UG ETD Template

Disease Maps and Geographic Map Projections - A Simulation Study and

Scoping Review

Inthuja Selvaratnam

A Thesis presented to The University of Guelph

In partial fulfilment of requirements for the degree of Master of Science in Population Medicine

Guelph, Ontario, Canada

ABSTRACT

DISEASE MAPS AND GEOGRAPHIC MAP PROJECTIONS - A SIMULATION STUDY

AND SCOPING REVIEW

Inthuja Selvaratnam Advisor(s): University of Guelph, 2020 Dr. Olaf Berke

Disease mapping is a common method in epidemiologic research. Although disease maps are ubiquitous and regarded as informative communication tools in public health, no widely accepted style guides and reporting guidelines have been established. This thesis presents a scoping review to broadly characterize of studies using disease maps and selected reporting practices of published disease maps in the scientific literature to identify any gaps and inconsistencies. Map projections were infrequently reported for disease maps. Varying the projection of disease maps and related spatial data can affect spatial statistical analyses and specifically the results of disease cluster detection methods. This is demonstrated here for spatial point pattern data and related point maps. Findings from this thesis support the development of a reporting guideline for disease maps. Such a reporting guideline can benefit and advance the research community by improving reproducibility and minimizing misinterpretations.

iii

ACKNOWLEDGEMENTS

First and foremost, I would like to thank my advisor Dr. Olaf Berke and my committee member

Dr. Jan Sargeant for their guidance and support. I appreciated your time and insight throughout the research and writing processes, especially amidst a pandemic. Thank you, Olaf, for the opportunity to do research and for supporting my project proposal, as well as providing guidance on conference presentations over the years.

Special thanks to my review team for agreeing to collaborate for work conducted in Chapter 3:

Abhinand Thaivalappil, Jamie Imada, Monica Vythilingam, Andrew Beardsall, Gillian

Hachborn, Mohamed Ugas, and Russell Forrest. iv

STATEMENT OF WORK

Inthuja Selvaratnam, under the guidance of Dr. Olaf Berke, developed the study objectives included in this thesis. The overall study design for Chapter 3 was conceived and designed by Inthuja Selvaratnam. Inthuja Selvaratnam contributed to the development of the study, review protocol, methodology, data collection, analysis, interpretation of results, and was the principal author of all chapters.

Dr. Olaf Berke provided assistance on study objectives, data analysis, interpretation as well as critical feedback and edits on all the chapters. The overall study design and methodology for Chapter 2 was conceived and designed by Dr. Olaf Berke. The base code for the simulation study was provided by Dr. Olaf Berke.

Dr. Jan Sargeant provided guidance for developing the methods in Chapter 3 and provided critical feedback and edits on all chapters.

Abhinand Thaivalappil, Jamie Imada, Monica Vythilingam, Andrew Beardsall, Gillian Hachborn, Mohamed Ugas, and Russell Forrest reviewed articles and extracted data for the scoping review conducted in Chapter 3. Monica Vythilingam assisted with analyzing data for the zoonotic diseases study characteristic in Chapter 3. Andrew Beardsall assisted with reference and study flow management in Chapter 3.

Inthuja Selvaratnam was responsible for the appropriate changes to each manuscript following review by the advisory committee and examination committee members: Drs. David Pearl, Jan Sargeant, and Olaf Berke.

TABLE OF CONTENTS

Abstract ...... ii

Acknowledgements ...... iii

Statement of Work ...... iv

Table of Contents ...... v

List of Tables ...... viii

List of Figures ...... ix

List of Abbreviations ...... x

List of Appendices ...... xi

Chapter 1: Introduction, Literature Review, Thesis Rationale, and Objectives ...... 1

1.1 Introduction and Background ...... 1

1.2 Disease Maps and Disease Mapping ...... 2

1.3 Mapping Health Data ...... 4

1.4 Cartographic Design and Features ...... 6

1.5 Spatial Statistics ...... 8

1.5.1. Spatial Autocorrelation ...... 9

1.5.2. Spatial Cluster Detection and Analysis ...... 11

1.5.3. Spatial Regression Models ...... 12

1.6 Reporting Guidelines...... 13

1.7 Thesis Rationale and Objectives ...... 15

1.8 References ...... 17

1.9 Figures ...... 21

Chapter 2: Effect of Map Projection on Spatial Point Data Analyses for Disease Cluster Detection and Clustering...... 26 vi

2.0 Abstract ...... 26

2.1 Introduction ...... 28

2.2 Methods ...... 32

2.2.1. Study Area ...... 32

2.2.2. Data Simulation ...... 32

2.2.3. Statistical Analysis ...... 33

2.3 Results ...... 35

2.3.1. Data Simulation ...... 35

2.3.2. Statistical Analysis ...... 35

2.4 Discussion ...... 37

2.4.1. Strengths and Limitations ...... 39

2.5 Conclusion ...... 40

2.6 References ...... 41

2.7 Figures ...... 43

2.8 Tables ...... 45

Chapter 3: Characteristics of Disease Maps of Zoonoses: A Scoping Review ...... 46

3.0 Abstract ...... 46

3.1 Introduction ...... 48

3.2 Methods ...... 54

3.2.1. Search Strategy ...... 55

3.2.2. Eligibility Criteria ...... 55

3.2.3. Study Selection ...... 56

3.2.4. Data Items and Extraction ...... 56

3.2.5. Study Characteristics ...... 57

3.2.6. Map Characteristics ...... 58 vii

3.2.7. Geospatial Methods and Analyses ...... 60

3.2.8. Geospatial Biases and Limitations ...... 61

3.2.9. Data Synthesis ...... 63

3.3 Results ...... 64

3.3.1. Study Selection and Data Extraction ...... 64

3.3.2. Study Characteristics ...... 64

3.3.3. Map Characteristics ...... 65

3.3.4. Geospatial Analyses, Biases, and Limitations ...... 65

3.4 Discussion ...... 67

3.4.1. Strengths and Limitations ...... 71

3.5 Conclusion ...... 73

3.6 References ...... 74

3.7 Figures ...... 78

3.8 Tables ...... 81

Chapter 4: Discussion and Conclusions ...... 85

4.0 Discussion and Summary of Key Findings ...... 85

4.1 Strengths and Limitations...... 88

4.2 Conclusions ...... 90

4.3 References ...... 91

Appendices ...... 93

viii

LIST OF TABLES

Table 2.1. Results from the Cuzick-Edward’s test for disease clustering (i.e. p-values) under the 5 map projections: Lambert Equal Area Conic, UTM 17North, Robinson, Mercator, and Albers ...... 45

Table 2.2. Results from the circular spatial scan test for the simulated case-control point data under 5 map projections ...... 45

Table 3.1 Search string inputted in Web of Science to identify published disease maps of zoonoses ...... 81

Table 3.2 Study characteristics of 302 included studies with zoonotic disease maps published in 2017-2018 ...... 82

Table 3.3 Disease map characteristics of 505 maps of human, animal, or vector zoonotic disease reported from 302 articles published in 2017-2018 ...... 83

Table 3.4 Categories of study author reported geospatial methods and analyses ...... 84

Table 3.5 Themes of study author reported geospatial biases and limitations ...... 84

LIST OF FIGURES

Figure 1-1. Showing 2,298,589 records from August 2020 Web of Science topic search string . 22

Figure 1-2. Diagram illustrating the different possible data layers that may be involved in a map making process for disease maps...... 23

Figure 1-3. Commonly used map types representing area health data: a) Choropleth Map b) Proportional Symbol c) Point Map d) Isopleth Map...... 24

Figure 1-4 Map Projection surfaces ...... 25

Figure 2-1 Simulated clusters: Cluster 1 with 22 controls and 13 cases (x=850km, y=2100km, radius=50km) in blue. Cluster 2 with 5 cases and 1 control (x=775km, y=2200km, radius=20km) in green...... 43

Figure 2-2: Spatial scan test for the same simulated data under different projections: a) Lambert Equal Area Conic, b) UTM 17 North, c) Robinson, d) Albers, e) Mercator ...... 44

Figure 3-1. PRISMA Study Flow diagram ...... 78

Figure 3-2. Stacked bar chart of zoonotic diseases presented in included studies (n=302) by pathogen type and region mapped in the disease maps ...... 79

Figure 3-3. The included map types (n=505) presented by the data format reported by authors in map legends or article text...... 80

LIST OF ABBREVIATIONS

CDC Centers for Disease Control and Preventions

GIS Geographical Information System

iKT Integrated Knowledge Translation

KT Knowledge Translation

KU Knowledge User

K-NN K-Nearest Neighbour

MAUP Modifiable Areal Unit Problem

MC Monte Carlo

RR Relative Risk

SMR Standardized Mortality Ratio

LIST OF APPENDICES

Appendix 1. Projection Parameters

Appendix 2. Study Protocol

Appendix 3. Search Strategy

Appendix 4. Zoonotic Diseases

Appendix 5. Relevance Screening and Extraction Forms

Appendix 6. Included Studies (n=302)

Appendix 7. Additional Figures

Chapter 1: Introduction, Literature Review, Thesis Rationale, and Objectives

1.1 Introduction and Background

The question of “where” is central to the study and practice of epidemiology and public health. Epidemiological investigations ranging from infectious diseases (Parpia et al., 2020), to chronic conditions like obesity (Osorio et al., 2020) highlight the importance of geographic and spatial (“geospatial”) factors concerning health events in a population.

Detailed visual information on the spatial distribution of disease, especially for unique spatially heterogeneous disease distributions that are specific to a disease, provide important insights for disease prevention and control programs (Wardrop et al., 2014). With recent advances in Geographic Information Systems (GIS) technology, and increased access to tools for management, synthesis, and display of spatial data, map-making processes and production are better enabled (Cromley & McLafferty, 2011; Rushton, 2003; Waller & Gotway, 2004).

Growing access also empowers public and citizen participation in the creation, evaluation, and analysis of spatial data and visualizations (Cromley & McLafferty, 2011). A search for geographic distribution map of disease or health in Web of Science (i.e. “health OR disease

AND geographic AND distribution AND map”) suggests there is a growing use for maps in health research (Figure 1.1).

However, despite their widespread use in epidemiological research and public health practice, there are no widely agreed upon and established guidelines for the reporting of disease

maps. A brief introduction is provided to the following areas: disease maps, spatial statistics, and reporting guidelines.

1.2 Disease Maps and Disease Mapping

Maps have been used for centuries to study the geographic distribution of disease and health in populations. Historically, geographically referenced health data have been used as the basis for many observational studies in epidemiology (Lawson, 2013). For example, John Snow in 1855 used mapping to show that cases of cholera resided in the vicinity of the Broad Street pump, acknowledging the difference between Euclidean distance proximity to the pump and real travel time in the neighborhood due to circuitous roads (Koch, 2017). In his analysis, Snow included the city streets, breweries, workhouses, and dockyards to eventually link the geographic extent of cholera deaths with the water sources within the urban environment (Koch, 2017).

Snow called this a “diagram of the topography of the outbreak” (Koch, 2017). His analysis led to the removal of the Broad street pump handle, which also coincided with a decline in the outbreak’s intensity (Koch, 2017). Over time, modern computing has made tremendous strides for mapping health outcomes. Today, disease mapping is not just mapping data from field observational studies but has widened its scope and also includes large clinical and registry dataset applications (e.g., survival time data), and thus disease mapping has wide applications

(Lawson, 2013; Rushton, 2003).

Maps are effective and powerful communication tools. Disease maps allow us to visualize the spatial distribution of health outcomes in a population and can reveal spatial patterns that are otherwise not recognized from tabulated data (Cromley & McLafferty, 2011).

This is especially useful for large amounts of data, where visual maps can be significant for 2

engaging policymakers and the public in planning and evaluation of health programs (Green,

2015).

Disease maps come in diverse formats and are used in a variety of applications (Bithell,

2000). The term “disease map” is not clearly defined in the literature and is used interchangeably with other terms such as health map and risk map. In this thesis, “disease map” will refer to any visualization of health data (not solely restricted to disease outcomes) that is geographically referenced within a geographic boundary. Geospatial health data, are labeled with spatial tags.

These can include regional identifiers, or longitude and latitude degrees, or Cartesian coordinates, or postal codes. The data are then mapped to a geographic boundary that will recognize the “spatial tag”.

“Disease mapping” refers to the practice of visualizing spatial variation in disease occurrence, which can help to generate hypotheses about disease causation in populations.

Disease maps may also be used at the knowledge dissemination stage of studies as a reporting tool to communicate research results. Disease maps can be generated with the help of

Geographic Information System (GIS) software and programs such as R Studio. A review of GIS use for public health and health promotion identified four predominant themes of applications: a) disease surveillance, b) risk analysis, c) health access and planning and d) community health profiling (Nykiforuk & Flaman, 2011). Under spatial epidemiological lenses, disease mapping also refers to the spatial statistics and analytical methods employed to generate visual maps.

These methods can include interpolation and regression methods, as well as global and local measures for cluster detection. As such, disease maps are also important tools to support data analysis and inferences.

1.3 Mapping Health Data

Disease maps are subjective, where inherent choices in their design and data mapped are important aspects for interpretation and inferences. Geospatial data come in different formats and can be mapped in various ways. Two basic forms of geospatial data are: point data, and areal or regional data. Point data refer to data where each data item has an exact spatial position associated with it (e.g. geographic coordinates), compared to regional data where observations are aggregated over a geographic area (e.g. public health unit, county, country) (Waller &

Gotway, 2004). Spatial data can be presented on disease maps in three different ways: 1) observed (raw) data, 2) smoothed data, 3) residual data. Smoothing data refers to a process in which “strength is borrowed” across small areas to improve local estimates and smooth extreme rates resulting from small local sample sizes (Berke, 2004). Smoothing methods include

Empirical Bayes smoothing for areal data, kriging methods for point data, and further interpolation-based methods, including splines. Residual data (i.e. from regression estimation) are also mapped to understand spatial variation in the residual process, i.e. the spatial variation in health data adjusted for known covariate effects (Dohoo et al., 2012).

A map can have multiple data sources or “layers” that are combined to produce a final map product as shown in Figure 1.2, for example a map predicting disease risk can use both health event data (i.e. the primary outcome) and covariate data (i.e. the putative risk factors). It is important to consider all data sources and the systematic errors that come with data collection and acquisition for each data source, and any additional errors that can arise during the map making process. For example, the modifiable areal unit problem (MAUP) (Cromley &

McLafferty, 2011; Waller & Gotway, 2004). The MAUP arises when the results of one analysis are different from another analysis for the same original data albeit for different aggregation schemes (e.g. a choropleth map of cancer cases by county level will look different from a map showing the same data by census tract and thus may lead to different inferences) (Cromley &

McLafferty, 2011). The MAUP is apparent in two forms: the scale effect and the zone effect.

The choice of areal scale is important as relationships between variables at one scale may be distorted when viewed from another scale, depending on the number and sizes of areal units used to describe scale. The zone effect occurs when there is variability as a result of alternative shape formations (boundaries) of the areal units at the same or similar scale. Furthermore, making inferences from maps using aggregated data to a lower level of aggregation can result in ecological biases (Lawson, 2013). Spatial sampling of populations can also give rise to selection biases and limitations of precision. One common limitation is the small numbers problem, where rates based on small populations tend to be less reliable and more variable compared to rates based on large populations (Beyer et al., 2012).

The temporal nature of health event data can also impact the applications and purposes of disease maps. Consider predictive disease maps for communicating risk: the predictive risk map can be based on observed data (retrospective), or on predicted data (prospective). Map outputs based on time-series data input in conjunction with spatial data inputs are referred to as spatiotemporal maps.

1.4 Cartographic Design and Features

The statistician George Box stated: “All models are wrong, but some are useful,” acknowledging that a model is a simplification of reality and can never be the “truth” (Box,

1976, 1979). It is worth recognizing that maps are also models of reality. Maps are a 2- dimensional representation of a 3-dimensional world, and it is impossible to create an entirely accurate map. The goal of map-making is to minimize inaccuracies as much as possible.

Cartographers use map design conventions and scientific research on human cognitive processes to communicate spatial information, which facilitates the exploration and insight of geographic phenomena (Centers for Disease Control and Prevention, 2012).

Maps that illustrate the geographic distribution of a single phenomenon, such as mortality rates or disease diffusion, are referred to as thematic maps (Centers for Disease Control and

Prevention, 2012). Commonly used thematic map types to present health data include: choropleth maps, dot/point maps, proportional symbol maps, and isopleth maps shown in Figure

1.3. Choropleth maps show geographic areas shaded in varying colours or intensities or patterns to represent the proportions or magnitudes of outcome data (Cromley & McLafferty, 2011).

Choropleth maps are a viable option when the availability of disaggregated individual data is lacking (Cromley & McLafferty, 2011). Choropleths maps also are preferred to address concerns of privacy and confidentiality of health data by not representing precise geographical locations

(Cromley & McLafferty, 2011). Point maps use point symbols within the study area to depict spatial locations of health events (Cromley & McLafferty, 2011). Isopleth maps depict smooth continuous data by using isolines or colour shades that connect points of same numerical values

(Centers for Disease Control and Prevention, 2012). Proportional symbol maps use symbols 6

mapped with their size proportional to the number of events or magnitude of outcomes (Cromley

& McLafferty, 2011).

Elementary reporting items across different map types include (Centers for Disease

Control and Prevention, 2012; Cromley & McLafferty, 2011):

1) the map scale: describes map distance in relation to earth distance,

2) data source: describes source, and date of mapped data,

3) legend: defines map symbols,

4) north arrow/compass: describes the map orientation,

5) map projection: describes how earth is projected onto the map

Map projections are used to transform the 3-dimensional earth surface onto a 2- dimensional plane, which distorts true distances, directions, and geographic shapes (Waller &

Gotway, 2004). To create a projection, a datum and a coordinate system overlay are used.

Datums are based on different ellipsoids that model the shape of the earth. WGS 1984 is the most widely used and recently developed datum (Centers for Disease Control and Prevention, 2012).

Coordinate system overlays are based on the way the datum ellipsoid is plotted onto a two- dimensional planar surface. Figure 1.4 shows three different developmental surfaces used to develop basic map projections. One of the most common coordinate systems is the Universal

Transverse Mercator projection. Depending on the intended use of the map, the choice of projection will matter. For example, two maps referenced to different datums and projections can show distances between the same point locations, but the distances can differ substantially

(Waller & Gotway, 2004). Map projections are often characterized as either equal area (favours preserving area), equidistant (favours preserving distance), or conformal (favours preserving shape). The choice and use of projection affect statistical data analyses, including resulting p- values and confidence intervals as will be presented in Chapter 2.

The scale of a map represents the relationship between a distance on the map and the corresponding distance on the ground and can help the map reader to visually measure distances

(Waller & Gotway, 2004). (Cromley & McLafferty, 2011). A scale bar can be represented in text

(e.g. “1 cm corresponds to 50 km”) or visually as a scale bar, or as a fraction (e.g. 1:25,000).

Depending on the projection used, the map scale varies throughout a map, and this may be more prominent for small-scale maps (Centers for Disease Control and Prevention, 2012).

In summary, map projections are important reporting items as they determine how a study area is visualized (including how distances are distorted and stretched) and scale bars are important to interpret distances.

1.5 Spatial Statistics

In addition to disease maps being used for descriptive and exploratory purposes of identifying and representing spatial patterns visually, disease maps are also used analytically to display the results of spatial statistical analyses, including spatial regression and cluster analyses

(Pfeiffer & Stevens, 2015; Robertson & Nelson, 2014; Rushton, 2003; Waller & Gotway, 2004).

In these instances, the choice and use of projection affects spatial statistics (Centers for Disease

Control and Prevention, 2012; Waller & Gotway, 2004), i.e., resulting tests and estimates and their related p-values and confidence intervals. The spatial relation between two points in space,

the distance, is central to spatial statistical analyses. This is further discussed for various statistics below (e.g., for the semivariogram, for the neighbourhood structure for Moran’s I, and for the scanning window of the spatial scan test).

1.5.1. Spatial Autocorrelation

Spatial data are characterized by spatial dependence or clustering (Waller & Gotway,

2004). Tobler’s first law of geography states that “everything is related to everything else, but near things are more related than distant things” (Tobler, 1970). Spatial statistics is mainly concerned with the modeling and analysis of this spatial dependence in a geospatial dataset

(Waller & Gotway, 2004). Several analytic methods can be used to measure spatial autocorrelation, such as Moran’s I statistic and provide important insights on population health outcomes. For example, positive spatial autocorrelation will be high when incidence values from regions close together in space are more similar than incidence values from regions further apart, which is often the case with infectious diseases (Wardrop et al., 2014). For aggregated data, spatial autocorrelation can be measured using the Moran I statistic (Moran, 1950) using the equation below:

푛 푛 푛 ∑푖=1 ∑푗=1 푤푖푗 (푦푖−푦̅)(푦푗−푦̅) 퐼 = 푛 2 푛 푛 Eq 1. ∑푖=1(푦−푦̅) (∑푖=1 ∑푗=1 푤푖푗 )

Where: n= number of spatial units indexed (observations) by i and j, yi and yj = variable of interest y and the respective values 9

푦̅ = the global mean wij= the spatial proximity matrix

The spatial proximity matrix (wij) in Moran’s I expresses the spatial arrangement, and is based on the neighbourhood or distance between locations (Dohoo et al., 2012). The proximity matrix is used to attribute weights to pairs of values. When neighbouring values are similar, I will be positive and when neighbouring values are different, I will be negative. A similarity index is expressed between local values of y and their neighbours, based on their deviation from a global mean (Dohoo et al., 2012).

Other methods for global measures of spatial autocorrelation or clustering include:

Geary’s C, Cuzick-Edward’s test, and the semivariogram. A semivariogram can be used to visualize spatial autocorrelation (Dohoo et al., 2012). An empirical semivariogram can show the covariation in attribute values between point observations separated at different distances

(Dohoo et al., 2012). If there is no spatial autocorrelation, the semivariogram curve will be an approximal horizontal line.

Measures of spatial autocorrelation utilize the geospatial distance between data points and thus are affected by the map projection used for a mapped dataset. Furthermore, not accounting for spatial dependence or autocorrelation could also present itself as a bias or limitation in studies, particularly for spatial regression model studies, ultimately leading to the misinterpretation of the relationships between observations and covariates (Wardrop et al.,

2014). Neglecting autocorrelation in spatial data analysis violates the underlying assumption of statistical independence in (ordinary) statistical methods and generally results in inaccurate

models, biased regression parameters, underestimated standard errors, falsely narrow confidence intervals, and overestimation of the significance of covariates (Cromley & McLafferty, 2011;

Wardrop et al., 2014).

1.5.2. Spatial Cluster Detection and Analysis

A spatial disease cluster is recognized as an unusually high number of cases that occur close together in space. Cluster analysis uses statistical methods to test the likelihood that an observed spatial or spatio-temporal pattern is a result of chance variation (Pfeiffer & Stevens, 2015).

Spatial cluster detection methods include spatial scan statistics, which are based on a comparison of the risk of disease within a window, usually circular, to that outside the window. The computing algorithm for the circular spatial scan test varies the radius of the windows as defined by the user’s upper bound and scans them across the study area (Dohoo et al., 2012). The radius of the window is changed continuously between 0 and a set upper limit, taking on an

(potentially) infinite number of distinct circles, each with different location and size, and the potential of being a cluster (Kulldorff et al., 2003). Other methods for cluster detection include:

LISA/Local Moran’s I, and local Getis-Ord G. Space-time clusters can also be detected during particular time intervals and particular places using space-time scan tests among others (Cromley

& McLafferty, 2011). Spatial scan statistics are widely used in disease surveillance and offer important timely information especially for rapidly evolving disease outbreak situations like

SARS-CoV-2 (Greene et al., 2020).

1.5.3. Spatial Regression Models

Regression models are used to quantify the cause and effect relationship between a dependent and one or more independent variables (Dohoo et al., 2012). Standard regression models (e.g. ordinary least squares for continuous data) follow assumptions that outcome data are independent of each other, their location, and their neighboring units, and that the residuals are normally distributed with constant variance (Kirby et al., 2017). Identifying spatial autocorrelation in the residuals can indicate that the above assumptions are violated (Kirby et al.,

2017). Spatial effects can be incorporated through spatial regression models or the geographically weighted regression (GWR) technique (Kirby et al., 2017). Spatial regression models are used to understand how independent variables affect the spatial distribution of a dependent variable (Dohoo et al., 2012; Pfeiffer & Stevens, 2015).

The diverse range of regression techniques and methods available, in addition to powerful computation, allow for creative and meaningful modelling of disease. It is important to note that many choices in regression model building are considered. Additionally, choices are also made for visualizing model outputs on a map (e.g., data aggregation scheme), and these choices can affect the inferences made from small or local areas on maps. Stating the choices made in spatial data modelling, and the choices made for their respective map visualizations can guide the reader and ensure transparency.

1.6 Reporting Guidelines

Evidence from systematic literature reviews and meta-analyses has identified shortcomings of reported research (Sargeant et al., 2019; Simera et al. 2010). To encourage standard and completeness of reporting around study design and outcomes, a series of reporting guidelines have been developed for various types of study designs (Moher et al., 2010). A reporting guideline is “a checklist, flow diagram, or explicit text to guide authors in reporting a specific type of research, developed using explicit methodology” (Moher et al., 2010). Some prominent guidelines for studies in human populations are the CONSORT statement (Moher et al., 2001), for reporting randomized clinical trials (RCTs), the PRISMA statement for reporting systematic reviews (Moher et al., 2009), and the STROBE statement for reporting observational studies (Vandenbroucke et al., 2014; von Elm et al., 2007), available on the Equator Network

(https://www.equator-network.org/) (Simera et al., 2010). Some prominent guidelines for studies in animal populations available on the Meridian network (https://meridian.cvm.iastate.edu/) are: the ARRIVE statement for reporting animal research (Kilkenny et al., 2010), the STROBE-Vet

Statement for reporting veterinary observational studies (O’Connor et al., 2016; Sargeant et al.,

2016), and the REFLECT statement for reporting randomized controlled trials for livestock and food safety (O’Connor et al., 2010; Sargeant et al., 2010).

Despite the widespread use of disease maps in academic research, journal publications, and use as decision support tools by public health agencies, to date no reporting guideline for disease maps accepted by the epidemiological research community exists (search conducted

August 2020 on the Equator Network website). While a plethora of publications and educational material exist for making disease maps, such as the “Cartographic Guidelines for Public Health”

by the Center for Disease Control and Prevention (CDC) for maps produced at the CDC (Centers for Disease Control and Prevention, 2012), a guideline that is widely established and agreed upon does not exist. Authors of spatial methods reviews have pointed out that standardizing reporting and a guideline for spatial methods in epidemiology can be useful (Robertson &

Nelson, 2014; Smith et al., 2015). Without comprehensive reporting of map elements and design decisions, geospatial statistics and analyses are not reproducible and may lead to misinterpretations and misinformation. Reproducibility allows independent investigators to subject the original data to their own analyses and interpretations (Peng et al., 2006). Scientific reproducibility calls for as much information as possible about all details of the data analysis process (Peng et al., 2006). The findings from disease maps may not be reproducible when map projections are not provided. Reproducibility of epidemiologic research studies permits: 1) findings to be verified, 2) alternative analyses of the same data, 3) disputing uninformed criticisms, and 4) expediting the exchange of ideas among investigators (Peng et al., 2006).

