The trends of Zika virus-related research from 2014 to 2018: a systematic review and meta-analysis

Abstract

forest of Uganda in 1947, ZIKV has circulated in Africa and Asia and did not appear to cause serious disease for almost 60 years (3)(4)(5).
The epidemiology of ZIKV changed in 2007 when the rst outbreak outside of Africa or Asia was reported on Yap Island (5,6). Later, in 2013-2014, the ZIKV epidemic spread eastward in the Paci c Ocean, and eventually, in 2015, it reached the Americas and caused the rst report of autochthonous transmission of ZIKV in Brazil (7,8). During the outbreaks from 2007 to the present, reports con rmed that ZIKV infection had a strong association with microcephaly in newborns, neurological disorders in adults (such as Guillain-Barré syndrome), as well as a potential for sexual transmission. Thus, ZIKV became a major concern for public health (5,(9)(10)(11)(12).
Due to the accumulation of reports that showed the signi cant consequences of the ZIKV outbreak in the Americas, the World Health Organization (WHO) declared a Public Health Emergency of International Concern (PHEIC) on 1 st February 2016 and agreed on the urgent need to coordinate international efforts to close the knowledge gaps associated with ZIKV potential to cause birth defects, mainly microcephaly (13)(14)(15).
Since 2015, interdisciplinary investigation and intersectoral cooperation regarding ZIKV have increased rapidly (16,17). This has resulted in a better understanding of the transmission, pathogenesis, and control of the previously little-known ZIKV (17). To characterize and visualize the rapid growth of global, multidisciplinary action against ZIKV, we decided to perform a bibliometric analysis of ZIKV research from 2014 to 2018.
Bibliometrics is a powerful tool that uses statistical analysis of publications to measure the "output" of individuals/research teams, institutions, and countries, to identify national and international networks (18). Bibliometrics has been widely used in the library and information science, especially for collection management and resource access (19). To date, four bibliometric studies on ZIKV have been identi ed (5,(20)(21)(22). All these previous studies were performed up to 2016 and pointed to the importance of bibliometric study in global emergencies to provide an accurate overview of scienti c output over time and its future impact. As a methodological tool, scienti c publication databases (e.g. Pubmed, Scopus, and Web of Science, etc) have been widely used for bibliometric analysis to assess the value of published scienti c output worldwide (23). Therefore, we chose a systematic review and meta-analysis method by using scienti c publication databases to follow the trend of ZIKV research. For analysis, we have collected publications for 2014-2018 and examined the trends of ZIKV research to observe the impact of the ZIKV outbreak in 2015 on the research area during the following ve years.

Methods
This review was performed according to the PRISMA (Preferred Reporting Items for Systematic Review and Meta-Analysis) guidelines, and the PRISMA ow diagram of this study was compiled in Additional le 1 (24).

Literature search
We designed a search strategy to access the literature related to ZIKV research from 2014 to 2018. The search strings were selected by referring to previous systematic reviews on ZIKV (20,21,25). A literature search was conducted in Scopus using the search string ((TITLE-ABS-KEY (zika) OR TITLE-ABS-KEY (zikv) OR KEY (zika AND virus) OR KEY (zika AND virus AND infection)).

Eligibility criteria
This systematic review was designed to follow the trend of ZIKV research in 2014-2018 using quantitative methods. Titles and abstracts were scrutinized to determine whether the articles addressed our review objectives and ful lled the inclusion and exclusion criteria. Differences between the reviewers were resolved through discussions.
All studies reported on ZIKV research in the natural science eld.
Studies were published between 2014 and 2018.
Studies were published as research articles, peer-reviewed, written in English.

Studies were excluded:
Neither title nor abstract is presented.
Type of publication -commentaries, letters, notes, short surveys, conference papers/reviews, articles in press, erratum, and retractions.
The author's name is not available.
The publication is not relevant to ZIKV related research.
Screening of the articles All papers obtained from the search were imported to Endnote (EndNote version X9 Clarivate Analytics) for storage and subsequent screening. Initially, duplicates were removed from the database. To remove the potential for selection bias, two authors also independently completed the screening of papers. The inclusion and exclusion criteria mentioned above were used for publication selection by reviewing abstracts and titles.

