One consistent theme of SBM is that the application of science to medicine is not easy. We are often dealing with a complex set of conflicting information about a complex system that is difficult to predict. That is precisely why we need to take a thorough and rigorous approach to information in order to make reliable decisions.
The same is true when applied to an individual patient. Often times we cannot make a single confident diagnosis based upon objective information. We have to be content with a diagnosis that is based partly on probability or on ruling out other possibilities. Sometimes we rely upon a so-called “therapeutic trial” to help confirm a diagnosis. If, for example, it is my clinical impression that a patient is probably having seizures, but I have no objective information to verify that (EEG and MRI scans are normal, which is often the case) I can help confirm the diagnosis by giving the patient an anti-seizure medication to see if that makes the episodes stop, or at least become less frequent. Placebo effects make therapeutic trials problematic, but if you have an objective outcome measure and a fairly dramatic response to treatment, that at least raises your confidence in the diagnosis.
We can apply the same basic principle on the population level. If a public health intervention is addressing the actual cause of one or more diseases, then we should see some objective markers of disease frequency or severity decrease over time. Putting fluoride in the public water supply decreased the incidence of tooth decay. Adding iodine to salt decreased the incidence of goiter. Fortifying milk with vitamin D decreased the incidence of rickets. However, removing thimerosal from the childhood vaccine schedule did not reduce the incidence of autism (or the rate of increase in autism diagnosis). That is because calcium deficiency causes rickets, but thimerosal (or the mercury it contains) does not cause autism.