Reporting guidelines can also increase critical thinking about design and analysis issues during map creation, leading to potentially higher quality maps. While reporting guidelines are intended for aiding in clarity for publication, they are not to be used to assess quality of studies

(Vandenbroucke, 2009). The heterogeneous nature of disease map visualizations, spatial data types, map purposes, and contexts make it difficult to articulate standard reporting items for disease maps. A set of standard reporting items would allow critical assessment of disease maps against objective criteria for comprehensive reporting. Standardization of reporting in the form of reporting guidelines for disease maps can also aid in rapid decision-making processes in epidemiological and public health investigations.

1.7 Thesis Rationale and Objectives

By presenting evidence that reporting characteristics are heterogenous and inconsistent in

Chapter 3, and presenting evidence that reporting items such as map projection are essential to the reproducibility of spatial analyses in Chapter 2, the overall work of this thesis is intended to advance the reporting of disease maps. The findings from this thesis can be used to inform the development and support of a reporting guideline for disease maps. Such a reporting guideline can benefit and advance the research community by minimizing misinterpretations and will also be useful to educators, public health practitioners, decision-makers, science journalists, and journal editors.

Chapter 2: Effect and Significance of Projection in Spatial Point Data Analyses

The work in this chapter aims to identify if varying map projections can affect the results of disease clustering and cluster detection methods. A simulation study for southern Ontario is performed. The spatial scan test and the Cuzick-Edward’s test was conducted for the same data under varying map projections.

Chapter 3: Characteristics of Disease Maps of Zoonoses: A Scoping Review

The work in this chapter aims to broadly characterize studies of disease maps and identify gaps, inconsistencies in the reporting of disease maps in the scientific literature. A scoping review of published journal articles was conducted. The search was limited to disease maps of zoonotic diseases to narrow search results. The search was also restricted to publications within

the years 2017 and 2018 to gauge reporting and map characteristics that are current and in practice at the time of investigation.

Thesis Goal and Objectives

The overall goal of this thesis is to present evidence that can be used to advance the reporting of disease maps for future guideline development. The objectives of this thesis were as follows:

1) Determine if map projections affect epidemiological inferences for spatial point data,

2) Assess the reporting characteristics of disease maps, such as map projection, in the

current scientific literature,

3) Broadly characterize studies of disease maps of zoonoses to understand and clarify

knowledge, gaps, and inconsistencies.

1.8 References

Berke, O. (2004). Exploratory disease mapping: kriging the spatial risk function from regional count data. International Journal of Health Geographics, 3(1), 18. https://doi.org/10.1186/1476- 072X-3-18

Beyer, K. M. M., Tiwari, C., & Rushton, G. (2012). Five Essential Properties of Disease Maps. Annals of the Association of American Geographers, 102(5), 1067–1075. https://doi.org/10.1080/00045608.2012.659940

Bithell, J. F. (2000). A classification of disease mapping methods. Statistics in Medicine, 19(17– 18), 2203–2215.

Box, G. E. (1976). Science and statistics. Journal of the American Statistical Association, 71(356), 791–799.

Box, G. E. (1979). Robustness in the strategy of scientific model building. In Robustness in statistics (pp. 201–236). Academic Press.

Centers for Disease Control and Prevention. (2012). Cartographic guidelines for public health. Geography and Geospatial Science Working Group. https://www.cdc.gov/dhdsp/maps/gisx/resources/cartographic_guidelines.pdf

Cromley, E. K., & McLafferty, S. L. (2011). GIS and public health. Guilford Press.

Dohoo, I., Martin, W., & Stryhn, H. (2012). Methods in Epidemiologic Research. VER Inc.

Green, C. (2015, March 31). Geographic Information Systems and Public Health: Benefits and Challenges. National Collaborating Centre for Infectious Diseases. https://nccid.ca/publications/geographic-information-systems-and-public-health-benefits-and- challenges/

Greene, S. K., Peterson, E. R., Balan, D., Jones, L., Culp, G. M., & Kulldorff, M. (2020). Detecting Emerging COVID-19 Community Outbreaks at High Spatiotemporal Resolution - New York City, June 2020. https://doi.org/10.1101/2020.07.18.20156901

Kilkenny, C., Browne, W. J., Cuthill, I. C., Emerson, M., & Altman, D. G. (2010). Improving Bioscience Research Reporting: The ARRIVE Guidelines for Reporting Animal Research. PLOS Biology, 8(6), e1000412. https://doi.org/10.1371/journal.pbio.1000412

Kirby, R. S., Delmelle, E., & Eberth, J. M. (2017). Advances in spatial epidemiology and geographic information systems. Annals of Epidemiology, 27(1), 1–9. https://doi.org/10.1016/j.annepidem.2016.12.001

Koch, T. (2017). Cartographies of Disease: Maps, Mapping, and Medicine (New expanded). ESRI Press.

Kulldorff, M., Tango, T., & Park, P. J. (2003). Power comparisons for disease clustering tests. Computational Statistics & Data Analysis, 42(4), 665–684. https://doi.org/10.1016/S0167- 9473(02)00160-3

Lawson. (2013). Bayesian disease mapping: hierarchical modeling in spatial epidemiology. Chapman and Hall/CRC.

Moher, D., Liberati, A., Tetzlaff, J., Altman, D. G., & Group, T. P. (2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLOS Medicine, 6(7), e1000097. https://doi.org/10.1371/journal.pmed.1000097

Moher, D., Schulz, K. F., & Altman, D. G. (2001). The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomised trials. The Lancet, 357(9263), 1191–1194. https://doi.org/10.1016/S0140-6736(00)04337-3

Moher, D., Schulz, K. F., Simera, I., & Altman, D. G. (2010). Guidance for Developers of Health Research Reporting Guidelines. PLoS Medicine, 7(2), e1000217. https://doi.org/10.1371/journal.pmed.1000217

Moran, P. A. P. (1950). Notes on Continuous Stochastic Phenomena. Biometrika, ,Vol. 37, No. 1/2, 8.

Nykiforuk, C. I. J., & Flaman, L. M. (2011). Geographic Information Systems (GIS) for Health Promotion and Public Health: A Review. Health Promotion Practice, 12(1), 63–73. https://doi.org/10.1177/1524839909334624

O’Connor, A. M., Sargeant, J. M., Dohoo, I. R., Erb, H. N., Cevallos, M., Egger, M., Ersbøll, A. K., Martin, S. W., Nielsen, L. R., Pearl, D. L., Pfeiffer, D. U., Sanchez, J., Torrence, M. E., Vigre, H., Waldner, C., & Ward, M. P. (2016). Explanation and Elaboration Document for the STROBE-Vet Statement: Strengthening the Reporting of Observational Studies in Epidemiology – Veterinary Extension. Zoonoses and Public Health, 63(8), 662–698. https://doi.org/10.1111/zph.12315

O’Connor, A. M., Sargeant, J. M., Gardner, I. A., Dickson, J. S., Torrence, M. E., Consensus Meeting Participants, Dewey, C. E., Dohoo, I. R., Evans, R. B., Gray, J. T., Greiner, M., Keefe, G., Lefebvre, S. L., Morley, P. S., Ramirez, A., Sischo, W., Smith, D. R., Snedeker, K., Sofos, J., … Wills, R. (2010). The REFLECT statement: methods and processes of creating reporting guidelines for randomized controlled trials for livestock and food safety by modifying the CONSORT statement. Zoonoses and Public Health, 57(2), 95–104. https://doi.org/10.1111/j.1863-2378.2009.01311.x

Osorio, M., Koziatek, C. A., Gallagher, M. P., Recaii, J., Weinstein, M., Thorpe, L. E., Elbel, B., & Lee, D. C. (2020). Concordance and Discordance in the Geographic Distribution of Childhood Obesity and Pediatric Type 2 Diabetes in New York City. Academic Pediatrics, 20(6), 809–815. https://doi.org/10.1016/j.acap.2020.03.012

Parpia, A. S., Skrip, L. A., Nsoesie, E. O., Ngwa, M. C., Abah Abah, A. S., Galvani, A. P., & Ndeffo-Mbah, M. L. (2020). Spatio-temporal dynamics of measles outbreaks in Cameroon. Annals of Epidemiology, 42, 64-72.e3. https://doi.org/10.1016/j.annepidem.2019.10.007

Peng, R. D., Dominici, F., & Zeger, S. L. (2006). Reproducible Epidemiologic Research. American Journal of Epidemiology, 163(9), 783–789. https://doi.org/10.1093/aje/kwj093

Pfeiffer, D. U., & Stevens, K. B. (2015). Spatial and temporal epidemiological analysis in the Big Data era. Preventive Veterinary Medicine, 122(1–2), 213–220. https://doi.org/10.1016/j.prevetmed.2015.05.012

Robertson, C., & Nelson, T. A. (2014). An Overview of Spatial Analysis of Emerging Infectious Diseases. The Professional Geographer, 66(4), 579–588. https://doi.org/10.1080/00330124.2014.907702

Rushton, G. (2003). Public Health, GIS, and Spatial Analytic Tools. Annual Review of Public Health, 24(1), 43–56. https://doi.org/10.1146/annurev.publhealth.24.012902.140843

Sargeant, J. M., O’Connor, A. M., Dohoo, I. R., Erb, H. N., Cevallos, M., Egger, M., Ersbøll, A. K., Martin, S. W., Nielsen, L. R., Pearl, D. L., Pfeiffer, D. U., Sanchez, J., Torrence, M. E., Vigre, H., Waldner, C., & Ward, M. P. (2016). Methods and Processes of Developing the Strengthening the Reporting of Observational Studies in Epidemiology – Veterinary (STROBE- Vet) Statement. Zoonoses and Public Health, 63(8), 651–661. https://doi.org/10.1111/zph.12314

Sargeant, J. M., O’Connor, A. M., Gardner, I. A., Dickson, J. S., Torrence, M. E., Dohoo, I. R., Lefebvre, S. L., Morley, P. S., Ramirez, A., & Snedeker, K. (2010). The REFLECT Statement: Reporting Guidelines for Randomized Controlled Trials in Livestock and Food Safety: Explanation and Elaboration. Journal of Food Protection, 73(3), 579–603. https://doi.org/10.4315/0362-028X-73.3.579

Sargeant, JM., O’Connor, AM., & Winder, CB. (2019). Editorial: Systematic reviews reveal a need for more, better data to inform antimicrobial stewardship practices in animal agriculture. Animal Health Research Reviews, 20(2), 103–105. https://doi.org/10.1017/S1466252319000240

Simera, I., Moher, D., Hirst, A., Hoey, J., Schulz, K. F., & Altman, D. G. (2010). Transparent and accurate reporting increases reliability, utility, and impact of your research: reporting guidelines and the EQUATOR Network. BMC Medicine, 8(1), 24.

Smith, C. M., Le Comber, S. C., Fry, H., Bull, M., Leach, S., & Hayward, A. C. (2015). Spatial methods for infectious disease outbreak investigations: systematic literature review. Eurosurveillance, 20(39), 30026. https://doi.org/10.2807/1560-7917.ES.2015.20.39.30026 19

Tobler, W. R. (1970). A Computer Movie Simulating Urban Growth in the Detroit Region. Economic Geography, 46, 234. https://doi.org/10.2307/143141

Vandenbroucke, J. P. (2009). STREGA, STROBE, STARD, SQUIRE, MOOSE, PRISMA, GNOSIS, TREND, ORION, COREQ, QUOROM, REMARK… and CONSORT: for whom does the guideline toll? Journal of Clinical Epidemiology, 62(6), 594–596. https://doi.org/10.1016/j.jclinepi.2008.12.003

Vandenbroucke, J. P., von Elm, E., Altman, D. G., Gøtzsche, P. C., Mulrow, C. D., Pocock, S. J., Poole, C., Schlesselman, J. J., & Egger, M. (2014). Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): Explanation and elaboration. International Journal of Surgery, 12(12), 1500–1524. https://doi.org/10.1016/j.ijsu.2014.07.014 von Elm, E., Altman, D. G., Egger, M., Pocock, S. J., Gøtzsche, P. C., & Vandenbroucke, J. P. (2007). The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for Reporting Observational Studies. Annals of Internal Medicine, 147(8), 573–577. https://doi.org/10.7326/0003-4819-147-8-200710160-00010

Waller, L. A., & Gotway, C. A. (2004). Applied spatial statistics for public health data (Vol. 368). John Wiley & Sons.

Wardrop, N. A., Geary, M., Osborne, P. E., & Atkinson, P. M. (2014). Interpreting predictive maps of disease: highlighting the pitfalls of distribution models in epidemiology. Geospatial Health, 237–246.

1.9 Figures

No. of Records of No.

Publication Year a)

Figure 1-1. Showing 2,298,589 records from August 2020 Web of Science topic search string: (health OR disease AND geographic AND distribution AND map) by a) Publication year and b) Web of Science Subject areas (Top 25). The bar graphs suggest that disease maps are increasingly being published and cited frequently in public health literature.

Figure 1-2. Diagram illustrating the different possible data layers that may be involved in a map making process for disease maps.

a) b)

c) d)

Figure 1-3. Commonly used map types representing area health data: a) Choropleth Map b) Proportional Symbol c) Point Map d) Isopleth Map.

Figure 1-4 Map Projection surfaces. From Geographic Information System Basics (v. 1.0), Campbell, and Shin 2012(Campbell & Shin, 2012) under a Creative Commons Attribution-NonCommercial- ShareAlike 3.0 License (CC BY-NC-SA).

Chapter 2: Effect of Map Projection on Spatial Point Data Analyses for Disease Cluster Detection and Clustering

2.0 Abstract

Background and Objectives: Disease maps are used to visualize the distribution and variation of disease outcomes geographically. Detecting clusters and clustering (spatial autocorrelation) of disease events is an important component of exploratory analyses in disease mapping. Map projections transform the 3-dimensional earth to a 2-dimensional planar surface. This study investigated the variability of cluster detection and clustering analyses under varying map projections for a simulated case-control point dataset mapped to Southern Ontario to identify whether map projections affected the spatial point pattern analysis of cluster detection and clustering.

Methods: A random case-control point dataset was simulated for Southern Ontario under a

Lambert Equal Area Conic projection with a Cartesian coordinate system. Two disease clusters were randomly generated within the simulated geospatial data. The simulated point data were analyzed using five different map projections: Lambert Equal Area Conic projection, UTM 17

North, Albers, Mercator, and Robinson projections. The Cuzick-Edwards test for disease clustering and the circular spatial scan test for disease cluster detection were evaluated under varying projection.

Results: A total of 655 point events including 69 cases were simulated, resulting in an incidence of approximately 11%. The Cuzick-Edwards test varied mildly with respect to the resulting p-

value and across the k parameters defining the size of the local neighbourhood of nearest neighbours (k=3, 5, 7, and 9 were used). Results from the circular spatial scan test, with p-values estimated from 999 Monte Carlo simulations indicated qualitative variability in the clusters detected; specifically under the Robinson projection two distinct disease clusters were detected, while for the other projections only one cluster was revealed. Under projections where a single cluster was detected, this cluster varied in size from 84 to 103 cases and controls.

Conclusions: This study demonstrated the importance of map projections in spatial statistical analyses of cluster detection, particularly for analyses measuring distances in Cartesian coordinates. The omission of reporting the projection of a spatial statistical map can have serious consequences for reproducibility of spatial statistical analyses and interpretation of results in a spatial epidemiological context.

2.1 Introduction

Disease maps allow us to visualize the spatial distribution of health outcomes in a population and can reveal spatial patterns that are otherwise not recognized from tabulated data

(Cromley & McLafferty, 2011). This is especially useful for large amounts of data, where visualizations by maps can be a significant step in engaging policymakers and the public in the planning and evaluation of health programs (Green, 2015). Disease maps are useful descriptive tools to convey and visualize observational data on disease frequency.

Disease maps are graphics where inherent choices in their design and the data mapped are important aspects for interpretation and inference. Geospatial data comes in different formats and can be mapped in various ways. Two basic forms of geospatial data are: point data, and areal or regional data. The term spatial point data refers to data where each observation or record has an exact spatial position associated with it (e.g., geographic coordinates). This differs from regional data where observations are aggregated over a geographic area (e.g., public health unit, county, country) (Waller & Gotway, 2004). Although point data can often be aggregated into spatial regional data, regional data generally cannot be disaggregated into point data for analysis. The present investigation focusses on analyses specific to point data.

The spatial relation between two points in space, the distance, is central to spatial statistical analyses. Map projections are used to transform the 3-dimensional earth surface onto a

2-dimensional plane, which distorts true distances, directions, and geographic shapes (Waller &

Gotway, 2004). A map projection consists of a datum and a coordinate system. Datums are based on different ellipsoids that model the shape of the earth. WGS 1984 is the most widely used and 28

recently developed datum (Centers for Disease Control and Prevention, 2012). Coordinate system overlays are created by projecting the three-dimensional datum ellipsoid onto a two- dimensional planar surface. The choice of the map projection affects the map and the distances portrayed on the map. Two maps referenced to different datums and projections can show distances between the same point locations with distances that differ substantially (Waller &

Gotway, 2004). For example, the distance between Atlanta and Seattle in the US increases by about 50% when the projection changes from Albers (2098 miles) to Mercator (2930 miles)

(Waller & Gotway, 2004).

In addition to the use of maps as descriptive tools, disease maps may be used to communicate the results of spatial data analysis, for reporting residual risks and disease cluster locations (Pfeiffer & Stevens, 2015; Robertson & Nelson, 2014; Rushton, 2003; Waller &

Gotway, 2004). In these instances, the choice and use of map projections affect both the resulting disease map graphically and the spatial data analysis results (Centers for Disease Control and

Prevention, 2012; Waller & Gotway, 2004).

Spatial data are characterized by spatial dependence or clustering (Waller & Gotway,

2004). Tobler’s first law of geography states that “everything is related to everything else, but near things are more related than distant things” (Tobler, 1970). Spatial statistics are mainly concerned with the modeling and analysis of this spatial dependence in a geospatial dataset

(Waller & Gotway, 2004). Several analytic methods can be used to measure spatial autocorrelation, sometimes referred to as global measures of clustering (e.g., Moran’s I statistic,

Geary’s C, D-function) and the measures of spatial autocorrelation provide insights on population health outcomes (Cromley & McLafferty, 2011; Hinrichsen et al., 2009). For 29

example, positive spatial autocorrelation will be high when incidence values from regions close together in space are more similar than incidence values from regions further apart, which is often the case with infectious diseases (Wardrop et al., 2014). For point data, spatial autocorrelation can be studied using the Cuzick-Edward’s statistic, which is based on the nearest- neighbour properties found in case-control point data (Waller & Gotway, 2004). The nearest- neighbour adjacency matrix W, is given by Eq 1.

1 푖푓 푙표푐푎푡푖표푛 푗 푖푠 푎푚표푛푔 푞 푛푒푎푟푒푠푡 푛푒푖푔ℎ푏표푢푟푠 표푓 푙표푐푎푡푖표푛 푖 푤 = { } Eq. 1 푖,푗 0 표푡ℎ푒푟푤푖푠푒

The test statistic is given by:

푁 푁 푇푞 = ∑푖=1 ∑푗=1 푤푤푖,푗, 훿푖훿푗 Eq. 2

Where 훿푖=1 if the ith location represents a case, and 훿푖=0 if the ith location represents a control.

Location j is among the q nearest neighbours of location i. Projections skew distances and directions; for instance, distances may be more skewed towards the north in a map projection, as compared to the east. Based on Eq 1., it can be posited that the measure of spatial autocorrelation utilizes the ranking of Euclidean distance between data points and thus could be affected by the projection parameters used for a mapped dataset.

A spatial disease cluster is recognized as an unusually high number of cases that occur close together in space. Disease cluster analysis uses statistical methods to test that an observed spatial pattern is a result of chance variation (Pfeiffer & Stevens, 2015). Spatial cluster detection methods include spatial scan statistics, which are based on a comparison of the risk of disease within a window, to that of outside the window. Spatial scan statistics are widely used in disease surveillance and offer important timely information, especially for rapidly evolving disease

outbreak situations like SARS-CoV-2 (Greene et al., 2020). Cluster detection methods have also been used to identify clusters of cancer incidence, for example clusters of mesothelioma and lung cancer among asbestos workers (Thun & Sinks, 2004).

One cluster detection method for point data is the circular spatial scan test (Kulldorff,

1997). In the circular spatial scan test, a circular window is imposed on the map, and the center of the circle moves across the study region (Kulldorff et al., 2003). The radius of the window is changed continuously between 0 and a user-defined upper limit, taking on a potentially infinite number of distinct circles, each with different location and size, and the potential of identifying a cluster (Kulldorff et al., 2003). The null hypothesis is rejected when at least one circle’s population has a higher underlying risk inside the circle relative to the population outside the circle (Kulldorff et al., 2003).

This study investigated the variability of cluster detection and clustering analyses under varying projections for a simulated case-control point dataset mapped to Southern Ontario. Study objectives were to (i) simulate a spatial point process of case and control locations with a cluster,

(ii) identify if map projections affected the Cuzick-Edward’s test for disease clustering, and (iii) assess the effect of changing map projections on the circular spatial scan test for cluster detection.

2.2 Methods

2.2.1. Study Area

The Ontario health region boundary file was downloaded in ArcInfo format from

Statistics Canada (Statistics Canada, 2013a) with a custom Lambert Conformal Conic projection.

The Ontario boundary file was modified to include only southern Ontario using the Southern

Ontario Health Regions 2015 (Map 7) definition from Statistics Canada, i.e., extending from the

North East Renfrew County and District Health Unit (3557) to the southern tip of Windsor-Essex

County Health Unit (3558) (Statistics Canada, 2013b). The region of southern Ontario spans approximately 600 km from the northeast to the southwest. All simulations and analyses were conducted in R Studio Version 1.1.456 (R Core Team, 2019; RStudio Team, 2020).

2.2.2. Data Simulation

A case-control dataset was simulated as a homogenous marked Poisson point process using the rmpoipp function from the R-package spatstat (Baddeley, 2020). The custom projection was transformed to the Lambert Equal Area Conic Projection in RStudio using the rgdal package’s spTransform function. Two disease clusters were induced by random re-labeling

(of controls to cases) in defined circular areas of a random sample of controls from the initially simulated homogenous Poisson process data.

The simulated point data were varied in RStudio using the rgdal package’s spTransform function. The projections were as follows: 1) Lambert Equal Area Conic Projection, 2) UTM 17

North, 3) Robinson, 4) Albers, and 5) Mercator. All projections were based on the WGS84 datum ellipsoid (projection strings used in this study are presented in Appendix 1).

2.2.3. Statistical Analysis

Cuzick-Edward’s test

The Cuzick-Edward’s test was conducted with the simulated data under each of the 5 projections using the qnn.test function from the smacpod package in R (French, 2020). The nearest neighbour parameter k was varied: k=3, 5, 7, 9, and p-values were estimated based on using 999 Monte Carlo (MC) simulations. The null hypothesis that cases and controls are sampled from a complete spatial random distribution is rejected at significance level α = 0.05.

Circular spatial scan test for point event data

The circular spatial scan test was conducted for the simulated dataset under each of the 5 projections using the spscan.test function from the smacpod package in R (French, 2020). The scan test was applied with a maximum population size for the cluster of less or equal to 50% than the total population. The null hypothesis of no clusters is rejected at the significance level α =

0.05, and the p-value was estimated using 999 MC simulations. The spatial scan test in R returns the centroid of the significant clusters, the radius of the window of the clusters, the total population in the cluster window, the standardized mortality ratio (SMR) in the cluster window

(the number of cases observed divided by the expected number of cases when the null hypothesis is true, i.e. when the risk is the same inside and outside the cluster), the relative risk (RR) in the

cluster window (relative to the estimated risk outside the window), and p-value (Kulldorff,

2018).

2.3 Results

2.3.1. Data Simulation

The resulting dataset consisted of 655 event locations with 586 controls and 69 cases.

Two circular clusters were defined by random relabelling of controls to increase the frequency of cases within two defined circular areas. Circular clusters were simulated under the Lambert

Equal Area Conic projection as follows (Figure 2.1):

Cluster 1:

(x=850km, y=2100km, radius=50km) 33 controls and 2 cases were randomly relabeled to 22 controls and 13 cases, i.e., risk changed from 6% (2/35) to 14% (13/35).

Cluster 2:

(x=775km, y=2200km, radius=20km) 6 controls and 0 case were randomly relabeled into 1 control and 5 cases, i.e. the risk changed from 0% (0/6) to 83% (5/6).

A total of 16 controls were relabeled into cases (11 in cluster 1, and 5 in cluster 2), resulting in an overall case event incidence of 11% (69/655) compared to the originally generated dataset with 8.1% (53/655).

2.3.2. Statistical Analysis

Cuzick-Edwards test

For all parameters of k nearest neighbours tested (k=3, 5, 7, 9), under all the projections, the Cuzick-Edward’s test (p-values less than 0.05) suggests the presence of clustering. Table 2.1 shows that the p-values changed slightly for the same data and same statistical test, under varying map projections. However, the overall inference of spatial clustering did not change. The p-values differed slightly across the projections for the same k, with the exception of k=9.

Circular Spatial Scan Test

Results from the circular spatial scan test for the simulated case-control point data under the 5 map projections are presented in Table 2.2 and Figure 2.2. Under the Robinson projection,

2 clusters were identified by the scan test, whereas the other projections identified only one larger cluster. While the cluster shape and size seem similar for the Lambert and UTM projections, they included different events locations. All clusters were significant. Differences in cluster sizes and relative risk were more apparent under the Robinson, Mercator, and Albers projection compared to the Lambert, and UTM 17N projections (Table 2.2). Apart from the

Robinson projection (SMRs were 5.7 and 5.9), cluster SMRs under other projections were relatively similar, ranging from 2.5 to 2.7. Similarly, cluster RRs were relatively similar ranging from 3.4-3.6 with the exception of the Robinson projection, where RRs were 6.4 and 6.8 due to the fact that two clusters were identified. Under projections where a single cluster was detected, this cluster varied in size from 84 to 103 cases and controls. The Mercator projection had the largest cluster radius of 120 km.

