From the Editor
Did the treatment work?
A producer had an outbreak of postweaning Escherichia coli diarrhea in an off-site nursery barn. He consulted with his veterinarian, feed salesman, and the owners of the sow herds supplying the pigs. Then he implemented the following new procedures: antibiotics and acidifiers were added to the water; egg yolk antibody was added to the feed; the antibiotic in the feed was changed; the disinfectant used in the rooms was changed; and the pigs were sorted into pens by sow herd source rather than by size. After these changes had been implemented, the next group of nursery pigs did not experience postweaning E coli diarrhea.
What worked? The veterinarian said it was separation of the pigs by source, and the water medication. The feed salesperson thought it was the egg yolk antibodies and the new in-feed antibiotic. The producer concluded that it was the new disinfectant. Perhaps nothing worked.
I was consulted by a nursery unit with classic postweaning E coli problems. The "best looking, large pigs" died suddenly, appeared dehydrated, and had discoloured extremities and fluid-filled intestines. Other pigs had diarrhea and the perianal region was scalded. The diagnostic laboratory cultured large numbers of K88 E coli. No changes were made to the medication or the management of the facility, and the problem did not appear in the next group of pigs weaned. If I had implemented a medication or management change, I would have been convinced that my treatment worked. However, because no change was implemented and the problem resolved itself, I realised that postweaning E coli is an intermittent disease. The affected farrow-to-feeder pig unit had six nursery rooms that were emptied in an all-out manner, cleaned and disinfected, and then filled over a 2-week time period. We followed pigs in this barn over a 2-year period. On four occasions, pigs experienced high morbidity and mortality due to postweaning E coli diarrhea. However, only once did the problem occur in consecutive groups of pigs. I do not mean to conclude that there is no useful treatment for postweaning E coli diarrhea, but I am sure that the clinical signs do not occur in each group of nursery pigs.
If we really want to determine "what works" in a scientific manner, we have to choose one treatment and apply that treatment in a randomized, repeated manner. The experimental unit is the smallest group in which we can apply a treatment.
An injectable treatment is easily applied to individual animals. To determine the effect of such a treatment, pigs would be individually identified with ear tags, randomly assigned to treatment, and then injected. Treated pigs and untreated pigs would be housed in pens together so that treatments would not be clustered by pen. In this case, the experimental unit is the pig.
If the treatment is applied in the feed or water, the experimental unit is the smallest group that can be given a particular feed. If each pen receives a separate feed, pen is the experimental unit; if not, the experimental unit becomes the room.
Let us assume that the outcome of interest is postweaning E coli diarrhea. You have been working with a producer who typically puts potassium penicillin in the water for the newly weaned pigs. He is complaining about ongoing postweaning E coli diarrhea problems that increase morbidity, mortality, and culls. For water medication, the smallest group that can be treated is the room, so that is the experimental unit. You flip a coin to randomly assign room to treatment group. The first room is the control room and the second room is the "treatment" room. In your study, the control will be the standard water medication (potassium penicillin) and the "treatment" group will receive apramycin in the water. During your study, the control pigs show no signs of postweaning E coli, but the second group experiences severe problems due to E coli. What do you conclude? #1, apramycin does not work; #2, penicillin does work; #3, neither treatment works; #4, the study needs to be repeated.
What if postweaning E coli diarrhea occurred in the pigs receiving penicillin but not in those receiving apramycin? Because we expected apramycin to work and did not expect penicillin to work, the results matched our preconceived concept of the problem. What would you conclude? #1, apramycin does work so we should use this in all subsequent nursery groups; #2, penicillin does not work; #3, neither treatment works; #4, the study needs to be repeated.
The correct answer for both scenarios is #4. Without replication, we cannot be certain of our conclusions. We do not know if the treatment really caused the observed change or was due to chance alone. The next step would be to continue to rotate rooms between penicillin and apramycin until you observed the "cause and effect" between treatment and outcome on numerous occasions.