The Adjuvant Efficacy of Angong Niuhuang Pill in the Treatment of Viral Encephalitis: A Meta-Analysis of Randomized Controlled Trials

Summary

Here, we present a protocol to analyze the effectiveness and safety of Angong Niuhuang Pill in the treatment of viral encephalitis.

Abstract

This meta-analysis aims to evaluate the efficacy of Angong Niuhuang Pill (ANP) as an adjuvant therapy in the treatment of viral encephalitis. Seven databases (PubMed, Cochrane Library, Embase, SinoMed, CNKI, VIP and WanFang) were included for literature retrieval from inception to July 2023. Randomized controlled trials comparing ANP plus conventional therapy with conventional therapy alone were eligible. Pooled effect sizes and 95% confidence intervals (CIs) were calculated for evaluating efficacy and safety. Sensitivity analysis and publication bias assessments were performed for analyzing the inconclusiveness of findings. 13 studies involving 1045 cases were included for meta-analysis. Adjuvant treatment with ANP increased the probability of the total effective rate by 17% compared with conventional treatment (Risk ratios (RR) = 1.17, 95%CI [1.08, 1.27]). The disappearance time of clinical syndromes and signs was significantly decreased after adjuvant treatment with ANP, including the time of defervescence (weighted mean difference (WMD) = -1.59, 95%CI [-2.09, -1.09]), the time of consciousness recovery (WMD = -1.79, 95%CI [-2.06, -1.51]), the time of headache disappearance (WMD = -1.51, 95%CI [-1.93, -1.08]), the time of tic disappearance (WMD = -1.88, 95%CI [-2.39, -1.36]). The adjuvant efficacy of ANP for treating viral encephalitis (VE) appears to improve the total effective rate and shorten the disappearance time of clinical syndromes. More high-quality randomized controlled trials (RCTs) are needed to support our findings.

Introduction

Encephalitis is defined as the inflammation of the brain parenchyma with high mortality, characterized by neurological pathological signs, fever, abnormal mental status, and altered electroencephalographic (EEG) and neuroimaging^1,2. The incidence of encephalitis ranges between 7 and 15 cases per 100,000 people^3,4,5. The cause of approximately half of encephalitis cases is unknown, and viral encephalitis (VE) accounts for about 20% to 50% of diagnosed cases^5,6. Common viruses include human enterovirus, human herpesvirus, herpes simplex virus, human immunodeficiency virus, and cytomegalovirus⁷. Patients with VE require intensive monitoring and supportive measures to maintain vital signs and alleviate cerebral inflammation⁸. Acyclovir is an empiric treatment for all cases of VE⁹. Evidence for the use of glucocorticoids, immunoglobulins, and interferon alfa are inconsistent⁸. Angong Niuhuang Pill (ANP), one of the well-known orifice-opening medicines, is widely used in mental and conscious abnormalities due to critical illness. China Medical Education Association recommended ANP as an A-level strength of recommendation and II-level quality of evidence (evidence from non-randomized studies or randomized controlled trials)¹⁰.

A systematic review (SR) is a comprehensive collection and rigorous evaluation of all clinical studies that meet the inclusion criteria. Meta-analysis is a methodology for quantitative analysis in SR. Distinguishing from the methods of basic meta-analysis, a trim and fill procedure was adopted to test the robustness of the effect size estimate to the assessment of publication bias¹¹. A small number of studies were removed until a new effect size made the plot symmetrical. Additional studies were added back to the plot based on imputed mirror-image studies. If the difference of the recalculated pooled estimate (PE) before and after the trim and fill procedure was less than 0.05, the effect size estimate of publication bias was robust. TSA 0.9.5.10 software was used to reduce the risk of false-positive results and calculate the required information size (RIS), which was beneficial to make our findings more robust. Given that no systematic review of the data on the improvement of symptoms, signs, biochemical indicators, and adverse reactions of ANP in the treatment of VE, we therefore performed this meta-analysis of the available randomized controlled trials.

Protocol

1. Protocol registration

1. Register the systemic review.
   NOTE: PROSPERO (Table of Materials) was built by National Institute for Health Research for the registration of systematic review. This systemic review protocol has been registered under the number CRD42022301182.

2. Software installation

1. EndNote software installation
   1. Purchase EndNote software and download the exe file as an installation program from EndNote's official website (Table of Materials).
   2. Decompress the ZIP file. Double-click to start the installer.
2. Stata software installation
   1. Purchase stata software from the Stata official website (Table of Materials).
   2. Double-click the Stata application program to start the installer.
3. TSA software installation
   1. Go to the website of The Copenhagen Trial Unit (Table of Materials) for registration and downloading a ZIP file.
   2. Double-click the TSA executable jar file for installation.