2.4 Discussion

The present investigation of a simulated homogenous Poisson point pattern over southern

Ontario (Canada) demonstrated that projection can affect the results of spatial statistical cluster detection in a Cartesian coordinate system. This can be expected considering a circle in one projection may be distorted to an ellipse in another projection. Thus, a circular scan test might cover a different area and thus capture a different population as a cluster, than the original circular cluster, which could be stretched to an ellipse by a map projection. Under the Robinson projection, a projection commonly used for world maps, two clusters were detected. Whereas for the other projections, used for continent or smaller regions, only one cluster was detected. This difference could be reflective of the geographic distortion differences from the projections on the simulated clusters across Southern Ontario. Southern Ontario is a northern region on a world map and possibly subject to distortion in shape and area, whereas other regional-based projections are conformal (favours preserving shape) and equal-area (favours preserving area) in the southern Ontario region. Future studies could investigate classes of projections further (e.g., equal-area, conformal, equidistant and compromised) to understand if cluster detection ability is modified by the type of projection.

Only slight changes in the p-values for the Cuzick-Edward’s test were found for the different projections. It is difficult to discern if the slight differences were a result of projection or chance variation from Monte Carlo p-value estimation. Future studies could investigate datasets expanding the number of MC simulations to further understand projection effects on the

Cuzick-Edward’s test p-values. The slight changes in the p-values may also be reflective of the

Cuzick-Edward’s test statistic dependence on rank-order of distances to nearest neighbours and 37

thus not as grossly affected by changes in distances; that is despite changes in projection, the rank orders may still be preserved. Future studies could investigate other measures of clustering for point data such as the D function (P. J. Diggle & Chetwynd, 1991; P.J. Diggle, 2003) which might be more sensitive to changes in distance resulting from projection.

The findings from this study suggest that the choice and use of map projections for disease mapping has consequences for epidemiological interpretations and in turn can have policy implications. Cluster alarms are used in disease surveillance, and since health officials are tasked with evaluating local disease cluster alarms, an important question is to determine if the cluster has occurred by chance or whether the excess is a result of a risk factor. Cluster detection methods are exploratory analyses (i.e. hypothesis generating) that can be used to conduct further investigations to explain the presence of clusters by examining potential risk factors such as through a regression analysis adjusted for known risk factors. Consequently, it becomes important to accurately identify clusters geographically, as subsequent analyses and decisions regarding resource allocation may rely on the exploratory findings.

Cluster identification can be dependent on the cluster detection method used and the method’s parameter settings. This study also shows the relevance of acknowledging the map projection in a cluster detection method and the respective epidemiological interpretation. Map projections matter but may not always be reported (as found in chapter 3 of this thesis). The reporting of projection, sample size, cluster test parameters, and significance level are all needed for reproducibility and critical appraisal in epidemiological research and public health decision making settings.

2.4.1. Strengths and Limitations

This study did not quantify the effect of projections on the spatial scan test but presents qualitatively that differences in analyses can result. This study was also a case study and limited in the number of scenarios evaluated. Future studies could evaluate the effect of projections across a range of different scenarios of varying different parameters (e.g., radius of the largest possible cluster to consider). This study also did not examine and compare projected map results to an unprojected map or a map with geographic coordinates (longitude, latitude), and this could provide further insight on the effect of projection in the analyses presented here. Given this study examined point data, the findings could be relevant to aggregated areal/regional data. Berke

(2019) identified that varying projection does affect spatial statistics of regional data (Berke,

2019). Variable results from the circular spatial scan test in Chapter 2 are consistent with findings from Berke 2019, where varying projection for regional data produced different results for another test for clustering, Moran’s I (Berke, 2019).

2.5 Conclusion

Maps while graphics have an important role in spatial epidemiological analysis (Waller,

2014), and should not be critiqued in isolation from the spatial analysis conducted. The findings from this investigation suggest the exploratory analyses of clustering and cluster detection in disease maps can be conditional on the specific map projection used. More research is needed to better understand the consequences of projection choice on spatial epidemiological inferences.

This study suggests that the omission of reporting the projection of a map displaying spatial statistics can have serious consequences for the interpretation of results in spatial epidemiological investigations, critical appraisal of research and for reproducibility of analyses.

Future studies can assess the quality and consistency of reporting projection in spatial statistical maps, among other characteristics.

2.6 References

Baddeley, A. (2020). Spatial Point Pattern Analysis, Model-Fitting, Simulation, Tests. CRAN Repository.

Berke, O. (2019). Dazed and confused: How map projections affect disease map analysis and perception. Frontiers in Veterinary Science, 6. https://doi.org/10.3389/conf.fvets.2019.05.00013

Cromley, E. K., & McLafferty, S. L. (2011). GIS and public health. Guilford Press.

Diggle, P. J., & Chetwynd, A. G. (1991). Second-Order Analysis of Spatial Clustering for Inhomogeneous Populations. Biometrics, 47(3), 1155–1163. JSTOR. https://doi.org/10.2307/2532668

Diggle, P. J. (2003). Statistical analysis of spatial point patterns. Edward Arnold.

French, J. (2020). Statistical Methods for the Analysis of Case-Control Point Data. CRAN Repository.

Hinrichsen, V. L., Klassen, A. C., Song, C., & Kulldorff, M. (2009). Evaluation of the performance of tests for spatial randomness on prostate cancer data. International Journal of Health Geographics, 8(1), 41. https://doi.org/10.1186/1476-072X-8-41

Kulldorff, M. (1997). A spatial scan statistic. Communications in Statistics - Theory and Methods, 26(6), 1481–1496. https://doi.org/10.1080/03610929708831995

Kulldorff, M. (2018, March). SaTScan Users Guide version 9.6. https://www.satscan.org/cgi- bin/satscan/register.pl/SaTScan_Users_Guide.pdf

R Core Team. (2019). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/.

Robertson, C., & Nelson, T. A. (2014). An Overview of Spatial Analysis of Emerging Infectious Diseases. The Professional Geographer, 66(4), 579–588. https://doi.org/10.1080/00330124.2014.907702

RStudio Team. (2020). RStudio: Integrated Development Environment for R. RStudio, PBC. http://www.rstudio.com/

Rushton, G. (2003). Public Health, GIS, and Spatial Analytic Tools. Annual Review of Public Health, 24(1), 43–56. https://doi.org/10.1146/annurev.publhealth.24.012902.140843

Statistics Canada. (2013a). Ontario Health Regions. https://www.statcan.gc.ca/access_acces/alternative_alternatif.action?l=eng&loc=data- donnees/boundary-limites/arcinfo/HRP035b11a_e.zip

Statistics Canada. (2013b). Southern Ontario Health Regions. https://www.statcan.gc.ca/pub/82- 402-x/2015002/maps-cartes/rm-cr07-eng.htm

Thun, M. J., & Sinks, T. (2004). Understanding Cancer Clusters. CA: A Cancer Journal for Clinicians, 54(5), 273–280. https://doi.org/10.3322/canjclin.54.5.273

Tobler, W. R. (1970). A Computer Movie Simulating Urban Growth in the Detroit Region. Economic Geography, 46, 234. https://doi.org/10.2307/143141

Waller, L. A. (2014). Putting spatial statistics (back) on the map. Spatial Statistics, 9, 4–19. https://doi.org/10.1016/j.spasta.2014.03.007

Waller, L. A., & Gotway, C. A. (2004). Applied spatial statistics for public health data (Vol. 368). John Wiley & Sons.

2.7 Figures

b) a)

c) d)

Figure 2-2: Spatial scan test for the same simulated data under different projections: a) Lambert Equal Area Conic, b) UTM 17 North, c) Robinson, d) Albers, e) Mercator

2.8 Tables

Table 2.1. Results from the Cuzick-Edward’s test for disease clustering (i.e. p-values) under the 5 map projections: Lambert Equal Area Conic, UTM 17North, Robinson, Mercator, and Albers.

Projection k=3 k=5 k=7 k=9 Lambert 0.001 0.005 0.001 0.001 UTM 17N 0.003 0.004 0.002 0.001 Robinson 0.001 0.001 0.001 0.001 Mercator 0.003 0.004 0.002 0.001 Albers 0.001 0.007 0.002 0.001

Table 2.2. Results from the circular spatial scan test for the simulated case-control point data under 5 map projections. P-values were estimated from 999 Monte Carlo simulations.

Projection cluster RR SMR m n p-value Radius (cases) (population) (km) Lambert 1 3.6 2.7 24 84 0.004 87 UTM 1 3.6 2.7 24 84 0.003 87 Robinson 1 6.8 5.9 10 16 0.003 32 2 6.4 5.7 9 15 0.010 24 Mercator 1 3.5 2.6 24 86 0.005 120 Albers 1 3.4 2.5 27 103 0.003 97

Chapter 3: Characteristics of Disease Maps of Zoonoses: A Scoping Review

3.0 Abstract

Objectives: The increasing availability of spatial data and mapping software makes disease mapping widely practiced and accessible. Although disease maps are regarded as informative communication tools in public health, there are no widely agreed upon and established guidelines for their reporting. This study investigated the characteristics and reporting of disease maps of zoonoses published in the scientific literature from 2017-2018. Study objectives were to characterize the literature that include disease maps, assess the characteristics of disease maps and their reporting.

Methods: Two reviewers conducted duplicate screening of titles and abstracts identified from a search in Medline, CAB Direct, Web of Science Core, and Agricola with the publication date restricted to 2017-2018. This was followed by full-text article screening in duplicate. Studies were included if they had a disease map figure describing a zoonotic disease. Information on study and map reporting characteristics were extracted and summarized from full-text articles meeting inclusion criteria.

Results: The search identified 1666 records. 302 articles meeting eligibility criteria were included, comprising 505 disease maps. Most published disease maps in this review were reported without information relevant for interpretations of geospatial analyses and their reproducibility, such as 92% of maps not reporting map projection.

Conclusions: The findings identified gaps and inconsistencies in the reporting of basic map information in the literature and support the development of a reporting guideline for disease maps.

3.1 Introduction

Disease maps have been used by epidemiologists for centuries to convey geographic distribution of disease and health outcomes (Cromley & McLafferty, 2011). Disease maps are presented in diverse ways and are used in a variety of applications (Bithell, 2000). The definition of a disease map across textbooks and journal articles is variable and may be used interchangeably with other terms, such as health map and risk map. In this study, a disease map refers to the visual representation of geographically referenced health data within a geographic boundary for an animal or human population(s). The word “disease” in “disease map” is not solely intended for disease outcomes such as mortality, incidence, or prevalence, but can refer to other health outcomes including vaccine status and health scores. Such health outcome data are labelled with spatial tags (e.g., geographic coordinates, regional identifiers, area codes). Disease maps can reveal spatial patterns that are otherwise not recognized from tabulated data (Cromley

& McLafferty, 2011).

Disease mapping as a practice is used to visualize spatial variation in disease occurrence and help to generate hypotheses about disease causation. Detailed visual information on the spatial distribution of disease, especially for unique spatially heterogeneous disease distributions that are specific to a disease, provides important insights for disease prevention and control programs (Wardrop et al., 2014). With recent advances in Geographic Information Systems

(GIS) technology, access to tools for management, synthesis, and display of spatial data, map- making processes and production are better enabled (Cromley & McLafferty, 2011; Rushton,

2003; Waller & Gotway, 2004). Growing access also empowers public and citizen participation in the creation, evaluation, and analysis of spatial data and visualizations (Cromley &

McLafferty, 2011). 48

The heterogeneous nature of disease map visualizations, their spatial data, purposes, and contexts makes it difficult to articulate standard reporting items for disease maps and critically assess them against other objective criteria. Basic elementary reporting items across different map types include the data source, scale bar, legend, and map projection (Centers for Disease

Control and Prevention, 2012; Cromley & McLafferty, 2011). Map projections, for example, are used to transform the 3-dimensional earth surface onto a 2-dimensional plane, which distorts true distances, directions, and geographic shapes (Waller & Gotway, 2004). Projections will use datums, which are based on different ellipsoids that model the shape of the earth and a coordinate system overlay. WGS 1984 is the most widely used and recently developed datum

(Centers for Disease Control and Prevention, 2012). One of the most common coordinate systems is the Universal Transverse Mercator projection. Depending on the intended use of the map, the choice of projection will matter. For example, two maps referenced to different datums and projections can show distances between the same point locations, but distances can differ substantially (Waller & Gotway, 2004).

In addition to disease maps being used for descriptive exploratory purposes of identifying and representing spatial patterns visually, disease maps are also used to display the results of spatial statistical analyses, including spatial regression and cluster analyses (Pfeiffer & Stevens,

2015; Robertson & Nelson, 2014; Rushton, 2003; Waller & Gotway, 2004). In these instances, the choice and use of projection parameters affect spatial statistics (Centers for Disease Control and Prevention, 2012; Waller & Gotway, 2004), and the resulting p-values and confidence intervals.

Spatial data are characterized by spatial dependence or clustering (Waller & Gotway,

2004). Tobler’s first law of geography states that “everything is related to everything else, but

near things are more related than distant things” (Tobler, 1970). Spatial statistics is mainly concerned with the modeling and analysis of this spatial dependence in a geospatial dataset

(Waller & Gotway, 2004). Several analytic methods can be used to measure spatial autocorrelation, such as Moran’s I statistic, and provide important insights on population health outcomes. For example, positive spatial autocorrelation will be high when incidence values from regions close together in space are more similar than incidence values from regions further apart, which is often the case with infectious diseases (Wardrop et al., 2014).

Measures of spatial autocorrelation utilize the geospatial distance between data points and thus are affected by the map projection used for a mapped dataset. Furthermore, not accounting for spatial dependence/autocorrelation could also present itself as a bias or limitation in studies, particularly for spatial regression model studies, ultimately leading to the misinterpretation of the relationships between observations and covariates (Wardrop et al.,

2014). Spatial autocorrelation, when not considered, could mean violating the underlying assumption of statistical independence in (ordinary) statistical methods and generally results in inaccurate models, biased regression parameters, underestimated standard errors, falsely narrow confidence intervals, and overestimation of the significance of covariates (Cromley &

McLafferty, 2011; Wardrop et al., 2014).

Another challenge with mapping health data is the modifiable areal unit problem

(MAUP) (Cromley & McLafferty, 2011)(Waller & Gotway, 2004). The MAUP occurs during geospatial analyses of aggregated data, where the results of one analysis are different from the same original data, albeit for different aggregation schemes (e.g., a choropleth map of cancer cases by county-level will look different from a map showing the same data by census tract)

(Cromley & McLafferty, 2011). The MAUP is apparent in two forms: the scale effect and the

zone effect. The choice of scale is important as relationships between variables at one scale may be distorted when viewed from another scale, depending on the number and sizes of areal units.

The zone effect considers the variability in results due to different formations of the areal units

(Waller & Gotway, 2004). Moreover, making inferences from maps using aggregated data to a lower level of aggregation can result in ecological biases (Lawson, 2013). Spatial sampling of populations can also give rise to selection biases and limitations of precision. One common limitation is the small numbers problem, where rates based on small populations tend to be less reliable and more variable compared to rates based on large populations (Beyer et al., 2012).

As such, maps are subjective graphics where inherent choices in design and data mapped are important aspects of interpretation and inference (Brewer, 2016; Cromley & McLafferty,

2011). Despite the widespread use of disease maps in research, journal publications, and use as decision support tools by public health agencies, to date, no current standard reporting guideline exists for geographic disease maps (search conducted December 2019 in EQUATOR- http://www.equator-network.org /). Without comprehensive reporting of map elements and creation, geospatial statistics and analyses are not reproducible and may lead to misinterpretations.

In this study, recently published journal articles that included disease maps of zoonoses are reviewed. As disease maps are used across disciplines of public health (e.g. non-communicable disease epidemiology, and environmental epidemiology), this review focused on disease maps of zoonoses to capture uses across animal and human populations, in addition to their utility in intersections of animal and human populations.

Zoonoses are diseases that are naturally communicable between humans and other vertebrate animals. There are over 200 zoonotic diseases, and emerging zoonoses such as Avian Influenza

and West Nile Virus have raised global public awareness around the importance of connecting animal and human populations for disease surveillance (Cromley & McLafferty, 2011). A majority of emerging infectious diseases (EID) ( e.g., severe acute respiratory virus, Ebola virus) are zoonotic or suspected to be of zoonotic origin and are significantly increasing over time with a significant burden on global economies and public health (Jones et al., 2008). Geographic foundations of disease emergence has been identified by five key drivers: cross-species transfer, spatial diffusion, genetic evolution, new detection capacity of newly recognized pathogens, and changes in human-environmental relationships (Mayer, 2000; Robertson & Nelson, 2014). In today’s interconnected global community, the geospatial information concerning zoonotic infectious diseases provides timely information important for their control and prevention across animal and human populations (Beard et al., 2018; Kraemer et al., 2016; Robertson & Nelson,

2014).

Narrowing the focus of this review to disease maps of zoonoses offered a wide breadth of applications to broadly characterize the literature that includes disease maps, especially with respect to disease surveillance. Communicating disease epidemiology research often requires the use of disease maps for knowledge translation to appropriate communities of users. The multidisciplinary nature of zoonotic infectious disease research benefits from collaborative efforts among ecologists, epidemiologists, geographers, public health practitioners and decision- makers. Examining zoonotic disease research offers a valuable starting point to understand reporting practices across disciplines and added value for creating a standard framework for communicating disease maps.

Using a systematized search of maps of zoonotic diseases in scientific journal databases, the objectives of this review were as follows:

1) Summarize disease map applications in studies of zoonotic diseases;

2) Evaluate the reporting characteristics of basic map elements (e.g. projection and scale); and

3) Summarize reported challenges with using spatial data and analyses for map representations,

inferences and interpretations.

3.2 Methods

To broadly characterize the applications, reporting characteristics, and challenges of disease maps of zoonoses and their studies, the scoping review methodological framework proposed by Arksey & O’Malley (Arksey & O’Malley, 2005) was used for this review. The research team revised a draft protocol after the search strategy was completed, and the final version of the review protocol is presented in Appendix 2.

This scoping review also included a consultation element applying integrated knowledge translation (iKT), a form of collaborative knowledge translation (KT) between researchers and knowledge users throughout the conduct of a study (Straus et al., 2013). Knowledge users are individuals or groups who will use and apply the knowledge generated. These can be decision- makers, policymakers, patients, or whole communities. Knowledge users in this work were identified (e.g., educators, spatial epidemiological researchers, statisticians, and public health practitioners) from previous research work or contact with study investigators (OB, IS). Ethics approval was asked for but not required by University of Guelph's Research Ethics Board because findings from this engagement were not used as results in this study and only informed method development. Knowledge users (KUs) were engaged prior to the conduct of the review via an email questionnaire. The purpose of their involvement was to validate the research questions posed for this review as relevant and provide added KT value to findings at the end of the research. Twenty-one knowledge users were contacted, and 12 responded. Contributors to the consultation provided additional insights on the relevance of data extraction items, confirmed that data extraction items pre-identified by the investigators were relevant and that items considered important by respective experts were not missed in this investigation.

3.2.1. Search Strategy

Articles with disease maps reporting on zoonoses were identified via searches in

MEDLINE via Ovid, Web of Science Core Collection, CAB-Direct, and Agricola via ProQuest databases. The search was limited to the years 2017 and 2018 to evaluate current reporting and study characteristics. The search strategy was limited to journal articles and did not include searching the grey literature. The search was conducted on December 18th, 2018, and again on

April 19th, 2019, to capture any additional publications from late 2018. Search terms and strategies were tailored to the requirements and structure of each database and consisted of

“zoono*” (e.g., zoonoses, zoonotic, zoonosis) AND “disease map*” OR other geographic map terms including “atlas”, “geographic distribution”, “risk map”. The search conducted in Web of

Science is presented in Table 3.1. The full detailed search strategy is presented in Appendix 3.

All articles identified from the search were imported into Endnote, and duplicate articles were removed.

3.2.2. Eligibility Criteria

For a journal article to be eligible, the subject of the title and abstract had to describe a zoonotic disease in a population with reference to geography. Reviewers verified if a disease was zoonotic against a list from Vbrova et al. (2009) (Appendix 4) and the Merck Veterinary Manual

(Merck Veterinary Manual, n.d.). Only journal articles published in English and published in

2017 and 2018, including electronic publications published ahead of print, were eligible for inclusion. Full-text articles with at least one zoonotic disease map figure were included. Zoonotic diseases in animal or human populations acquired through all transmission modes (vector-borne,

food-borne, indirect contact, and direct contact) were included. Maps that had no clearly demarked geographic boundaries of a study population (e.g., clipped study boundaries, no geographic boundary at all), or maps with arbitrary boundaries drawn by study authors for the purposes of a study area were excluded (e.g., geographic boundaries that are not otherwise recognized administratively outside of the study).

3.2.3. Study Selection

The review team pilot-tested screening forms on a random sample of articles (20 titles and abstracts, and 10 full-text articles), and all had an agreement score of >80% with the team consensus rating (i.e., reviewer rating compared and scored against the consensus rating) before screening the remaining records.

Articles were screened by two reviewers using pre-specified eligibility criteria (Appendix

5). Any disagreements on article selection were resolved by consensus or by a third reviewer.

Articles that met eligibility criteria for abstracts and titles were included for subsequent screening by inspecting the full-texts of the articles. Full-text articles were located and screened visually for the presence of a disease map of zoonotic disease. Article screening was done using the Covidence software for literature review (Covidence, 2017). Articles were verified for inclusion by the lead author (IS) following screening. Eligible articles underwent data extraction.

3.2.4. Data Items and Extraction

A data collection form was designed and informed by a combination of literature search

(e.g., educational texts, and grey literature), expert consultation, and a pre-data charting exercise.

To assure that the data collection form reflected the research questions and was appropriate for the heterogeneous nature of disease maps, two reviewers engaged in a data charting exercise using a sample of 50 articles. The data collection form was modified to improve the data extraction process as needed. The final data extraction form collected information related to study characteristics, map characteristics, and geospatial analyses. Data extraction was conducted in Microsoft Excel (Microsoft Corporation, n.d.) after the review team consisting of 9 reviewers pilot-tested the extraction forms on a random sample of six included articles. After pilot-testing and refining the data collection form further to improve clarity on extraction items,

1 reviewer independently extracted data for an included article.

3.2.5. Study Characteristics

Information on the study population, zoonotic disease, and geographic region were collected at the study level. Study populations were categorized from a list based on the population with the health outcome: human population, animal population, invertebrate vector population, or a combination of these populations.

For map utility in a study, the use of maps was determined to be either descriptive or analytical by reviewers. Studies with maps only used by authors for visualizing observational disease distribution were categorized as descriptive. Study articles with at least one map that represented outputs from geospatial methods and analyses (e.g., spatio-temporal map predicting risk, or any distance-based/ spatial statistics) were categorized as analytical.

To gain some insights on data sharing practices for the purposes of transparency, and reproducibility, data availability statements when provided, were extracted into prescribed

categories. These categories included: Availability in a repository, Available from corresponding author or request, Available in this published article/supplementary file, Available from a third party with restrictions, Not publicly available, Publicly available and open access, Other.

3.2.6. Map Characteristics

Given a journal article could report multiple maps, reporting characteristics were collected at the map level. Studies including multiple maps in one figure or a set of maps were counted as one composite map as characteristics and features were assumed to be the same for individual maps within one composite figure. Information was extracted from each map for the map type, and these included: choropleth maps, point and symbol maps, proportional symbol maps, maps with a continuous gradient feature, and combination maps. Choropleth maps show geographic areas shaded in varying colours, intensities, or patterns to represent the proportions or magnitudes of outcome data (Cromley & McLafferty, 2011). Isopleth maps and other maps with continuous gradient representation of data were categorized as maps with a continuous gradient feature. Isopleth maps depict smooth continuous data by using isolines that connect points of same numerical values (Centers for Disease Control and Prevention, 2012). Point maps and symbol maps represent geographically referenced health data as points and symbols on specific locations. Proportional symbol maps have symbols mapped with their size proportional to the number of events/magnitude of outcome (Cromley & McLafferty, 2011). And finally, a combination map was defined as a map with more than one map feature to represent the health outcome data (e.g., a point map overlaid on a choropleth map).

The health outcome represented in each map, as described by authors were categorized as follows: 1) measures of disease frequency, 2) molecular typing/phylogeographic, and 3) other.

Measures of disease frequency included: cases, prevalence, incidence, and mortality risk as outcomes reported by authors. Maps that reported on the genetic identity of a pathogen as a map outcome were categorized into molecular typing or phylogeographic. Map outcomes such as exposure to a health intervention (e.g., vaccination status) were categorized as “other”.

The format of data as presented in map legends and article text was categorized from a list of categories adapted from Mitchell (Mitchell, 1999) and were as follows:

1) Counts and amounts provide total numbers and values associated with mapped features

(e.g., 50 cases).

2) Ratios represent map data created by dividing one quantity over another (e.g., a disease

incidence risk of 50 /1,000,000 people as a legend symbol).

3) Ranks represent mapped data with a hierarchal order, from high to low, and show values

relative to each other instead of measured values (e.g. low risk, to high risk).

4) Categories represent mapped data that are described by categories (e.g., serostatus).

The reporting of basic map design elements from map figures was evaluated as “yes” or

“no". These elements included:

1) Scale Bar represents the relationship between a distance on the map and the corresponding

distance on the ground (Waller & Gotway, 2004). The scale of a map provides important

information on the level of detail that a map can portray (Cromley & McLafferty, 2011). A

scale bar can be represented in text such as (e.g., 1 inch to 50km) or visually as a scale bar, or

as a fraction (e.g., 1:25,000).

2) Compass/ North Arrow describes the orientation of the map and is useful when the map is

not oriented in the North direction or for large-scale maps (e.g., shows an area with more

detail) (Centers for Disease Control and Prevention, 2012).

3) Legend defines and explains the map symbols and features.

4) Projection transforms the three-dimensional surface of the earth onto a two-dimensional

plane. The datum and coordinate system are reported to convey the projection (e.g., WGS

1984, UTM Zone 18M).

5) Time refers to an indication of the date (at the minimum the year) in the title of the map or

figure caption.

Elaboration and further description of extracted items are presented in Appendix 5.

3.2.7. Geospatial Methods and Analyses

Geospatial methods and analyses were extracted as open text by a reviewer from studies that reported them and were categorized into the following by the lead author (IS):

1) Disease mapping methods and spatial regression models

This included regression models with a spatial component such as generalized linear mixed models, conditional autoregressive (CAR) models, geographical weighted regression (GWR).

This category also included model-based approaches, which “borrow strength” across small areas to improve local estimates to smooth extreme rates resulting from small local sample sizes, including Empirical Bayes smoothing, kriging, and other interpolation methods.

2) Spatial cluster analysis

This included methods for cluster detection, also known as “local measures” for spatial clusters

such as LISA/Local Moran’s I, local Getis-Ord G, and spatial and space-time scan statistics

using various distributions.

