Take a Close Look at Company Claims

Companies list research on their website to make their tests sound scientific. When you look closer, you will find:

“Loosely-related” studies from prestigious journals: For example, a company that sells a food sensitivity test based on white blood cells listed an article from The New England Journal of Medicine about white blood cells during allergic reactions. It looked impressive, but the study had little to do with the test they were selling.   

Group testimonials A group of people use their product or service and report feeling better, but there isn’t a placebo group or blinding. Reporting information from a group sounds scientific, but it is no different than individual testimonials (i.e., My name is Sara, and I felt great ten minutes after restricting these foods).

Scientific formatting:  Companies often write their group testimonials in a scientific journal format, giving the impression their “research” has been published in a reputable journal.

Cherry-picking their evidence: Companies list studies that support their theory and ignore any that do not.  

Lies: Many advertisements include “confirmed by studies from the University of XYZ.” I’ve called several food sensitivity testing companies, and not one has been able to support their claims.

Poorly defined measurement outcomes: The immune system is an elaborate orchestration of hundreds of cells and chemicals. Well-conducted studies measure a few clinically meaningful immune cells/chemicals. Dubious companies often measure many immune cells. Any change is reported as evidence of their product’s effectiveness. These are not proper studies because:

  • Clinically meaningless results: Changes in obscure or poorly understood immune cells/chemicals are clinically useless (until more is learned about that cell).
  • Statistical results may be inaccurate: When many cells/chemicals are measured, they will probably find a difference by chance alone. As you know, statistical differences between the treatment and control groups are usually qualified with p<0.05 (meaning there is a 5% chance that the statistical difference was an error). For every twenty cells/chemicals measured, at least one incorrect significant difference will likely be found (1/20 = 5%).

Companies could easily show evidence for their FS test with this inexpensive, simple study.

  • Recruit a group with a condition the test purports to help (e.g., migraine headache).
  • Measure a clinically meaningful outcome (ideally an objective measurement by a blinded researcher).
  • Randomize the subjects. The treatment group restricts based on their test results, and the control group uses someone else’s test results (e.g., scrambled results).

To my knowledge, no companies have done this simple study. Perhaps, they know the results would be unfavorable!