ± 6.63)<3 weeksGao et al[38]injection of 5mg Dexoral prednisoneCCPG60 vs 60(42.52 ± 9.47) vs (43.61 ± 9.54)(8.53 ± 3.56) vs (8.74 ± 3.62)8 weeksGu et al 2003injection of 5mg Dexoral prednisoneCCPG30 vs 30n.a.n.a.n.a.Guo et al 2018injection of 5mg Dexoral prednisoneCCPG47 vs 4745.8 ± 3.516.7 ± 3.28 weeksLi[42]injection of 5mg Dexoral prednisoneCCPG34 vs 27(44.35 ± 5.08) vs (46.72 ± 5.62)(7.9 ± 2.8) vs (8.3 ± 2.9)8 weeksLi[43]injection of 5mg Dexoral prednisoneCCPG400 vs 400(48.27 ± 3.16) vs (48.30 ± 3.22)(12.30 ± 1.34) vs (12.20 ± 1.14)n.a.Liu[44]injection of 5mg Dexoral prednisoneCCPG60 vs 6022–62n.a.8 weeksLiu et al[45]injection of 5mg Dexoral prednisoneCCPG36 vs 3642.0 ± 9.52–188 weeksMao[48]injection of 3mg Dexoral prednisoneCCPG56 vs 57(42.0 ± 5.78) vs (42.2 ± 5.82)(10–37) vs (10–35)n.a.Qu[50]injection of 3mg Dexoral prednisoneCCPG34 vs 3453 ± 8.61 weeks-2 months8 weeksWang et al[52]injection of 3mg Dexoral prednisoneCCPG49 vs 4942 ± 12.67–608 weeksXie[55]injection of 5mg Dexoral prednisoneCCPG41 vs 41(46.3 ± 2.5) vs (45.2 ± 2.4)(9.4 ± 1.3) vs (9.5 ± 1.2)8 weeksYang et al[56]injection of 5mg Dexoral prednisoneCCPG20 vs 2044.4 ± 5.17.7 ± 2.7n.a.Zhang et al[58]injection of 5mg Dexoral prednisoneCCPG42 vs 43(45.04 ± 1.05) vs (44.77 ± 0.94)n.a.8 weeksHuo et al[41]injection of Dex 5mg plus 1mL 2% lidocaineoral prednisoneJCPG48 vs 45(45.56 ± 8.89) vs (44.86 ± 6.54)(6.19 ± 4.95) vs (7.56 ± 3.24)n.a.Liu et al[45]injection of Dex 5mg plus 1mL 2% lidocaineoral prednisoneACGP70 vs 7036.8 ± 2.3n.a.3 weeksMa[47]injection of Dex 5mg plus 1mL 2% lidocaineoral prednisoneCCPG55 vs 55(37.67 ± 1.67) vs (36.87 ± 1.98)(9.43 ± 1.43) vs (9.87 ± 1.76)8 weeksQin[49]injection of Dex 5mg plus 1mL 2% lidocaineoral prednisoneCCPG64 vs 64(37.12 ± 9.03) vs (37.29 ± 9.16)n.a.n.a.Wang[51]injection of Dex 5mg plus 1mL 2% lidocaineoral prednisoneCCPG40 vs 40(45.3 ± 3.5) vs (45.8 ± 3.4)n.a.8 weeksWen et al[53]injection of Dex 5mg plus 1mL 2% lidocaineoral prednisoneCCPG35 vs 3530~50n.a.8 weeksWu[54]injection of Dex 5mg plus 1mL 2% lidocaineoral prednisoneCCPG30 vs 3030~55n.a.8 weeksYang[57]injection of Dex 5mg plus 1mL 2% lidocaineoral prednisoneCCPG34 vs 34(41.2 ± 3.5) vs (42.3 ± 3.8)n.a.7 weeksZhao[59]injection of Dex 5mg plus 1mL 2% lidocaineoral prednisoneCCPG21 vs 21(36.45 ± 8.47) vs (36.25 ± 8.64)n.a.n.a.ACGP = American Thyroid Association Guidelines for Diagnosis and Management of Hyperthyroidism and Other Causes of Thyrotoxicosis CCPG = Thyroiditis: a guide to diagnosis and treatment of thyroid diseases in China, Dex = dexamethasone, JCPG = Japanese Guidelines for the diagnosis and management of thyroid disease and their utility, n.a. = not applicable. Basic characteristics of included studies (n = 23). ACGP = American Thyroid Association Guidelines for Diagnosis and Management of Hyperthyroidism and Other Causes of Thyrotoxicosis CCPG = Thyroiditis: a guide to diagnosis and treatment of thyroid diseases in China, Dex = dexamethasone, JCPG = Japanese Guidelines for the diagnosis and management of thyroid disease and their utility, n.a. = not applicable. Network maps of all comparisons for all outcomes. 3.3. Methodological quality Nine studies[37–40,42,43,46,57,58] correctly generated random allocation methods using appropriate techniques, such as computer-based randomization sequence. Only one study[39] clearly described how to conceal the random sequence and blind participants, personnel, and outcome assessors. All studies[37–59] were evaluated as low risk in attrition bias, reporting bias, and other bias domains. Overall, the methodical quality of all eligible studies was at a moderate level. Details of quality assessment were shown in Figure S1, Supplemental Digital Content 3. 