Page 99 - Read Online

Jaswant et al. One Health Implement Res 2024;4:15-37 https://dx.doi.org/10.20517/ohir.2023.61 Page 19

Screening
We reviewed the title and abstracts of all articles that met the inclusion criteria. Where detail was lacking,
the entire article was reviewed before its inclusion or exclusion was determined. A first screening phase
based on titles and abstracts was conducted independently by three independent reviewers (GJ, CB and
MM), and out-of-topic studies excluded. During this phase, studies that were not covering rabies from Asia,
Africa or Latin America were also excluded. A second screening phase based on the full texts was conducted
by GJ and CB using a standardised eligibility form. Discrepancies observed between reviewers were resolved
through discussion. Duplicate studies were removed.

Data extraction
A form designed for this review was used for data extraction [Supplementary Table 4]. The fields extracted
included: authors, year of publication, country, study aim, study design, species from which samples were
collected, numbers of samples (tested and confirmed for rabies and RABV sequences), sample type,
sequence type [Whole Genome Sequencing (WGS) or partial genome, indicating the length and section
sequenced], sequencing platform, type of phylogenetic analysis, outcome of the study, and the main study
findings, including any recommendations for control measures derived from analysis of sequencing data.

Data synthesis
The main characteristics of the studies were summarised in tabular form (as per the data extraction
[49]
proforma). Data analysis and visualisation was carried out in R (version 4.0.3) .

RESULTS
Study selection
The database search identified 1,558 publications, of which 161 were excluded as duplicates using
[50]
automation software . Manual searches identified an additional 23 relevant publications. After screening
and assessing eligibility, 220 were retained for systematic review [Figure 1].

Study Characteristics
The 220 articles generated new RABV sequences from 94 countries [Figure 2], with most from China (n =
54 publications) and Brazil (n = 19) while six undertook large-scale meta-analyses with additional
sequencing from multiple countries [51-56] . An average of two studies presenting new RABV sequences were
published per year [Figure 2], with most in 2013 (n = 16) and 2015 (n = 17). All studies generated RABV
sequences from brain tissue samples, with some on Flinders Technology Associates (FTA) cards [21,57-59] and
four including alternative sample types (nuchal biopsy, cerebrospinal fluid and salivary glands). Most
publications (n = 188) reported results from partial sequences only, using 1st generation sequencing, mostly
the N gene (n = 119). Other studies sequenced the G, P or M genes or the G-L intergenic region, and 54
were multi-gene [Figure 3]. Twenty-nine studies generated WGS, with hotspots in China (n = 10) and
Tanzania (n = 5), and nine used multiple platforms (1st and 2nd [15,52,60-63] or 1st and 3rd generation
platforms [64,65] ). In the last decade, 3rd generation sequencing (Nanopore) increased, as did sequencing
output.

Sequences were generated from 94 of 149 countries in endemic regions (61.7%). Among the endemic
regions, countries in Africa conducted the least in-country sequencing (42.8%, 14/27) whereas countries in
Asia conducted the most (76.6%, 23/30, Figure 3). In terms of sequencing output, Asia led with 6,715
sequences (381 WGS), followed by Africa with 3,757 sequences (315 WGS) then Latin America with 1,143
sequences (26 WGS). Tree-building methods used for analyses in publications included Neighbour Joining
(n = 76), Maximum Likelihood (n = 63), Bayesian Inference (n = 27) and a combination of these methods
(n = 48). The main study objectives reported were to identify circulating RABV, describe transmission