Characteristics of registered systematic reviews on traditional Chinese medicine for COVID-19

Rongna Lian Lanzhou University First A liated Hospital Ya Gao Lanzhou University School of Basic Medical Sciences Ruinian Zhang Lanzhou University First A liated Hospital Dairong Xie Lanzhou University First A liated Hospital Yi Zhang Lanzhou University First A liated Hospital Mei Zhang Gansu Provincial Cancer Hospital Junhua Zhang Tianjin University of Traditional Chinese Medicine Jinhui Tian (  tianjh@lzu.edu.cn ) Lanzhou University https://orcid.org/0000-0002-0054-2454


Introduction
In late December 2019, an outbreak of pneumonia of unknown origin characterized by strong interpersonal transmission. Then scientists found that the Coronavirus Disease 2019 (COVID-19) caused by Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), an enveloped RNA virus [1][2][3][4][5] . Within three months, the COVID-19 outbreak had already affected six continents 6 , and the WHO upgraded its status from epidemic to pandemic on March 11, 2020 7 . As of August 1, 2020, a total of 17396943 cases were reported, including 675060 deaths 8 .
PROSPERO is an international database of prospectively registered systematic reviews in health and social care, welfare, public health, education, crime, justice, and international development. China's rst novel coronavirus pneumonia diagnosis and treatment plan (trial version third) issued on January 23, 2020 was rst written into the Chinese medicine treatment plan 9 . Novel coronavirus pneumonia diagnosis and treatment plan (Fourth Edition) issued on January 27, 2020 recommended the Chinese her 10 . Medical workers and scienti c researchers actively carry out research and have registered numerous Chinese medicine related COVID-19 systematic reviews (SRs). However, no research has focused on the Page 3/19 characteristics of these registered SRs. This study was designed to evaluate the cooperation between institution and the distribution of outcome measures in registered SRs of Chinese medicine related to COVID-19, to provide a reference for future researchers to register and carry out COVID-19 SRs.

Data sources
We systematically searched the PROSPERO registration platform (https://www.crd.york.ac.uk/prospero/) to identify all COVID-19 SRs related to TCM. The retrieval time limit is from the establishment of the database to July 1, 2020.

Inclusion and exclusion criteria
The type of included study was registered SRs on PROSPERO. The study population was patients diagnosed with COVID-19, and there were no restrictions on age, gender, race, and course of disease.
Intervention was TCM treatment, including traditional Chinese medicine treatment, integrated Chinese and Western medicine, acupuncture, Taijiquan and so on. We excluded basic science, diagnostic tests, empirical studies, and health services. Duplicate records and record records were also excluded.

Study selection and data extraction
Two researchers independently reviewed the records and screened out eligible registrations according to the inclusion and exclusion criteria, and then proceeded to a cross-check. Con icts were settled through discussions with a third reviewer.
We developed a data extraction form using Microsoft Excel 2016 (Microsoft Corp, Redmond, WA, www.microsoft.com) through discussions with the review team. Then, one author extracted data from the included SRs using the pre-de ned form and a second reviewer checked the extracted data. The detailed data included: title, author, registration time, start time and expected completion time, whether conducted literature search, names of databases searched, intervention, control, countries, provinces, institutions, primary outcomes, secondary outcomes, software used for data analyses, and other information.

Data management and analysis
We used standardized names to replace different expressions of institutions, interventions, and outcome indicators. After that, the data extraction information table was designed by Microsoft Excel 2016 (Microsoft Corp, Redmond, WA, www.microsoft.com), and the information was extracted. After sorting out the data in a recognizable format, VOSviewer 1.6.14 (Leiden University, Leiden, Netherlands) software was used to generate the cooperation network maps of institutions, countries, and provinces. In the network maps, the size of nodes indicates the frequency of analysis elements, and the color of nodes and lines represents the cooperative relationship or co-occurrence relationship 11-14. The parameters of the VOSviewer are as follows: counting method (fractional counting), ignoring documents with multiple authors (the maximum number of authors per document is 25). Among the 80 included SRs, 1 (1.25%) SR was performed by only one author, 56 (70.00%) SRs had 2-5 participants, 21 (26.25%) SRs had 6-10 participants, and 2 (2.50%) SRs had more than 10 participants. All 80 studies were systematic reviews, including meta-analysis (64, 80.00%), network meta-analysis (8,10.00%), review of reviews (3, 3.75%), narrative synthesis (3, 3.75%), synthesis of quali ed studies (1, 1.25%), individual patient data (IPD) meta-analysis (1, 1.25%). The types of studies included in these SRs are diverse. 3 (3.75%) SRs did not limit the types of included studies. RCTs (77, 96.25%) was the most involved type. The second to the fourth were observational studies (10, 12.50%), case-control studies (7,8.75%), and quasi randomized studies (4, 5.00%).