3.4. Clinical efficacy Sixteen studies reported clinical efficacy, including 7 studies[37–39,43,46,52,58] comparing monotherapy and oral prednisone, and 9 studies[41,44,47,49,51,53,54,57,59] comparing combination therapy and oral prednisone. Pooled results revealed that monotherapy (OR, 2.50; 95% CrI, 1.62–3.85) and combination therapy (OR, 2.79; 95% CrI, 1.53–5.07) significantly improved clinical efficacy related to oral prednisone. However, monotherapy was comparable to combination therapy in terms of clinical efficacy (OR, 1.11; 95% CrI, 0.53–2.33). The relative efficacy of all comparisons is displayed in Figure 3. Results of SUCRA analysis indicated a probability of 70.0% for monotherapy and 80.0% for combination therapy (Fig. 4). Relative efficacy of all treatment protocols for each outcome. The results of SUCRA for all treatment protocols for all outcomes. N/A indicated not applicable. 3.5. Recurrence rate Seven studies reported the recurrence rate, including 3 studies[43,45,46] comparing monotherapy and oral prednisone, and 4 studies[41,44,53,54] comparing combination therapy and oral prednisone. Pooled results revealed that monotherapy (OR, 0.24; 95% CrI, 0.06–1.06) and combination therapy (OR, 0.34; 95% CrI, 0.10–1.18) numerically decreased the recurrence rate related to oral prednisone. However, monotherapy was comparable to combination therapy in terms of recurrence rate (OR, 1.38; 95% CrI, 1.20–9.61). The relative efficacy of all comparisons is displayed in Figure 3. Results of SUCRA analysis indicated a probability of 79.9% for monotherapy and 66.5% for combination therapy (Fig. 4). 3.6. Adverse reactions Eleven studies reported adverse reactions, including 8 studies[37,39,42,43,45,50,52,56] comparing monotherapy with oral prednisone, and 3 studies[41,49,59] comparing combination therapy with oral prednisone. Pooled results revealed that, compared with oral prednisone, monotherapy (OR, 0.19; 95% CrI, 0.06–0.58) was significantly associated with a decreased risk of adverse reactions, but combination therapy (OR, 0.34; 95% CrI, 0.08–1.36) was numerically associated with decreased adverse reactions. Moreover, monotherapy was comparable to combination therapy in terms of adverse reactions (OR, 1.79; 95% CrI, 0.30–10.65). The relative efficacy of all comparisons is displayed in Figure 3. Results of SUCRA analysis indicated a probability of 87.1% for monotherapy and 59.6% for combination therapy (Fig. 4). 3.7. Remission time of clinical symptoms Fourteen studies reported the remission time of fever, including 10 studies[37–40,43,48,50,52,55,56] comparing monotherapy and oral prednisone, and 4 studies[51,53,57,59] comparing combination therapy and oral prednisone. Pooled results revealed that monotherapy (MD, −1.69; 95% CrI, −2.04 to −1.34) and combination therapy (MD, −1.54; 95% CrI, −2.07 to −1.01) significantly shortened the remission time of fever related to oral prednisone. However, no difference was detected between monotherapy and combination therapy (MD, 0.15; 95% CrI, −0.49 to 0.78). The relative efficacy of all comparisons is displayed in Figure 3. Results of SUCRA analysis indicated a probability of 83.6% for monotherapy and 66.4% for combination therapy (Fig. 4). Sixteen studies reported the remission time of thyroid pain, including 12 studies[37–40,42,43,45,48,50,52,55,56] for comparison between monotherapy and oral prednisone and 4 studies[51,53,57,59] for comparisons between combination therapy and oral prednisone. Pooled results revealed that monotherapy (MD, −2.31; 95% CrI, −2.75 to −1.88) and combination therapy (MD, −1.86; 95% CrI, −2.59 to −1.14) significantly shortened the remission time of thyroid pain related to oral prednisone. However, no difference was detected between monotherapy and combination therapy (MD, 0.45; 95% CrI, −0.40 to 1.30). Relative