Towards a more humane intracerebral pathogenicity index
by
Albert Charles Trajstman
Applied Bioinformatics Group, CSIRO Mathematical & Information Sciences

Diagnostic testing of animals for certain viruses often requires an assessment of the pathogenicity of the virus. Ordered measures of virus pathogenicity exist and one is the intracerebral pathogenicity index (ICPI). Assessments based on the ICPI are of fundamental importance to livestock trade and to animal health worldwide consequently it is important that any assessments based on the ICPI avoid ad hoc techniques. A rigorous approach is to set the ICPI in a statistically sound framework;

By modelling the progression of the avian disease caused by Newcastle disease virus (NDV) as a one-step, time homogeneous Markov Chain (MC) it is possible to set the intracerebral pathogenicity index (ICPI) in a statistical context. Previous assessments of ICPI values for different viruses ignore the fact that the same ICPI value may occur via different sequences of disease status. However, the MC model approach takes into account the state transitions by which an animal came to obtain its particular ICPI value. Because the probability that an animal remains normal approaches zero as time increases, the MC model suggests a pathogenicity index based on the multiplying factor that determines this probability’s approach to zero.

The determination of a virus’s intracerebral pathogenicity index (ICPI) requires tracking sick animals until death or until the eighth day of the standard observation period. A more humane procedure would be to euthanase sick animals however such intervention in the standard protocol could compromise the rating of the virus. A modified scoring system is proposed whereby euthanased animals receive a score that on average does not alter the ICPI. The variance of an animal’s contribution to the optimum modified ICPI is never greater than the variance of its contribution to the standard ICPI.

Date received: March 26, 2002