Spatio-Temporal Modeling of Animal Diseases

By Dr. Zubair Khalid, DVM, MS, PhD · 2026-08-11

Key Takeaways

• Spatio-temporal modeling integrates geographic location and time to analyze animal disease dynamics, focusing on space-time event data (case location, date, population at risk) to detect clusters and estimate transmission.
• The space-time scan statistic, implemented in SaTScan, is a primary tool for cluster detection, utilizing cylindrical windows to assess deviations from random distribution and generating metrics like relative risk and p-values.
• Disease spread occurs at characteristic spatial and temporal scales influenced by host movement, vector dispersal, and environmental factors; matching data resolution (e.g., county vs. farm, monthly vs. daily) to these scales is critical to avoid aggregation bias.
• Regression models extend cluster detection by linking case counts to environmental and socio-economic risk factors, employing methods like zero-inflated negative binomial regression and spatial analysis to quantify drivers of disease.
• Phylodynamic approaches, using whole-genome sequencing and Bayesian evolutionary analysis, reconstruct pathogen movement and divergence over time, distinguishing local transmission from repeated introductions, as demonstrated in lumpy skin disease virus studies.
• Data sources vary in reliability, with diagnostic laboratory data requiring careful consideration of testing behavior and spatial/temporal sampling intensity, while participatory epidemiology offers valuable insights in data-sparse regions.

Spatio-temporal modeling integrates geographic location and time into a unified analytical framework for understanding how animal diseases arise, persist, and spread across populations and landscapes. This article provides veterinary researchers with a structured account of the conceptual foundations, analytical methods, and practical applications of spatio-temporal epidemiology in animal populations. It addresses the question of how space-time data can be used to detect clusters, estimate transmission dynamics, identify environmental drivers, and inform surveillance design across species and production systems.

The readership for this material includes graduate students in veterinary epidemiology, researchers designing observational studies of livestock or wildlife diseases, and diagnosticians interpreting surveillance data. The scope covers models that explicitly incorporate both spatial and temporal dimensions, including scan statistics, regression-based approaches, phylodynamic methods, and simulation frameworks. Purely spatial methods that ignore time fall outside this treatment, as do purely temporal time-series analyzes that disregard location.

The evidentiary basis draws on published studies of lumpy skin disease, highly pathogenic avian influenza, rabies, anthrax, bovine trichomoniasis, and livestock diseases in semi-arid rangelands, alongside international surveillance standards from the World Organization for Animal Health and foundational epidemiologic methods from the United States Centers for Disease Control and Prevention.

At a Glance

Conceptual Foundations of Space-Time Analysis

The Space-Time Interaction

Disease events that occur close in both space and time provide stronger evidence of transmission than events that are close in only one dimension. This interaction between spatial proximity and temporal proximity is the central quantity that spatio-temporal models estimate. A case occurring 5 km from a previous case within 3 days suggests local transmission, whereas the same spatial distance across a 6-month interval may reflect independent introductions from a common source.

The CDC principles of epidemiology in public health practice describe the classic triad of person, place, and time as the descriptive foundation of outbreak investigation. In veterinary applications, the equivalent triad is host species, geographic location, and calendar time. Spatio-temporal modeling formalizes this descriptive triad into quantitative inference about the processes that generate observed case patterns.

Epidemic Waves and Transmission Scales

Diseases spread at characteriztic spatial and temporal scales determined by host movement, vector dispersal, environmental persistence, and husbandry practices. The Nigerian highly pathogenic avian influenza H5N1 epidemic between 2006 and 2008 illustrates this principle: three distinct epidemic waves occurred, with peaks from January to March in northern regions during 2006 and 2007, followed by a July to September peak in southern regions in 2007. Space-time scan statistics identified three clusters extending across state and international borders, consistent with both local and long-distance transmission mechanisms.

The scale mismatch between surveillance data and transmission processes creates the central analytical challenge. County-level aggregation may capture regional spread but obscure farm-to-farm transmission. Monthly aggregation may reveal seasonal patterns but hide explosive short-term outbreaks. The analyst must match data resolution to the hypothesized transmission mechanism or explicitly model multiple scales.