3) Spatial Autocorrelation analysis

This included methods for global measures of spatial autocorrelation or clustering, for example:

Moran’s I, Geary’s C, Cuzick-Edwards K-NN-test, Average nearest neighbour, Knox test,

Semivariogram, or Empirical variogram.

4) GIS-based spatial analysis

This included any geospatial analyses that involved using GIS tools and databases such as zonal

statistics, multicriteria evaluation, and network analysis.

5) Point sets and distance statistics

This included analyses where distance was central to the input and output of the analyses such as

source or risk-factor proximity analyses, case-case distances.

6) Other

All other analyses and methods that did not fit in the above categories but involved a geospatial

element in an analysis such as dynamic spatial disease models of transmission.

3.2.8. Geospatial Biases and Limitations

Biases and limitations as described by the authors with respect to the geospatial analyses

conducted were extracted as open text. The extracted items were then categorized by the lead

investigator (IS) into themes adapted from Fotheringham and Rogerson’s (1993) “Impediments that arise in spatial analysis”. Themes were as follows:

1) Spatial Scale and Aggregation of data

This included discussion and recognition of ecological biases such as MAUP, or limitations on making inferences from mapped data as result of spatial scale and aggregation scheme of mapped data.

2) Boundary problems

This included discussion and recognition of study boundaries potentially biasing the study results or recognized as limitations on map inferences.

3) Model assumptions and goodness of fit

This included a discussion and recognition of assumptions likely to be violated as a result of spatial data, or discussion of spatial assumptions that are limitations or biases for model fit.

4) Spatial Autocorrelation

This included a discussion and recognition of spatial autocorrelation not incorporated into a model framework or analyses resulting in biases or as a study limitation.

5) Geolocation problems and spatial biases in sampling, geocoding, data collection

This included any discussions pertaining to biased estimates or study limitations from how geographic data were collected (e.g., sampling bias) and mapped (e.g., mapping migratory animals that do not have a residential address).

3.2.9. Data Synthesis

The extracted data were descriptively analyzed by frequencies and percentages to summarize study and map characteristics. Geospatial analyses and methods collected as open text were categorized by the lead author using an inductive approach of content analysis (Elo &

Kyngäs, 2008). Reported biases and limitations were also collected as open text, and later categorized by the lead author from identified themes in the literature. A bar chart figure comprising the zoonotic pathogens by study area, and a bar chart of map types was created in R

Studio using ggplot2 (R Core Team, 2019; RStudio Team, 2020; Wickham, 2016).

3.3 Results

3.3.1. Study Selection and Data Extraction

The search identified 1666 unique citations. The study flow is presented in Figure 3.1. A total of 302 articles met the eligibility criteria after full-text review and were included in this review representing 505 individual and composite disease maps. Citations for the included studies are listed in Appendix 6.

3.3.2. Study Characteristics

A majority of the included studies were studies mapping animal populations (43%) compared to human populations (30%), invertebrate vector populations (3%), or maps with both animal and human populations (21%), and both invertebrate vector and human populations (2%)

(Table 3.2). Viral (32%) and bacterial pathogens (34%) were the most common group of zoonoses (66%) among the included studies (Table 3.2). A breakdown of pathogen type by the geographic region of the maps for the included studies is presented as a bar graph in Figure 3.2.

Most disease maps were studies from Asia (25%) and Europe (20%), and across all regions virus and bacterial pathogens were common. South America had the most studies with helminth pathogens, and Asia had the most studies with protozoa pathogens. A list of zoonotic pathogen species of included studies is presented in Appendix 7.

Included studies were evenly distributed between 2017 (48%) and 2018 (52%) for the year of publication. Most studies utilized the map figures for descriptive (66%) compared to analytical purposes (34%). Although most studies did not include a data availability statement,

25% of studies indicated all data were available in the main article or supplementary file, and 9% indicated the data were publicly available or open access.

3.3.3. Map Characteristics

Within the 302 included publications, 505 maps of zoonotic disease maps were identified.

Of these 505 maps, 88% were individual maps, and the remaining 12% were composite figures of maps. The most commonly used map type were choropleth maps (38%), followed by point maps (23%) (Table 3.3). A majority of maps were found to report measures of disease frequency

(88%) as outcomes. A small number of maps (4%) reported on a molecular typing or phylogeographic outcome. Data in the format of ratios (25%) were most commonly presented in choropleth maps (Figure 3.3, Table 3.3). Choropleth maps were also used to present count and categorical data (Figure 3.3). Less than half of the disease maps (47%) included a scale (Table

3.3). A majority of maps did not report map projection (92%). Most maps did include a compass

(53%), a legend (85%), and a date associated with the mapped data (68%).

3.3.4. Geospatial Analyses, Biases, and Limitations

Of the 302 included studies, 103 studies reported on 144 geospatial data analyses (Table

3.4). Most commonly reported geospatial analyses included disease mapping and spatial regression models (42%), spatial autocorrelation (15%), and cluster detection (21%).

Of the 302 included studies, 50 studies reported on geospatial specific biases and limitations 60 times (Table 3.5). Most commonly identified themes of biases and limitations associated with geospatial analyses included: geolocation problems and spatial bias in sampling

(58%), problems arising from the spatial scale and aggregation of data (17%), and lack of accounting for spatial autocorrelation (8%).

3.4 Discussion

The results of this review suggest that there may be gaps in reporting practices of basic map information in published disease maps. Some basic map design elements, such as the legend and compass, were commonly reported. However, design elements such as the map projection and scale were only reported by a low percentage of included maps (8% and 47%, respectively).

Reporting the projection and scale can be important for: (1) immediate transparent interpretation of the map, (2) subsequent data analysis or reproduction of results presented by a map, and (3) documentation purposes for prospective and retrospective comparisons. For example, sequential analysis of disease maps retrospectively and prospectively can be important to determine whether an outbreak is taking place or has occurred (Lawson & Rotejanaprasert, 2018).

Although un-projected maps can give misleading impressions of geographic distribution of data, if maps do not require a high level of accuracy for location, then it may not be necessary to transform data to a projected coordinate system or to report projection (Waller & Gotway,

2004). The choice of a projected coordinate system and reporting of projection are important for maps requiring accurate visual depiction or for maps displaying spatial statistics, where it is essential to preserve and represent relative distances (Waller & Gotway, 2004). In this review, most studies (66%) utilized maps for descriptive purposes. In these instances, perhaps projection is an aesthetic and design choice, though it can still affect geographic interpretations of disease patterns. However, in maps displaying spatial statistical analysis, projection can affect how both the visual geography and the spatial data analysis are interpreted.

An example of mapped datasets for spatial analysis are maps presenting cluster detection.

This review identified spatial cluster analysis as a relatively popular geospatial analysis among the methods categorized in this review. A systematic review by Smith et al. (2015) on spatial 67

methods for infectious disease outbreak investigations also identified spatial scan statistics (i.e. cluster detection) as a frequently adopted type of analytic method in their review (Smith et al.,

2015). Accurately identifying clusters geographically is important for public health decision making and subsequent analyses. The basic reporting of items such as projection and scale can be important for interpreting and critically appraising maps displaying cluster detection analysis.

Standardization of reporting basic map information can aid in rapid decision-making processes and follow-up of spatial epidemiological investigations. One possible way standardization of reporting items could be achieved is through the development of a reporting guideline or an extension to the “Strengthening the Reporting of Observational Studies in

Epidemiology” STROBE statement and STROBE-Vet for animal populations with items specific to spatial data (Smith et al., 2015; O’Connor et al., 2016; Sargeant et al., 2016; Vandenbroucke et al., 2014; von Elm et al., 2007). Journal editors can also initiate the process by providing their authors and reviewers with some basic map reporting guidelines, particularly for journals specializing in medical geography, spatial epidemiology, and spatial statistics.

Although disease maps and map-making processes are heterogenous, the current investigation identified choropleth maps as a common map type in the infectious zoonotic disease literature, and this can be an impetus for starting guidelines for this particular map type.

Choropleth maps are a preferred choice in health mapping as they can protect the privacy of individual health data and are more amenable to mapping the aggregated regional data that are most available and disclosed by disease surveillance systems (Cromley & McLafferty, 2011).

Although this study did not examine choropleth map characteristics in-depth such as evaluating the classification methods used to display data (e.g., natural, manual, quintile breaks), future

studies and guideline development could address the reporting of items unique to choropleth maps such as classification methods.

Disease maps should also be considered with the appropriate background population, which provide the basis for denominators in epidemiological metrics of disease (e.g., incidence rate or risk, prevalence, mortality rate). This review identified choropleth maps presenting both ratio data and count/amount data. Standardized values rather than absolute values are recommended to allow for comparisons across a map (Cromley & McLafferty, 2011). However, absolute values are still meaningful when background population data are not available, as might be the case with animal population data. A reporting guideline or a consensus statement on how disease maps should be reported can bring greater awareness to issues such as these.

The current investigation suggests that most studies may not reflect on geospatial biases and limitations. This may be related to incomplete reporting or may be because the biases and limitations are not present, due to the study design or control at the analysis stages of a study. In a review of spatial epidemiological approaches to inform leptospirosis surveillance, Dhewantara et al.(2019) found that even though studies evaluated spatial autocorrelation, they did not necessarily incorporate the spatial autocorrelation into the modelling framework. Future studies can assess how specific geospatial biases and limitations are reflected in the research design and analysis stages, in addition to reporting.

A common theme of biases and limitations identified in this study was related to spatial scale and aggregation of data. Fotheringham and Rogerson (1993) recommend that practitioners using methods of spatial analysis within GIS, should : 1) provide an introductory discussion of the spatial problem, 2) provide a discussion of alternative ways of evaluating and dealing with

the problem, and 3) discuss the ability to interpret the output associated with the alternative ways. To increase transparency and accountability, a map reporting guideline could include a recommendation for data tables to be presented together with disease maps. The reporting of data sources, the scale of the spatial data, and analytical methods and limitations can improve the interpretations and utility of disease maps. Examining data availability statements in this review suggests that data sharing for maps is not a regular or common practice, yet. A reporting guideline can also help foster an environment where data sources are reported adequately and encourage open access to geographical datasets where appropriate (i.e., in accord with ethic guidelines of releasing potentially identifying information). While reporting guidelines are not appropriate for use in assessing quality, they are intended to aid in clarity for publication

(Vandenbroucke, 2009). Guidelines can also increase critical thinking about design and analysis issues during map creation, leading to potentially higher quality maps.

Most disease maps mapped regions of Asia (25%). This is consistent with literature findings suggesting hotspots for the emergence of wildlife and vector-borne diseases are regions of lower latitude (Jones et al., 2008). In a review of global patterns of zoonotic disease in mammals by Han et al. (2016), most zoonoses were caused by bacteria followed by viruses, helminths, protozoa, and fungi (Han et al., 2016); consistent with the sample of disease maps captured in this review. Some maps in this study also mapped multiple zoonoses (13%).

Studying and mapping a composition of diseases that burden a population can help to scale up interventions and reduce global burden of disease (Kraemer, 2016). Most of the maps included in this study reported outcomes in animal populations (43%) compared to human populations

(30%). This could be a positive indication of research interest in animal population data as sentinel information in understanding human health risks. Further, it is important to understand 70

the spatial variation of risk factors, especially for infectious diseases with strong environmental and sociodemographic determinants, that are also spatially heterogenous (Mayfield et al., 2018).

In one included study in this review, Lawson and Rotejanaprasert (2018) advocate for the use of

Bayesian joint models for zoonotic diseases using the information of both human and animal populations in statistical modelling of disease patterns in space and time, producing a spatially correlated shared effect between human and animal populations.

3.4.1. Strengths and Limitations

The review evaluated a small subset of disease maps in the published scientific literature, which allowed the review to be manageable and provide a census of the literature through a scoping review. The review focussed on the evaluation of the recently published literature to reflect current practices. The review also included disease maps of all types (e.g., choropleth map, dot map, isopleth map), providing a broad characterization of disease maps in the literature assessed. The findings from this review can be used to support and inform the development of a reporting guideline for disease maps. While other reviews have examined spatial analytic methods and systems (Beard et al., 2018; Carroll et al., 2014; Dhewantara et al., 2019; Fritz et al., 2013; Smith et al., 2015), this review is the first to describe reporting practices of basic information associated with diseases maps; adding to the literature new evidence on characteristics of published disease maps.

This study followed a meta-research process, where included studies were assessed for eligibility based on the visual inspection of figures opposed to solely manuscript text. As a scoping review, the quality of the included studies and the reporting quality of disease maps were

not critically appraised. One limitation of this study was that data were not extracted in duplicate nor verified by a second reviewer as recommended for scoping reviews (Arksey & O’Malley,

2005).

3.5 Conclusion

The present study findings suggest that reporting and characteristics of disease maps are inconsistent, and readers should not accept disease maps uncritically. There are significant gaps in the reporting of map elements such as projection, necessary for reproducibility, and informed decision making. As such, there is a need for increasing capacities and educational training among user groups for map-making as important reporting items can be omitted. The findings from this study support the need for developing guidelines and recommendations related to reporting, understanding, and creating maps for spatial epidemiological analysis to inform diverse user groups.

3.6 References

Arksey, H., & O’Malley, L. (2005). Scoping studies: towards a methodological framework. International Journal of Social Research Methodology, 8(1), 19–32.

Beard, R., Wentz, E., & Scotch, M. (2018). A systematic review of spatial decision support systems in public health informatics supporting the identification of high risk areas for zoonotic disease outbreaks. International Journal of Health Geographics, 17(1), 38. https://doi.org/10.1186/s12942-018-0157-5

Bithell, J. F. (2000). A classification of disease mapping methods. Statistics in Medicine, 19(17– 18), 2203–2215.

Brewer, C. A. (2016). Designing Better Maps: A Guide for GIS Users (2nd ed.). ESRI Press.

Covidence. (2017). Covidence. Veritas Health Innovation.

Cromley, E. K., & McLafferty, S. L. (2011). GIS and public health. Guilford Press.

Dhewantara, P. W., Lau, C. L., Allan, K. J., Hu, W., Zhang, W., Mamun, A. A., & Magalhães, R. J. S. (2019). Spatial epidemiological approaches to inform leptospirosis surveillance and control: A systematic review and critical appraisal of methods. Zoonoses and Public Health, 66(2), 185– 206. https://doi.org/10.1111/zph.12549

Elo, S., & Kyngäs, H. (2008). The qualitative content analysis process. Journal of Advanced Nursing, 62(1), 107–115. https://doi.org/10.1111/j.1365-2648.2007.04569.x

Fotheringham, A., & Rogerson, P. A. (1993). GIS and spatial analytical problems. International Journal of Geographical Information Science, 7(1), 3–19.

Han, B. A., Kramer, A. M., & Drake, J. M. (2016). Global Patterns of Zoonotic Disease in Mammals. Trends in Parasitology, 32(7), 565–577. https://doi.org/10.1016/j.pt.2016.04.007

Jones, K. E., Patel, N. G., Levy, M. A., Storeygard, A., Balk, D., Gittleman, J. L., & Daszak, P. (2008). Global trends in emerging infectious diseases. Nature, 451(7181), 990–993. https://doi.org/10.1038/nature06536

Kraemer, M. U. G., Hay, S. I., Pigott, D. M., Smith, D. L., Wint, G. R. W., & Golding, N. (2016). Progress and Challenges in Infectious Disease Cartography. Trends in Parasitology, 32(1), 19–29. https://doi.org/10.1016/j.pt.2015.09.006

Lawson. (2013). Bayesian disease mapping: hierarchical modeling in spatial epidemiology. Chapman and Hall/CRC.

Lawson, A. B., & Rotejanaprasert, C. (2018). Bayesian spatial modeling for the joint analysis of zoonosis between human and animal populations. Spatial Statistics, 28, 8–20. https://doi.org/10.1016/j.spasta.2018.08.001

Mayer, J. D. (2000). Geography, ecology and emerging infectious diseases. Social Science & Medicine, 50(7), 937–952. https://doi.org/10.1016/S0277-9536(99)00346-9

Mayfield, H. J., Smith, C. S., Lowry, J. H., Watson, C. H., Baker, M. G., Kama, M., Nilles, E. J., & Lau, C. L. (2018). Predictive risk mapping of an environmentally-driven infectious disease using spatial Bayesian networks: A case study of leptospirosis in Fiji. Plos Neglected Tropical Diseases, 12(10). https://doi.org/10.1371/journal.pntd.0006857

Merck Veterinary Manual. (n.d.). Merck Veterinary Manual. Retrieved July 24, 2020, from https://www.merckvetmanual.com/

Microsoft Corporation. (n.d.). Microsoft Excel (Version 2018) [Computer software]. https://office.microsoft.com/excel.

Mitchell, A. (1999). The ESRI Guide to GIS Analysis,: Vol. Volume 1: Geographic Patterns and Relationships. ESRI Press.

O’Connor, A. M., Sargeant, J. M., Dohoo, I. R., Erb, H. N., Cevallos, M., Egger, M., Ersbøll, A. K., Martin, S. W., Nielsen, L. R., Pearl, D. L., Pfeiffer, D. U., Sanchez, J., Torrence, M. E., Vigre, H., Waldner, C., & Ward, M. P. (2016). Explanation and Elaboration Document for the STROBE-Vet Statement: Strengthening the Reporting of Observational Studies in Epidemiology - Veterinary Extension. Zoonoses and Public Health, 63(8), 662–698. https://doi.org/10.1111/zph.12315

R Core Team. (2019). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/.

Robertson, C., & Nelson, T. A. (2014). An Overview of Spatial Analysis of Emerging Infectious Diseases. The Professional Geographer, 66(4), 579–588. https://doi.org/10.1080/00330124.2014.907702

RStudio Team. (2020). RStudio: Integrated Development Environment for R. RStudio, PBC. http://www.rstudio.com/

Rushton, G. (2003). Public Health, GIS, and Spatial Analytic Tools. Annual Review of Public Health, 24(1), 43–56. https://doi.org/10.1146/annurev.publhealth.24.012902.140843

Straus, S., Tetroe, J., & Graham, I. D. (2013). Knowledge translation in health care: moving from evidence to practice. John Wiley & Sons.

Tobler, W. R. (1970). A Computer Movie Simulating Urban Growth in the Detroit Region. Economic Geography, 46, 234. https://doi.org/10.2307/143141

Vrbova, L., Stephen, C., Kasman, N., Boehnke, R., Doyle-Waters, M., Chablitt-Clark, A., Gibson, B., Brauer, M., & Patrick, D. (2009). Systematic Review of Surveillance Systems for Emerging Zoonotic Diseases (p. 57).

Waller, L. A., & Gotway, C. A. (2004). Applied spatial statistics for public health data (Vol. 368). John Wiley & Sons. 76

Wickham, H. (2016). ggplot2: elegant graphics for data analysis. Springer.

3.7 Figures

Figure 3-1. PRISMA Study Flow diagram.

Figure 3-2. Stacked bar chart of zoonotic diseases presented in included studies (n=302) by pathogen type and region mapped in the disease maps. Global maps refer to maps of the world. Multiregional maps refer to maps with more than one region mapped.

Figure 3-3. The included map types (n=505) presented by the data format reported by authors in map legends or article text.

3.8 Tables

Table 3.1 Search string inputted in Web of Science to identify published disease maps of zoonoses Date April 19, 2019 Platform/Interface Web of Science Databases Indexes=SCI-EXPANDED, SSCI, A&HCI, CPCI-S, CPCI-SSH Institution University of Guelph Search String: (TS=(zoono* AND (disease map* OR geographic information system OR GIS OR geospatial OR spatial OR geographic OR geomap* OR cartograph* OR cartogram* OR atlas OR geographic distribution OR disease distribution OR spatio?temporal OR spatial epidemiology OR geostatistic OR risk map OR health map OR map) )) Limits English, Document Types: Article, 2017-2018

Table 4.2 Study characteristics of 302 included studies with zoonotic disease maps published in 2017-2018 Study Characteristic (n=302) Count (%) Study Population Animal 130 (43.1) Human 90 (29.8) Both Animal - Human 65 (21.5) Invertebrate Vector 10 (3.3) Both Animal Invertebrate Vector - Human 7 (2.3) Zoonotic Pathogen type Virus 98 (32.4) Bacteria 102 (33.8) Helminths 31 (10.3) Protozoa 29 (9.6) Fungi 4 (1.3) Multiple Zoonoses 38 (12.6) Mapped regions of studies Africa 41 (13.6) Asia 76 (25.2) Europe 62 (20.5) North America 45 (14.9) South America 41 (13.6) Australia/Oceania 4 (1.3) Multiregional (if more than one map) 9 (3.0) Global (for world maps) 24 (7.9) Year published 2017 146 (48.3) 2018 156 (51.7) Map utility in Study Descriptive 199 (66) Analytical 103 (34) Data Availability Statement Availability in a repository 7 (2.3) Available from corresponding author or request 13 (4.3) Available in this published article/supplementary file 76 (25.2) Available from a third party with restrictions 12 (4.0) Not publicly available 13 (4.3) Publicly available and open access 27 (8.9) Other 4 (1.3) Not reported 150 (49.7)

Table 5.3 Disease map characteristics of 505 maps of human, animal, or vector zoonotic disease reported from 302 articles published in 2017-2018 Map Characteristic (n=505) Frequency %

Map Type Individual 442 (88.0) Composite 63 (12.0) Map figure type Choropleth 191 (37.8) Point or Symbol map 118 (23.4) Combination 85 (16.8) Continuous gradient 35 (6.9) Proportional Symbol 22 (4.4) Other 54 (10.7) Map Health Outcome Measures of Disease Frequency 446 (88.4) Molecular or Phylogeographic Typing 18 (3.6) Other 41 (8.0) Format of Mapped Data Categories 202 (40.0) Counts and Amounts 139 (27.5) Ratios 128 (25.4) Ranks 36 (7.1) Map Design Elements Projection Parameters reported 42 (8.3) Compass reported 268 (53.1) Legend reported 430 (85.1) Scale Bar reported 236 (46.7) Time reported 345 (68.3)

Table 6.4 Categories of study author reported geospatial methods and analyses Count (%) Geospatial Methods and Analyses (n=144)* Disease mapping and spatial regression models 61 (42.4) Spatial cluster analysis 30 (20.8) Spatial Autocorrelation analysis 22 (15.3) GIS based spatial analysis 9 (6.2) Point Sets and Distance Statistics 4 (2.8) Other 18 (12.5)

*From 103 studies reporting geospatial methods. An included study could report more than one category.

Table 7.5 Themes of study author reported geospatial biases and limitations Count (%) Geospatial Biases and Limitations Themes (n=60)* Spatial Scale and Aggregation of data versus disaggregate data 10 (16.7) Boundary problems 2 (3.3) Model Assumptions and goodness of fit 8 (13.3) Spatial autocorrelation 5 (8.3) Geolocation problems and spatial biases in sampling, geocoding, data collection 35 (58.3)

*From 50 included studies reporting geospatial specific biases and limitations. An included study could report more than one category

Chapter 4: Discussion and Conclusions

The overall work of this thesis is intended to advance the reporting of disease maps in epidemiological research. In chapter 2, it was demonstrated that the choice of map projections could be essential to the reproducibility of spatial analyses like the spatial scan test. Using the same data and the same spatial analysis for cluster detection, varying the map projection showed different results. In chapter 3, evidence was presented that reporting characteristics of disease maps are inconsistent. A total of 302 studies comprising 505 disease maps were reviewed for characteristics of published disease maps. The review identified that characteristics such as projection, scale, and compass are not always reported. The overall work of this thesis supports the development of a future guideline for disease maps.

4.0 Discussion and Summary of Key Findings

Disease maps are used in epidemiological investigations to communicate the geographic distribution of health events. Disease maps are also used to display the results of spatial statistical analyses, including spatial regression and cluster analyses (Pfeiffer & Stevens, 2015;

Robertson & Nelson, 2014; Rushton, 2003; Waller & Gotway, 2004). The spatial relation between two points in space, especially their distance, is central to most spatial statistical analyses.

Map projections are used to transform the 3-dimensional earth surface onto a 2- dimensional plane, which distort true distances, directions, and geographic shapes (Waller &

Gotway, 2004). Chapter 2 of this thesis presented a simulation study to assess tests for disease clustering and cluster detection and their related p-values under varying map projections. This 85

proof of concept study identified that different projections could affect the spatial statistics, particularly the circular spatial scan test. Results from the circular spatial scan test, with p-values estimated from 999 Monte Carlo simulations, indicate variability in the clusters detected, including cluster radius size, case counts and loci in the cluster, relative risk, standardized mortality risk. This can be expected considering a circle in one projection may be distorted to ellipse (or deviate from a circular shape) in another projection and thus capture different populations and cover different areas where a cluster is located. For example, in Chapter 2, under the Lambert Equal Area Conic projection, the spatial scan test detected a cluster with a population count of 84. Under the Albers projection, the spatial scan test detected a cluster with a population count of 103.

This study identified the importance of map projections in spatial statistical analyses of cluster detection, particularly for analyses based on distance measures in Cartesian coordinates.

The omission of reporting the projection of a disease map can have serious consequences for the reproducibility of analyses and interpretation of results in spatial epidemiological contexts.

Health officials are tasked with evaluating local disease cluster alarms and an important question is to determine if the cluster has occurred by chance or whether the excess is a result of a risk factor. Consequently, it becomes important to accurately identify clusters. Cluster detection methods are an increasingly common method in spatial epidemiological studies, as identified in

Chapter 3 of this thesis and previous literature (Dhewantara et al., 2019; Kirby et al., 2017;

Robertson & Nelson, 2014; Smith et al., 2015).

In chapter 3, a scoping review was conducted to broadly characterize studies of disease maps of zoonoses to understand and clarify knowledge, gaps, and inconsistencies. A total of 302

articles meeting inclusion criteria were reviewed, comprising 505 disease maps. The results of this review indicate that reporting practices of disease maps in publications may not be consistent. It was found that basic map design elements such as the legend and compass are commonly but not always reported. However, design elements such as the map projection were only reported in approximately 8% of disease maps included in the review. Reporting items such as the map scale and map projection can be important for: (1) Immediate transparent interpretation of the map; (2) subsequent data analysis or reproduction of results presented by a map; and (3) documentation purposes for prospective and retrospective comparisons. For example, sequential analysis of disease maps can be important to determine whether an outbreak is occurring (Lawson & Rotejanaprasert, 2018).

Of the 302 included studies, 103 studies reported on 144 geospatial data analyses. The most reported geospatial analyses included: spatial regression models and disease mapping analyses (e.g., interpolation) (42%), spatial autocorrelation analysis (15%), and local cluster detection (21%). Of the 302 included studies, 50 studies reported on geospatial biases and limitations. Most commonly identified themes of biases and limitations associated with geospatial analyses included: geolocation problems and spatial bias in sampling (58%); problems arising from spatial scale and aggregation of data (17%); and lack of accounting for spatial autocorrelation (8%).