Discussion
The PROSPERO platform opened COVID-19 retrieval channels and made a reasonable and meticulous classi cation and management of the literature so that all the documents retrieved were in line with inclusion criteria. As of July 24, 2020, 27 (33.75%) registrations have reached the expected completion time. However, only two registrations were completed on time, and the completion rate was only 7.41%. The shortest study was decided to be completed within 28 days. It can be inferred that the reasons for the low completion rate on time may be that the amount of literature in the early stage is less, it is di cult to obtain enough data, the expected completion time is too short to complete the research, and so on. This shows that in the future research, we should evaluate the feasibility of the study, reasonably plan the research progress, and strive to complete it within the expected registration time.
China's rst COVID-19 diagnosis and treatment plan (trial version third) issued on January 23, 2020 was rst written into the Chinese medicine treatment plan. Novel coronavirus pneumonia diagnosis and treatment plan (Fourth Edition) issued on January 27, 2020 recommended the Chinese herb, Huo Xiang Zheng Qi capsule, Jinhua Qinggan Granule, Lianhua Qingwen capsule, Shu Feng Jiedu Capsule, and Feng Feng Tong Sheng pill. The rst registration was on February 5, 2020, and the second was on March 20, 2020. Registration was concentrated in April when the outbreak had spread worldwide. At the same time, as of April 30, 2020, the global cumulative number of cases exceeded 3.13 million. The registered concentration months are associated with the severity of the epidemic worldwide. With the expansion and severity of the epidemic, scholars began to pay more attention to this aspect of research, making SRs to produce high-quality evidence to support clinical practice.
More representative data and more advanced data can be obtained by searching the Chinese and English databases of 77 (96.25%) studies. Only English databases were searched for the 3 (3.75%) registrations. Each study did not use a single database. The most commonly used database combination is CNKI combined with PubMed/MEDLINE (75,93.75%) and PubMed/MEDLINE combined with EMBASE (75,93.75%).
A total of 6 countries have registered COVID-19 SRs of traditional Chinese medicine, of which 76 (95.00%) SRs were undertaken by China. Only 4 (5.00%) studies were conducted by cooperation between countries. A total of 21 provinces in China have registered COVID-19 SRs of TCM, of which 7 (33.33%) have formed cooperative relations. A total of 81 institutions involved in the included SRs, but only 10(12.35%) institutions had collaborations. Sichuan Province participated in 25 SRs, Chengdu University of traditional Chinese medicine participated in 18 (31.25%) SRs, and the Chengdu University of traditional Chinese medicine hospital participated in 5 (6.25%) studies. Beijing has 13 (16.25%) SRs with the second largest number of registrations, including 6 (7.5%) from Beijing University of traditional Chinese medicine and 5 (6.25%) from Guang'anmen Hospital. In the social network of provinces, Beijing is in the central position and has more cooperative relations. In the social network of the institutions, Beijing University of traditional Chinese medicine is at the center and has cooperative relations with eight institutions. In general, there is little cooperation among provinces and institutions. Therefore, researchers in the future should strengthen more comprehensive research and carry out extensive cooperation between provinces and institutions.
The main clinical manifestations of COVID-19 are fever, dry cough, and fatigue. Laboratory examination can be found in the early detection of normal or reduced peripheral WBC, most patients with C-reactive protein and erythrocyte sedimentation rate increased. Novel coronavirus nucleic acids can be detected in specimens of nasopharyngeal swabs and other samples by RT-PCR or NGS10.
The description of intervention measures is very brief. Most of the registrations did not indicate the TCM treatment methods used. Only 17(21.25%) registrations gave the speci c drug names. The primary outcome measure used most in the included literature was clinical effective rate (45,56.25%), followed by clinical improvement time (31,38.75%) and clinical symptoms improvement (31,38.75%). The most frequently used secondary outcome measure was adverse events, which was used by 62.5% of the included literatures. Compared with the primary outcome measures of the top 20, the secondary outcome measures of the top 20 increased adverse events, safety measurements, WBC, quality of life, mechanical ventilation, blood gas analysis, blood routine, and canceled clinical e ciency, lung function, the incidence of adverse events, serum cytokine levels, nucleic acid detection, cure rate, and oxygenation index. In the top 20 primary outcome measures, clinical symptoms, laboratory tests, and signs were associated. The secondary outcome measures focused more on the quality of life, safety measures, and adverse drug events. TCM symptom score was used in both indicators, and the frequency of use was relatively low (main outcome index, 2.15%, secondary outcome index, 3.17%). This index can be considered to improve the characteristics of traditional Chinese medicine in future research. The description of primary and secondary outcome indicators in the inclusion registry was not coherent.
We conducted a comprehensive analysis of the Chinese medicine related COVID-19 SRs registered in PROSPERO using the bibliometric analysis method and presented collaborations of provinces and institutions by using visual network maps. However, this study also has some limitations. Due to the lack of registered research, the cooperation between countries and institutions is not close enough to re ect the future cooperation trend. Although we have standardized some institutions, interventions, and outcomes in different ways, there may still be bias.

Conclusions
Many COVID-19 SRs of TCM have been registered, but the completion rate was low. China was the country with the largest number of registrations, Sichuan was the province with the largest output, and Chengdu University of traditional Chinese medicine was the institution with the largest output. Collaborations between countries, provinces, and institutions were not close enough. More comprehensive and extensive collaborations between different provinces and different regions should be further strengthened to strengthen communication, share information, and obtain more representative experimental results. More attention should be paid to the de ciency of interventions and outcome measures, and the standardization of results should be strengthened.   The social network analysis of provinces Figure 3 The social network analysis of institutions