Space-Time Cluster Detection Methods

The Scan Statistic

The space-time scan statistic, developed by Kulldorff and implemented in the SaTScan software, remains the most widely used method for cluster detection in veterinary epidemiology. The method constructs a cylindrical window in three dimensions: a circular geographic base with a height representing a time interval. The cylinder moves across the study region and time period, comparing the observed case count inside each window against the expected count under a null hypothesis of random distribution.

The Nigerian avian influenza study applied this approach using the Cuzick-Edwards test and the SaTScan space-time scan statistic to analyze 1,654 suspected and 299 confirmed H5N1 outbreaks. The analysis identified clusters that crossed administrative boundaries, providing evidence that control strategies limited to individual states would be insufficient.

Retrospective versus Prospective Scanning

Retrospective scanning analyzes the complete historical dataset to identify clusters that have already occurred. This approach suits descriptive studies, hypothesis generation, and evaluation of past control measures. The Texas bovine trichomoniasis study used retrospective scan statistics on 31,202 diagnostic laboratory test results aggregated at county level with monthly time steps, identifying a spatial cluster in southeastern Texas that could only partially be explained by cattle herd density.

Prospective scanning analyzes data sequentially, testing for emerging clusters as new cases accrue. This method supports early warning systems and real-time surveillance. The choice between retrospective and prospective analysis depends on whether the research question concerns historical patterns or ongoing detection.

Directional and Network-Based Extensions

Standard scan statistics assume isotropic spread, meaning equal probability of transmission in all directions. Disease spread along rivers, roads, or livestock movement corridors violates this assumption. Directional tests, as applied in the hippopotamus anthrax study in Uganda's Queen Elizabeth Protected Area, examine whether epidemic movement follows a preferred bearing. That study analyzed 317 hippo cases from 2004 to 2005 and 137 from 2010, using permutation models of the spatial scan statistic and directional tests to characterize epidemic propagation patterns.

Regression Models for Spatio-Temporal Risk Factors

Zero-Inflated and Spatial Regression Approaches

Cluster detection identifies where and when disease aggregates, but does not explain why. Regression models extend the analysis by relating case counts or incidence rates to candidate risk factors measured across space and time. The South African rabies study applied univariate zero-inflated negative binomial regression followed by multivariable spatial analysis using integrated nested Laplace approximation for two time periods, 1998 to 2002 and 2008 to 2012. This two-stage approach first identified candidate ecological and socio-economic variables, then modeled their joint effects while accounting for spatial correlation.

Environmental Correlates

The northern Kenya study of livestock diseases in semi-arid rangelands demonstrated how environmental variables structure disease risk. Participatory epidemiology combined with spatial analysis identified peste des petits ruminants, foot and mouth disease, and camel trypanosomiasis as the highest-impact diseases for pastoral livelihoods. Correlations between mean annual rainfall and disease occurrence were significant for East Coast fever, cattle helminthiasis, cattle anaplasmosis, and camel pox, with correlation coefficients ranging from 0.63 to 0.77 in absolute value. Disease hotspots corresponded closely to seasonal herd distributions, indicating that livestock movement patterns mediate environmental risk.

Phylodynamic Approaches to Space-Time Reconstruction

Whole-Genome Phylogeny and Temporal Signal

Genomic sequence data add a molecular dimension to spatio-temporal analysis. Phylogenetic trees reconstructed from pathogen genomes can be dated using molecular clock models, allowing researchers to estimate when lineages diverged and where they likely originated. The lumpy skin disease virus study compared whole-genome-sequence-based phylogeny against single-gene phylogenies and found that single-gene approaches lacked phylogenetic and spatiotemporal resolution. Strains from multiple clades within cluster 1.2 corresponded with recorded outbreaks across Eurasia and South Asia, while cluster 2.5 strains spread in Southeast Asia.