The choice of projecting a coordinate system is important for maps, especially for maps visualizing spatial statistics. In such maps (e.g. where spatial statistics are represented) preserving and representing relative distances is a priority (Waller & Gotway, 2004) as this affects both the visual and statistical interpretations. However, reporting projection is good practise even for disease maps where the representation of spatial distances is not a priority. In 87

chapter 3, most included studies (66%) utilized maps for descriptive purposes compared to studies that utilized maps for presenting spatial analyses. Reporting map projections could be advocated as a standard practice for adoption in the scientific community. Standardizing this practice can result in easier comparisons between maps used to represent descriptive data and maps used to display statistical data for the same phenomena and/or comparisons of the same geographic regions.

Chapter 3 also presents evidence that disease map characteristics and applications are heterogeneous. Despite the heterogeneous nature of disease maps, select map characteristics are foundational in how maps are created, such as map projection. Map projection affects both the visual interpretation of a map and the spatial statistical analyses of a map as demonstrated in

Chapter 2. The overall work of this thesis supports and calls for the development of a reporting guideline for disease maps that includes a recommendation for reporting items such as the map projection. Authors of infectious disease epidemiology reviews have also called for similar standardizing and guidelines for spatial epidemiological analysis (Dhewantara et al., 2019;

Robertson & Nelson, 2014; Smith et al., 2015). A reporting guideline with reporting items that are agreed upon, unique to map features, and accepted by the epidemiological community can increase transparency, educational training, accountability and help foster critical thinking about design and analysis issues during map creation, leading to potentially higher quality maps, and better evidence-informed decision making.

4.1 Strengths and Limitations

The study in Chapter 2 complements work by Berke (2019). These studies are the first to demonstrate differences in cluster analyses as a result of map projection. In Chapter 2, a case- 88

control dataset was simulated, that is a homogenous Poisson marked point process. A disease cluster was subsequently simulated within this dataset. The simulated maps were analyzed using the Cuzick-Edward’s test (k nearest neighbours statistic) for clustering and the spatial scan test for cluster detection under varying projection. These statistics are commonly used with spatial point data in public health (Hinrichsen et al., 2009). Since the study investigated point data, the findings suggest they could be relevant to aggregate areal or regional data. Berke (2019) identified that varying projection affects spatial statistics of regional data. Variable results from the circular spatial scan test in Chapter 2 are consistent with findings from Berke (2019), where varying projection for regional data produced different results for another test for clustering,

Moran’s I.

The review in Chapter 3 is the first of its kind to synthesize the reporting characteristics of disease maps. While other reviews have examined spatial analytic methods and systems

(Beard et al., 2018; Carroll et al., 2014; Dhewantara et al., 2019; Fritz et al., 2013; Smith et al.,

2015), the review in Chapter 3 examined reporting of basic information associated with diseases maps. The review also included disease maps of all types (e.g., choropleth map, dot map, isopleth map), providing a broad characterization of studies of disease maps.

The choice to study disease maps of “zoonoses” proved to be difficult due to the large variety of zoonotic diseases (e.g., vector-borne, parasitic, bacteria, viruses) and lack of a definitive register of zoonoses, particularly for emerging zoonoses. However, studying zoonoses also poses the opportunity to gain insights to mapping practices across disciplines (e.g., ecology, human and veterinary epidemiology, public health).

The research conducted also has some limitations. While the proof of concept study in chapter 2 illustrated that cluster detection could be conditional on the specific map projection used for disease maps, the study was not able to quantify the effect of projection, and only presented that differences in analyses result qualitatively. Future studies can investigate more scenarios and different parameter settings for the spatial scan test under varying projection.

In Chapter 3, the broad nature of the review objectives did not allow for collecting data on finer details of reporting practices pertaining to specific map types, specific spatial analytic methods, and other reporting characteristics. Future studies can focus on reporting practices pertaining to specific areas such as map type (e.g., a review of choropleth maps), or spatial analytic method (e.g. review of studies reporting the spatial scan test) to inform the development of reporting guidelines better.

4.2 Conclusions

The research described in this thesis aims to advance the reporting of disease maps for reproducibility and critical appraisal. In chapter 2, it was demonstrated that the projection of a disease map can affect spatial statistical analyses of disease cluster detection. In Chapter 3, it was shown that map projections are not often reported in the published literature with disease maps.

Chapter 3 also presented evidence that reporting characteristics are heterogeneous. The overall work of this thesis supports and calls for the development of a reporting guideline for disease maps. Such a reporting guideline can benefit and advance the epidemiological community by minimizing misinterpretations and will also be useful to educators, researchers, public health practitioners, decision-makers, science journalists, and journal editors.

4.3 References

Berke, O. (2019). Dazed and confused: How map projections affect disease map analysis and perception. Frontiers in Veterinary Science, 6. https://doi.org/10.3389/conf.fvets.2019.05.00013

Carroll, L. N., Au, A. P., Detwiler, L. T., Fu, T., Painter, I. S., & Abernethy, N. F. (2014). Visualization and analytics tools for infectious disease epidemiology: A systematic review. Journal of Biomedical Informatics, 51, 287–298. https://doi.org/10.1016/j.jbi.2014.04.006

Cromley, E. K., & McLafferty, S. L. (2011). GIS and public health. Guilford Press.

Fritz, C. E., Schuurman, N., Robertson, C., & Lear, S. (2013). A scoping review of spatial cluster analysis techniques for point-event data. Geospatial Health, 7(2), 183–198. https://doi.org/10.4081/gh.2013.79

Robertson, C., & Nelson, T. A. (2014). An Overview of Spatial Analysis of Emerging Infectious Diseases. The Professional Geographer, 66(4), 579–588. https://doi.org/10.1080/00330124.2014.907702

Rushton, G. (2003). Public Health, GIS, and Spatial Analytic Tools. Annual Review of Public Health, 24(1), 43–56. https://doi.org/10.1146/annurev.publhealth.24.012902.140843

Waller, L. A., & Gotway, C. A. (2004). Applied spatial statistics for public health data (Vol. 368). John Wiley & Sons.

APPENDICES

Appendix 1. Projections

Projection strings used in R studio to transform spatial dataset using the rgdal package’s spTransform function.

Original Projection proj=lcc +lat_1=49 +lat_2=77 +lat_0=63.390675 +lon_0=-91.86666666666666 +x_0=6200000 +y_0=3000000 +datum=NAD83 +units=m +no_defs +ellps=GRS80 +towgs84=0,0,0"

Lambert Equal Area Conic proj=leac +lat_1=49 +lat_2=77+lat_0=63 +lon_0=-91+units=km +datum=WGS84"

UTM 17N proj=utm +zone=17N +units=km +datum=WGS84"

Robinson proj=robin +units=km +datum=WGS84

Mercator proj=merc +units=km +datum=WGS84"

Albers

+proj=aea +lat_1=20 +lat_2=60 +lat_0=40 +lon_0=-96 +x_0=0 +y_0=0 +units=km +ellps=GRS80 +datum=NAD83"

Appendix 2. Study Protocol

The following investigation seeks to assess, and broadly characterize the use of disease maps for reporting health outcomes in the literature. Disease maps are used in many population health studies such as ones examining environmental exposures, social determinants of health, and public health interventions. In this investigation, we investigate disease maps with respect to zoonotic infectious diseases. Zoonotic health outcomes are sensitive to geospatial and temporal changes, and so disease map representations are a valuable tool for deciphering any space and place based effects.

The recent use of disease maps in the peer-reviewed literature on zoonotic diseases will be reviewed with respect to the following questions:

1) What are map applications? 2) Which basic map elements (e.g. projection and scale) are reported? 3) What are reported challenges with using spatial data and analyses for map representations, inferences and interpretations?

Inclusion Criteria

All study designs reporting on a zoonotic infectious disease in a population

Studies reporting and using disease map(s)

Time: 2017-2018

Methods:

Information sources: Medline-Ovid, Web of Science, CAB Direct Abstracts, and Agricoloa databases

Data management and collection: Screening, and data extraction of studies will be conducted by two reviewers in duplicate using forms constructed in Excel. Any disagreements on study selection and data extraction will be cleared by a third reviewer and discussion.

Data synthesis: Study characteristics will be quantitatively and qualitatively summarized. Data extraction form will be informed by a data charting pilot and in consultation with experts.

Integrated and end of KT plan: Experts in the field e.g. spatial epidemiologists, medical geographers, health service researchers, zoonotic and infectious disease experts will be consulted to guide research questions and data collection. Study findings are expected to be disseminated through presentations at conferences and an end of study publication available to knowledge users and researchers.

Appendix 3. Search Strategy

Date April 19, 2019 Platform/Interface Web of Science Databases Indexes=SCI-EXPANDED, SSCI, A&HCI, CPCI-S, CPCI-SSH Institution University of Guelph Search String: (TS=(zoono* AND (disease map* OR geographic information system OR GIS OR geospatial OR spatial OR geographic OR geomap* OR cartograph* OR cartogram* OR atlas OR geographic distribution OR disease distribution OR spatio?temporal OR spatial epidemiology OR geostatistic OR risk map OR health map OR map) )) Limits English, Document Types: Article, 2017-2018 Hits 595

Date April 19, 2019 Platform/Interface Proquest Databases Agricola Institution University of Guelph Search String: zoono* AND (disease map* OR geographic information system OR GIS OR geospatial OR spatial OR geographic OR geomap* OR cartograph* OR cartogram* OR atlas OR geographic distribution OR disease distribution OR spatio#temporal OR spatial epidemiology OR geostatistic OR risk map OR health map OR map) Limits English, Scholarly Journals, Document Types: Journal Article or Journal, 2017-2018 Hits 272

Date April 19, 2019 Platform/Interface CAB-Direct Institution University of Guelph Search String: ((zoono*) AND ((disease map* ) OR (geographic information system ) OR (GIS ) OR (geospatial) OR (spatial) OR ( geographic ) OR (geomap* ) OR (cartograph* ) OR (cartogram* ) OR (atlas ) OR (geographic distribution ) OR (spatio?temporal ) OR (spatial epidemiology ) OR (geostatistic ) OR (risk map ) OR (health map ) OR (map)) Limits English, Item Types: Journal Article or Journal Issue, 2017-2018 Hits 509

Date April 19, 2019 Platform/Interface Ovid Databases Medline Institution University of Guelph Search String: 1 Zoonoses/ 2 zoono*.mp. 3 geographic information systems.mp. or Geographic Information Systems/ 4 GEOGRAPHY, MEDICAL/ or GEOGRAPHY/ or geography.mp. 5 Topography, Medical/ 6 Environmental Medicine/sn, td [Statistics & Numerical Data, Trends] 7 geospatial.mp.

SPATIAL REGRESSION/ or MODELS, SPATIAL INTERACTION/ or spatial.mp. or SPATIAL 8 ANALYSIS/ 9 GEOGRAPHIC MAPPING/ or geographic.mp. 10 geomap*.mp. 11 cartograph*.mp. 12 cartogram*.mp. 13 atlas.mp. (Geographic adj distribution).mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub-heading word, keyword heading word, 14 organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] 15 spatial regression.mp. 16 spatio?temporal.mp. 17 Cluster Analysis/ 18 spatial epidemiology.mp. 19 Geostatistic*.mp. 20 risk map.mp. 21 health map.mp. 22 map*.mp. 23 MAPS AS TOPIC/ 24 disease map*.mp. (disease adj distribution).mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub-heading word, keyword heading word, 25 organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] 26 Disease Reservoirs/ 27 disease outbreaks/ or epidemics/ or pandemics/ or endemic diseases/ 28 disease transmission, infectious/ disease notification/ or epidemiological monitoring/ or population surveillance/ or 29 biosurveillance/ or public health surveillance/ or sentinel surveillance/ 30 1 or 2 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 or 17 or 18 or 19 31 or 20 or 21 or 22 or 23 or 24 32 25 or 26 or 27 or 28 or 29 or 31 33 30 and 32 34 limit 33 to (yr="2017 - 2018" and english)

Appendix 4. Zoonotic Diseases

Adapted from Vbrova 2009, Appendix 1: Emerging and re-emerging zoonoses listed by agent

Viruses and Prions (n=54) Bacteria and Rickettsia (N=30) Parasites (N=10) Andes Aeromonas caviae Anisakis simplex Australian bat lyssavirus A. hydrophila Echinococcus granulosus Bagaza A. veronii (var. sobria) Loa loa Banna Anaplasma phagocytophila Metorchis conjunctus Barmah Forest Bacillus anthracis Onchocerca volvulus California encephalitis Borrelia burgdorferi Strongyloides stercoralis Cercopithecine herpes Brucella melitensis Taenia solium Chikungunya Campylobacter fetus Trichinella spiralis Crimean-Congo hemorrhagic fever C. jejuni Wuchereria bancrofti Dengue Coxiella burnetiid (q fever) Eastern equine encephalitis Clostridium botulinum Tickborne encephalitis Ehrlichia chaffeensis Protozoa (N=11) Guama E. ewingii Guanarito Escherichia coli Babesia microti Hantaan Francisella tularensis Cryptosporidium hominis Hendra Leptospira interrogans (Leptospirosis) C. parvum Influenza A (zoonotic influenzas only) Listeria monocytogenes Giardia duodenalis Japanese encephalitis Mycobacterium avium Leishmania donovani Junin M. bovis L. infantum Laguna Negra M. marinum Plasmodium falciparum Lassa Rickettsia prowazekii P. vivax Machupo Salmonella enteritidis Toxoplasma gondii Marburg S. typhi Trypanosoma brucei Mayaro S. typhimurium T. cruzi Menangle Shigella dysenteriae Middle East respiratory syndrome Vibrio cholerae coronavirus (MERS-CoV) V. parahaemolyticus Fungi (N=9) Monkeypox V. vulnificus Murray Valley encephalitis Yersinia enterocolitica Histoplasma capsulatum Nipah Y. pestis Malassezia pachydermatis O'nyong-nyong Penicillium marneffei Oropouche Picobirnavirus Encephalitozoon cuniculi Powassan virus E. hellem Puumala E. intestinalis Rabies Enterocytozoon bieneusi Reston Ebola Nosema connori Rift Valley fever Trachipleistophora hominis Ross River Sabia Salehabad Sandfly fever Naples Severe acute respiratory syndrome coronavirus Seoul Sin Nombre Sindbis St. Louis encephalitis Venezuelan equine encephalitis Wesselsbron West Nile Western equine encephalitis Yellow fever Zaire Ebola Zika Bovine spongiform encephalopathy agent

Appendix 5. Relevance Screening and Extraction Forms

Relevance Screening Inclusion Criteria:

Inclusion Criteria Reasons for Exclusion • Map of a zoonotic disease in a • Is not a research article from a journal population • Article has no map • Indication of geographically • Not a disease map (e.g. exclude maps of non- referenced health data within a health outcomes such as economic outcomes, geographic boundary sampling sites unless test results reported) • Not a zoonotic disease of a population (e.g. map of chicken pox disease incidence) • Is a dynamic map (e.g. interactive, non-static, real-time, web maps) • Non-English • No disease map of zoonoses by way of: o No zoonotic pathogen/disease/health outcome in a defined population o map of vector agents that does not report on any health outcomes (e.g. disease risk) related to the zoonotic disease/pathogen in a population o No geospatial health data AND appropriate geographic boundary • Exclude maps that characterize pathogenic agents (e.g. bacteria) in abiotic sources such as water and not in a biotic population

Data extraction guide

Study characteristics

First Author Surname Surname of the first author listed Ex. For Joe Carlie Smith, type in Smith

Pub Year The publication year of the article. Note this is not the year the article was downloaded.

Animal/Human Population Select one of: • Animal • Human • Invertebrate Animal – Vector • Invertebrate Animal - Other

If article disease maps only report on vertebrate animal populations, select animal.

If article disease maps only report on invertebrate animal populations that are established vectors, select “Invertebrate Animal – Vector” • Invertebrate animals-vectors can transmit zoonotic pathogen (e.g. arthropod insects that bite susceptible hosts).

If article maps only report on invertebrate animal populations that are not established vectors, select “Invertebrate Animal – other” • Other invertebrate animals-other can include a population (e.g. worms, small sea animals) tested for a pathogen but not established vector for pathogen to humans/vertebrate hosts.

If article disease maps only report on human populations, select human.

If article disease maps report on both animal and human populations, select the most appropriate options for both:

• Both Animal – Human • Both Animal Invertebrate vector -Human • Both Animal Invertebrate other - Human

If the study population is not explicitly defined in text or by geographic boundaries, the article is excluded.

Indicates the zoonoses pathogen agent specified by the authors. Where this is not available indicate the zoonotic disease name/transmission route, or disease term specified by the authors. Zoonoses

For unspecified zoonoses, type in “unspecified zoonoses”. For a group of zoonoses, type in “multiple zoonoses” if they don’t have a group name (e.g. genus). 99

Ex. Echinococcus is the genus name for the tape worm Echinococcus multilocularis (pathogenic agent). Echinococcosis is the disease name.

Ex. Malaria is the disease name, and the pathogenic agents are Plasmodium parasites- P. falciparum, P. vivax, P. ovale, and P. malariae

Study utility of disease map(s) Select one of: • Descriptive • Analytical

Select Descriptive if: A disease map is solely used for descriptive purposes. For example, incidence/mortality/other disease outcome is presented visually in a map only to establish study context. Such maps can be hypothesis generating but do not give rise to significant conclusions/inferences in the article.

Ex. A dot map displaying positive cases of x pathogen in the background section of an article; remainder of article investigates behaviour patterns of population that are not spatially/geographically relevant.

• No casual inferences from map are made but are valued for reporting noteworthy spatial pattern

Select Analytical if: Any one disease map in the article is used to conduct/the result of spatial analyses (i.e. derived from spatial analyses or the basis for geospatial analyses reported in the methods/results). Such spatial analyses can be hypothesis testing and can report statistical significance.

Ex. Spatial regression analyses, Bayesian modelling to forecast risk estimates, cluster detection to make statistical inferences

Spatial data analysis: “general ability to manipulate spatial data into different forms and extract additional meaning as a result” (Bailey 1994)

Ex. A map of the population is presented with spatially modelled risk estimates for disease X spatially and temporally prospectively

• Makes spatial comparisons between groups of populations to make inferences between exposures and outcomes of interests

The number of figures with disease maps in the article.

Maps that share the same legend in a composite figure are to be counted as No. of Disease Map (DM) figures 1.

Ex. An article with two separate maps, and 1 composite figure of maps that share same map types/legends, type in: 2 maps, 1 map series. 100

Ex. If a composite figure is comprised of maps of different types, count them individually. 3 maps in a composite figure that are different by map type and legend, type “3”

Map reporting characteristics Identify the following for each map figure (Disease Map 1 is denoted DM1 and so on in chronological order as they appear cited in the paper). One composite figure will be considered as one map

Indicate the map purpose stated by authors. Indicate the figure number in data field. If unclear, copy and paste the figure description or caption of the Map Purpose map.

Ex. Figure 3b= risk distribution of disease x

Indicate the health outcome represented by the map. If more than one, list them all separated by a comma

Ex: Incidence, Prevalence, mortality Map health outcome For health outcomes, since there are sometimes discrepancies in how the author refers to the health outcome (incidence/prevalence/case), we will go with what the legend specifies> if not, then the caption> and if not, then the article text.

Map Type Select one of: • Point/Dot • Choropleth • Isopleth • Diffusion • Cartogram • Dasymetric • Graduated Symbol • Combination (e.g. point and choropleth map) • Other Combination maps will include more than one type of map visualization categories. If none of the categories fit a map visualization for a given figure, select ‘other”.

Select one of: Compass/North arrow • Yes • Noe Select Yes if you can identify a north arrow/compass on the map

101

Projection/Transformation/Geo Indicates the projection parameters reported. graphic Coordinate System Parameters Reported Ex. WGS 1984, UTM zone 18N

Typically reported on map, in figure caption, or the methods section. Should be specified by the projection datum and coordinate system, e.g. NAD83 or WGS84 or UTM or Albers or Mercator or Lambert ect.

NR for not reported

Select one of: • Visual scale bar Scale • Map figure text (includes figure caption) • X and Y axes (metric or coordinates) • Not reported

Geographical Region Indicates the region of the disease map Select one of: • North America • South America • Europe • Africa • Asia • Australia/Oceania • Multiregional (if more than one region) • Global (for world maps)

See https://unstats.un.org/unsd/methodology/m49/ for country classification

Format of mapped data Select the format of the data value listed in the maps. This should be presented in the legend: • Categories Ex. Nominal

• Ranks Ex. Ordinal; features in order, from high to low

• Counts and amounts Ex. Frequency; Counts and amounts show you total numbers

• Ratios Ex. people per household, population per square mile,%

Legend Select one of: • Yes 102

• No

Select yes if time of map data collection is reported (e.g. a date/duration provided) or time dimension considered, select no if no timepoint/ time frame reported. • Yes Time • No For modelled data, this refers to the time period for which the map is considered (e.g. forecasted projections for geographic range in year 2030 or present time)

Map interpretations and geospatial analytical maps

What geospatial analyses and Indicate any geospatial method and analyses conducted in the study. methods were conducted? Examples of analyses and methods: • Spatial regression analysis • Moran’s I test • Spatial scan (scan test, flex scan test) • Spatial autocorrelation algorithms • Space-time k-function • Knox test • Network modeling • K-nearest neighbor • Nearest neighbor analysis • Getis-Ord Gi(d) statistic • Smoothing/Empirical Bayes smoothing • Kriging • Environmental risk prediction model

Was there a discussion of bias/ Select one of: limitations pertaining to disease • Yes map (geospatial) inferences? • No

If yes, what biases and Indicate any biases/limitations reported by authors with respect to map limitations were inferences reported/explained? Ex. MAUP bias – modifiable areal unit problem bias, also known as CSOP Change of Support Problem. Edge Effect Visual bias Spatial misclassification

103

Appendix 6. Included Studies (n=302)

1. Abedi AA, Shako JC, Gaudart J, et al. Ecologic features of plague outbreak areas, Democratic Republic of the Congo, 2004-2014. Emerg Infect Dis. 2018;24(2):210-220. doi:10.3201/eid2402.160122 2. Abuzaid AA, Abdoon AM, Aldahan MA, et al. Cutaneous Leishmaniasis in Saudi Arabia: A Comprehensive Overview. Vector-Borne Zoonotic Dis. 2017;17(10):673-684. doi:10.1089/vbz.2017.2119 3. Adeola OA, Olugasa BO, Emikpe BO. Molecular detection of influenza A(H1N1)pdm09 viruses with M genes from human pandemic strains among Nigerian pigs, 2013-2015: Implications and associated risk factors. Epidemiol Infect. 2017;145(16):3345-3360. doi:10.1017/S0950268817002503 4. Afelt A, Lacroix A, Zawadzka-Pawlewska U, Pokojski W, Buchy P, Frutos R. Distribution of bat-borne viruses and environment patterns. Infect Genet Evol. 2018;58(September 2017):181-191. doi:10.1016/j.meegid.2017.12.009 5. Ahmadkhani M, Alesheikh AA, Khakifirouz S, Salehi-Vaziri M. Space-time epidemiology of Crimean-Congo hemorrhagic fever (CCHF) in Iran. Ticks Tick Borne Dis. 2018;9(2):207-216. doi:10.1016/j.ttbdis.2017.09.006 6. Akkou M, Bouchiat C, Antri K, et al. New host shift from human to cows within Staphylococcus aureus involved in bovine mastitis and nasal carriage of animal’s caretakers. Vet Microbiol. 2018;223(June):173-180. doi:10.1016/j.vetmic.2018.08.003 7. Aleanizy FS, Mohmed N, Alqahtani FY, El Hadi Mohamed RA. Outbreak of Middle East respiratory syndrome coronavirus in Saudi Arabia: A retrospective study. BMC Infect Dis. 2017;17(1):1-7. doi:10.1186/s12879-016-2137-3 8. Alegria-Moran R, Miranda D, Barnard M, Parra A, Lapierre L. Characterization of the epidemiology of bat-borne rabies in Chile between 2003 and 2013. Prev Vet Med. 2017;143:30-38. doi:10.1016/j.prevetmed.2017.05.012 9. Allen T, Murray KA, Zambrana-Torrelio C, et al. Global hotspots and correlates of emerging zoonotic diseases. Nat Commun. 2017;8(1):1124. doi:10.1038/s41467-017-00923-8 10. Alvarez J, Whitten T, Branscum AJ, et al. Understanding Q Fever Risk to Humans in Minnesota Through the Analysis of Spatiotemporal Trends. Vector-Borne Zoonotic Dis. 2018;18(2):89-95. doi:10.1089/vbz.2017.2132 11. Álvarez-Hernández G, Roldán JFG, Milan NSH, Lash RR, Behravesh CB, Paddock CD. Rocky Mountain spotted fever in Mexico: past, present, and future. Lancet Infect Dis. 2017;17(6):e189-e196. doi:10.1016/S1473-3099(17)30173-1 12. Alves WC, Rossi GAM, Lopes WDZ, et al. Geospatial distribution and risk factors for bovine cysticercosis in the state of Rondônia, Brazil. Pesqui Vet Bras. 2017;37(9):931-936. doi:10.1590/s0100-736x2017000900006 13. Amarilla ACF, Pompei JCA, Araujo DB, et al. Re-emergence of rabies virus maintained by canid populations in Paraguay. Zoonoses Public Health. 2018;65(1):222-226. doi:10.1111/zph.12392 14. Amato B, Di Marco Lo Presti V, Gerace E, et al. Molecular epidemiology of Mycobacterium tuberculosis complex strains isolated from livestock and wild animals in Italy suggests the need for a different eradication strategy for bovine tuberculosis. Transbound Emerg Dis. 2018;65(2):e416-e424. doi:10.1111/tbed.12776 15. Anderson TC, Marsden-Haug N, Morris JF, et al. Multistate Outbreak of Human Salmonella Typhimurium Infections Linked to Pet Hedgehogs – United States, 2011–2013. Zoonoses Public Health. 2017;64(4):290-298. doi:10.1111/zph.12310 16. Anker JCH, Koch A, Ethelberg S, Mølbak K, Larsen J, Jepsen MR. Distance to pig farms as risk factor for community- onset livestock-associated MRSA CC398 infection in persons without known contact to pig farms—A nationwide study. Zoonoses Public Health. 2018;65(3):352-360. doi:10.1111/zph.12441 17. Aprea G, Amoroso MG, Di Bartolo I, et al. Molecular detection and phylogenetic analysis of hepatitis E virus strains circulating in wild boars in south-central Italy. Transbound Emerg Dis. 2018;65(1):e25-e31. doi:10.1111/tbed.12661 18. Artun O, Kavur H. Investigation of the spatial distribution of sandfly species and cutaneous leishmaniasis risk factors by using geographical information system technologies in Karaisali district of Adana province, Turkey. J Vector Borne Dis. 2017;54(3):233-239. doi:10.4103/0972-9062.217614 19. Atherstone C, Smith E, Ochungo P, Roesel K, Grace D. Assessing the Potential Role of Pigs in the Epidemiology of Ebola Virus in Uganda. Transbound Emerg Dis. 2017;64(2):333-343. doi:10.1111/tbed.12394

