CutL: an alternative to Kulldorff's scan statistics for cluster detection with a specified cut-off level

Area of Science:

• Epidemiology
• Biostatistics
• Geographic Information Systems (GIS)

Background:

• Identifying epidemiological clusters is crucial for public health surveillance.
• Existing methods for cluster detection often involve complex statistical approaches.
• Establishing reference incidence rates from literature is a key component in cluster detection.

Purpose of the Study:

• To present a novel method for detecting epidemiological clusters with significantly elevated incidence rates.
• To offer an alternative to computationally intensive methods like Monte Carlo simulations.
• To provide a statistically sound and cartographically presentable approach to cluster identification.

Main Methods:

• Utilizes the binomial exact test for one proportion to compare observed incidence rates against literature-derived reference levels.
• Employs an algorithm to aggregate adjacent areas with potential clusters, thereby minimizing multiple comparisons.
• Leverages statistical software for cartographic analysis and visualization of results.

Main Results:

• The proposed method effectively identifies clusters with significantly higher incidence rates.
• It maintains high sensitivity and specificity in cluster detection.
• Results are comparable to established methods such as Kulldorff's scan statistics.

Conclusions:

• The developed method offers an efficient and accurate approach to epidemiological cluster detection.
• It provides a valuable tool for public health researchers and practitioners.
• Cartographic presentation enhances the interpretability and utility of cluster findings.