Bayesian Evolutionary Analysis

The South African rabies study constructed phylogenetic trees from rabies virus genomes isolated from dogs, jackals, and an African civet, then applied Bayesian evolutionary analysis using a strict time clock model. This approach estimated the reproductive number for identified disease clusters and linked molecular diversity to the spatial and temporal distribution of cases. Phylodynamic methods are particularly valuable for distinguishing between sustained local transmission and repeated introductions from external sources, a distinction that purely statistical cluster detection cannot make.

Data Sources and Surveillance Infrastructure

Diagnostic Laboratory Data

State-wide diagnostic laboratory systems provide a rich but biased source of spatio-temporal data. The Texas trichomoniasis study relied on the Texas Veterinary Medical Diagnostic Laboratory system, which performs at least 95% of all T. fetus testing in the state. Testing prevalence peaked in summer at 5.5%, and the overall proportion of positives was 3.7% among 31,202 test results. Diagnostic data reflect testing behavior as much as disease occurrence, and the analyst must account for spatial and temporal variation in sampling intensity.

Participatory Epidemiology

In data-sparse regions, participatory epidemiology methods generate valuable information through structured interviews with livestock keepers. The northern Kenya study selected key informants purposively with the help of local leaders, identifying goats as the most economically important livestock species and ranking diseases by their impact on pastoral livelihoods. This approach captures local knowledge about disease occurrence and seasonality that formal surveillance systems may miss.

International Reporting Standards

The World Organization for Animal Health maintains surveillance standards and reporting frameworks that govern how member countries collect, analyze, and notify disease information. The WOAH terrestrial animal health code provides the international standards for animal health, welfare, surveillance, and trade-related disease control. Spatio-temporal analyzes that identify new clusters or unusual disease patterns may trigger notification obligations under these standards, and researchers should be aware of the regulatory context in which their findings will be interpreted.

Model Selection in Applied Settings

Choosing a spatio-temporal modeling approach begins with the surveillance question, not the dataset. For outbreak detection in ongoing surveillance, prospective scan statistics applied to weekly or monthly aggregated case counts provide the most direct answer to "is something occurring now that should not be." For retrospective investigation of a completed epidemic, the same scan statistic applied to historical data identifies where and when clusters occurred, as demonstrated in the Nigerian HPAI H5N1 epidemic of 2006 to 2008, where three distinct spatio-temporal clusters extended across state and international borders Nigerian HPAI spatio-temporal outbreak analysis.

When the objective shifts to risk factor identification, regression-based approaches become necessary. Zero-inflated negative binomial models accommodate the excess zeros common in livestock disease data, where many spatial units report no cases in most time periods. The rabies investigation in northern South Africa used univariate zero-inflated models followed by multivariable spatial regression with integrated nested Laplace approximation, a computationally efficient Bayesian approach that handles spatial random effects without Markov chain Monte Carlo sampling rabies spatio-temporal epidemiology in northern South Africa.

Decision Framework for Model Choice

The semi-arid rangelands of northern Kenya illustrate the last row. Participatory epidemiology identified Peste des Petits Ruminants, foot and mouth disease, and camel trypanosomiasis as the highest-impact diseases, and spatial analysis showed that disease hotspots tracked seasonal herd distributions instead of static geographic features livestock disease epidemiology in northern Kenya rangelands. This finding has direct control implications: vaccination and treatment campaigns should be scheduled to coincide with wet-season and dry-season congregation sites, not delivered at fixed locations year-round.

Case Study: Bovine Trichomoniasis in Texas

The Texas bovine trichomoniasis surveillance program provides a worked example of diagnostic laboratory data applied to spatio-temporal analysis. The study population comprised bulls tested through the Texas Veterinary Medical Diagnostic Laboratory system, which performs at least 95% of all T. fetus testing in the state. Preputial samples were cultured and confirmed by real-time PCR, and results were aggregated at the county level with monthly time steps Tritrichomonas fetus spatio-temporal analysis in Texas bulls.

