The Bell Curve of Diabetes Treatment

Patient Expert

Your doctor has just read a research study that tested 10,000 people with type 2 diabetes and found that the new drug Flabaway was very effective. She prescribes Flabaway, and it doesn't do a thing for you.

What's going on here?

A nonscientific explanation is YMMV (your mileage may vary). What works for one person won't work for another.

A more technical term is "normal distribution," or "the bell curve." If you want to understand the mathematics behind it, see here . Otherwise, you just need a general understanding of what's involved.

When you study the effect of anything on a large population, you can calculate the average effect. But you can never predict how that factor will affect any one individual.

The average effect is what will happen with most people. For example, let's say that the average effect of 1500 mg of metformin is to reduce the hemoglobin A1c by 1 point. Most people taking 1500 mg of metformin will see their A1c go down by about 1 point.

But not everyone will see that effect. Some people might see their A1c go down by 3 points, and some people might see their A1c go down by zero points, or even go up.The farther from the average the results are, the fewer people will see this effect. Hence the curve has the approximate shape of a bell.

The late biologist and author Stephen Jay Gould once wrote a famous essay discussing the bell curve. He was diagnosed with a rare cancer. His doctor wouldn't give him a prognosis, so he went to the library and looked it up. He was stunned to find that the average life expectancy with this type of cancer was only 8 months.

But then he remembered the bell curve. Just because the average life expectancy was short didn't mean his life expectancy had to be short. Some people lived a long time with that cancer, and he was determined to be among them. In fact, he was. He lived for another 20 years.

You can see an illustration of individual variation here . Look at Figures 1-3. The pink or red lines show the mean for the group, and the conclusions are mostly based on the mean. But it's clear that different individuals had results that were different from the mean.

In terms of total cholesterol, the authors wrote, "There were no changes in total cholesterol observed in either group" (those eating eggs and the controls). And there weren't, on average. But you can see from Figure 1 that some people saw their total cholesterol increase, some saw it decrease, and some saw little effect.

A recent article illustrateds how type 2 diabetes variability can arise.

We've known for some times that type 2 is a multigenic disease, meaning there's not just one gene that causes it. Scientists recently found 13 genes that affect blood glucose regulation, insulin resistance, and beta cell function. This is in addition to the genes that have previously been found.

When they're mutated, five of those genes seem to increase the risk of type 2. Most of these genes affect beta cell function; only one affects insulin resistance (IR).

It's generally accepted that in order to get type 2 diabetes, you need to have both IR and defective beta cells. But each of us may have a slightly different combination of beta cell deficiency and IR caused by modifications of different genes or slightly different modifications of the same genes.

And even this research, which began by looking at 2.5 million genetic variants in populations of European descent, explains only about 10% of the genetic contribution to fasting glucose. More genes contributing to type 2 will be found in the future.

So we can have different permutations and combinations of diabetic genes and variations in how we respond to different treatments.

Doctors have to prescribe according to the greatest probability of success. So if some drug (or supplement or diet) works for 80% of people taking it, that drug should be the first one the doctor tries. But it may not work for you.

Conversely, some treatment that doesn't work for most people might work wonderfully for you.

We don't yet have the means to determine ahead of time which drugs or other treatments will be best for each individual patient. It's mostly a question of trial and error. But we need to understand that if something doesn't work, we haven't failed. The treatment has.

If some treatment doesn't work for us, we have to insist on getting another one and not let the health care people tell us to keep trying it because some study "proved" that worked.

YMMV. Like Stephen Jay Gould, we might be at the far edges of the bell curve.