Online Program

Return to main conference page
Wednesday, September 25
Wed, Sep 25, 1:15 PM - 2:30 PM
Challenges in Pain Assessment in Human and Animal Studies: Perspectives from FDA, Industry, and Academia

Minimizing Bias and Placebo Effects in Chronic Pain Studies (300955)

*Dorothy Cimino Brown, Elanco Animal Health 

Keywords: selection bias, observation bias, placebo effects, randomization, blinding, regression-to-the-mean, chronic pain

The use of patient-oriented outcomes in clinical research highlights the challenges of minimizing biases & placebo effects in chronic pain studies.

Selection bias occurs when subjects with influencing factors appear more frequently in one study group. To minimize selection bias, randomization includes 2 processes. (1) generation of an unpredictable allocation sequence based on a random procedure. (2) concealment of that sequence which ensures those screening animals for study inclusion have no knowledge of group assignment prior to its randomization. This is distinct from blinding, which is being unaware of an animals’ treatment group during outcomes data collection.

Observation bias occurs when knowledge of a study subject’s group allocation influences identification of relevant events during the study. This leads to exaggerated estimates effects. The potential for observation bias is related to the subjectivity of the outcomes used. When the outcome of interest is subjective, such as owner assessment of pain/function, knowledge of the animal’s group allocation can affect the score attributed to that animal. To obtain unbiased estimates of intervention effect, the effect documented in the control group is removed from the effect documented in the intervention group.

Placebo effects are improvements documented in a control group with no active intervention. Commonly, the assumption is that the animals are not improved, but merely a bias of owners or investigators who want to believe that animals are improving. However, animals in a control group can actually improve, even with no active intervention due to regression-to-the-mean. There are a variety of approaches to minimizing placebo effects in the design and implementation of a study. In addition, understanding the factors associated with placebo ‘responders’, as well as treatment ‘non-responders’, allows for innovative study designs that can increase the sensitivity for detecting a positive effect.