May 7, 2009

Smoking Gun: Did Avandia Kill those ACCORD Patients Who Attained 6.5% A1cs?

Doctors in the US have been told the ACCORD study "proved" it was dangerous for people with diabetes to shoot for a 6.5% A1c. ACCORD study found that a higher rate of heart attack was seen in the group with lower blood sugars. This is leading many doctors to give the tragically flawed advice to patients that they should keep their A1cs up, closer to 7% to preserve health.

In contrast, another, larger and longer study, ADVANCE, found no harm and slightly fewer heart attacks in the intensive control group who attained those A1cs of 6.5% over a period of five years.

The full text publications describing both the ACCORD and ADVANCE studies are now available for free and it seems to me they make it crystal clear what killed people in the ACCORD study.

ACCORD is the study that found excess deaths in the group that attained A1cs below 6.5%. ADVANCE did not. You'll find both studies here:

Intensive Blood Glucose Control and Vascular Outcomes in Patients with Type 2 Diabetes. The ADVANCE Collaborative Group. NEJM, Volume 358:2560-2572, June 12, 2008, Number 24

Effects of Intensive Glucose Lowering in Type 2 Diabetes The Action to Control Cardiovascular Risk in Diabetes Study Group.[ACCORD] NEJM Volume 358:2545-2559, June 12, 2008 Number 24

So let's look at what the differences were in how these studies lowered A1c.

You can find out exactly what drugs people took in the ADVANCE study in the supplementary material published with this study.

As you'll see, ADVANCE relied mostly on a sulfonylurea drug, Gliclazide, which isn't available in the US, though in the UK, Australia, and Canada it is considered the best of the sulfonylurea drugs.

Patients who did not get to goal with gliclazide were put on metformin and several other drugs including basal insulin, usually NPH (33% vs 18% on Lantus or Levemir). Only 19% of the participants in the tight control group used fast acting insulin, half of these used R and half analogs.

The actual breakdown of who took what in ADVANCE shows that the preponderance of study subjects (77%) were taking the gliclazide, metformin, or a combination of the two. Only 32 (.6%) of the 5,571 people in the intensive control arm of ADVANCE were taking Avandia or Actos.

Contrast this with ACCORD. You can see what drugs people were taking in ACCORD in this table.

It turns out 4,702 of the 5,128 people in the intensive treatment arm of ACCORD were taking at TZD drug. That's 91.7% of all of them. But here's the kicker: almost all of them--4,677 or 91.2%--were taking Avandia.

I read this and found myself baffled at how many reports on the differences between these two studies I'd read claimed there was no one drug associated with the excess deaths. The disparity between the two studies could not be clearer. One study used almost no Avandia and people who attained tight control had no negative outcomes compared to those with standard control. The other gave Avandia to 9 out of 10 participants in the tight control arm and saw excess deaths.

Yet ACCORD--the Avandia study--is the study doctors and insurers are using to argue that tight control is dangerous and causes heart attacks.

I'm flabbergasted and you should be too. We already knew that Avandia killed people before this study was published. So why wasn't the "Danger" of lowering A1c pinned to the drug people were taking in the study, a drug already known to kill people? And why did the ADA urge patients on Avandia to keep taking it, even after its danger became evident?

I'll let you draw your conclusions, but my guess it has something to do with what drug companies donate what to the American Diabetes Association.

Another bit of information that got lost in the reporting of ACCORD is that the increased risk of excess death in the tight control group was largely found in people who already had experienced heart attacks before the study started--those most prone to heart failure, the condition both Actos and Avandia promote. Among those who had not had heart attacks before the study, the risk of heart disease dropped with tight control.

Two more factors associated with increased risk of negative outcomes with tight control in ACCORD was age over 65 and an A1c at the start of the study that was greater than 8%.

In contrast ACCORD found improvements with tight control for every subgroup of participants.

So now you know why "tight control" CAN be dangerous: It's dangerous when it's achieved using drugs we know can kill people, especially people prone to heart failure, i.e. Avandia and Actos. If you aren't taking these drugs, relax. Lower your blood sugar as much as you want using diet, metformin, gliclazide, and insulin and be secure in the knowledge that you are not going to worsen your risk of having a heart attack.

If your doctor tells you "tight control is dangerous" ask him if he realizes that Avandia was given to over 91% of the subjects in the study that "proved" tight control dangerous.

