SpillOver: Viral Risk Ranking
METHODS

Because of the vast and specific knowledge-base for viruses that are primarily human-to-human spread, especially those cases where the origin of those viruses is now believed to be primarily human, we have intentionally not included them in this ranking of viruses for which much less is known. For example, influenza A would be number one in this ranking; however, because the existing wealth of data on influenzas allows more detailed analyses of risk, including strains and variants, they are not included in this ranking tool. The goal of the SpillOver tool is to identify those viruses that should be prioritized for surveillance and studied as intensively as influenzas have been. An additional summary of methods is below, and full descriptive methods are available in the associated manuscript and supplementary information published by Grange, Mazet et al.

Developed by infectious disease scientists, SpillOver: Viral Risk Ranking (aka SpillOver) explores and directly compares hundreds of viruses of wildlife-origin to identify viruses with the highest risk of zoonotic transmission from wildlife to humans (spillover) and spread in human populations.

The SpillOver website was developed as an outreach platform to increase accessibility to evidence-based risk ranking of viruses for scientists, policymakers, public health managers and communities, especially in areas with higher risk of emerging infectious disease. By combining a large number of risk factors and acquiring a broad scope of expert opinion, our approach is one of the most comprehensive assessments of zoonotic viral spillover risk. The innovative design, based on a credit report used by money lenders, is fully customizable for future developments, including updating new and existing risk factors and data sets.

Currently, the application is designed for wildlife-origin viruses with host species in the classes Mammalia, Reptilia, Amphibia and Aves, and data can auto-populate for 26 virus families of concern to human health (Anelloviridae, Arenaviridae, Bornaviridae, Coronaviridae, Filoviridae, Flaviviridae, Hepadnaviridae, Hepeviridae, Orthomyxoviridae, Paramyxoviridae, Picornaviridae, Reoviridae, Retroviridae, Rhabdoviridae, Togaviridae, Astroviridae, Caliciviridae, Picobirnaviridae, Arteriviridae, Herpesviridae, Poxviridae, Parvoviridae, Adenoviridae, Papillomaviridae, Polyomaviridae, Nairoviridae). It is not intended for viruses that only infect humans, vector-borne viruses or viruses of domestic species (animals that have been domesticated to be kept as pets or used as food) origin.

A selection of risk factors determined to contribute to the risk of viral spillover from animals to humans were identified by the SpillOver scientists through an extensive literature review and research. Risk factors were categorized as either belonging to the Host (animal infected by the virus), the Virus (the infective agent), or the Environment (location in which the host animal was found). A panel of international experts in the field of virology, epidemiology, ecology, molecular biology, public health, veterinary and human medicine, and One Health reviewed the risk factors thought to contribute to the risk of zoonotic virus spillover and spread. Each expert assigned a Spillover Risk (options = high (3), medium (2), low (1), or not relevant for spillover(0)), alongside their self-identified Level of Expertise (options = novice (1), competent (2), proficient (4), expert (8), master (16)) for each risk factor.

To account for variation and uncertainty in expert opinion, we calculated a weighted average score for each risk factor from the sum of expert responses to Spillover Risk, accounting for the Level of Expertise of each expert within each subject. If a participant did not declare a Spillover Risk or Level of Expertise, a value of 0 and 8 was assigned respectively. A weighted average score (Risk Factor Contribution – aka ‘Risk Factor Influence’ in the manuscript) was then calculated for each risk factor from the sum of expert responses (i) using the following formula:

$$Risk\ Factor\ Contribution\ (0-3) = {\sum Level\ of\ Risk\ (0-3)x\ Level\ of\ Expertise\ (1-16)i \over \sum Level\ of\ Expertise\ (1-16)i}$$

Risk factors with a Risk Factor Contribution score of two or above (medium to high) were selected for inclusion in the analysis. Two additional risk factors, virus segmentation and virus envelope, with scores lower than two were included due to scientific evidence supporting their role in human-to-human transmission of viruses. Each risk factor has multiple categorical Risk Level options (e.g. DNA or RNA for “Virus Genome” risk factor). These options were assigned a Risk Level Score on a scale between 0 and 5, with 5 being the riskiest, based on scientific evidence and expert opinion. For example, the Risk Levels for the risk factor “Livestock density” were High (5), Medium (3), Low (1), None (0), or Unknown (2.5). A central Risk Level Score of 2.5 was assigned to all Risk Levels categorized as Unknown. For viruses with more than one detection, the assessment uses the principle of the worst-case scenario, i.e. the most severe possible outcome is assigned using the highest Risk Level Score for that virus. For example, a virus may have been detected in two locations, one which had high livestock densities and the other low livestock density. The assigned Risk Level for this virus’ risk factor would be ‘High livestock density’ or 5. The highest risk score for each risk factor is then combined with the expert assigned weight for that risk factor. Using the livestock density example, the calculation would be:

$$Virus\ Data\ Risk\ = {Risk\ Factor\ Influence\ x\ Risk\ Level\ Score\over 3}\ = {(2.174603175)\ x\ (5) \over 3} = 3.62433862$$

For each virus, a relative Spillover Risk Score is calculated as the sum of the Virus Data Risk scores. Finally, virus ranking positions are assigned in descending order according to each virus’ Spillover Risk Score, thus relative spillover potential.