The dataset contained 31,202 test results with a 3.7% positive proportion. Testing prevalence peaked in summer at 5.5%, a seasonal signal consistent with the breeding season and the practice of testing bulls before sale or turnout. The scan statistic identified a spatial cluster in southeastern Texas that could only partially be explained by cattle herd density, suggesting that additional factors, such as herd movement patterns or regional testing practices, contributed to the observed distribution.

Several lessons transfer to other production systems. First, diagnostic laboratory data are convenience samples shaped by testing behavior, not random samples of the underlying population. Seasonal testing peaks may reflect management calendars instead of true incidence peaks. Second, the spatial resolution of the analysis is constrained by the aggregation unit. County-level aggregation smooths local heterogeneity and can obscure clusters that operate at the ranch or watershed scale. Third, the absence of a cluster does not prove absence of disease, it may indicate insufficient testing density or poor spatial coverage.

For practitioners interpreting such outputs, the critical question is whether the cluster represents a true excess of disease or an excess of testing. Comparing the spatial distribution of test volume against the distribution of positives provides a partial check. Where testing density is low, cluster detection has limited power, and negative results should be interpreted cautiously.

Species and Production System Modifiers

The correct analytical choice changes with the host species, production system, and data environment. Wildlife systems present different constraints than livestock systems. The hippopotamus anthrax outbreaks in Uganda's Queen Elizabeth Protected Area required methods suited to a free-ranging population with no census data and no owner-reported cases. The investigators used permutation models of the spatial scan statistic, which do not require population-at-risk denominators, and directional tests to determine epidemic movement hippo anthrax spatio-temporal analysis in Uganda. These methods are appropriate wherever denominator data are unreliable, including many wildlife and pastoral systems.

In contrast, intensively managed livestock populations with registration databases and movement records support network-based approaches that model transmission along animal movement pathways. These methods are less applicable in wildlife systems where movement data are absent. The choice between permutation-based and population-based scan statistics hinges on denominator availability, and the choice between areal and network models hinges on whether transmission follows administrative boundaries or trade and movement connections.

Production system also determines temporal resolution. In intensive poultry systems, where flock turnover occurs every 5 to 8 weeks, weekly aggregation captures the relevant transmission dynamics. In extensive beef systems, where animals move seasonally between grazing areas, monthly or seasonal aggregation may be more appropriate. The Nigerian HPAI analysis identified three epidemic waves with peaks in January to March in northern regions and July to September in southern regions, a pattern that reflects both poultry production cycles and the timing of introduction events Nigerian HPAI spatio-temporal outbreak analysis.

Documentation and Reporting Standards

Spatio-temporal analyzes intended for regulatory or trade purposes must align with international reporting frameworks. The World Organization for Animal Health maintains surveillance standards and notification requirements that define what constitutes a reportable event and how surveillance data should be structured WOAH animal health surveillance standards. The Terrestrial Animal Health Code provides additional standards for surveillance and trade-related disease control WOAH terrestrial animal health code.

Analytical outputs should document the case definition, the population at risk and its source, the aggregation units and time steps, the statistical methods and software, the cluster detection criteria including significance thresholds and maximum cluster size, and the sensitivity of results to these choices. Where phylogenetic methods are used, the sequence accession numbers, alignment methods, and clock model assumptions must be reported. The lumpy skin disease virus analysis demonstrated that single-gene phylogenies lack the resolution of whole-genome approaches, and the authors recommended generating at least one whole-genome sequence whenever possible to anchor the phylogenetic interpretation lumpy skin disease virus whole-genome phylogeny. This recommendation applies broadly: the cost of whole-genome sequencing continues to fall, and the phylogenetic resolution it provides is often decisive for distinguishing local persistence from repeated introduction.

Limitations and Interpretation Pitfalls

Spatio-temporal models share common failure modes. The ecological fallacy operates when inferences about individual-level transmission are drawn from areal aggregates. The modifiable areal unit problem means that changing the aggregation boundaries can change the cluster locations. Temporal aggregation choices similarly affect results: monthly aggregation may obscure clusters that last days, while daily aggregation may fragment a sustained outbreak into multiple small clusters.

