Field evaluation of an equine influenza ELISA used in New South Wales during the 2007 Australian outbreak response.

This research article assesses the effectiveness of a commonly-used equine influenza test during the Australian horse flu epidemic in 2007. Analysing data from infected and uninfected horses, the report found the test to be accurate and informative.

Research Context

• The research conducted arises from the 2007 equine influenza outbreak in Australia, where tens of thousands of horses were infected. During this outbreak, the commonly-used bELISA test for influenza A was utilized to determine infection.

Data Collection

• Post-outbreak examinations provided field data with results from 475 known infected horses and 1323 known uninfected ones.
• This served in assessing the performance of the test under typical field conditions rather than in a controlled laboratory environment.

Methods of Evaluation

• The investigators employed various statistical techniques to analyse the test’s efficiency. This includes the Receiver Operating Characteristic (ROC) plots, the area under the curve (AUC) method, and hypothesis testing.
• ROC plots are broadly used in the evaluation of medical diagnostic tests where sensitivity and specificity are regarded. The AUC technique is another statistical method used to measure a test’s accuracy. The closer the AUC is to one, the better the test is at distinguishing between the different conditions.
• Hypothesis testing was used to determine if the results were significantly meaningful or due to chance.

Test Performance

• The bELISA test was found to be highly accurate. This was demonstrated by its AUC score of 0.993 (with a standard error of +/-0.003), where a rating of 1 represents a perfect test.
• Hypothesis testing found the results significantly informative, represented by a z-score of -32.0, and p-value of less than 0.0001. In statistics, a z-score indicates how many standard deviations an element is from the mean. A p-value below 0.05 usually indicates strong evidence against the null hypothesis, thus making the results statistically significant.
• Lastly, the researchers found the test’s sensitivity (the rate of true positives) to be 0.992, and the specificity (the rate of true negatives) at 0.967, both fairly close to 1, indicating high accuracy.