A logistic regression of risk factors for disease occurrence on Asian shrimp farms

Serious shrimp-disease outbreaks have reduced shrimp production and slowed industry growth since 1991. This paper tests factors such as farm siting an d design, and farm-management practices for relationships with disease occu rrence. Logistic regression is used to analyze farm-level data from 3951 sh rimp farms in 13 Asian countries. Disease occurrence is modeled as a 0-1 va riable where 1 = disease loss of greater than or equal to 20% to any 1 crop , and 0 = losses of < 20%. Logistic regression is performed for each of 3 l evels of shrimp culture intensity, i.e. extensive, semi-intensive, and inte nsive. Attempts to apply logistic regression models to each country were no t successful due to insufficient data for most countries. Factors affecting disease occurrences were quite different for different farming intensities . Farms that had larger pond production areas, with larger number of farms discharging effluent into their water supply canals, and removed silt had g reater disease occurrence. On the other hand, farms that practiced polycult ure and took water from the sea through a canal had lower disease occurrenc e.