efficacy of all comparisons is displayed in Figure 3. Results of SUCRA analysis indicated a probability of 92.5% for monotherapy and 57.5% for combination therapy (Fig. 4). Sixteen studies reported the remission time of thyroid enlargement, including 12 studies[37–40,42,43,45,48,50,52,55,56] for comparison between monotherapy and oral prednisone and 4 studies[51,53,57,59] for comparisons between combination therapy and oral prednisone. Pooled results revealed that monotherapy (MD, −4.48; 95% CrI, −5.20 to −3.76) and combination therapy (MD, −5.38; 95% CrI, −6.73 to −4.04) significantly shortened the remission time of thyroid pain related to oral prednisone. However, no difference was detected between monotherapy and combination therapy (MD, −0.90; 95% CrI, −2.42 to 0.63). Relative efficacy of all comparisons is displayed in Figure 3. Results of SUCRA analysis indicated a probability of 56.3% for monotherapy and 93.7% for combination therapy (Fig. 4). 3.8. Determination of optimal protocol As shown in Figure 5, combination therapy was close to the outer boundary of the radar plot in terms of clinical efficacy and remission time of thyroid enlargement, indicating that it was a relatively optimal treatment protocol for these 2 outcomes. However, monotherapy was close to the outer boundary of the radar plot in terms of recurrence rate, adverse reactions, remission time of fever and thyroid pain, indicating that it was a relatively optimal treatment protocol for these outcomes. Radar plot of all treatment protocols for all outcomes. 3.9. Publication bias As shown in Figure S2, Supplemental Digital Content 4, the distribution of funnel plots in terms of all outcomes was asymmetric, indicating that our results suffered from the negative influence of publication bias. 4. Discussion This network meta-analysis indicated that, compared with oral prednisone, monotherapy and combination therapy significantly improved clinical efficacy and shortened remission time of clinical symptoms, as well as were associated with significantly or numerically decreased recurrence rate and adverse reactions. However, no statistical difference between monotherapy and combination therapy was detected for all outcomes. Moreover, the radar plot indicated that monotherapy might be the optimal treatment protocol for recurrence rate, adverse reactions, and remission time of fever and thyroid pain, but combination therapy might be the optimal treatment protocol for clinical efficacy and remission time of thyroid enlargement. Although oral administration of GCs and NSAIDs was recognized as a common treatment strategy for SAT, several adverse reactions significantly limited the use of this method.[15,61] As a long-term GC, intrathyroidal injection of Dex has therefore been extensively used to reduce inflammation in order to address the drawbacks of oral administration of GCs and NSAIDs.[62,63] It’s noted that numerous studies have demonstrated the advantages of intrathyroidal injection of Dex in the treatment of SAT,[37–40,42,43,45,46,48,50,52,55,56,58] which were further confirmed by our findings. Moreover, as a local anesthetic, lidocaine was suggested to have an anti-inflammatory effect.[64] Therefore, intrathyroidal injection of combined Dex with lidocaine has also been used for the treatment of SAT in clinical practice. Similarly, combination therapy has also been confirmed to effectively improve clinical efficacy and be associated with a reduced risk of adverse reactions,[41,44,47,49,51,53,54,57,59] which was consistent with our findings in this network meta-analysis. Compared with previously published studies, this network meta-analysis first determined no difference between monotherapy and combination therapy, although additional lidocaine was injected. However, monotherapy was confirmed to be associated with a lower recurrence rate and adverse reactions, although combination therapy was related to higher clinical efficacy. Certainly, as an invasive approach, intrathyroidal injection of Dex alone or Dex plus lidocaine for the treatment of SAT should be cautiously used in clinical practice.