In physics, we have a useful notion: "to first approximation".
For example, Newton's Laws are correct, to first approximation. Einstein gives us corrections, but, for most folks, in most normal situations, Newton is good enough.
That's a useful term to know. There are many things that people try to make complicated and model, when in reality, the "first approximation" is good enough understanding upon which to make a decision.
In general barring wierd disease, fat people eat more than skinny people. When most fat people eat less, they become skinny.
I've left room for exceptions in my statement. The exceptions of people who gain weight despite reduced calorie intake requires further medical study. But the first approximation concept says "calories in, calories out" works for most fat people. Put them in a secure weight loss camp, and they'll lose weight and be cured of some diseases. Funny how that works.
I heard part of a Freakonomics Radio cast, where a skeptic argues that there's not conclusive science that being fat causes diseases like diabetes. Sure, it's POSSIBLE that it's not the fat that causes it, but the extra substance that fat people eat that skinny people don't that's causing it.
But the simplistic observation that people who become fat then tend to get diabetes and it goes when they diet and become skinny leads us to the simple conclusion that being fat causes the disease.
That may not be absolutely technically correct as the true cause of the disease, but it's close enough for government work, and is functionally effective as a treatment for most patients.