How Spatial Epidemiology Helps Understand Infectious Human Disease Transmission
Introduction
- Infectious diseases can be transmitted:
- Directly (e.g., influenza)
- Indirectly (e.g., through water, soil)
- Spatial dimensions are crucial for understanding and managing diseases.
- Advancements in computing and statistical methods enhance disease visualization and analysis.
Figure 1: Visualization of spatial dimensions in disease spread.
Spatial Data Types
- Point Data:
- Raw data such as incidences, deaths, physical locations (e.g., hospitals)
- Aggregate Data:
- Summaries of individual points like incidence rates by area
- Point Data provides detailed and precise information, allowing for pinpointing specific outbreak locations.
- Aggregate Data helps in understanding broader trends but can obscure local variations.
Disease Mapping Techniques
Location Mapping
- Historical examples:
- Yellow fever in New York (1798)
- Cholera in London (1854)
- Advantages: Quick observation of disease spread.
- Limitations: Privacy issues, difficulty in reflecting population distributions.
Surface Mapping
- Choropleth Maps:
- Example: Filariasis in India, Q fever in the Netherlands.
- Advantages: Visualize incidence rates clearly.
- Limitations: May hide demographic factors, sensitive to spatial scales.
Figure 2: Example of a choropleth map showing disease incidence.
Advanced Mapping Techniques
Bayesian Smoothing
- Reduces bias in local risk estimation.
- Example: Applied in studies of dengue, influenza, and tuberculosis.
Kernel Density Estimation (KDE)
- Identifies dense points and visualizes as smooth surfaces.
- Example: Dengue cases in Delhi.
Figure 3: Illustration of Bayesian Smoothing applied to disease data.
Figure 4: KDE showing the density of dengue cases.
Overall Spatial Patterns
Clustering Methods
- Nearest Neighbour Ratio:
- Assesses clustering degree (e.g., rabies in China).
- Ripley’s K Function:
- Measures spatial clustering over distance (e.g., leishmaniasis in Brazil).
Figure 5: Visualization of clustering methods.
- Clustering methods help identify hotspots and inform targeted interventions.
Statistical Tests for Clustering
Point Data
- Nearest Neighbour Ratio
- Cuzick and Edwards’ test
- Ripley’s K function
Aggregate Data
- Spatial autocorrelation statistics (e.g., Moran’s I)
- These tests are crucial for confirming whether observed patterns are statistically significant.
Hot Spot Detection
Local Indicators of Spatial Association (LISA)
- Identifies clusters of high or low values.
Spatial Scan Statistics
- Detects clusters in space and time (e.g., leishmaniasis in Brazil).
Figure 6: Example of hot spot detection using LISA and spatial scan statistics.
Identifying Risk Factors
Neighbourhood Effect
- Disease risk influenced by nearby cases.
Spatial Heterogeneity
- Variation in disease risk across different areas.
Spatial Regression Methods
- Adjust for neighbourhood effects and spatial heterogeneity.
Figure 7: Illustration of neighbourhood effect and spatial heterogeneity.
Conclusion
- Spatial epidemiology provides insights into the spread and control of infectious diseases.
- Future Directions: Enhancing GPS methods, optimizing Bayesian estimations.
- Key Impact: Improved disease surveillance, hypothesis generation, and strategic control measures.
Figure 8: The impact of spatial epidemiology on disease control.
References
- Lin, C.-H., & Wen, T.-H. (2022). How spatial epidemiology helps understand infectious human disease transmission.