[41] It should be noted that Li et al[10] developed a systematic review and meta-analysis in 2019 to compare the efficacy and safety of routine oral glucocorticoid administration for SAT patients to ultrasound-guided intrathyroidal injection of glucocorticoids; however, the results of this systematic review and meta-analysis have not yet been published (10). In light of this, the current network meta-analysis first compares the efficacy and safety of intrathyroidal GC injection vs oral prednisone, and the merged findings supported the benefits of intrathyroidal GC injection vs oral prednisone in the treatment of SAT patients. Furthermore, we showed that monotherapy was equally effective as combination therapy in the treatment of SAT patients. Our network meta-analysis had several limitations. First, although 23 eligible studies were included in the final analysis, an extremely small sample size was enrolled in individual studies, except for one study. Therefore, our findings could not avoid the negative impact of inadequate statistical power. Second, all eligible studies were performed in China, and thus, caution should be exercised when considering using our findings for SAT in other clinical settings. Meanwhile, publication bias could not be avoided. Comparison-adjusted funnel plots have confirmed the presence of publication bias. Therefore, our findings should be further investigated in future studies with large-scale and high quality. Third, we only evaluated outcomes related to clinical efficacy and safety in this network meta-analysis for the purpose of facilitating clinical decision-making; however, laboratory indicators such as erythrocyte sedimentation rate and free triiodothyronine were not considered. Fourth, variations were available in disease duration and treatment duration; we could not perform subgroup analysis to eliminate the negative influence of these variations on the pooled results because some studies did not report data. Additionally, the specific protocol of intrathyroidal injection differed from one to another; sensitivity analysis was also not conducted due to limited data. Fourth, the manual screening of reference lists from included studies, while improving comprehensiveness, may introduce citation bias and limit the full replicability of the search process. Finally, we did not register the formal protocol of this network meta-analysis. However, we strictly conducted literature retrieval, methodological quality assessment, data extraction, and data analysis to enhance the reliability of our findings. Intrathyroidal injection is better than oral administration of GCs for the treatment of SAT patients, and monotherapy is comparable to combination therapy. Furthermore, because the intrathyroidal treatment option in SAT patients is not yet widely accepted, there is still a need for multicenter, randomized controlled studies with a high level of evidence to examine the safety and efficacy of intrathyroidal injection. Acknowledgments We would like to deeply appreciate all authors who performed all eligible studies which have been included in the present network meta-analysis. Author contributions Data curation: Hongxia Wei. Formal analysis: Hongxia Wei. Investigation: Hongxia Wei. Methodology: Hongxia Wei. Writing – review & editing: Hongxia Wei, Yong Huang, Binyi Li. Conceptualization: Yong Huang, Binyi Li. Writing – original draft: Binyi Li.