Surveillance intensity is the most pervasive confounder. Areas with more active surveillance generate more case reports regardless of true disease incidence. The Texas trichomoniasis analysis illustrates this: testing prevalence varied seasonally, and the identified cluster could not be fully explained by herd density, leaving open the possibility that testing behavior contributed to the pattern Tritrichomonas fetus spatio-temporal analysis in Texas bulls.

Phylogenetic interpretations carry their own uncertainties. Sampling density strongly influences tree topology, and unsampled lineages can create apparent geographic structure that reflects sampling gaps instead of true transmission barriers. Bayesian evolutionary analyzes require explicit priors on clock rates and population models, and results should be examined for sensitivity to these priors. The rabies study in South Africa combined spatio-temporal cluster detection with phylogenetic reconstruction, an approach that triangulates evidence from independent data sources and is more robust than either method alone rabies spatio-temporal epidemiology in northern South Africa.

Where the evidence base is thin, acknowledge it. Participatory epidemiology provides valuable hypothesis generation but relies on informant recall and local disease terminology, which may not map cleanly onto laboratory-confirmed diagnoses. The Kenya study identified disease hotspots through community reporting and correlated them with rainfall, but the authors noted that confirmatory laboratory data were limited livestock disease epidemiology in northern Kenya rangelands. In such settings, the analytical output should be framed as a prioritization tool for further investigation, not as a definitive disease map.

Recognized Failure Modes in Space-Time Analysis

Spatio-temporal models fail in characteriztic ways that are identifiable before results are interpreted. The most common failure is the ecological fallacy, where inference about individual-level transmission is drawn from aggregated data. County-level or district-level case counts obscure within-unit heterogeneity, and a cluster detected at coarse resolution may represent several independent introductions instead of local spread. Discriminating between these possibilities requires examination of phylogenetic relatedness among isolates from the cluster, as demonstrated in analyzes of lumpy skin disease virus where whole-genome sequencing distinguished multiple clades circulating within the same outbreak period whole-genome phylogeny of lumpy skin disease virus.

Second, edge effects distort cluster detection near study boundaries. The scan statistic compares observed cases against expected counts generated within a defined window, and windows truncated by administrative borders underestimate expected values, producing spurious clusters. Detection is improved by buffering study regions, using network-based distances instead of Euclidean distances, and interpreting clusters near borders with caution. The Nigerian HPAI H5N1 analysis identified clusters extending across state and international borders, which the authors interpreted as evidence of long-distance transmission instead of artefact, but such interpretations require supporting data on animal movements spatio-temporal epidemiology of HPAI H5N1 outbreaks in Nigeria.

Third, temporal aggregation choices alter cluster geometry. Monthly aggregation smooths short epidemic waves, while daily aggregation fragments sustained outbreaks into multiple small clusters. Sensitivity analysis across aggregation levels is the standard corrective, and results that change materially with aggregation should be reported as unstable.

Common Analytical Errors and Corrections

Less experienced analysts frequently confuse incidence with prevalence when interpreting cluster outputs. The scan statistic identifies areas of elevated case counts relative to the population at risk, not necessarily areas of high disease frequency. A cluster in a densely populated livestock region may reflect testing intensity instead of biological risk. The Texas bovine trichomoniasis study addressed this by using testing data from a laboratory system performing at least 95% of state testing, thereby reducing ascertainment bias, but still noted that identified clusters could only partially be explained by cattle density spatio-temporal epidemiology of Tritrichomonas fetus in Texas bulls.

A second recurring error is over-interpretation of p-values from multiple testing. Retrospective scanning evaluates many candidate windows, and the reported p-value is adjusted for the multiple testing inherent in the procedure. Analysts who treat the scan statistic p-value as a conventional hypothesis test overstate significance. The correct interpretation is that the cluster is unlikely to have arisen by chance given the entire scanning procedure, not that the cluster is necessarily epidemiologically meaningful.

