BMI does correlate with cardiovascular risk, and is useful when applied on a population basis, but less so to some individuals who are muscular. From experience most people who use the "BMI isn't useful, look at weightlifters/bodybuiilders" comment, are far from having a muscular physique.
Ways around this: DEXA scanning for body morphometry (http://en.wikipedia.org/wiki/Body_fat_percentage), is an accurate way of determining body fat composition, but is time consuming and would add cost.
A better predictor of cardiovasulcar risk (and other manifestations of the metaboic syndrome such as diabetes) is waist circumference. Rather than taxing healthcare or unhealthy foods (which are fine in moderation), why not heavily tax big pants (>37 inch waist is the cut off for risk in a male).