Is It Time To Abandon BMI Criteria For Bariatric Surgery?Friday, January 6, 2012
As regular readers will be well aware, our recent publications on the Edmonton Obesity Staging System clearly show that BMI alone is a rather poor measure of mortality risk associated with excess weight.
Thus, according to our analyses of the NHANES population, 20-30% of obese individuals (Stage 0/1) had virtually no increased mortality risk even over the almost 20 year observation period, compared to Stage 2/3 individuals, who had substantially increased risk. Conversely, even in the ‘overweight’ category (BMI 25-30), almost 50% of individuals had a mortality risk as high of that of obese Stage 2/3 individuals.
This means that indications for obesity treatment based on BMI alone will overtreat a substantial number of obese individuals, who may have little benefit in terms of mortality, and miss an even greater number of individuals, who may well benefit from such treatments.
These observations are directly relevant to yesterday’s post on the findings of the SOS study.
As readers will recall, not only was there no relationship between BMI levels and cardiovascular outcomes in the SOS population but the overall cardiovascular risk of these participants – despite the majority being ‘severely’ obese – was surprisingly low. In fact, the annual risk for experiencing an adverse cardiovascular outcome for SOS participants was well under 1% per year!
This risk level is highly reminiscent of the overall risk of Stage 0 obese individuals in our NHANES analyses.
Thus, it is readily apparent why it took almost 15 years to demonstrate any cardiovascular benefit of bariatric surgery in the SOS study – clearly this was a very ‘low-risk’ obese population.
Contrast this to the almost 3% annual cardiovascular event rate for the participants in the SCOUT trial, which, by definition, consisted exclusively of Stage 2/3 individuals. In this population, it took less than 3.5 years of even very modest weight-loss (3 to 10 kg) to significantly reduce cardiovascular outcomes.
But did surgery prove more beneficial in higher-risk participants in the SOS trial? It certainly did!
In fact, the only predictor of greater benefit of having bariatric surgery in the SOS paper proved to be having an elevated plasma insulin level – a rather crude marker of insulin resistance. It is fair to assume that these participants were in fact those with higher obesity Stages (elevated fasting insulin levels alone, would already suggest at least Stage 1 obesity).
The importance of this difference is reflected in the numbers-needed-to-treat (NNT): as low as 21 in participants with baseline plasma insulin concentrations above the median (>17.0 mU/L) and as high as 173 in individuals below or at the median (≤17.0 mU/L) insulin concentration.
This essentially means that you would need to operate only 21 patients with Stage 1+ obesity to ‘save’ one life (over 15 years) but 173 Stage 0 patients for the same benefit. It does not take a financial genius to figure out that from a ‘cost-per-life-saved’ perspective, operating on Stage 1+ patients is a ‘no-brainer’ whereas operating on Stage 0 patients would (and should) probably raise some eyebrows (especially in a publicly funded healthcare system).
Thus, as we have argued before (and argued by Livingstone in an editorial accompanying the SOS paper in JAMA), it is high time we fully appreciate the “inadequacy of BMI as an indication for bariatric surgery” and begin adopting more sophisticated criteria (such as those of the Edmonton Obesity Staging System) to ensure that this treatment is available to those who are likely to benefit the most.
Padwal RS, Pajewski NM, Allison DB, & Sharma AM (2011). Using the Edmonton obesity staging system to predict mortality in a population-representative cohort of people with overweight and obesity. CMAJ : Canadian Medical Association journal = journal de l’Association medicale canadienne, 183 (14) PMID: 21844111
Livingston EH (2012). Inadequacy of BMI as an indicator for bariatric surgery. JAMA : the journal of the American Medical Association, 307 (1), 88-9 PMID: 22215170