104

20. Awada L, Tizzani P, Noh SM, et al. Global dynamics of highly pathogenic avian influenza outbreaks in poultry between 2005 and 2016—Focus on distance and rate of spread. Transbound Emerg Dis. 2018;65(6):2006-2016. doi:10.1111/tbed.12986 21. Azcona-Gutiérrez JM, de Lucio A, Hernández-De-Mingo M, et al. Molecular diversity and frequency of the diarrheagenic enteric protozoan Giardia duodenalis and Cryptosporidium spp. in a hospital setting in Northern Spain. PLoS One. 2017;12(6):1-21. doi:10.1371/journal.pone.0178575 22. Becker DJ, Bergner LM, Bentz AB, Orton RJ, Altizer S, Streicker DG. Genetic diversity, infection prevalence, and possible transmission routes of Bartonella spp. in vampire bats. PLoS Negl Trop Dis. 2018;12(9):1-21. doi:10.1371/journal.pntd.0006786 23. Beerens N, Heutink R, Bergervoet SA, Harders F, Bossers A, Koch G. Multiple reassorted viruses as cause of highly pathogenic avian influenza A(H5N8) virus epidemic, the Netherlands, 2016. Emerg Infect Dis. 2017;23(12):1966- 1973. doi:10.3201/eid2312.171062 24. Benavides JA, Rojas Paniagua E, Hampson K, Valderrama W, Streicker DG. Quantifying the burden of vampire bat rabies in Peruvian livestock. PLoS Negl Trop Dis. 2017;11(12):1-17. doi:10.1371/journal.pntd.0006105 25. Benitez A do N, Martins FDC, Mareze M, et al. Spatial and simultaneous representative seroprevalence of anti- Toxoplasma gondii antibodies in owners and their domiciled dogs in a major city of southern Brazil. PLoS One. 2017;12(7):1-18. doi:10.1371/journal.pone.0180906 26. Berger KA, Pigott DM, Tomlinson F, et al. The geographic variation of surveillance and zoonotic spillover potential of influenza viruses in domestic poultry and swine. Open Forum Infect Dis. 2018;5(12):1-9. doi:10.1093/ofid/ofy318 27. Betsem E, Kaidarova Z, Stramer SL, et al. Correlation of West Nile virus incidence in donated blood with West Nile neuroinvasive disease rates, United States, 2010–2012. Emerg Infect Dis. 2017;23(2):212-219. doi:10.3201/eid2302.161058 28. Birch CPD, Goddard A, Tearne O. A new bovine tuberculosis model for England and Wales (BoTMEW) to simulate epidemiology, surveillance and control. BMC Vet Res. 2018;14(1):1-16. doi:10.1186/s12917-018-1595-9 29. Biscornet L, Dellagi K, Pagès F, et al. Human leptospirosis in Seychelles: A prospective study confirms the heavy burden of the disease but suggests that rats are not the main reservoir. PLoS Negl Trop Dis. 2017;11(8):1-22. doi:10.1371/journal.pntd.0005831 30. Blackburn JK, Matakarimov S, Kozhokeeva S, et al. Modeling the ecological niche of bacillus anthracis to map anthrax risk in Kyrgyzstan. Am J Trop Med Hyg. 2017;96(3):550-556. doi:10.4269/ajtmh.16-0758 31. Bold B, Hattendorf J, Shagj A, et al. Patients with cystic echinococcosis in the three national referral centers of Mongolia: A model for CE management assessment. PLoS Negl Trop Dis. 2018;12(8):1-14. doi:10.1371/journal.pntd.0006686 32. Bond KA, Franklin L, Sutton B, Stevenson MA, Firestone SM. Review of 20 years of human acute Q fever notifications in Victoria, 1994–2013. Aust Vet J. 2018;96(6):223-230. doi:10.1111/avj.12704 33. Boroduske A, Trofimova J, Kibilds J, et al. Coxiella burnetii (Q fever) infection in dairy cattle and associated risk factors in Latvia. Epidemiol Infect. 2017;145(10):2011-2019. doi:10.1017/S0950268817000838 34. Braae UC, Devleesschauwer B, Sithole F, Wang Z, Willingham AL. Mapping occurrence of Taenia solium taeniosis/cysticercosis and areas at risk of porcine cysticercosis in Central America and the Caribbean basin. Parasites and Vectors. 2017;10(1):424. doi:10.1186/s13071-017-2362-7 35. Braae UC, Hung NM, Satrija F, Khieu V, Zhou XN, Willingham AL. Porcine cysticercosis (Taenia solium and Taenia asiatica): Mapping occurrence and areas potentially at risk in East and Southeast Asia. Parasites and Vectors. 2018;11(1):1-11. doi:10.1186/s13071-018-3203-z 36. Brand RF, Rostal MK, Kemp A, et al. A phytosociological analysis and description of wetland vegetation and ecological factors associated with locations of high mortality for the 2010-11 Rift Valley fever outbreak in South Africa. PLoS One. 2018;13(2):1-26. doi:10.1371/journal.pone.0191585 37. Brasil P, Zalis MG, de Pina-Costa A, et al. Outbreak of human malaria caused by Plasmodium simium in the Atlantic Forest in Rio de Janeiro: a molecular epidemiological investigation. Lancet Glob Heal. 2017;5(10):e1038-e1046. doi:10.1016/S2214-109X(17)30333-9

105

38. Bravo-Vasquez N, Karlsson EA, Jimenez-Bluhm P, et al. Swine influenza virus (H1N2) characterization and transmission in ferrets, Chile. Emerg Infect Dis. 2017;23(2):241-251. doi:10.3201/eid2302.161374 39. Brehony C, Cullinan J, Cormican M, Morris D. Shiga toxigenic Escherichia coli incidence is related to small area variation in cattle density in a region in Ireland. Sci Total Environ. 2018;637-638:865-870. doi:10.1016/j.scitotenv.2018.05.038 40. Brooke RJ, Teunis PFM, Kretzschmar MEE, Wielders CCH, Schneeberger PM, Waller LA. Use of a Dose–Response Model to Study Temporal Trends in Spatial Exposure to Coxiella burnetii: Analysis of a Multiyear Outbreak of Q Fever. Zoonoses Public Health. 2017;64(2):118-126. doi:10.1111/zph.12288 41. Brookes VJ, Gill GS, Singh CK, et al. Exploring animal rabies endemicity to inform control programmes in Punjab, India. Zoonoses Public Health. 2018;65(1):e54-e65. doi:10.1111/zph.12409 42. Bruhn FRP, Morais MHF, Cardoso DL, Bruhn NCP, Ferreira F, Rocha CMBM Da. Spatial and temporal relationships between human and canine visceral leishmaniases in Belo Horizonte, Minas Gerais, 2006-2013. Parasites and Vectors. 2018;11(1):1-11. doi:10.1186/s13071-018-2877-6 43. Bui CM, Gardner L, MacIntyre CR, Sarkar S. Erratum: Correction: Influenza A H5N1 and H7N9 in China: A spatial risk analysis (PloS one (2017) 12 4 (e0174980)). PLoS One. 2017;12(4):e0176903. doi:10.1371/journal.pone.0176903 44. Cabal A, Vicente J, Alvarez J, et al. Human influence and biotic homogenization drive the distribution of Escherichia coli virulence genes in natural habitats. Microbiologyopen. 2017;6(3):e00445. doi:10.1002/mbo3.445 45. Čabanová V, Kocák P, Víchová B, Miterpáková M. First autochthonous cases of canine thelaziosis in Slovakia: A new affected area in Central Europe. Parasites and Vectors. 2017;10(1):179. doi:10.1186/s13071-017-2128-2 46. Camp J V., Haider R, Porea D, Oslobanu LE, Forgách P, Nowotny N. Serological surveillance for Tahyna virus (California encephalitis orthobunyavirus, Peribunyaviridae) neutralizing antibodies in wild ungulates in Austria, Hungary and Romania. Zoonoses Public Health. 2018;65(4):459-463. doi:10.1111/zph.12457 47. Campos R, Santos M, Tunon G, et al. Epidemiological aspects and spatial distribution of human and canine visceral leishmaniasis in an endemic area in northeastern Brazil. Geospat Health. 2017;12(1). doi:10.4081/gh.2017.503 48. Cárdenas NC, Galvis JOA, Farinati AA, Grisi-Filho JHH, Diehl GN, Machado G. Burkholderia mallei: The dynamics of networks and disease transmission. Transbound Emerg Dis. 2019;66(2):715-728. doi:10.1111/tbed.13071 49. Cassan C, Diagne CA, Tatard C, et al. Leishmania major and Trypanosoma lewisi infection in invasive and native rodents in Senegal. PLoS Negl Trop Dis. 2018;12(6):1-21. doi:10.1371/journal.pntd.0006615 50. Chadsuthi S, Bicout DJ, Wiratsudakul A, et al. Investigation on predominant Leptospira serovars and its distribution in humans and livestock in Thailand, 2010-2015. Day NP, ed. PLoS Negl Trop Dis. 2017;11(2):e0005228. doi:10.1371/journal.pntd.0005228 51. Chadsuthi S, Chalvet-Monfray K, Wiratsudakul A, Suwancharoen D, Cappelle J. A remotely sensed flooding indicator associated with cattle and buffalo leptospirosis cases in Thailand 2011-2013. BMC Infect Dis. 2018;18(1):1-9. doi:10.1186/s12879-018-3537-3 52. Chai C, Wang Q, Cao S, et al. Serological and molecular epidemiology of Japanese encephalitis virus infections in swine herds in China, 2006-2012. J Vet Sci. 2018;19(1):151-155. doi:10.4142/jvs.2018.19.1.151 53. Chastagner A, Hervé S, Bonin E, et al. Spatiotemporal Distribution and Evolution of the A/H1N1 2009 Pandemic Influenza Virus in Pigs in France from 2009 to 2017: Identification of a Potential Swine-Specific Lineage. J Virol. 2018;92(24):1-22. doi:10.1128/jvi.00988-18 54. Chaudhry M, Ahmad M, Rashid H Bin, et al. Prospective study of avian influenza H9 infection in commercial poultry farms of Punjab Province and Islamabad Capital Territory, Pakistan. Trop Anim Health Prod. 2017;49(1):213-220. doi:10.1007/s11250-016-1159-6 55. Clarke LL, Ruder MG, Mead DG, Howerth EW. Heartland virus exposure in white-tailed deer in the Southeastern United States, 2001-2015. Am J Trop Med Hyg. 2018;99(5):1346-1349. doi:10.4269/ajtmh.18-0555 56. Coetzer A, Anahory I, Dias PT, et al. Enhanced diagnosis of rabies and molecular evidence for the transboundary spread of the disease in Mozambique. J S Afr Vet Assoc. 2017;88(1):1-10. doi:10.4102/jsava.v88i0.1397

106

57. Cortes MC, Cauchemez S, Lefrancq N, et al. Characterization of the spatial and temporal distribution of Nipah virus spillover events in Bangladesh, 2007–2013. J Infect Dis. 2018;217(9):1390-1394. doi:10.1093/infdis/jiy015 58. Costa DNCC, Blangiardo M, Rodas LAC, et al. Canine visceral leishmaniasis in Araçatuba, state of São Paulo, Brazil, and its relationship with characteristics of dogs and their owners: A cross-sectional and spatial analysis using a geostatistical approach. BMC Vet Res. 2018;14(1):229. doi:10.1186/s12917-018-1550-9 59. Creanga A, Hang NLK, Cuong VD, et al. Highly pathogenic avian influenza a(h5n1) viruses at the animal-human interface in Vietnam, 2003-2010. J Infect Dis. 2017;216(suppl_4):S529-S538. doi:10.1093/infdis/jix003 60. Crim SM, Chai SJ, Karp BE, et al. Salmonella enterica Serotype Newport Infections in the United States, 2004-2013: Increased Incidence Investigated Through Four Surveillance Systems. Foodborne Pathog Dis. 2018;15(10):612-620. doi:10.1089/fpd.2018.2450 61. da Silva MS, Silveira S, Caron VS, et al. Backyard pigs are a reservoir of zoonotic hepatitis E virus in southern Brazil. Trans R Soc Trop Med Hyg. 2018;112(1):14-21. doi:10.1093/trstmh/try017 62. Daly ER, Fredette C, Mathewson AA, Dufault K, Swenson DJ, Chan BP. Tick bite and Lyme disease-related emergency department encounters in New Hampshire, 2010–2014. Zoonoses Public Health. 2017;64(8):655-661. doi:10.1111/zph.12361 63. Das Neves LB, Teixeira PEF, Silva S, et al. First molecular identification of Echinococcus vogeli and Echinococcus granulosus (sensu stricto) G1 revealed in feces of domestic dogs (Canis familiaris) from Acre, Brazil. Parasites and Vectors. 2017;10(1):1-6. doi:10.1186/s13071-016-1952-0 64. Davies S, Abdullah S, Helps C, Tasker S, Newbury H, Wall R. Prevalence of ticks and tick-borne pathogens: Babesia and Borrelia species in ticks infesting cats of Great Britain. Vet Parasitol. 2017;244(May):129-135. Doi:10.1016/J.Vetpar.2017.07.033 65. De Aquino Fm, Soares Ve, Rossi Gam, et al. Analysis of bovine cysticercosis in the state of Goiás, Brazil and economical losses for beef farms. Parasitol Open. 2017;3:e12. doi:10.1017/pao.2017.13 66. De Aquino FM, Soares VE, Rossi GAM, et al. Prevalence of bovine fascioliasis, areas at risk and ensuing losses in the state of Goiás, Brazil. Rev Bras Parasitol Vet. 2018;27(2):123-130. doi:10.1590/S1984-296120180024 67. de Jong W, Rusli M, Bhoelan S, et al. Endemic and emerging acute virus infections in Indonesia: an overview of the past decade and implications for the future. Crit Rev Microbiol. 2018;44(4):487-503. doi:10.1080/1040841X.2018.1438986 68. De Morais EGF, Rodrigues Magalhães FJ, De Lima Filho CDF, et al. Geo-epidemiological study of Leptospira spp. Infection in cattle, feral cats and rodents of the Fernando de Noronha Island, Brazil. Acta Sci Vet. 2018;46(1):1-9. doi:10.22456/1679-9216.79176.88400 69. De Vries SG, Visser BJ, Stoney RJ, et al. Leptospirosis among returned travelers: A geosentinel site survey and multicenter analysis-1997-2016. Am J Trop Med Hyg. 2018;99(1):127-135. doi:10.4269/ajtmh.18-0020 70. Delabouglise A, Choisy M, Phan TD, et al. Economic factors influencing zoonotic disease dynamics: Demand for poultry meat and seasonal transmission of avian influenza in Vietnam. Sci Rep. 2017;7(1):5905. doi:10.1038/s41598- 017-06244-6 71. Dermauw V, Dorny P, Braae UC, et al. Epidemiology of Taenia saginata taeniosis/cysticercosis: A systematic review of the distribution in southern and eastern Africa. Parasites and Vectors. 2018;11(1):1-12. doi:10.1186/s13071-018- 3163-3 72. Devleesschauwer B, Allepuz A, Dermauw V, et al. Taenia solium in Europe: Still endemic? Acta Trop. 2017;165:96- 99. doi:10.1016/j.actatropica.2015.08.006 73. Dhewantara PW, Mamun AA, Zhang WY, et al. Epidemiological shift and geographical heterogeneity in the burden of leptospirosis in China. Infect Dis Poverty. 2018;7(1):1-14. doi:10.1186/s40249-018-0435-2 74. Dhewantara PW, Mamun A Al, Zhang WY, et al. Geographical and temporal distribution of the residual clusters of human leptospirosis in China, 2005–2016. Sci Rep. 2018;8(1):16650. doi:10.1038/s41598-018-35074-3

107

75. Dietrich M, Gomard Y, Lagadec E, et al. Biogeography of Leptospira in wild animal communities inhabiting the insular ecosystem of the western Indian Ocean islands and neighboring Africa article. Emerg Microbes Infect. 2018;7(1):1-12. doi:10.1038/s41426-018-0059-4 76. Drewes S, Turni H, Rosenfeld UM, et al. Reservoir-Driven Heterogeneous Distribution of Recorded Human Puumala virus Cases in South-West Germany. Zoonoses Public Health. 2017;64(5):381-390. doi:10.1111/zph.12319 77. Drewes S, Sheikh Ali H, Saxenhofer M, et al. Host-associated absence of human puumala virus infections in northern and Eastern Germany. Emerg Infect Dis. 2017;23(1):83-86. doi:10.3201/eid2301.160224 78. Du P, Zheng H, Zhou J, et al. Detection of multiple parallel transmission outbreak of streptococcus suis human infection by use of genome epidemiology, China, 2005. Emerg Infect Dis. 2017;23(2):204-211. doi:10.3201/eid2302.160297 79. Duarte CTD, Pinto PSA, Silva LF, Santos TO, Bevilacqua PD, Nieto ECA. Epidemiological aspects of cysticercose in relation to hydrographic net at triângulo mineiro, MG, Brazil. Semin Agrar. 2018;39(1):221-230. doi:10.5433/1679- 0359.2018v39n1p221 80. Dumitrache MO, Györke A, Mircean M, Benea M, Mircean V. Ocular thelaziosis due Thelazia callipaeda (Spirurida: Thelaziidae) in Romania: first report in domestic cat and new geographical records of canine cases. Parasitol Res. 2018;117(12):4037-4042. doi:10.1007/s00436-018-6122-1 81. Durski KN, Mccollum AM, Nakazawa Y, Petersen BW, Reynolds MG. Emergence of Monkeypox — West and Central Africa , 1970 – 2017. MMWR. 2018;67(10). 82. Duscher T, Hodžić A, Glawischnig W, Duscher GG. The raccoon dog (Nyctereutes procyonoides) and the raccoon (Procyon lotor)—their role and impact of maintaining and transmitting zoonotic diseases in Austria, Central Europe. Parasitol Res. 2017;116(4):1411-1416. doi:10.1007/s00436-017-5405-2 83. Dwiyati Pujimulyani, Wisnu Adi Yulianto A. Hot-spot and cluster analysis on legal and illegal dumping sites as the contributors of leptospirosis in a flood hazard area in Pahang, Malaysia. Asian J Agri Biol. 2017;5(2):56-59. 84. Eddens T, Kaplan DJ, Anderson AJM, Nowalk AJ, Campfield BT. Insights From the Geographic Spread of the Lyme Disease Epidemic. Clin Infect Dis. 2019;68(3):426-434. doi:10.1093/cid/ciy510 85. Eisen RJ, Eisen L. The Blacklegged Tick, Ixodes scapularis: An Increasing Public Health Concern. Trends Parasitol. 2018;34(4):295-309. doi:10.1016/j.pt.2017.12.006 86. Eisen RJ, Kugeler KJ, Eisen L, Beard CB, Paddock CD. Tick-borne zoonoses in the United States: Persistent and emerging threats to human health. ILAR J. 2017;58(3):319-335. doi:10.1093/ilar/ilx005 87. El-Shesheny R, Barman S, Feeroz MM, et al. Genesis of influenza A(H5N8) viruses. Emerg Infect Dis. 2017;23(8):1368-1371. doi:10.3201/eid2308.170143 88. Erickson T, da Silva J, Nolan MS, Marquez L, Munoz FM, Murray KO. Newly recognized pediatric cases of typhus group rickettsiosis, Houston, Texas, USA. Emerg Infect Dis. 2017;23(12):2068-2071. doi:10.3201/eid2312.170631 89. Esmaeilnejad M, Bazrafshan E, Ansari-Moghaddam A. Effect of Climatic Changes on Spatial Distribution of Zoonoses: A Case Study from South Khorasan Province, Iran. Heal Scope. 2018;7(1):1-7. doi:10.5812/jhealthscope.56045 90. Faber M, Heuner K, Jacob D, Grunow R. Tularemia in Germany-A re-emerging zoonosis. Front Cell Infect Microbiol. 2018;8(February):1-10. doi:10.3389/fcimb.2018.00040 91. Failloux AB, Bouattour A, Faraj C, et al. Surveillance of Arthropod-Borne Viruses and Their Vectors in the Mediterranean and Black Sea Regions Within the MediLabSecure Network. Curr Trop Med Reports. 2017;4(1):27-39. doi:10.1007/s40475-017-0101-y 92. Fernandes NCC de A, Cunha MS, Guerra JM, et al. Outbreak of yellow fever among nonhuman primates, Espirito Santo, Brazil, 2017. Emerg Infect Dis. 2017;23(12):2038-2041. doi:10.3201/eid2312.170685 93. Fernández de Mera IG, Blanda V, Torina A, et al. Identification and molecular characterization of spotted fever group rickettsiae in ticks collected from farm ruminants in Lebanon. Ticks Tick Borne Dis. 2018;9(1):104-108. doi:10.1016/j.ttbdis.2017.10.001

108

94. Fiorillo G, Bocchini P, Buceta J. A Predictive Spatial Distribution Framework for Filovirus-Infected Bats. Sci Rep. 2018;8(1):7970. doi:10.1038/s41598-018-26074-4 95. Fischer S, Freuling CM, Müller T, et al. Defining objective clusters for rabies virus sequences using affinity propagation clustering. PLoS Negl Trop Dis. 2018;12(1):1-17. doi:10.1371/journal.pntd.0006182 96. Flies EJ, Weinstein P, Anderson SJ, Koolhof I, Foufopoulos J, Williams CR. Ross river virus and the necessity of multiscale, eco-epidemiological analyses. J Infect Dis. 2018;217(5):807-815. doi:10.1093/infdis/jix615 97. Forrester JD, Apangu T, Griffith K, et al. Patterns of human plague in Uganda, 2008–2016. Emerg Infect Dis. 2017;23(9):1517-1521. doi:10.3201/eid2309.170789 98. Fouskis I, Sandalakis V, Christidou A, et al. The epidemiology of Brucellosis in Greece, 2007-2012: A “One Health” approach. Trans R Soc Trop Med Hyg. 2018;112(3):124-135. doi:10.1093/trstmh/try031 99. Funke S, Anker JCH, Ethelberg S. Salmonella Dublin patients in Denmark and their distance to cattle farms. Infect Dis (Auckl). 2017;49(3):208-216. doi:10.1080/23744235.2016.1249024 100. Fyfe J, Picozzi K, Waiswa C, Bardosh KL, Welburn SC. Impact of mass chemotherapy in domestic livestock for control of zoonotic T. b. rhodesiense human African trypanosomiasis in Eastern Uganda. Acta Trop. 2017;165:216- 229. doi:10.1016/j.actatropica.2016.08.022 101. Gambino-Shirley K, Stevenson L, Concepción-Acevedo J, et al. Flea market finds and global exports: Four multistate outbreaks of human Salmonella infections linked to small turtles, United States—2015. Zoonoses Public Health. 2018;65(5):560-568. doi:10.1111/zph.12466 102. García-Bocanegra I, Jurado-Tarifa E, Cano-Terriza D, Martínez R, Pérez-Marín JE, Lecollinet S. Exposure to West Nile virus and tick-borne encephalitis virus in dogs in Spain. Transbound Emerg Dis. 2018;65(3):765-772. doi:10.1111/tbed.12801 103. Gardner EG, Kelton D, Poljak Z, von Dobschuetz S, Greer AL. A rapid scoping review of Middle East respiratory syndrome coronavirus in animal hosts. Zoonoses Public Health. 2019;66(1):35-46. doi:10.1111/zph.12537 104. Garofolo G, Di Giannatale E, Platone I, et al. Origins and global context of Brucella abortus in Italy. BMC Microbiol. 2017;17(1):28. doi:10.1186/s12866-017-0939-0 105. Gasmi S, Ogden N, Lindsay L, et al. Surveillance for Lyme disease in Canada: 2009–2015. Canada Commun Dis Rep. 2017;43(10):194-199. doi:10.14745/ccdr.v43i10a01 106. Georgi E, Walter MC, Pfalzgraf MT, et al. Whole genome sequencing of Brucella melitensis isolated from 57 patients in Germany reveals high diversity in strains from Middle East. PLoS One. 2017;12(4):1-15. doi:10.1371/journal.pone.0175425 107. Ghebremariam MK, Michel AL, Vernooij JCM, Nielen M, Rutten VPMG. Prevalence of bovine tuberculosis in cattle, goats, and camels of traditional livestock raising communities in Eritrea. BMC Vet Res. 2018;14(1):1-13. doi:10.1186/s12917-018-1397-0 108. Gibb R, Moses LM, Redding DW, Jones KE. Understanding the cryptic nature of Lassa fever in West Africa. Pathog Glob Health. 2017;111(6):276-288. doi:10.1080/20477724.2017.1369643 109. Glennon EE, Restif O, Sbarbaro SR, et al. Domesticated animals as hosts of henipaviruses and filoviruses: A systematic review. Vet J. 2018;233:25-34. doi:10.1016/j.tvjl.2017.12.024 110. Golpayegani AA, Moslem AR, Akhavan AA, Zeydabadi A, Mahvi AH, Allah-Abadi A. Modeling of environmental factors affecting the prevalence of zoonotic and anthroponotic cutaneous, and zoonotic visceral leishmaniasis in foci of Iran: A remote sensing and GIS based study. J Arthropod Borne Dis. 2018;12(1):41-66. 111. Gonwong S, Chuenchitra T, Khantapura P, et al. Nationwide seroprevalence of leptospirosis among young Thai men, 2007-2008. Am J Trop Med Hyg. 2017;97(6):1682-1685. doi:10.4269/ajtmh.17-0163 112. Govindaraj G, Sridevi R, Nandakumar SN, et al. Economic impacts of avian influenza outbreaks in Kerala, India. Transbound Emerg Dis. 2018;65(2):e361-e372. doi:10.1111/tbed.12766 113. Grayzel SE, Martínez-López B, Sykes JE. Risk Factors and Spatial Distribution of Canine Coccidioidomycosis in California, 2005–2013. Transbound Emerg Dis. 2017;64(4):1110-1119. doi:10.1111/tbed.12475

