A new study finds that individuals with diabetes and obesity were about half as likely to have a heart attack if they underwent weight loss surgery compared to similar patients who did not have the surgery.

This week – bariatric surgery, and the similarities between Danny DeVito and Arnold Schwarzenegger as we discuss this paper, appearing in the Journal of the American Medical Association.

Bariatric surgery is on the rise with over 200,000 operations performed last year.

But until now, we had no strong data linking bariatric surgery with improved macrovascular outcomes in diabetic patients.

OK – macrovascular disease is just another way of saying heart attacks and strokes – the major killers in patients with diabetes. We’ve got ample data that medical management of diabetes can improve microvascular outcomes like retinopathy and diabetic nephropathy.

But macrovascular outcomes are a harder nut to crack. Remember the ACCORD trial? That one showed that patients with diabetes randomized to intensive glucose control had a higher risk of death from any cause.

But bariatric surgery does more than just lower glucose levels. It reduces weight. It improves blood pressure and cholesterol levels. Heck, it improves sleep apnea. There are a lot of ways bariatric surgery could reduce heart attacks.

Does it?

This study suggests it does, and in a big way – reducing the rate of heart attack by 50% in fact, and cutting the all-cause mortality rate by more than 60% at 5-years. Crazy good numbers.

But this was not a randomized trial. It compared patients who chose to undergo bariatric surgery to other patients who didn’t get bariatric surgery.  Here’s how it worked.

Across 4 large health systems, the authors identified 5,301 patients with diabetes and a BMI above 35 who underwent bariatric surgery – the vast majority roux-en-Y gastric bypass.

They then found other patients with diabetes and a BMI above 35 that didn’t get bariatric surgery. From that pool they tried to find “statistical twins” to the surgery patients: individuals of the same sex, roughly the same age, hemoglobin A1C level, interaction with the healthcare system, duration of diabetes, and BMI.

The big question in a study like this, regardless of the results, is how well do we think this match worked? Are these controls really the same in every way to the people who got surgery except for the fact that they didn’t get surgery?

Are they twins like Parent Trap?

Or twins like Twins?

Looking at Table 1 in the paper, we get some hints the match wasn’t perfect.

Surgery patients were more likely to have private insurance and less likely to have Medicare and they had, on average, more contact with the health system.

And that makes sense, right? Getting bariatric surgery is not just about having a certain BMI. You have to cross a number of hurdles – some insurance companies require you to prove you have tried (and failed) multiple weight loss programs, or medical regimens. Some require psychiatric evaluations. And let’s not forget that in the end, some bariatric surgeon has to agree to do the operation in the first place. You want selection bias, cause that’s how you get selection bias.   

So let’s agree that a randomized trial would be better. But observational data is still data. Do we think that all those imbalances between the groups is enough to account for this effect:

My opinion? Probably not. This data is fairly compelling. I am cautiously optimistic here that bariatric surgery will save lives in this population.

The question now, it seems to me, is whether all those hurdles that caused this study to be more “Twins” than “Parent Trap” are really necessary in the first place.