3. Database retrieval

1. Construct retrieval strategy in the following databases: PubMed, Cochrane Library, Embase, Chinese Biomedical Literature Service System (SinoMed), China National Knowledge Infrastructure (CNKI), Chinese Scientific Journals Database (VIP), and WanFang database (Table of Materials).
2. Retrieval time: From the inception of the database to July 2023. Take PubMed as an example:
   1. Use PubMed advanced search builder (Supplementary Figure 1A).
   2. Combine the search terms as follows: Angong Niuhuang OR Angong pill OR Bezoar pill) AND (Viral encephalitis.
   3. Download all the retrieved literature: Send To > Citation Manager > Create File (Supplementary Figure 1B).

4. Inclusion and exclusion criteria

1. Inclusion criteria
   1. Study design: Include randomized controlled trials (RCTs).
   2. Interventions: Include the comparison between Angong Niuhuang Pill (ANP) plus conventional therapy versus conventional therapy alone.
      NOTE: Conventional therapies involved antibiotics, antiviral drugs, intracranial decompression, antipyretics, and maintaining water-electrolyte balance, etc.
   3. Patients: Include patients with viral encephalitis based on the diagnostic criteria.
      1. Altered mental status: Include patients with no other cause found for altered consciousness, drowsiness, or personality changes lasting ≥24 h.
      2. Ensure at least 3 of the following associated manifestations are present: (a) Body temperature higher than 38 °C within 72 h before or after onset; (b) Generalized or partial seizures that cannot be entirely attributed to pre-existing epilepsy; (c) New symptoms of neurological defects; (d) White blood cell count ≥ 5/mm² in cerebrospinal fluid; (e) Neuroimaging showed abnormal brain parenchyma; (f) EEG abnormalities consistent with encephalitis that cannot be attributed to other causes.
   4. Outcomes: Include outcomes such as total effective rate, time of defervescence, time of consciousness recovery, time of headache disappearance, and time of tic disappearance.
2. Exclusion criteria
   1. Complete the literature retrieval and download the full texts. Read the full texts and exclude studies that meet the exclusion criteria. Record the reasons for each excluded study.
      1. Additional drug: Exclude studies in which, besides ANP, the treatment group uses additional drugs different from the control group.
      2. Data missing: Exclude literature with incomplete data.

5. Literature management and screening

1. Importing literature and removing the duplicates
   1. Open Endnote and create a new group by clicking My Group > Create Group (Supplementary Figure 2A).
   2. Import the retrieved literature by clicking on File > Import > Options > Import Options > EndNote Import > Import (Supplementary Figure 2B).
   3. Remove the duplicates by clicking Reference > Find Duplicates > Keep this Record (Supplementary Figure 2C).
2. Screening literature and data extraction
   1. Create an Excel extraction spreadsheet.
   2. According to the inclusion and exclusion criteria, have two authors screen the included literature and extract data of included articles independently as follows: 1) first author; 2) publication year; 3) baseline data of a patient with VE (age, gender, ANP usage, duration); 3) outcome measures.
      NOTE: Record the reasons for each eliminated study.
   3. Data types: Extract data from dichotomous variables (effective rate), which used sample size and composition ratio, and continuous variables (time of defervescence, time of consciousness recovery, time of headache disappearance, time of tic disappearance), which used mean and standard deviation.
   4. Create a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart, visualizing the number of studies processed at each step and the reasons for removing them.

6. Quality assessment

1. Use the Cochrane risk of bias tool to evaluate the methodology. Include the terms of bias assessment: random sequence generation, allocation concealment, blinding of participants and personnel, incomplete outcome data, selective reporting, blinding of outcome assessment, and other biases.
   NOTE: The third party would intervene to resolve discrepancies by discussions.

