Designing Field Trials to Compare Vaccines or Antibiotics

June 16, 2009 — 

Maintaining healthy pigs, improving productivity, and minimizing the effects of disease through the use of effective management procedures or pharmaceuticals (vaccines and antimicrobials) are important. Field trials conducted on commercial farms are a useful way of testing the ability of a new product to improve production or to decrease illness and death.

Pigs should be allocated to treatment groups using a formal random process, which means that each pig in the trial has the same chance of being selected to be in either the treatment or control (untreated or standard treatment) group. It is essential to have control pigs in order to determine the difference in health and production between the treated pigs and the untreated pigs that are present in the barn at the same point in time. When there are no controls, performance can only be compared to historical data.

Ideally, the people working in the barn will be "blind", i.e. they will not know which pigs received which treatment, so they will not care for the pigs differently.

The experimental unit is the smallest independent group of pigs that can be assigned to a treatment group. With vaccines or injectable antibiotics, the experimental unit is usually the pig. If the treatment cannot be randomly assigned to an individual pig, for example with feed trials, then the calculations will be based on pens of pigs rather than individual pigs. The treatment should be randomly assigned to pens of pigs and the growth rate should be compared on a per pen basis. In order to reduce the total number of pigs required, the number of pens can be increased and the number of pigs per pen decreased.

In order for a field trial to be valid, the observed association between the treatment and the outcome must be due to the treatment and not the result of chance or bias. Bias is a systematic error in measurement or a systematic difference between groups. Random assignment of pigs should remove bias. Thus there will be no association between the weight of the pigs and any other factor.

In order to determine whether or not there is a significant difference between two groups, the number of pigs must be large enough to measure the expected difference. Sample size equations can be used to determine the number of pigs needed. A sample that is too small will result in finding no difference between the groups when a difference does exist. A sample that is too large is expensive. The sample size is determined by how variable the outcome (ADG or death rate) is expected to be, the expected difference between treatment and control groups, and how confident one wants to be of the results (usually set at a 95% level).

At the completion of the trial, statistical tests are used to determine if there was a significant difference in performance between the treated and control groups. The trial should be repeated over time to ensure that the results are consistent.

Summarized by Janet Alsop - Veterinarian Disease Prevention - Swine/OMAFRA Presented by Dr. Cate Dewey at the 2009 London Swine Conference. http://www.omafra.gov.on.ca/english/livestock/swine/news/mayjun09.htm

Source: Atlantic Swine Research Partnership, June 16, 2009