Data extraction and visualization
Following the screening of titles and abstracts, eligible studies were analyzed and the following details were extracted and summarized.
To extract data from selected articles, we used R programming language (Rstudio, Version 1.1.463) with the open-source tool "Bibliometrix" which is a unique R package for quantitative research in scientometrics and bibliometrics that includes all the main bibliometric methods of analysis (26). Based on the extracted data, we performed tabulation and visualization. Brie y, we designed the tables containing speci c information of the articles which were sorted annually or numeric order (authors, title, journal source, institute, year, citation, country of the corresponding author). We assumed that the corresponding author is usually the author who provides the intellectual input and approves protocols in the study. The corresponding author's country was noted, extracted from the publications, and sorted to generate the geographical distribution and number of publications in different countries during 2014-2018.
The visual graphic is useful to present such data with easy to catch at a glance. Therefore, we chose a geographic information system (GIS) to display the distribution of the publication and number of publications in a different country (27). To generate and illustrate the GIS maps, we used Rstudio with "rworldmap" plug-in which is a package for visualizing global-scale data, concentrating on data referenced by country codes or gridded at half-degree resolution (28).
Graph data visualization was performed by using GraphPad 7.0 (CA, USA) and Adobe Illustrator CC 2017 (CA, USA).

Statistical analysis
We compared the retrieved result from 2014-2015 (low active stage) with the result from 2016-2018 (high active stage) with the unpaired t-test (parametric continuous variables). Statistical signi cance was set at 2-tailed P < 0.05. Statistical analyses were performed using GraphPad 7.0 (CA, USA).

Results
The ZIKV research eld was driven by public interest The comparison of publication numbers is a way to observe an active/inactive stage of the research area. So, we compared the number of ZIKV related publications to follow the movement of research in the last ve years.
In 2014-2015, the average number of ZIKV related publications was 23.5 and the number then increased rapidly in 2016, 2017, and 2018 (39.2, 56.5, and 58.5 times, respectively). The conceivable reason behind this rapid increment of publication is an accumulation of the public interest after the Brazilian authority publicized the rst report of autochthonous transmission of ZIKV in 2015 and the association with new clinical manifestations such as congenital Zika syndrome and Guillain-Barré syndrome. That propelled ZIKV research to a subject of international concern and led to efforts to better understand ZIKV and suggest measures to liberate the world from the ZIKV emergency declared by WHO (Fig 1).
Martinez-Pulgarin et al, (2016) previously showed that only 325 ZIKV related publications were found over six decades, inferring that research on ZIKV was at a very low level until 2015 (29). We could see that the publication number was increased tremendously in the consecutive three years (average number = 1209.3 publications/year). When comparing 2014-2015 (low active stage) vs 2016-2018 (high active stage) there was a P value = 0.0078, 95% con dence interval = 594.9 -1777, and difference between means = 1186 ± 185.7, R 2 = 0.9315. This result showed how the research on ZIKV was changed to a highly active stage after the emergence of ZIKV in the Americas (Fig 1). In addition, the declaration of a public health emergency by WHO did not only bring public interest to ZIKV research but also increment funding opportunities for researchers (described in the timeline of ZIKV in Figure 1). We could observe that the public interest in emerging diseases leads to increased funding, which then results in a surge of publications in the research eld. Also, we extracted the 20 most cited articles in an attempt to determine which articles were in uential in ZIKV research, based on the respective research eld. The result showed that mainly case reports were the most cited in 2014. In 2015, case reports, diagnostic methods, epidemiological studies, and potential sexual transmission of ZIKV were the most cited. In 2016, the articles regarding the association between ZIKV infection and new clinical manifestations (such as microcephaly, Guillain-Barré Syndrome), and the development of ZIKV animal models were highly cited. Studies of the ZIKV vaccine, ZIKV antiviral treatment, and understanding of ZIKV pathogenicity and tropism were most cited in 2017. In 2018, articles regarding viral persistence, and diagnostic approaches using advanced techniques such as CRISPR-Cas9 (Clustered regularly interspaced short palindromic repeats-CRISPR associated protein 9) and graphene, as well as vaccine development, and real-time surveillance system were most cited (Table  1). Overall, the trend of ZIKV research was shifted from epidemiological research to the development of clinical applications (e.g. advanced diagnostic techniques, antiviral treatment, vaccine) during the ve years.
The geographic distribution of ZIKV publications points to the globalization of ZIKV research.

