On March 29, The Australian published an article under the heading “Demand for obesity surgery soars, but lifestyle training the best value“, reporting results of an economic evaluation, commissioned by Medibank Private and carried out by KPMG Econtech, which compared lifestyle, drug, and surgical interventions for obesity.

The study was a cost-benefit analysis, where health benefits are converted to dollar values, allowing to report the net lifetime cost/benefit of an intervention. The study outcomes are that gastric banding produces a net cost of $3366, pharmaceutical interventions a net benefit of $608, and lifestyle interventions a net benefit of $1764.

Now this is rather odd, for two reasons. The first is that bariatric surgery is indicated only for (morbidly) obese persons for whom lifestyle and drug interventions have failed. So the target population for surgery is different from that for lifestyle and drug interventions, making the comparison invalid. This is not mentioned in the report.

The second reason is that from the literature it is known that surgery has large and mostly lasting effects, while lifestyle and drug interventions have small and largely vanishing effects. Granted, surgery is more expensive, but if that produces a lifelong stream of health benefits, compared to a short trickle from lifestyle and drug interventions, this should be well worth the investment.

So we decided to see how KPMG Econtech had obtained the results (KPMG Econtech: Economic Modelling of the Impact of Obesity and Obesity Interventions. March 26, 2010).

The key to understanding the KPMG results is the definition of effect size. Usually the effect size of an obesity intervention is expressed as the average weight change, but here the KPMG study introduces an innovation: the “success rate”. An additional innovation is that the criterion for success differs by type of intervention. For gastric banding, success is defined as the proportion of patients achieving a loss of baseline weight of 20% or more after 10 years of follow-up. For lifestyle and drug interventions, success is defined as the proportion achieving a loss of baseline weight of 10% or more after one year of follow-up.

This creates a rather tilted playing field. The length of  follow-up matters because a meta-analysis of lifestyle interventions has found that weight regain is such that on average after about five years patients are back on their baseline weight. KPMG, in contrast, assumes that the success rate measured at the end of year one is permanent — an assumption that is clearly much more optimistic than the evidence warrants. In the Swedish bariatric surgery trial, where KPMG got the surgery data from, the control group, which had lifestyle interventions of various kinds, had actually gained weight after 10 years of follow-up.

The KPMG report explains why for surgery success requires a 20% weight loss, while lifestyle and drug interventions only need 10% as follows (on page 49): “The success rate associated with a 20% weight gain [sic!] is recorded in research papers for surgical intervention, but not pharmacological or lifestyle intervention. This implies that a higher weight loss of 20 percent is more prevalent among those who undertake the option of surgery.”

So KPMG seems to be saying that because drug and lifestyle interventions are less effective than surgery, they should be kept to a lower standard.

We would certainly qualify the KPMG study as “innovative”. In the article in The Australian, Medibank’s Julie Andrews qualifies the growth in bariatric surgery as “alarming”.

Could there, whisper it, be a relation with the innovative methods of KPMG?

Jan J Barendregt, MA, PhD, is Associate Professor of Epidemiology, School of Population Health, University of Queensland.