The Ranking Comparison page allows users to compare and contrast the relative risk of ranked viruses. The Quick View provides a selection of information regarding the risk scores, ranking position, key virus and host information, as well as the number of records and data accuracy. Users have the opportunity to search and filter viruses on a selection of attributes to customize the ranking comparison. SpillOver produces a spillover risk report for each virus detailing each of the risk factors, its Risk Factor Contribution, and the Virus Data Risk score. The virus report also estimates the relative risk of host, viral, and environmental factors that contribute to a virus’ overall spillover risk score. The risk report produces a virus distribution map based on the countries of detection contained within the database. Virus distribution in animals is calculated based on publicly available evidence of virus detection in wildlife. Users can privately, in their own browser, rank a new virus or add to existing virus reports using the Rank your virus tool. For each report of virus detection, the following information is collected at a minimum: virus taxonomy, virus sequence or reference number, host species and sample type, the estimated timeline of host species divergence from humans, host location, and the primary high-risk disease transmission interface where the host was sampled. Metadata for other host, virus, and environmental factors are auto-populated based on data provided by the user. This metadata is derived from 19 publicly-available resources on: the host (IUCN Red List¹, Birdlife International², Catalogue of Life³, TimeTree⁴, diet database⁵), the virus (International Committee on Taxonomy of Viruses (ICTV)⁶, ViralZone database⁷, reference virology textbooks^8-10, published human/zoonotic virus databases^7,11-13, the NCBI nucleotide database), and the environment (LADA land-use systems¹⁴, NASA Sedac Gridded population of the World UN Adjusted population density v4 2015¹⁵, Global Grid of Probabilities of Urban Expansion to 2030¹⁶, adapted HYDE land conversion 2005^17,18, Global Forest Change 2017¹⁹).

When appropriate, users can edit risk factor answers to improve accuracy of information collated from external resources. Users are encouraged to review and submit their virus ranking data to the database. New submissions of virus detection data are subject to quality control and verification by scientific administrators prior to inclusion in the database. A discussion forum is available for users to provide feedback, comments, and suggestions to improve SpillOver. Updates and notifications will be posted by the SpillOver administrators when appropriate.

The authors claim no responsibility for inaccuracies contained in the SpillOver database. To report errors, please leave a comment on the discussion forum or email spillover@ucdavis.edu.

1.The IUCN Red List of Threatened Species. 2017. (Accessed August, 2017, at http://www.iucnredlist.org.)
2.Bird species distribution maps of the world. In: World BIaHotBot, ed. 6.0 ed2016.
3.Roskov Y, Abucay L, Orrell T, et al. Species 2000 & ITIS Catalogue of Life, 2018 Annual Checklist. Naturalis, Leiden, the Netherlands: Species 2000; 2018.
4.Kumar S, Stecher G, Suleski M, Hedges SB. TimeTree: A Resource for Timelines, Timetrees, and Divergence Times. Mol Biol Evol 2017;34:1812-9.
5.Wilman H, Belmaker J, Simpson J, de la Rosa C, Rivadeneira MM, Jetz W. EltonTraits 1.0: Species-level foraging attributes of the world's birds and mammals. Ecology 2014;95:2027-.
6.Lefkowitz EJ, Adams MJ, Davison AJ, Siddell SG, Simmonds P. Virus Taxonomy: The Classification and Nomenclature of Viruses2017.
7.Hulo C, de Castro E, Masson P, et al. ViralZone: a knowledge resource to understand virus diversity. Nucleic Acids Res 2011;39:D576-82.
8.Dubovi EJ, Maclachlan NJ, eds. Fenner's Veterinary Virology. 5 ed. Boston: Academic Press; 2016.
9.Tidona C, Darai G. The Springer Index of Viruses. 2 ed. New York, USA: Springer; 2011.
10.Burrell CJ, Howard CR, Murphy FA. Fenner and White's Medical Virology. Fifth Edition ed: Academic Press; 2017.
11.Woolhouse MEJ, Brierley L. Epidemiological characteristics of human-infective RNA viruses. Sci Data 2018;5:180017.
12.Geoghegan JL, Senior AM, Di Giallonardo F, Holmes EC. Virological factors that increase the transmissibility of emerging human viruses. Proc Natl Acad Sci U S A 2016;113:4170-5.
13.Johnson CK, Hitchens PL, Evans TS, et al. Spillover and pandemic properties of zoonotic viruses with high host plasticity. Sci Rep-Uk 2015;5.
14.Nachtergaele F, Pertri M. Mapping land use systems at global and regional scales for land degradation assessment analysis. Rome, Italy: FAO; 2011.
15.Center for International Earth Science Information Network - CIESIN - Columbia University. Gridded Population of the World, Version 4 (GPWv4): Population Density Adjusted to Match 2015 Revision UN WPP Country Totals. Palisades, NY: NASA Socioeconomic Data and Applications Center (SEDAC); 2016.
16.Seto K, Güneralp B, Hutyra LR. Global Grid of Probabilities of Urban Expansion to 2030. Palisades, NY: NASA Socioeconomic Data and Applications Center (SEDAC); 2015.
17.Kees KG, Arthur B, Gerard vD, Martine dV. The HYDE 3.1 spatially explicit database of human-induced global land-use change over the past 12,000 years. Global Ecology and Biogeography 2011;20:73-86.
18.Klein Goldewijk K, Beusen A, Janssen P. Long-term dynamic modeling of global population and built-up area in a spatially explicit way: HYDE 3.1. The Holocene 2010;20:565-73.
19.Hansen MC, Potapov PV, Moore R, et al. High-Resolution Global Maps of 21st-Century Forest Cover Change. Science 2013;342:850-3.