One of the pillars upon which modern medicine is based is objectively
determining if a certain procedure, treatment, etc. is the cause of a certain
outcome.
For example, when the meningococcal conjugate vaccine was
first introduced, there were reports of Guillain-Barre syndrome (a rare, temporary
paralysis) in a few persons who had received the vaccine and some concern that
the vaccine may have caused it. It turned out that when the data was analyzed,
it was found that people who had received the vaccine were no more likely to
get GBS than those who had not received the vaccine. So the few cases of GBS
after vaccination were a coincidence. This concept is often summed up with the
saying “Correlation is not causation,” meaning that just because two things
have a temporal relationship doesn’t mean one caused the other. Now years
later, we know that the vaccine does not cause GBS and we have been able to
protect a multitude of young people against deadly meningococcal disease with
the vaccine.Similarly, it is important to have control groups to see if treatments work. If 80% of people taking Treatment A get better, it must work, right? What if 80% of people with the same disease who did not get Treatment A (the control group) also get better? Suddenly Treatment A doesn’t look so great.
In medical trials, treatments are usually compared to a placebo. A placebo is something that is made to look, taste, etc. like the treatment in question but has no medication in it (ie. a “sugar pill”). In an ideal trial, the patients and doctors do not know who is receiving a placebo and who is receiving the treatment being investigated. Then the results are compared to see if the treatment in question worked any better than the placebo.
This approach is what allows us to provide treatments that provide benefits to patients and avoid treatments that don’t work or may even be harmful. Stories of individuals’ experiences can guide further investigation, but it is often only when the experiences of many individuals are compiled in a systematic way which minimizes bias as much as possible that one can separate the wheat from the chaff.