A model for determining optimal sampling protocols targeting detection of new disease introduction into expected negative animal populations

For animal production sites (especially those of very high value, such as genetic production and breeding/reproduction sites), that are expected to be and remain negative for a particular disease agent, an appropriate detection plan for new introduction of undesired disease agents must include both continuous clinical observation and well designed diagnostic sampling/testing protocols. Whereas basic sample size determination methods for disease detection from single samplings are generally understood, the factors that contribute to appropriatly sized and timed sampling are less well understood and frequently poorly applied in the design and execution of protocols. A stochastic model was developed to improve sampling protocol development targeting detection of new disease introduction into expected negative animal populations.

An algorithm described by Rothman and Greenland (1998) was modified and incorporated into a tool built to stochastically model onset of detection of new disease agent introduction in expected negative animal populations. An animal isolation scenario was modeled where a new cohort of 500 replacement females are moved into an empty site every 60 days, where no live animals exit to a downstream site during the first 30 days and animals are moved from the isolation to the downstream site as needed over the second 30 days of the overall 60 day period, after which the isolation site is empty and sanitized in preparation for the next incoming group of replacement females. Model scenarios were compared by varying index positives at entry (1 or 5 animals), contact probability (30% or 70%), transfer probability (30% or 70%), detection onset lag (2 or 3 days) and detection duration (14 or 21 days). Detection probability curves were generated across a 60 day period for each scenario at sample sizes of 15, 30, 45 and 60. For the sample sizes evaluated, the cohort day at which > 95% of model runs were detected as positive was used as the criteria for comparing scenarios.

Two scenarios comparing 5 or 1 index positives (both with 30%/70% contact/transfer probabilities and 2/21 day detection onset/duration) at a sample size of 30, the 95% detection threshold was achieved at the 15th and 23rd cohort day, respectively.

This stochastic sampling protocol model can be used to derive more informed and appropriate detection sampling protocols for the detection of new disease agent introduction into expected negative animal populations, as well as generate tables to be used as references for disease detection sampling that take into account the dynamics of exposure and transmission in animal cohorts.

Disclosure of Interest: None Declared.