Third, analysts often ignore the population-at-risk denominator entirely. When denominator data are unavailable, the scan statistic defaults to a Bernoulli model, which requires case and non-case locations. Using case-only data with a Poisson model without a proper denominator produces clusters that reflect the spatial distribution of the sampled population, not disease risk. Corrective action is to obtain livestock population estimates, even coarse ones, or to restrict interpretation to relative comparisons within the same dataset.

Evidence Gaps and Divergent Expert Opinion

The evidence base for spatio-temporal methods in veterinary medicine is uneven across species and production systems. Wildlife disease applications are constrained by incomplete detection, as illustrated by anthrax surveillance in hippopotamus populations where carcass detection depends on accessibility and scavenger activity spatio-temporal epidemiology of anthrax in hippos in Uganda. Expert opinion differs on whether such incomplete data justify complex modeling or whether descriptive approaches are more honest. Some analysts argue that Bayesian hierarchical models can accommodate detection bias explicitly, while others contend that the assumptions required exceed what field data can support.

Divergence also exists on the value of phylodynamic reconstruction. Whole-genome sequencing provides superior resolution for reconstructing transmission networks, but the cost and technical requirements limit routine application. The lumpy skin disease virus analysis concluded that single-gene sequencing lacks sufficient phylogenetic resolution and recommended at least one whole-genome sequence per outbreak where possible whole-genome phylogeny of lumpy skin disease virus. This recommendation is not universally adopted, particularly in resource-limited settings where routine surveillance relies on cheaper methods.

Escalation and Reporting Thresholds

Referral to a specialist spatial epidemiologist is warranted when clusters cross administrative borders, when results will inform regulatory action, or when the analysis combines multiple data sources with different spatial resolutions. Laboratory involvement is required when phylogenetic interpretation is needed to distinguish reintroduction from sustained local transmission. Regulatory reporting obligations are defined by international standards, and veterinarians should consult the World Organization for Animal Health surveillance standards and the terrestrial animal health code for current notification requirements WOAH animal health surveillance standards and WOAH terrestrial animal health code. Where results are intended to guide control policy, the analysis should be documented with sufficient methodological detail for independent replication, including data sources, aggregation choices, model parameters, and sensitivity analyzes.

Frequently Asked Questions

What Is the Minimum Data Quality Needed for a Defensible Space-Time Analysis?

At minimum, you need reliable case counts with consistent spatial identifiers and time stamps. Diagnostic laboratory datasets, such as the state-wide bovine trichomoniasis testing records from Texas, can support cluster detection when case definitions and testing volumes are stable across the reporting period. If case ascertainment changes mid-study, for example after a new diagnostic test is introduced, interpret temporal trends with caution. For phylodynamic work, whole-genome sequences provide substantially better spatio-temporal resolution than single-gene data, as demonstrated in lumpy skin disease virus investigations. When only partial data exist, restrict the analysis window and state the limitation explicitly in the methods.

How Should I Proceed When Spatial Data Are Only Available at Coarse Administrative Levels?

County-level or district-level aggregation is workable but reduces power to detect small clusters and can obscure local transmission. The Nigerian H5N1 analyzes successfully identified clusters spanning state and international borders using province-level data, so coarse resolution does not preclude useful inference. Use the smallest administrative unit consistently available, aggregate time in monthly or weekly intervals, and avoid interpreting cluster boundaries as precise farm-level locations. If herd-level coordinates exist for only a subset of cases, consider a sensitivity analysis comparing results with and without the precise locations. Report the spatial unit in all maps and tables so readers can judge the resolution.

What Are the Practical Cost and Computational Constraints for Routine Surveillance Programs?

Retrospective scan statistics and descriptive mapping run on standard desktop computers with open-source software. Prospective scanning for early outbreak detection requires more infrastructure because data must be updated and reanalyzed on a fixed schedule. Whole-genome sequencing remains more expensive than single-gene typing, but the improved phylogenetic and spatio-temporal resolution justifies the cost when investigating emerging or genetically variable pathogens. For routine surveillance in resource-limited settings, participatory epidemiology combined with hotspot mapping, as applied in northern Kenya, offers a low-cost alternative that captures seasonal movement patterns. Prioritize investment in consistent case reporting over analytical sophistication, because models cannot compensate for missing or biased surveillance data.