The other hazard often cited when tight control is discussed is the possibility of serious hypos. So it's worth noting that ADVANCE put people who needed insulin largely on NPH, the notoriously hypo-causing insulin. Even so, during the five years of the study, the researchers found far fewer hypos than they were expecting based on UKPDS. Had they used the more easily controlled newer long acting insulins, they might have seen even fewer hyps than they did.

The breakdown of insulins used wasn't published in the ACCORD study.

We'll return to what ELSE these studies found in future blog posts.



renegadediabetic said...

Big pharma isn't going to admit its drugs kill people until too many people have died. This is just another example of apalling sloppy science in medical research. They get fixated on one study, misinterpret the results, and ignore the big picture. The same thing holds true for heart disease, cholesterol, and statins. Just shows the bottom line that $$$$$ is the prime consideration, not health.

Jenny said...

I've been asked why do I include Actos here as a dangerous drug when the people in ACCORD were mostly taking Avandia?

Simple answer: Because there's a bunch of research evidence showing that Actos also raises the incidence of heart failure in those who take it.

The only reason, IMHO, Actos looks "better" than Avandia is that it was rarely prescribed until last year so we don't have the 10,000s of people taking it over 5 year to look at to see what it really does.

Now that everyone's been switched to Actos, it will take 7 more years for it's physical cost to be assessed.

You can read about the many studies documenting the very serious problems that are associated with Actos including a doubling of heart failure risk HERE

Anonymous said...

Jenny it is disappointing you decided not to post my comments in full. actos has been on the market for a decade, was the #1 prescribed diabetes drug in 2005 and has over 10 million prescriptions to date...hardly the small patient cohorts you are claiming. You are claiming heart failure as reasons why ACCORD failed while the ACCORD study had no difference in heart failure between groups/ You claim patients with previous events are worse off on actos yet never mention a published study (in diabetes care)showing actos patients with heart failure had better outcomes than those on other diabetic medications; even in high risk, previous MI patients. Please help me understand your position.

Anonymous said...

thanks Jenny for posting. What are your thoughts on this study? Here is the link again:
Also what is the likely fall out from the Januvia UCLA study. My doctor has said it is creating quite a stir.

Jenny said...

I think this paragraph from that study says it all:
Reversibility of serious heart failure

Only 34 patients (out of 149 [22.8%]) in the pioglitazone group and 17 (out of 108 [15.7%]) in the placebo group had a serious heart failure event that resulted in permanent discontinuation of the study medication (P = 0.1602). The number of patients with serious heart failure for whom the heart failure event resolved during follow-up was 116 (77.9%) for pioglitazone and 80 (74.1%) for placebo (P = 0.4822).
Removing the spin, this says that a much higher percentage of Actos patients developed serious heart failure and though very slightly more of them reversed it going off the drug, one third didn't.

And this was only a 3 year study, so the death toll doesn't mean that much because people with heart failure typically drag on for a while. The five or ten year mortality is likely to show excess deaths in the Actos group.

Remember, Actos has been shown to cause heart failure in young people with no signs of it before taking the drug.

I don't have time to comb over the statistics, but if your removed the excess people who developed heart failure from Actos and survived, and then compared the two groups, the death rate in the Actos might look less like an improvement.

Trinkwasser said...

Astonishing! (not it's not)

My take on TZDs used to be that they were good drugs for those not in the category for which they should not be prescribed anyway, now I'm starting to change my mind.

Seen this?

not content with publishing crap papers in prestigious journals, some of them are now inventing their *own* journals

Venkat said...


I know this is off topic. I am a Type II since 1999. I take Low car (<40g net carbs) for the past 1 year and reaping benefits. Though all parameters are text book numbers, my HDL is either 39 or 40. it does not increase beyond that. Can you suggest what I can do to get the HDL in 50 ranges?

Last week, I have started taking vitamin B12, D and Omega supplements daily. I consume 2 eggs per day as well.



Jenny said...


I'm not an expert on lipids, but Dr. Davis, the cardiologist over at the heart scan blog is always recommending large doses of fish oil, along with Vitamin D and sometimes niacin. NOT the no-flush kind, which he says doesn't work.

Trinkwasser said...

You could also try adding more saturated fats (The Horror!) and exercise. Without the carbs and high insulin required to deal with them I convert sat fats into HDL rather than LDL. A lot of people do this *but not everyone* so check your results