An awful lot of health issues go away by themselves over time.
It isn't a good idea to assume that a treatment worked just because a patient got better. That's what you need statistics and statistically meanful numbers of test subjects for.