How Do Space-Time Methods Differ for Wildlife Reservoirs Compared with Domestic Livestock?

Wildlife systems require explicit attention to host movement, seasonal aggregation, and environmental drivers. Anthrax outbreaks in hippos at Queen Elizabeth Protected Area showed distinct epidemic curves and directional spread that reflected water-dependent aggregation and carcass-mediated transmission. Domestic livestock analyzes can assume relatively stable farm locations and focus on animal movements, but wildlife analyzes must incorporate habitat use and migration corridors. Sampling is also more opportunistic in wildlife, so detection probability varies across space and time. When wildlife and domestic species interact, as in rabies transmission between dogs and jackals in South Africa, joint analysis of both host populations improves inference about maintenance and spillover dynamics.

What Records Should I Maintain to Allow Future Spatio-Temporal Reanalysis?

Record the case definition, diagnostic method, sampling date, and the spatial reference for every case. Distinguish between the date of onset, date of sample collection, and date of laboratory confirmation, because these can differ by weeks. Store coordinates in a standard projection and retain the original administrative identifiers. Document changes in surveillance intensity, testing capacity, and reporting protocols, since these affect both cluster detection and regression modeling. For molecular data, archive sequence files with collection metadata. Following international surveillance standards, such as those published by the World Organization for Animal Health, ensures that records remain interoperable across jurisdictions and useful for retrospective studies.

How Do I Explain Space-Time Cluster Results to Producers or Regional Authorities?

Frame results in terms of actionable risk periods and locations instead of statistical mechanics. State that a cluster means more cases occurred in that place and time than expected by chance, and describe what that implies for movement controls or vaccination timing. Reference the seasonal patterns identified in the analysis, such as the dry-season and wet-season grazing hotspots documented in northern Kenya. Acknowledge uncertainty honestly, noting that absence of a detected cluster does not prove absence of transmission. Provide maps with simple legends and avoid presenting cluster boundaries as exact farm-level predictions. Direct producers to species-specific clinical resources for disease recognition and to the relevant animal health authority for control requirements.

Related Clinical & Scientific Guides

• Evaluating Veterinary Surveillance System Attributes
• Network Analysis for Infectious Disease Spread in Animal Populations
• Randomized Controlled Trials in Veterinary Field Settings

References and Further Reading

• Lumpy Skin Disease Virus Genome Sequence Analysis: Putative Spatio-Temporal Epidemiology, Single Gene versus Whole Genome Phylogeny and Genomic Evolution.. 2023.
• Spatio-temporal epidemiology of highly pathogenic avian influenza (H5N1) outbreaks in Nigeria, 2006-2008.. 2012.
• Spatio-temporal epidemiology of animal and human rabies in northern South Africa between 1998 and 2017.. 2022.
• Spatio-temporal epidemiology of livestock diseases in the variable semi-arid rangelands of northern Kenya.. 2023.
• Spatio-temporal epidemiology of Tritrichomonas fetus infection in Texas bulls based on state-wide diagnostic laboratory data.. 2012.
• Spatio-temporal epidemiology of anthrax in Hippopotamus amphibious in Queen Elizabeth Protected Area, Uganda.. 2018.
• WOAH Animal Health Surveillance Standards. WOAH.
• CDC Principles of Epidemiology in Public Health Practice. CDC.
• MSD Veterinary Manual, Professional Edition. MSD Veterinary Manual.

Related Articles

• Using Simulation Models in Veterinary Epidemiology
• Basic Reproductive Ratio (R0) in Veterinary Epidemiology
• Effect Modification and Interaction in Veterinary Epidemiology
• Understanding Bias in Veterinary Epidemiological Studies
• Understanding Ecological Studies in Veterinary Epidemiology

This article is educational professional reference material for veterinary audiences. It is not a substitute for veterinary diagnosis, individual clinical judgment, current product labeling, or applicable regulatory requirements.