BMI is a comparatively crude predictor of mortality. Measures that take into account fat around the waist are better predictors.
The idea is that there are fats we deposit around our waist and organs that are likely indicative of metabolic disorders, diabetes and heart disease.
Here's a paper that constructs a measure of body shape, using both BMI and waist circumference, and uses it to predict all-cause mortality:
https://journals.plo...al.pone.0039504
And there is a calculator based on their measure "a body shape index" ABSI which gives your probability of mortality compared to average. Be careful with the waist measure if you do it, as small differences can affect the outcome strongly.
https://www.fatcalc.com/absiAs far as manipulating your numbers go, diet is the best predictor followed by exercise. Eat lots of whole plants, and eat relatively little meat. If you eat meat, eat something that ate a natural diet to get the omega 3 to 6 ratio in your favor. Omega 3s are anti-inflammatory, and 6s pro-inflammatory, and they both compete for the same receptors. Ancestrally it is estimated that we would have evolved with close to a 1:1 ratio of these fats. Now that number is heavily skewed in favor of omega 6s.