7. Data synthesis and analysis

1. Constructing Forest plots using STATA software
   1. For continuous data, follow steps 7.1.2-7.1.4.
   2. For data entry, click User > Meta-Analysis of Binary and Continuous > Main > Continuous. Set the sample size, mean, and standard deviation of the treatment group as n1, mean1, and sd1, and set the sample size, mean, and standard deviation of the control group as n2, mean2, and sd2 (Supplementary Figure 3A).
   3. Effect model: Appy the effect model (fixed/random) based on I² statistics among the reported trials. I² > 50% indicates a high heterogeneity; choose the Random model. Click User > Meta-Analysis of Binary and Continuous > Continuous > Random I-V Heterogeneity. Use Fixed Inverse Variance if I² ≤ 50%.
   4. Calculate the outcomes of the same unit as weighted mean difference (WMD); under Statistics, choose noStandard. For different units of outcomes, use the standardized mean difference (SMD); unser Statistics, choose Cohen (Supplementary Figure 3B).
   5. For dichotomous data, follow steps steps 7.1.6-7.1.7.
   6. For data entry, use Risk ratios (RR). Click on User > Meta-Analysis of Binary and Continuous > Main > Count. Set the effective number and non-effective number of the treatment group as E1 and NE1, and for the control group as E2 and NE2 (Supplementary Figure 3C).
   7. Effect model: Apply the effect model (fixed/random) based on I² statistic among the reported trials. I² > 50% indicates a high heterogeneity; choose the Random model. Click User > Meta-Analysis of Binary and Continuous > Binary > Fixed. Use Fixed Inverse Variance if I² ≤ 50%.
      NOTE: With small data and few studies, the M-H test is more robust. But M-H can only process binary variables. I-V can be adopted for both continuous and binary variables.
   8. Performing sensitivity analysis using STATA software
      1. If I² > 50% in heterogeneity test, perform sensitive analysis to identify the possible cause of heterogeneity. Click on User > Meta-Analysis > influence analysis, metan-based (metainf) > Continuous (Supplementary Figure 3D).
   9. Checking publication bias using STATA software
      1. For continuous data, select Egger's linear regression test to assess the publication bias of continuous variables. Use the command db metabias. Click Main, choose _ES _seES, and select Egger (Supplementary Figure 3E).
      2. For binary data, use RR or OR as the effect indicators of binary variable Harbord's weighted linear regression test. Use the command metabias e1 ne1 e2 ne2, or harbord.
   10. Trimming and filling procedure using STATA software
       1. Visualized by a funnel plot, adopt the trim and fill procedure to test the robustness of the effect size estimate to assess publication bias. Use the command db metatrim.
       2. Click Main and choose _es_sees. Then, click Linear > Fixed > Funnel (Supplementary Figure 3F).
          NOTE: Binary variables should make a logarithmic transformation to generate a new dataset for metatrim, such as gen logES = log(_ES).
   11. Trial sequential analysis
       1. Build a new meta-analysis. Click on File > New Meta-Analysis. Choose Data Type, import Name, Group1, and Group2 and choose the Outcome Type (Supplementary Figure 4A).
          NOTE: For dichotomous data, negative outcome measures such as death, cancer incidence; positive outcome measures such as survival, viral clearance, or smoking cessation. For continuous variables, a negative mean effect increase represents a negative event (e. g., increase in depression score), and a positive mean increase in mean effect is a positive event (e.g., platelet count).
       2. Set effect measure and model according to Forest plot results. Use the Constant continuous correction method and Value = 0.5 (Supplementary Figure 4B).
       3. Click on Trial, fill in the Study, Year, and Effect and Total number of Intervention and Control groups. Then, click Add Trial (Supplementary Figure 4C).
          NOTE: Supplementary Figure 4D shows the data entry tab that is used to distinguish between continuous and binary variables.
       4. Click Tsa > Alpha-Spending Boundary. Set Boundary Type as Two-Sided, the Type Error rate α = 0. 05 and Power = 0.80. Click Perform Calculation, then Graph to see the figure.
       5. For the meta-analysis of dichotomous outcome, fill the expected event rate and intervention effect, including Relative Risk Reduction (RRR) and Incidence in the control group. According to precious forest plot of total effective rate, calculate RRR (Supplementary Figure 4E).
       6. For continuous data, click Meta-analysis and choose Mean difference and Randon-effect(BT) according the previos results of forest plot.
       7. Click on Trial, fill in the Study, Year, Mean response, Standard Deviation and Group size of Intervention and Control groups. Then, click Add Trial.
       8. Click on Tsa. Select Alpha-Spending Boundary. Then, setting the Boundary Type as Two-Sided, the Type ⅠError rate α to 0.05 and Power to 0.8. Choose Empirical in mean difference and variance.Click Perform Calculation, then click Graph to see the figure.

Results

Search results and study characteristics
Systematic retrieval yielded 122 studies and manual searching added 9 relevant literatures. After removing 49 duplicates, 82 studies remained to be screened by abstract and full-text. Ultimately, 13 studies^{12,13,14,15,16,17,}

Discussion

This meta-analysis provides evidence that the adjunctive ANP can improve the total effective rate by 17% in patients with VE. Adjuvant therapy with ANP can significantly shorten the disappearance time of positive pathological signs and symptoms. Xingnaojing (XNJ) injection extracted from ANP is widely used for abnormal consciousness associated with encephalopathy. A recent meta-analysis verified the adjunctive effect of XNJ for treating VE, contributing to an increase in the number of cure cases and shortening the time o...

Materials

Name	Company	Catalog Number	Comments
PROSPERO	National Institute for Health and Care Research	https://www.crd.york.ac.uk/PROSPERO/
STATA	StataCorp	https://www.stata.com/
TSA	Copenhagen Trial Unit, Centre for Clinical Intervention Research	https://ctu.dk/tsa/downloads/
PubMed	National Library of Medicine	https://pubmed.ncbi.nlm.nih.gov/
Cochrane Library	Cochrane Library	https://www.cochranelibrary.com/
Embase	Elsevier	https://www.embase.com/search/quick
SinoMed	SinoMed	http://www.sinomed.ac.cn/index.jsp
CNKI	China National Knowledge Infrastructure	https://www.cnki.net/
VIP	Chongqing VIP Information Co., Ltd.	http://www.cqvip.com/
WanFang database	Wanfang Data Knowledge Service Platform	https://www.wanfangdata.com.cn/index.html