Discussion
In this study, we retrieved data to follow the ZIKV research trends by using a bibliometric analysis of ZIKV related publications during 2014-2018. Our results highlighted the importance of gathering public interest in global health issues, and how public interest can act as a powerful catalyzer to trigger the research eld (30,31). Likewise, we found that the efforts to overcome the global emergency caused by the ZIKV outbreak in the Americas led to globalization in the research eld (32).
Although the connection between ZIKV and more severe clinical manifestations had been suggested already in 2007 and the following years. ZIKV research was still at a low activity stage until the outbreak in the Americas occurred, with microcephaly in newborn infants that alerted public attention (16, 33, 34).
For example, photographs of a mother holding her newborn baby with microcephaly, most likely caused by ZIKV infection during pregnancy, prompted sympathy and concerned authorities (35). Once the public interest was gathered, it led to global awareness and actions to combat the outbreak, and various research elds were fuelled to investigate and better understand all aspects of ZIKV.
During 2014-2015, the ZIKV research trend mainly focused on epidemiological aspects and public health, such as source and spread of ZIKV, case reports, and diagnostic methods. Then it shifted to the However, despite the progress in ZIKV research many questions remain to be addressed to continue the development of effective ZIKV countermeasures. For instance: how long-lasting is the immune response in a ZIKV exposed patient; could pre-existing immunity against ZIKV lead the antibodydependent enhancement of new ZIKV variants, or other aviviruses (such as dengue fever); what treatment options are there for the babies born with a neurological disorder caused by ZIKV infection during pregnancy; and ZIKV re-emergence could potentially occur in naïve regions, and global efforts should be directed towards surveillance, vaccines, and antiviral drug development.

Limitations
There are also some limitations to the study. Although we used Scopus for our bibliometric analysis (we believe that no scienti c literature database is considered better), there are other public and commercially available scienti c literature databases, such as PubMed. The decision to use Scopus was based on three reasons 1. The retrieved data from Scopus has compatibility to R programming language, particularly opensource R package "Bibliometrix", while the data from PubMed has not.
2. There are wide variations in citation data in each database (40,41).
Because this study was limited to articles that were published only in English, we missed articles published in Portuguese and Spanish which were native languages in most of the outbreak area.
We showed the trend of ZIKV related research based on publication number, citation number, and geographic information. However, a limitation of our study is that we could not correlate between the shifting of trends in ZIKV research and the size of research funding. It would be interesting to merge the actual ZIKV research budget of each country for 2014-2018 into our GIS (Fig 3) to investigate whether the ZIKV research eld has been driven by not only public interest but also actual nancial support to each researcher in different countries.

Conclusions
Bibliometric assessments in infectious diseases are of utmost importance, not only to correlate the needs for research in a global emergency but also, as shown here, to provide an accurate overview of scienti c output over time and its future impact (43,44). Through this study, we have provided an analysis of the Authors' contributions GYD designed and performed the experiments; analyzed, compiled, and interpreted the data; and wrote the manuscript. ME designed, interpreted, and compiled the data; wrote the manuscript; and provided funding. All authors read and approved the nal manuscript. (1) The major events in the gure were selected from "WHO -The History of Zika Virus" (30). Statistical analysis was performed by unpaired, two-tailed t-test, 2014-2015 (low active stage) vs 2016-2018 (high active stage); P value = 0.0078, 95% con dence interval = 594.9 -1777, difference between means = 1186 ± 185.7, R2 = 0.9315. P value = 0.0688, 95% con dence interval = -13. 14 -196.8