109

114. Gremião IDF, Miranda LHM, Reis EG, Rodrigues AM, Pereira SA. Zoonotic Epidemic of Sporotrichosis: Cat to Human Transmission. Sheppard DC, ed. PLoS Pathog. 2017;13(1):e1006077. doi:10.1371/journal.ppat.1006077 115. Gurley ES, Hegde ST, Hossain K, et al. Convergence of humans, bats, trees, and culture in Nipah virus transmission, Bangladesh. Emerg Infect Dis. 2017;23(9):1446-1453. doi:10.3201/eid2309.161922 116. Hammerl JA, Ulrich RG, Imholt C, et al. Molecular Survey on Brucellosis in Rodents and Shrews – Natural Reservoirs of Novel Brucella Species in Germany? Transbound Emerg Dis. 2017;64(2):663-671. doi:10.1111/tbed.12425 117. Han R, Yang J, Niu Q, et al. Molecular prevalence of spotted fever group rickettsiae in ticks from Qinghai Province, northwestern China. Infect Genet Evol. 2018;57(October 2017):1-7. doi:10.1016/j.meegid.2017.10.025 118. He J, Christakos G, Wu J, et al. Spatiotemporal variation of the association between climate dynamics and HFRS outbreaks in Eastern China during 2005-2016 and its geographic determinants. PLoS Negl Trop Dis. 2018;12(6):1-22. doi:10.1371/journal.pntd.0006554 119. Herrador Z, Fernandez-Martinez A, Gomez-Barroso D, et al. Mediterranean spotted fever in Spain, 1997-2014: Epidemiological situation based on hospitalization records. PLoS One. 2017;12(3):1-13. doi:10.1371/journal.pone.0174745 120. Hill EM, House T, Dhingra MS, et al. Modelling H5N1 in Bangladesh across spatial scales: Model complexity and zoonotic transmission risk. Epidemics. 2017;20:37-55. doi:10.1016/j.epidem.2017.02.007 121. Hogerwerf L, Holstege MMC, Benincà E, Dijkstra F, van der Hoek W. Temporal and spatial analysis of psittacosis in association with poultry farming in the Netherlands, 2000-2015. BMC Infect Dis. 2017;17(1):1-9. doi:10.1186/s12879- 017-2608-1 122. Holakouie-Naieni K, Mostafavi E, Boloorani AD, Mohebali M, Pakzad R. Spatial modeling of cutaneous leishmaniasis in Iran from 1983 to 2013. Acta Trop. 2017;166:67-73. doi:10.1016/j.actatropica.2016.11.004 123. Hurd J, Berke O, Poljak Z, Runge M. Spatial analysis of Leptospira infection in muskrats in Lower Saxony, Germany, and the association with human leptospirosis. Res Vet Sci. 2017;114(October 2016):351-354. doi:10.1016/j.rvsc.2017.06.015 124. Hussain G, Roychoudhury S, Singha B, Paul J. Incidence of Cryptosporidium andersoni in diarrheal patients from southern Assam, India: a molecular approach. Eur J Clin Microbiol Infect Dis. 2017;36(6):1023-1032. doi:10.1007/s10096-016-2887-2 125. Hutter SE, Käsbohrer A, González SLF, et al. Assessing changing weather and the El Niño Southern Oscillation impacts on cattle rabies outbreaks and mortality in Costa Rica (1985-2016). BMC Vet Res. 2018;14(1):1-14. doi:10.1186/s12917-018-1588-8 126. Ibarra-Cerdeña CN, Valiente-Banuet L, Sánchez-Cordero V, Stephens CR, Ramsey JM. Trypanosoma cruzi reservoir- triatomine vector co-occurrence networks reveal meta-community effects by synanthropic mammals on geographic dispersal. PeerJ. 2017;2017(4):1-26. doi:10.7717/peerj.3152 127. Jiang H, Zheng X, Wang L, Du H, Wang P, Bai X. Hantavirus infection: a global zoonotic challenge. Virol Sin. 2017;32(1):32-43. doi:10.1007/s12250-016-3899-x 128. Jiménez-Rocha AE, Argüello-Vargas S, Romero-Zuñiga JJ, et al. Environmental factors associated with Dictyocaulus viviparus and Fasciola hepatica prevalence in dairy herds from Costa Rica. Vet Parasitol Reg Stud Reports. 2017;9(May):115-121. doi:10.1016/j.vprsr.2017.06.006 129. Johnson TL, Boegler KA, Clark RJ, et al. An acarological risk model predicting the density and distribution of host- Seeking ixodes scapularis nymphs in Minnesota. Am J Trop Med Hyg. 2018;98(6):1671-1682. doi:10.4269/ajtmh.17- 0539 130. Johnson TL, Graham CB, Maes SE, et al. Prevalence and distribution of seven human pathogens in host-seeking Ixodes scapularis (Acari: Ixodidae) nymphs in Minnesota, USA. Ticks Tick Borne Dis. 2018;9(6):1499-1507. doi:10.1016/j.ttbdis.2018.07.009 131. Judson SD, LeBreton M, Fuller T, et al. Translating Predictions of Zoonotic Viruses for Policymakers. Ecohealth. 2018;15(1):52-62. doi:10.1007/s10393-017-1304-3

110

132. Kärssin A, Häkkinen L, Niin E, et al. Trichinella spp. biomass has increased in raccoon dogs (Nyctereutes procyonoides) and red foxes (Vulpes vulpes) in Estonia. Parasites and Vectors. 2017;10(1):609. doi:10.1186/s13071- 017-2571-0 133. Karthikeyan A, Shanmuganathan S, Pavulraj S, et al. Japanese Encephalitis, Recent Perspectives on Virus Genome, Transmission, Epidemiology, Diagnosis and Prophylactic Interventions. J Exp Biol Agric Sci. 2017;5(6):730-748. doi:10.18006/2017.5(6).730.748 134. Kasem S, Qasim I, Al-Hufofi A, et al. Cross-sectional study of MERS-CoV-specific RNA and antibodies in animals that have had contact with MERS patients in Saudi Arabia. J Infect Public Health. 2018;11(3):331-338. doi:10.1016/j.jiph.2017.09.022 135. Kerins JL, Dorevitch S, Dworkin MS. Spotted Fever Group Rickettsioses (SFGR): Weather and incidence in Illinois. Epidemiol Infect. 2017;145(12):2466-2472. doi:10.1017/S0950268817001492 136. Kimura T, Fukuma A, Shimojima M, et al. Seroprevalence of severe fever with thrombocytopenia syndrome (SFTS) virus antibodies in humans and animals in Ehime prefecture, Japan, an endemic region of SFTS. J Infect Chemother. 2018;24(10):802-806. doi:10.1016/j.jiac.2018.06.007 137. Kinkar L, Laurimäe T, Acosta-Jamett G, et al. Global phylogeography and genetic diversity of the zoonotic tapeworm Echinococcus granulosus sensu stricto genotype G1. Int J Parasitol. 2018;48(9-10):729-742. doi:10.1016/j.ijpara.2018.03.006 138. Kinkar L, Laurimäe T, Balkaya I, et al. Genetic diversity and phylogeography of the elusive, but epidemiologically important Echinococcus granulosus sensu stricto genotype G3. Parasitology. 2018;145(12):1613-1622. doi:10.1017/S0031182018000549 139. Kinross P, Petersen A, Skov R, et al. aureus ( MRSA ) among human MRSA isolates , European Union / European Economic Area countries , 2013. Surveill Outbreak Rep. 2014. doi:10.2807/1560-7917. ES.2017.22.44.16-00696 140. Knox KK, Thomm AM, Harrington YA, Ketter E, Patitucci JM, Carrigan DR. Powassan/Deer Tick Virus and Borrelia Burgdorferi Infection in Wisconsin Tick Populations. Vector-Borne Zoonotic Dis. 2017;17(7):463-466. doi:10.1089/vbz.2016.2082 141. Koffi J, Lindsay R, Ogden N. Surveillance for Lyme Disease in Canada, 2009-2012. Online J Public Health Inform. 2016;8(1):194-199. doi:10.5210/ojphi.v8i1.6477 142. Koffi SK, Meite S, Ouattara A, et al. Geographic distribution of anti-Leptospira antibodies in humans in Côte d’Ivoire, West Africa. Eur J Clin Microbiol Infect Dis. 2018;37(11):2177-2180. doi:10.1007/s10096-018-3359-7 143. Kracalik I, Malania L, Broladze M, et al. Changing livestock vaccination policy alters the epidemiology of human anthrax, Georgia, 2000–2013. Vaccine. 2017;35(46):6283-6289. doi:10.1016/j.vaccine.2017.09.081 144. Krow-Lucal ER, Lindsey NP, Fischer M, Hills SL. Powassan Virus Disease in the United States, 2006-2016. Vector- Borne Zoonotic Dis. 2018;18(6):286-290. doi:10.1089/vbz.2017.2239 145. Kuhn KG, Nielsen EM, Mølbak K, Ethelberg S. Epidemiology of campylobacteriosis in Denmark 2000–2015. Zoonoses Public Health. 2018;65(1):59-66. doi:10.1111/zph.12367 146. Kumar K, Arshad SS, Selvarajah GT, et al. Japanese encephalitis in Malaysia: An overview and timeline. Acta Trop. 2018;185(February):219-229. doi:10.1016/j.actatropica.2018.05.017 147. Kumar R, Patil RD. Cryptic etiopathological conditions of equine nervous system with special emphasis on viral diseases. Vet World. 2017;10(12):1427-1438. doi:10.14202/vetworld.2017.1427-1438 148. Lai S, Zhou H, Xiong W, et al. Changing epidemiology of human brucellosis, China, 1955-2014. Emerg Infect Dis. 2017;23(2):184-194. doi:10.3201/eid2302.151710 149. Lancelot R, Béral M, Rakotoharinome VM, et al. Drivers of Rift Valley fever epidemics in Madagascar. Proc Natl Acad Sci U S A. 2017;114(5):938-943. doi:10.1073/pnas.1607948114 150. Lashnits E, Correa M, Hegarty BC, Birkenheuer A, Breitschwerdt EB. Bartonella Seroepidemiology in Dogs from North America, 2008–2014. J Vet Intern Med. 2018;32(1):222-231. doi:10.1111/jvim.14890 151. Lawson AB, Rotejanaprasert C. Bayesian spatial modeling for the joint analysis of zoonosis between human and animal populations. Spat Stat. 2018;28:8-20. doi:10.1016/j.spasta.2018.08.001 111

152. Lawson B, Franklinos LHV, Rodriguez-Ramos Fernandez J, et al. Salmonella Enteritidis ST183: Emerging and endemic biotypes affecting western European hedgehogs (Erinaceus europaeus) and people in Great Britain. Sci Rep. 2018;8(1):2449. doi:10.1038/s41598-017-18667-2 153. Le Rutte EA, van Straten R, Overgaauw PAM. Awareness and control of canine leishmaniosis: A survey among Spanish and French veterinarians. Vet Parasitol. 2018;253(January):87-93. doi:10.1016/j.vetpar.2018.01.013 154. Lee SH, Kim HJ, Byun JW, et al. Molecular detection and phylogenetic analysis of severe fever with thrombocytopenia syndrome virus in shelter dogs and cats in the Republic of Korea. Ticks Tick Borne Dis. 2017;8(4):626-630. doi:10.1016/j.ttbdis.2017.04.008 155. Lemos TS, Cequinel JC, Costa TP, et al. Outbreak of human brucellosis in Southern Brazil and historical review of data from 2009 to 2018. PLoS Negl Trop Dis. 2018;12(9):1-12. doi:10.1371/journal.pntd.0006770 156. Li L, Guo X, Zhao Q, et al. Investigation on Mosquito-Borne Viruses at Lancang River and Nu River Watersheds in Southwestern China. Vector-Borne Zoonotic Dis. 2017;17(12):804-812. doi:10.1089/vbz.2017.2164 157. Li Y, Yin W, Hugh-Jones M, et al. Epidemiology of human anthrax in China, 1955-2014. Emerg Infect Dis. 2017;23(1):14-21. doi:10.3201/eid2301.150947 158. Liang W, Gu X, Li X, et al. Mapping the epidemic changes and risks of hemorrhagic fever with renal syndrome in Shaanxi Province, China, 2005-2016. Sci Rep. 2018;8(1):1-10. doi:10.1038/s41598-017-18819-4 159. Lim SM, Geervliet M, Verhagen JH, et al. Serologic evidence of West Nile virus and Usutu virus infections in Eurasian coots in the Netherlands. Zoonoses Public Health. 2018;65(1):96-102. doi:10.1111/zph.12375 160. Linden J V., Prusinski MA, Crowder LA, et al. Transfusion-transmitted and community-acquired babesiosis in New York, 2004 to 2015. Transfusion. 2018;58(3):660-668. doi:10.1111/trf.14476 161. Liu L, Guo B, Li W, et al. Geographic distribution of echinococcosis in Tibetan region of Sichuan Province, China. Infect Dis Poverty. 2018;7(1):1-9. doi:10.1186/s40249-018-0486-4 162. Liu Y, Lund RB, Nordone SK, Yabsley MJ, McMahan CS. A Bayesian spatio-temporal model for forecasting the prevalence of antibodies to Ehrlichia species in domestic dogs within the contiguous United States. Parasites and Vectors. 2017;10(1):1-14. doi:10.1186/s13071-017-2068-x 163. Ma G, Holland C V., Wang T, et al. Human toxocariasis. Lancet Infect Dis. 2018;18(1):e14-e24. doi:10.1016/S1473- 3099(17)30331-6 164. Machado A, Rito T, Ghebremichael S, et al. Genetic diversity and potential routes of transmission of Mycobacterium bovis in Mozambique. PLoS Negl Trop Dis. 2018;12(1):1-17. doi:10.1371/journal.pntd.0006147 165. Maciel ALG, Loiko MR, Bueno TS, et al. Tuberculosis in Southern Brazilian wild boars (Sus scrofa): First epidemiological findings. Transbound Emerg Dis. 2018;65(2):518-526. doi:10.1111/tbed.12734 166. Madeira S, Manteigas A, Ribeiro R, et al. Factors that Influence Mycobacterium bovis Infection in Red Deer and Wild Boar in an Epidemiological Risk Area for Tuberculosis of Game Species in Portugal. Transbound Emerg Dis. 2017;64(3):793-804. doi:10.1111/tbed.12439 167. Madinga J, Kanobana K, Lukanu P, et al. Geospatial and age-related patterns of Taenia solium taeniasis in the rural health zone of Kimpese, Democratic Republic of Congo. Acta Trop. 2017;165:100-109. doi:10.1016/j.actatropica.2016.03.013 168. Mallawarachchi CH, Nilmini Chandrasena TGA, Premaratna R, Mallawarachchi SMNSM, De Silva NR. Human infection with sub-periodic Brugia spp. in Gampaha District, Sri Lanka: A threat to filariasis elimination status? Parasites and Vectors. 2018;11(1):18-23. doi:10.1186/s13071-018-2649-3 169. Martins-Melo FR, Ramos AN, Cavalcanti MG, Alencar CH, Heukelbach J. Reprint of “Neurocysticercosis-related mortality in Brazil, 2000–2011: Epidemiology of a neglected neurologic cause of death.” Acta Trop. 2017;165:170- 178. doi:10.1016/j.actatropica.2016.11.009 170. Mateo M, de Mingo MH, de Lucio A, et al. Occurrence and molecular genotyping of Giardia duodenalis and Cryptosporidium spp. in wild mesocarnivores in Spain. Vet Parasitol. 2017;235:86-93. doi:10.1016/j.vetpar.2017.01.016

112

171. Mathole MA, Muchadeyi FC, Mdladla K, Malatji DP, Dzomba EF, Madoroba E. Presence, distribution, serotypes and antimicrobial resistance profiles of Salmonella among pigs, chickens and goats in South Africa. Food Control. 2017;72(August):219-224. doi:10.1016/j.foodcont.2016.05.006 172. Mayer FQ, Dos Reis EM, Bezerra AVA, et al. Pathogenic Leptospira spp. in bats: Molecular investigation in Southern Brazil. Comp Immunol Microbiol Infect Dis. 2017;52(May):14-18. doi:10.1016/j.cimid.2017.05.003 173. Mayfield HJ, Lowry JH, Watson CH, Kama M, Nilles EJ, Lau CL. Use of geographically weighted logistic regression to quantify spatial variation in the environmental and sociodemographic drivers of leptospirosis in Fiji: a modelling study. Lancet Planet Heal. 2018;2(5):e223-e232. doi:10.1016/S2542-5196(18)30066-4 174. Mehmood K, Zhang H, Sabir AJ, et al. A review on epidemiology, global prevalence and economical losses of fasciolosis in ruminants. Microb Pathog. 2017;109:253-262. doi:10.1016/j.micpath.2017.06.006 175. Melo HA, Rossoni DF, Teodoro U. Effect of vegetation on cutaneous leishmaniasis in Paraná, Brazil. Mem Inst Oswaldo Cruz. 2018;113(6):1-7. doi:10.1590/0074-02760170505 176. Menghistu HT, Hailu KT, Shumye NA, Redda YT. Mapping the epidemiological distribution and incidence of major zoonotic diseases in South Tigray, North Wollo and Ab’ala (Afar), Ethiopia. Arez AP, ed. PLoS One. 2018;13(12):1- 19. doi:10.1371/journal.pone.0209974 177. Methner U, Moog U. Occurrence and characterisation of Salmonella enterica subspecies diarizonae serovar 61: K: 1, 5, (7) in sheep in the federal state of Thuringia, Germany. BMC Vet Res. 2018;14(1):1-8. doi:10.1186/s12917-018-1741- 4 178. Miguel E, Chevalier V, Ayelet G, et al. Risk factors for MERS coronavirus infection in dromedary camels in Burkina Faso, Ethiopia, and Morocco, 2015. Eurosurveillance. 2017;22(13). doi:10.2807/1560-7917.ES.2017.22.13.30498 179. Mohammadinia A, Alimohammadi A, Saeidian B. Efficiency of geographically weighted regression in modeling human leptospirosis based on environmental factors in Gilan province, Iran. Geosci. 2017;7(4). doi:10.3390/geosciences7040136 180. Mohebali M, Moradi-Asl E, Rassi Y. Geographic distribution and spatial analysis of Leishmania infantum infection in domestic and wild animal reservoir hosts of zoonotic visceral leishmaniasis in Iran: A systematic review. J Vector Borne Dis. 2018;55(3):184-188. doi:10.4103/0972-9062.249125 181. Monteiro KJL, dos Reis ERC, Nunes BC, et al. Focal persistence of soil-transmitted helminthiases in impoverished areas in the State of Piaui, northeastern Brazil. Rev Inst Med Trop Sao Paulo. 2018;60:1-10. doi:10.1590/s1678- 9946201860024 182. Montoya A, García M, Gálvez R, et al. Implications of zoonotic and vector-borne parasites to free-roaming cats in central Spain. Vet Parasitol. 2018;251(January):125-130. doi:10.1016/j.vetpar.2018.01.009 183. Mosimann L, Traoré A, Mauti S, et al. A mixed methods approach to assess animal vaccination programmes: The case of rabies control in Bamako, Mali. Acta Trop. 2017;165:203-215. doi:10.1016/j.actatropica.2016.10.007 184. Mrljak V, Kuleš J, Mihaljević Z, et al. Prevalence and geographic distribution of vector-borne pathogens in apparently healthy dogs in Croatia. Vector-Borne Zoonotic Dis. 2017;17(6):398-408. doi:10.1089/vbz.2016.1990 185. Murphy DS, Lee X, Larson SR, Johnson DKH, Loo T, Paskewitz SM. Prevalence and Distribution of Human and Tick Infections with the Ehrlichia muris-Like Agent and Anaplasma phagocytophilum in Wisconsin, 2009-2015. Vector- Borne Zoonotic Dis. 2017;17(4):229-236. doi:10.1089/vbz.2016.2055 186. Murray KO, Evert N, Mayes B, et al. Typhus Group Rickettsiosis, Texas, USA, 2003–2013. Emerg Infect Dis. 2017;23(4):645-648. doi:10.3201/eid2304.160958 187. Muturi M, Gachohi J, Mwatondo A, et al. Recurrent Anthrax Outbreaks in Humans, Livestock, and Wildlife in the Same Locality, Kenya, 2014-2017. Am J Trop Med Hyg. 2018;99(4):833-839. doi:10.4269/ajtmh.18-0224 188. Muvhali M, Smith AM, Rakgantso AM, Keddy KH. Investigation of Salmonella Enteritidis outbreaks in South Africa using multi-locus variable-number tandem-repeats analysis, 2013-2015. BMC Infect Dis. 2017;17(1):1-9. doi:10.1186/s12879-017-2751-8

113

189. Napp S, Chevalier V, Busquets N, et al. Understanding the legal trade of cattle and camels and the derived risk of Rift Valley Fever introduction into and transmission within Egypt. PLoS Negl Trop Dis. 2018;12(1):1-25. doi:10.1371/journal.pntd.0006143 190. Nascimento MSJ, Pereira SS, Teixeira J, et al. A nationwide serosurvey of hepatitis e virus antibodies in the general population of Portugal. Eur J Public Health. 2018;28(4):720-724. doi:10.1093/eurpub/ckx213 191. Naufal Spir PR, Prestes-Carneiro LE, Fonseca ES, Dayse A, Giuffrida R, D’Andrea LAZ. Clinical characteristics and spatial distribution of Visceral leishmaniasis in children in São Paulo state: an emerging focus of Visceral leishmaniasis in Brazil. Pathog Glob Health. 2017;111(2):91-97. doi:10.1080/20477724.2017.1289666 192. Nelson MI, Culhane MR, Trovão NS, et al. The emergence and evolution of influenza A (H1α) viruses in swine in Canada and the United States. J Gen Virol. 2017;98(11):2663-2675. doi:10.1099/jgv.0.000924 193. Ning C, Shuyi G, Tao Y, Hao Z, Zhang X. Epidemiological survey of human brucellosis in Inner Mongolia, China, 2010–2014: A high risk groups-based survey. J Infect Public Health. 2018;11(1):24-29. doi:10.1016/j.jiph.2017.02.013 194. Nyirenda SS, Hang’ombe BM, Simulundu E, et al. Molecular epidemiological investigations of plague in Eastern Province of Zambia. BMC Microbiol. 2018;18(1):1-7. doi:10.1186/s12866-017-1146-8 195. Olayemi A, Oyeyiola A, Obadare A, et al. Widespread arenavirus occurrence and seroprevalence in small mammals, Nigeria. Parasites and Vectors. 2018;11(1):1-9. doi:10.1186/s13071-018-2991-5 196. Oleaga A, Zanet S, Espí A, Pegoraro de Macedo MR, Gortázar C, Ferroglio E. Leishmania in wolves in northern Spain: A spreading zoonosis evidenced by wildlife sanitary surveillance. Vet Parasitol. 2018;255(March):26-31. doi:10.1016/j.vetpar.2018.03.015 197. Otalu OJ, Kwaga JKP, Okolocha EC, Islam MZ, Moodley A. High Genetic Similarity of MRSA ST88 Isolated From Pigs and Humans in Kogi State, Nigeria. Front Microbiol. 2018;9(December):1-8. doi:10.3389/fmicb.2018.03098 198. Paden CR, Yusof MFBM, Al Hammadi ZM, et al. Zoonotic origin and transmission of Middle East respiratory syndrome coronavirus in the UAE. Zoonoses Public Health. 2018;65(3):322-333. doi:10.1111/zph.12435 199. Pandit PS, Doyle MM, Smart KM, Young CCW, Drape GW, Johnson CK. Predicting wildlife reservoirs and global vulnerability to zoonotic Flaviviruses. Nat Commun. 2018;9(1):1-10. doi:10.1038/s41467-018-07896-2 200. Papadopoulos E, Komnenou A, Thomas A, Ioannidou E, Colella V, Otranto D. Spreading of Thelazia callipaeda in Greece. Transbound Emerg Dis. 2018;65(1):248-252. doi:10.1111/tbed.12626 201. Pedersen K, Anderson TD, Bevins SN, et al. Evidence of leptospirosis in the kidneys and serum of feral swine (Sus scrofa) in the United States. Epidemiol Infect. 2017;145(1):87-94. doi:10.1017/S0950268816002247 202. Pereira MN, Rossi GAM, Lopes WDZ, et al. Spatial analysis of bovine cysticercosis in the state of Mato Grosso do Sul, Brazil — The needs of interventions in animal and human populations. Vet Parasitol Reg Stud Reports. 2017;8:94- 98. doi:10.1016/j.vprsr.2017.03.001 203. Pigott DM, Deshpande A, Letourneau I, et al. Local, national, and regional viral haemorrhagic fever pandemic potential in Africa: a multistage analysis. Lancet. 2017;390(10113):2662-2672. doi:10.1016/S0140-6736(17)32092-5 204. Pijnacker R, Reimerink J, Smit LAM, et al. Remarkable spatial variation in the seroprevalence of Coxiella burnetii after a large Q fever epidemic. BMC Infect Dis. 2017;17(1):1-8. doi:10.1186/s12879-017-2813-y 205. Pinheiro Leite ACC, Dos Anjos DM, Simões EM, et al. Spatial characterization and identification of chiroptera shelters and their relation to cases of rabies in production animals in semi-arid, Brazil, from 2007 to 2015. Semin Agrar. 2018;39(6):2875-2882. doi:10.5433/1679-0359.2018v39n6p2875 206. Pisarenko S V., Kovalev DA, Volynkina AS, et al. Global evolution and phylogeography of Brucella melitensis strains. BMC Genomics. 2018;19(1):1-10. doi:10.1186/s12864-018-4762-2 207. Poester VR, Mattei AS, Madrid IM, et al. Sporotrichosis in Southern Brazil, towards an epidemic? Zoonoses Public Health. 2018;65(7):815-821. doi:10.1111/zph.12504 208. Pohlmann A, Starick E, Harder T, et al. Outbreaks among Wild Birds and Domestic. Emerg Infect Dis. 2017;23(4):633-636. doi:http://dx.doi.org/10.3201/eid2304.161949

