When training doctors are taught science involving research rather than memorization at all, it is usually epidemiology, which can NEVER establish anything more than a *chance* that there might be a statistically significant correlation between one thing and another. Correlation can never establish causation at all (Did you know that over 80% of child rapists drink water?! See what I mean? Correlation never implies causation.) This is the kind of study all over the news that makes people think that eggs are a wonder food one decade and a Silent Killer the next. BTW, when the correlation is lower that 200%, it means there is no relationship between the studied factors better than chance.
So, "Men who ate kiwi fruit more than twice a month were 198% more likely to have their penises shrivel up and fall off." means exactly the same thing as "Men who ate kiwi fruit more than twice a month were no more likely than other men to have their penises shrivel up and fall off." So feel free to ignore any such media "proof" that there is a magic food or diet. There isn't. We're omnivores who have lived on every possible type and range of foods. Actual malnutrition syndromes like beri beri or scurvy, and toxic exposures as with lead poisoning from food storage jars painted inside with lead based pigments are the exception, not the rule.
Epidemiology is a perfectly sound scientific method: it identifies out of the endless range of possibilities which correlations *might* actually have a causal relationship, and warrant a true scientific test with controlled diverse populations, double blinded so neither researchers nor participants know who gets what. And even when a cause and effect relationship is established in that test, the direction of cause and effect is not established. To do so we must do another true test with proper methods.
There are many things we truly do not know, and it's not malpractice to do your best to treat people with illnesses based on existing observations, even if they are technically guesses and hunches. Qualitative research is just as important as the above quantitative methods, because they can yield important "your mileage may vary" case evidence about individual differences and group differences that can't be quantitatively captured. Quant research can only reflect populations and are biased toward the average - evidence for outliers or any other individuals can't be adequately captured.
Studies linking BMI to cancer in populations can't yield any information at all about an individual's risk of getting cancer based on her BMI. Medical practice is not currently scientfically based for the most part, even when there's good evidence.
For example, in reproductive medicine there would be essentially no episiotomies or circumcisions performed in hospitals if medical practice were truly scientifically based. Episiotomies (cutting the vagina with a scalpel to make more room) CAUSE vaginal/labial/anal tears 90% of the time and worsen them when cut after a tear starts, while outcomes for the babies are just as good for the the mothers without episiotomies, meaning that episiotomies aren't "saving babies' lives". Circumcision has no scientifically significant health benefit at all, it's basically just a cultural genital mutilation ritual. (Which is admittedly up to the parents, many of whom don't regard it in this light.) Some obstetricians simply won't believe the science on either common practice because their approach to medicine uses faith based thinking rather than scientific thinking.
Faith based thinking is wonderful in the proper context, of course. But it has deadly results when used in medicine, where the faith based belief that fat people are less healthy because of their own moral turpitude, the "fact" that they are chronically non-compliant pathological liars because they report that they followed treatment but did not lose weight and that they are less deserving of care because they are perceived as (these are quotes from medical professionals) "repulsive" and "lazy" and "sloppy" and "lacking in self control." If fat people lived morally upright lives eating less food than the WHO and UN call starvation levels, they wouldn't be fat, and moreover they would enjoy all the health benefits that malnutrition brings. Like the scurvy, beri beri, and brain damage caused by malabsorption of necessary nutrients.