114

209. Pohlmann A, Starick E, Harder T, et al. Outbreaks among wild birds and domestic poultry caused by reassorted influenza a(H5n8) clade 2.3.4.4 viruses, Germany, 2016. Emerg Infect Dis. 2017;23(4):633-636. doi:10.3201/eid2304.161949 210. Poulsen MN, Pollak J, Sills DL, et al. Residential proximity to high-density poultry operations associated with campylobacteriosis and infectious diarrhea. Int J Hyg Environ Health. 2018;221(2):323-333. doi:10.1016/j.ijheh.2017.12.005 211. Quiner CA, Nakazawa Y. Ecological niche modeling to determine potential niche of Vaccinia virus: A case only study. Int J Health Geogr. 2017;16(1):1-12. doi:10.1186/s12942-017-0100-1 212. Raharinosy V, Olive MM, Andriamiarimanana FM, et al. Geographical distribution and relative risk of Anjozorobe virus (Thailand orthohantavirus) infection in black rats (Rattus rattus) in Madagascar. Virol J. 2018;15(1):1-11. doi:10.1186/s12985-018-0992-9 213. Rainova I, Harizanov R, Kaftandjiev I, Tsvetkova N, Mikov O, Kaneva E. Human parasitic diseases in Bulgaria in between 2013-2014. Balkan Med J. 2018;35(1):61-67. doi:10.4274/balkanmedj.2017.0167 214. Rajabi M, Pilesjö P, Bazmani A, Mansourian A. Identification of Visceral Leishmaniasis-Susceptible Areas using Spatial Modelling in Southern Caucasus. Zoonoses Public Health. 2017;64(7):e5-e22. doi:10.1111/zph.12325 215. Rajabi M, Mansourian A, Pilesjö P, Shirzadi MR, Fadaei R, Ramazanpour J. A spatially explicit agent-based simulation model of a reservoir host of cutaneous leishmaniasis, Rhombomys opimus. Ecol Modell. 2018;370:33-49. doi:10.1016/j.ecolmodel.2017.12.004 216. Ramey AM, Hill NJ, Cline T, et al. Surveillance for highly pathogenic influenza A viruses in California during 2014- 2015 provides insights into viral evolutionary pathways and the spatiotemporal extent of viruses in the Pacific Americas Flyway. Emerg Microbes Infect. 2017;6(9). doi:10.1038/emi.2017.66 217. Ran X, Chen X, Wang M, et al. Brucellosis seroprevalence in ovine and caprine flocks in China during 2000-2018: A systematic review and meta-analysis. BMC Vet Res. 2018;14(1):1-9. doi:10.1186/s12917-018-1715-6 218. Ran X, Cheng J, Wang M, et al. Brucellosis seroprevalence in dairy cattle in China during 2008–2018: A systematic review and meta-analysis. Acta Trop. 2019;189(August 2018):117-123. doi:10.1016/j.actatropica.2018.10.002 219. Rauch J, Eisermann P, Noack B, et al. Typhus group rickettsiosis, Germany, 2010–2017. Emerg Infect Dis. 2018;24(7):1213-1220. doi:10.3201/eid2407.180093 220. Redding DW, Tiedt S, Lo Iacono G, Bett B, Jones KE. Spatial, seasonal and climatic predictive models of rift valley fever disease across Africa. Philos Trans R Soc B Biol Sci. 2017;372(1725):1-9. doi:10.1098/rstb.2016.0165 221. Reilly S, Sanderson WT, Christian WJ, Browning SR. Geographical clusters and predictors of rabies in three southeastern states. Vector-Borne Zoonotic Dis. 2017;17(6):432-438. doi:10.1089/vbz.2016.2061 222. Restrepo AMC, Yang YR, McManus DP, et al. Spatiotemporal patterns and environmental drivers of human echinococcoses over a twenty-year period in Ningxia Hui Autonomous Region, China. Parasites and Vectors. 2018;11(1). doi:10.1186/s13071-018-2693-z 223. Retkute R, Jewell CP, Van Boeckel TP, et al. Dynamics of the 2004 avian influenza H5N1 outbreak in Thailand: The role of duck farming, sequential model fitting and control. Prev Vet Med. 2018;159(March):171-181. doi:10.1016/j.prevetmed.2018.09.014 224. Retmanasari A, Widartono BS, Wijayanti MA, Artama WT. Prevalence and Risk Factors for Toxoplasmosis in Middle Java, Indonesia. Ecohealth. 2017;14(1):162-170. doi:10.1007/s10393-016-1198-5 225. Rettinger A, Broeckl S, Fink M, et al. The Region of Difference Four is a Robust Genetic Marker for Subtyping Mycobacterium caprae Isolates and is Linked to Spatial Distribution of Three Subtypes. Transbound Emerg Dis. 2017;64(3):782-792. doi:10.1111/tbed.12438 226. Ribeiro J, Staudacher C, Martins CM, et al. Bat rabies surveillance and risk factors for rabies spillover in an urban area of Southern Brazil. BMC Vet Res. 2018;14(1):173. doi:10.1186/s12917-018-1485-1 227. Rizzo F, Edenborough KM, Toffoli R, et al. Coronavirus and paramyxovirus in bats from Northwest Italy. BMC Vet Res. 2017;13(1):1-11. doi:10.1186/s12917-017-1307-x

115

228. Romero-Alvarez D, Escobar LE. Oropouche fever, an emergent disease from the Americas. Microbes Infect. 2018;20(3):135-146. doi:10.1016/j.micinf.2017.11.013 229. Romha G, Gebru G, Asefa A, Mamo G. Epidemiology of Mycobacterium bovis and Mycobacterium tuberculosis in animals: Transmission dynamics and control challenges of zoonotic TB in Ethiopia. Prev Vet Med. 2018;158(June):1- 17. doi:10.1016/j.prevetmed.2018.06.012 230. Rood EJJ, Goris MGA, Pijnacker R, Bakker MI, Hartskeerl RA. Environmental risk of leptospirosis infections in the Netherlands: Spatial modelling of environmental risk factors of leptospirosis in the Netherlands. Kunze G, ed. PLoS One. 2017;12(10):1-11. doi:10.1371/journal.pone.0186987 231. Rosenberg R, Lindsey NP, Fischer M, et al. Vital Signs: Trends in Reported Vectorborne Disease Cases — United States and Territories, 2004–2016. MMWR. 2018;67(17):496-501. doi:10.15585/mmwr.mm6717e1 232. Rossi GAM, Lopes WDZ, de Souza Almeida HM, et al. Spatial distribution, prevalence and epidemiological risk factors of cysticercosis in cattle from state of São Paulo, Brazil, slaughtered for human consumption. Vet Parasitol Reg Stud Reports. 2017;8(April 2016):117-122. doi:10.1016/j.vprsr.2017.03.007 233. Rossi GAM, Martins IVF, Campos RF de, Soares LFS, Almeida HM de S, Mathias LA. Spatial distribution of bovine cysticercosis—A retrospective study in Brazil from 2010 through 2015. Prev Vet Med. 2017;145:145-149. doi:10.1016/j.prevetmed.2017.06.013 234. Rotejanaprasert C, Lawson A, Rossow H, et al. Towards integrated surveillance of zoonoses: Spatiotemporal joint modeling of rodent population data and human tularemia cases in Finland. BMC Med Res Methodol. 2018;18(1):72. doi:10.1186/s12874-018-0532-8 235. Rouffaer LO, Baert K, Van Den Abeele AM, et al. Low prevalence of human enteropathogenic Yersinia spp. in brown rats (Rattus norvegicus) in Flanders. Cloeckaert A, ed. PLoS One. 2017;12(4):e0175648. doi:10.1371/journal.pone.0175648 236. Roy AN, Straub MH, Stephenson N, Sholty KE, Foley J. Distribution and Diversity of Borrelia burgdorferi Sensu Lato Group Bacteria in Sciurids of California. Vector-Borne Zoonotic Dis. 2017;17(11):735-742. doi:10.1089/vbz.2017.2134 237. Ruan S. Modeling the transmission dynamics and control of rabies in China. Math Biosci. 2017;286:65-93. doi:10.1016/j.mbs.2017.02.005 238. Sabeta CT, Janse Van Rensburg DD, Phahladira B, et al. Rabies of canid biotype in wild dog (Lycaon pictus) and spotted hyaena (Crocuta crocuta) in Madikwe Game Reserve, South Africa in 2014-2015: Diagnosis, possible origins and implications for control. J S Afr Vet Assoc. 2018;89(0):1-13. doi:10.4102/jsava.v89i0.1517 239. Sadeuh-Mba SA, Momo JB, Besong L, Loul S, Njouom R. Molecular characterization and phylogenetic relatedness of dog-derived Rabies Viruses circulating in Cameroon between 2010 and 2016. Rupprecht CE, ed. PLoS Negl Trop Dis. 2017;11(10):e0006041. doi:10.1371/journal.pntd.0006041 240. Saghafipour A, Vatandoost H, Zahraei-Ramazani AR, et al. Spatial Distribution of Phlebotomine Sand Fly Species (Diptera: Psychodidae) in Qom Province, Central Iran. J Med Entomol. 2017;54(1):35-43. doi:10.1093/jme/tjw147 241. Sakkas H, Bozidis P, Franks A, Papadopoulou C. Oropouche fever: A review. Viruses. 2018;10(4):175. doi:10.3390/v10040175 242. Salimi M, Jesri N, Javanbakht M, Farahani LZ, Shirzadi MR, Saghafipour A. Spatio-temporal distribution analysis of zoonotic cutaneous leishmaniasis in Qom Province, Iran. J Parasit Dis. 2018;42(4):570-576. doi:10.1007/s12639-018- 1036-5 243. Salines M, Andraud M, Rose N. Combining network analysis with epidemiological data to inform risk-based surveillance: Application to hepatitis E virus (HEV) in pigs. Prev Vet Med. 2018;149(August 2017):125-131. doi:10.1016/j.prevetmed.2017.11.015 244. Santos N, Nunes T, Fonseca C, et al. Spatial analysis of wildlife tuberculosis based on a serologic survey using dried blood spots, Portugal. Emerg Infect Dis. 2018;24(12):2169-2175. doi:10.3201/eid2412.171357 245. Schneider MC, Leonel DG, Hamrick PN, et al. Leptospirosis in Latin America: exploring the first set of regional data. Rev Panam Salud Pública. 2017;41:1. doi:10.26633/rpsp.2017.81

116

246. Self SCW, McMahan CS, Brown DA, Lund RB, Gettings JR, Yabsley MJ. A large-scale spatio-temporal binomial regression model for estimating seroprevalence trends. Environmetrics. 2018;29(8):1-17. doi:10.1002/env.2538 247. Sevá A da P, Mao L, Galvis-Ovallos F, Tucker Lima JM, Valle D. Risk analysis and prediction of visceral leishmaniasis dispersion in São Paulo State, Brazil. PLoS Negl Trop Dis. 2017;11(2):1-17. doi:10.1371/journal.pntd.0005353 248. Shah SZ, Jabbar B, Ahmed N, et al. Epidemiology, pathogenesis, and control of a tick-borne disease- Kyasanur forest disease: Current status and future directions. Front Cell Infect Microbiol. 2018;8(MAY). doi:10.3389/fcimb.2018.00149 249. Shiravand B, Tafti AAD, Hanafi-Bojd AA, Almodaresi SA, Mirzaei M, Abai MR. Modeling spatial risk of zoonotic cutaneous leishmaniasis in Central Iran. Acta Trop. 2018;185(April):327-335. doi:10.1016/j.actatropica.2018.06.015 250. Shittu I, Meseko CA, Gado DA, et al. Highly pathogenic avian influenza (H5N1) in Nigeria in 2015: evidence of widespread circulation of WA2 clade 2.3.2.1c. Arch Virol. 2017;162(3):841-847. doi:10.1007/s00705-016-3149-4 251. Siddle KJ, Eromon P, Barnes KG, et al. Genomic Analysis of Lassa Virus during an Increase in Cases in Nigeria in 2018. N Engl J Med. 2018;379(18):1745-1753. doi:10.1056/NEJMoa1804498 252. Silva ERD de FS, Castro V, Mineiro ALBB, et al. Sociodemographic and environmental analysis for the occurrence of anti-leptospira antibodies in dogs of Teresina, Piauí, Brazil. Cienc e Saude Coletiva. 2018;23(5):1403-1414. doi:10.1590/1413-81232018235.19532016 253. Simón F, González-Miguel J, Diosdado A, Gómez PJ, Morchón R, Kartashev V. The Complexity of Zoonotic Filariasis Episystem and Its Consequences: A Multidisciplinary View. Biomed Res Int. 2017;2017:1-10. doi:10.1155/2017/6436130 254. Sindičić M, Bujanić M, Štimac I, et al. First identification of Echinococcus multilocularis in golden jackals in Croatia. Acta Parasitol. 2018;63(3):654-656. doi:10.1515/ap-2018-0076 255. Singh P, Lingala MAL, Sarkar S, Dhiman RC. Mapping of malaria vectors at District Level in India: Changing scenario and identified gaps. Vector-Borne Zoonotic Dis. 2017;17(2):91-98. doi:10.1089/vbz.2016.2018 256. Singh RK, Dhama K, Malik YS, et al. Ebola virus - Epidemiology, diagnosis, and control: Threat to humans, lessons learnt, and preparedness plans - An update on its 40 year’s journey. Vet Q. 2017;37(1):98-135. doi:10.1080/01652176.2017.1309474 257. Smallbone WA, Chadwick EA, Francis J, et al. East-West Divide: Temperature and land cover drive spatial variation of Toxoplasma gondii infection in Eurasian otters (Lutra lutra) from England and Wales. Parasitology. 2017;144(11):1433-1440. doi:10.1017/S0031182017000865 258. Smith KA, Oesterle PT, Jardine CM, et al. Tick infestations of wildlife and companion animals in Ontario, Canada, with detection of human pathogens in Ixodes scapularis ticks. Ticks Tick Borne Dis. 2019;10(1):72-76. doi:10.1016/j.ttbdis.2018.08.018 259. Smith K, Oesterle PT, Jardine CM, et al. Powassan virus and other arthropod-borne viruses in wildlife and ticks in Ontario, Canada. Am J Trop Med Hyg. 2018;99(2):458-465. doi:10.4269/ajtmh.18-0098 260. Sofizadeh A, Hanafi-Bojd A, Shoraka H. Modeling spatial distribution of Rhombomys opimus as the main reservoir host of zoonotic cutaneous leishmaniasis in northeastern Iran. J Vector Borne Dis. 2018;55(4):297-304. doi:10.4103/0972-9062.256565 261. Spina-Tensini T, Muro MD, Queiroz-Telles F, et al. Geographic distribution of patients affected by Cryptococcus neoformans/Cryptococcus gattii species complexes meningitis, pigeon and tree populations in Southern Brazil. Mycoses. 2017;60(1):51-58. doi:10.1111/myc.12550 262. Springer A, Montenegro VM, Schicht S, Pantchev N, Strube C. Seroprevalence and current infections of canine vector- borne diseases in Nicaragua. Parasites and Vectors. 2018;11(1):1-9. doi:10.1186/s13071-018-3173-1 263. Stalb S, Polley B, Danner K, et al. Detection of tularemia in European brown hares (Lepus europaeus) and humans reveals endemic and seasonal occurrence in BadenWuerttemberg, Germany. Berl Munch Tierarztl Wochenschr. 2017;130(7/8):293-299. doi:10.2376/0005-9366-16079

117

264. Stein E, Elbadawi LI, Kazmierczak J, Davis JP. Babesiosis Surveillance — Wisconsin, 2001–2015. MMWR Morb Mortal Wkly Rep. 2017;66(26):687-691. doi:10.15585/mmwr.mm6626a2 265. Strakova P, Kubankova M, Vasickova P, Juricova Z, Rudolf I, Hubalek Z. Hepatitis E virus in archived sera from wild boars (Sus scrofa), Czech Republic. Transbound Emerg Dis. 2018;65(6):1770-1774. doi:10.1111/tbed.12950 266. Straub MH, Roy AN, Martin A, Sholty KE, Stephenson N, Foley JE. Distribution and prevalence of vector-borne diseases in California chipmunks (Tamias spp.). Pal U, ed. PLoS One. 2017;12(12):1-19. doi:10.1371/journal.pone.0189352 267. Štrbac M, Ristić M, Petrović V, et al. Epidemiological characteristics of brucellosis in Vojvodina, Serbia, 2000-2014. Vojnosanit Pregl. 2017;74(12):1140-1147. doi:10.2298/VSP160212311S 268. Sule WF, Oluwayelu DO, Hernández-Triana LM, Fooks AR, Venter M, Johnson N. Epidemiology and ecology of West Nile virus in sub-Saharan Africa. Parasites and Vectors. 2018;11(1):1-10. doi:10.1186/s13071-018-2998-y 269. Suttie A, Yann S, Phalla Y, et al. Detection of low pathogenicity influenza a(H7n3) virus during duck mortality event, Cambodia, 2017. Emerg Infect Dis. 2018;24(6):1103-1107. doi:10.3201/eid2406.172099 270. Suttie A, Yann S, Y P, et al. Detection of Low Pathogenicity Influenza. Emerg Infect Dis. 2018;24(6):1103-1107. 271. Thiry D, Mauroy A, Pavio N, et al. Hepatitis E Virus and Related Viruses in Animals. Transbound Emerg Dis. 2017;64(1):37-52. doi:10.1111/tbed.12351 272. Thomas-Bachli AL, Pearl DL, Berke O, Parmley EJ, Barker IK. A geographic study of West Nile virus in humans, dead corvids and mosquitoes in Ontario using spatial scan statistics with a survival time application. Zoonoses Public Health. 2017;64(7):e81-e89. doi:10.1111/zph.12350 273. Thompson M, Berke O. Evaluation of the Control of West Nile Virus in Ontario: Did Risk Patterns Change from 2005 to 2012? Zoonoses Public Health. 2017;64(2):100-105. doi:10.1111/zph.12285 274. Tian GZ, Cui BY, Piao DR, et al. Multi-locus variable-number tandem repeat analysis of Chinese Brucella strains isolated from 1953 to 2013. Infect Dis Poverty. 2017;6(1):89. doi:10.1186/s40249-017-0296-0 275. Tian H, Feng Y, Vrancken B, et al. Transmission dynamics of re-emerging rabies in domestic dogs of rural China. PLoS Pathog. 2018;14(12):1-19. doi:10.1371/journal.ppat.1007392 276. Tian H, Hu S, Cazelles B, et al. Urbanization prolongs hantavirus epidemics in cities. Proc Natl Acad Sci U S A. 2018;115(18):4707-4712. doi:10.1073/pnas.1712767115 277. Tiee MS, Harrigan RJ, Thomassen HA, Smith TB. Ghosts of infections past: Using archival samples to understand a century of monkeypox virus prevalence among host communities across space and time. R Soc Open Sci. 2018;5(1):171089. doi:10.1098/rsos.171089 278. Tolsá MJ, García-Peña GE, Rico-Chávez O, Roche B, Suzán G. Macroecology of birds potentially susceptible to West Nile virus. Proc R Soc B Biol Sci. 2018;285(1893). doi:10.1098/rspb.2018.2178 279. Torres AG. Escherichia coli diseases in Latin America-a “One Health” multidisciplinary approach. Pathog Dis. 2017;75(2). doi:10.1093/femspd/ftx012 280. Trout Fryxell RT, Hendricks BM, Pompo K, et al. Investigating the Adult Ixodid Tick Populations and Their Associated Anaplasma, Ehrlichia, and Rickettsia Bacteria at a Rocky Mountain Spotted Fever Hotspot in Western Tennessee. Vector-Borne Zoonotic Dis. 2017;17(8):527-538. doi:10.1089/vbz.2016.2091 281. Tsui CK-M, Miller R, Uyaguari-Diaz M, et al. Beaver Fever: Whole-Genome Characterization of Waterborne Outbreak and Sporadic Isolates To Study the Zoonotic Transmission of Giardiasis. mSphere. 2018;3(2):1-17. doi:10.1128/msphere.00090-18 282. Ukhovskyi V V., Vydayko NB, Aliekseieva GB, Bezymennyi M V., Polupan IM, Kolesnikova IP. Comparative analysis of incidence of leptospirosis among farm animals and humans in Ukraine. Regul Mech Biosyst. 2018;9(3):409-416. doi:10.15421/021861 283. VanderWaal K, Deen J. Global trends in infectious diseases of swine. Proc Natl Acad Sci U S A. 2018;115(45):11495- 11500. doi:10.1073/pnas.1806068115

118

284. Venter M, Pretorius M, Fuller JA, et al. West Nile virus lineage 2 in horses and other animals with neurologic disease, South Africa, 2008–2015. Emerg Infect Dis. 2017;23(12):2060-2064. doi:10.3201/eid2312.162078 285. Vhoko K, Iannetti S, Burumu J, Ippoliti C, Bhebhe B, De Massis F. Estimating the prevalence of brucellosis in cattle in Zimbabwe from samples submitted to the central veterinary laboratory between 2010 and 2014. Vet Ital. 2018;54(1):21-27. doi:10.12834/VetIt.1111.6191.2 286. Villa-Mancera A, Reynoso-Palomar A. Bulk tank milk ELISA to detect IgG1 prevalence and clustering to determine spatial distribution and risk factors of Fasciola hepatica-infected herds in Mexico. J Helminthol. 2019;93(6):704-710. doi:10.1017/S0022149X18000792 287. Vlasova AN, Amimo JO, Saif LJ. Porcine rotaviruses: Epidemiology, immune responses and control strategies. Viruses. 2017;9(3):48. doi:10.3390/v9030048 288. Walsh MG, De Smalen AW, Mor SM. Climatic influence on anthrax suitability in warming northern latitudes. Sci Rep. 2018;8(1):1-9. doi:10.1038/s41598-018-27604-w 289. Walsh MG, Wiethoelter A, Haseeb MA. The impact of human population pressure on flying fox niches and the potential consequences for Hendra virus spillover. Sci Rep. 2017;7(1):1-13. doi:10.1038/s41598-017-08065-z 290. Walsh MG, Willem de Smalen A, Mor SM. Wetlands, wild Bovidae species richness and sheep density delineate risk of Rift Valley fever outbreaks in the African continent and Arabian Peninsula. Althouse B, ed. PLoS Negl Trop Dis. 2017;11(7):e0005756. doi:10.1371/journal.pntd.0005756 291. Wang X, Wei X, Song Z, et al. Mechanism study on a plague outbreak driven by the construction of a large reservoir in southwest china (surveillance from 2000-2015). Vinetz JM, ed. PLoS Negl Trop Dis. 2017;11(3):e0005425. doi:10.1371/journal.pntd.0005425 292. Wells K, Gibson DI, Clark NJ, Ribas A, Morand S, McCallum HI. Global spread of helminth parasites at the human– domestic animal–wildlife interface. Glob Chang Biol. 2018;24(7):3254-3265. doi:10.1111/gcb.14064 293. White AM, Zambrana-Torrelio C, Allen T, et al. Hotspots of canine leptospirosis in the United States of America. Vet J. 2017;222:29-35. doi:10.1016/j.tvjl.2017.02.009 294. Wise EL, Marston DA, Banyard AC, et al. Passive surveillance of United Kingdom bats for lyssaviruses (2005-2015). Epidemiol Infect. 2017;145(12):2445-2457. doi:10.1017/S0950268817001455 295. Wróblewska-Łuczka P, Chmielewska-Badora J, Zwoliński J, et al. Seroepidemiologic evaluation of exposure to infection with hantavirus (Serotype puumala) among forestry workers in Poland. Balt For. 2017;23(3):612-618. https://dx.plos.org/10.1371/journal.pntd.0006579. 296. Wu FT, Bányai K, Jiang B, et al. Novel G9 rotavirus strains co-circulate in children and pigs, Taiwan. Sci Rep. 2017;7:1-11. doi:10.1038/srep40731 297. Xu X, Cui Y, Xin Y, et al. 2018 Xu Cui PLoS Path Genetic diversity and spatial-temporal distribution of Yersinia pestis in Qinghai Plateau, China.pdf. Attwood SW, ed. PLoS Negl Trop Dis. 2018;12(6):1-17. doi:10.1371/journal.pntd.0006579 298. Yaglom HD, Nicholson WL, Casal M, Nieto NC, Adams L. Serologic assessment for exposure to spotted fever group rickettsiae in dogs in the Arizona–Sonora border region. Zoonoses Public Health. 2018;65(8):984-992. doi:10.1111/zph.12517 299. Zhang H, Zhang E, He J, Li W, Wei J. Genetic characteristics of Bacillus anthracis isolated from northwestern China from 1990 to 2016. Yang R, ed. PLoS Negl Trop Dis. 2018;12(11):1-11. doi:10.1371/journal.pntd.0006908 300. Zhou L, Ren R, Yang L, et al. Sudden increase in human infection with avian influenza A(H7N9) virus in China, September-December 2016. West Pacific Surveill response J WPSAR. 2017;8(1):6-14. doi:10.5365/WPSAR.2017.8.1.001 301. Zhou L, Tan Y, Kang M, et al. Preliminary Epidemiology of Human Infections with Highly Pathogenic Avian Influenza A(H7N9) Virus, China, 2017. Emerg Infect Dis. 2017;23(8):1355-1359. doi:10.3201/eid2308.170640 302. Zohaib A, Saqib M, Athar MA, et al. Countrywide Survey for MERS-Coronavirus Antibodies in Dromedaries and Humans in Pakistan. Virol Sin. 2018;33(5):410-417. doi:10.1007/s12250-018-0051-0

119

Appendix 7. Additional Figures

Figure 1. Mapped zoonotic pathogens from included studies (n=302)

Zoonotic Pathogen Count Multiple zoonoses 38 Leptospira interrogans 21 Leishmania species 19 Brucella species 13 Rabies virus 13 Avian influenza virus 10 Rickettsia species 10 Taenia species 11 Influenza A virus 9 Mycobacterium species 9 Borrelia species 7 Echinococcus species 7 Hantavirus 7 Salmonella species 7 Bacillus anthracis 6 Hepatitis E virus 6 Middle East Respiratory Syndrome Coronavirus 6 Yersinia species 6 Coxiella burnetii 5 Rift Valley fever virus 5 West Nile virus 5 Fasciola 4 120

Lyssavirus 4 Escherichia coli 3 Francisella tularensis 3 Japanese encephalitis virus 3 Lassa virus 3 Powassan virus 3 Thelazia callipaeda 3 Toxoplasma gondii 3 Babesia microti 2 Bartonella 2 Ebola virus 2 Methicillin-resistant Staphylococcus aureus 2 Monkeypox virus 2 Nipah Virus 2 Oropouche virus 2 Plasmodium falciparum, Plasmodium vivax, Plasmodium ovale, and 2 Plasmodium malariae Rotavirus 2 Severe fever with thrombocytopenia syndrome virus 2 Sporothrix 2 Staphylococcus aureus 2 Trypanosoma species 2 Yellow fever virus 2 Anjozorobe virus 1 Brugia malayi 1 Burkholderia mallei 1 Campylobacter 1 Chlamydia species 1 Coccidioides species 1 Crimean-Congo haemorrhagic fever virus 1 Cryptococcus species 1 Cryptosporidium species 1 Dirofilaria species 1 Ehrlichia species 1 Filovirus 1 Flavivirus 1 Giardia 1 Heartland virus 1 Hendra virus 1 Kyasanur forest disease virus 1 Ross river virus 1 Streptococcus suis 1

121

Strongyloides stercoralis 1 Swine influenza virus 1 Tahyna virus 1 Toxocara 1 Trichinella 1 Vaccinia virus 1

122