Preface: When I saw my endo last week, she praised me on my latest A1c level (6.2, Baby!), but said that my standard deviation score was “unacceptable.” Yikes! What does this mean for my health going forward? …
You’ve all heard of standard deviation, correct? It’s that mathematical measurement that became popular a few years ago as a “back up” for showing how well a diabetes patient was doing with their glucose control.
Sure, you can have a good A1c, but we all know that a low A1c often comes with a lot of lows, and a lot of lows can come with rebound highs. So, having a low A1c may not the perfect measure of good D-health. The theory with using standard deviation in diabetes went so far as to assert that it could actually help predict how likely you were to develop future complications.
But now it turns out that our trust in standard deviation might be unfounded — while our trust in A1c remains inflated (ugh).
For those who struggled in statistics class, standard deviation, mathematically speaking, shows how much variation there is from the average. In diabetes, a low standard deviation indicates that blood sugar tends to be very close to the ideal mid-range most of the time. A high standard deviation, on the other hand, shows that a patient has a wide swing of blood sugar values, regardless of their A1c result.
The theory went that low standard deviation, i.e. less blood sugar swings, would be a good indications that a person’s risk of complications is lower than someone with a high standard deviation, even if their A1c was higher than recommended (above 7%). This is supposedly so because the wild swings in blood sugar levels supposedly take a toll on a body — possibly even more so than having slightly elevated BG levels that remain steady.
“The fundamental question is does variability give you an independent risk for getting diabetes complications? It’s the 500 pound gorilla and we just don’t know,” says Dr. Irl Hirsch, endocrinologist at University of Washington and a lifetime type 1 himself, who’s been one of the most famous proponents of applying standard deviation to diabetes care. What’s that? We don’t know? Not enough evidence here?!
According to an April 2008 Diabetes article, analysis of the participants in DCCT found that A1c was only responsible for 11% of the overall risk for developing microvascular complications. That means 89% of our risk for complications is coming from somewhere else. But whether that’s standard deviation, environment, genetics, or something else entirely, no one knows.
In a recent Journal of the American Medical Association article entitled, “Beyond HbA1c: Need for Additional Markers of Risk for Diabetic Microvascular Complications,” co-authored by Dr. Hirsch and Dr. Michael Brownlee of the Albert Einstein College of Medicine, the authors state that it is “crucial” to find out what causes the remaining 89% of microvascular risk. They write, “Physicians will have to realize that much remains to be done in identifying important factors contributing to microvascular complications risk, which are not captured by the HbA1c.”
And it’s still unclear whether or not standard deviation actually plays a role in complications risk. Dr. Eric Kilpatrick, a doctor in the UK who contributed to a July 2006 Diabetes article, says, “We looked at the DCCT database and found that within-day glucose variability did not seem to contribute to the risk and therefore did not seem to help make up the remaining 89%.” He adds that their findings were based at “face-value” of the DCCT study, and that “this is still speculation.” Oy vey.
So the A1c may only contribute 11% of our overall risk of complications? Does that mean we should stop paying so much attention to it? Not exactly, no. It seems that patients whose A1c levels are all over the map may be at highest risk for complications. This notion is supported by a July 2008 Diabetes article that states, “The DCCT data has shown that in patients with type 1 diabetes, increasing variability in A1C adds to the risk of microvascular complications over and above that predicted by the mean A1C value alone.” So variation in A1c levels over the years could be a real risk factor.
Confused yet? Yeah me too. The researchers from that last study cited admit that their findings seem to conflict with other results indicating that day-to-day blood sugar variations do not play a huge role in complications.
At the end of the day, what we learned from our conversations with Dr. Hirsch and Dr. Kilpatrick is that researchers really just don’t know what role standard deviation plays in the risk for complications because this hasn’t been studied as intensely as needed. Luckily, Dr. Hirsch was recently approved for a grant to study just this: the effects of standard deviation in 12 clinics around the United States starting in 2011. So we have some clarity to look forward to in the New Year, and will certainly keep you all (and my endo!) posted as results unfold.
Thank you for this post! It’s been on my mind lately…this standard deviation stuff. I asked my endo about it a few years ago and he said it is “irrelevant”. It makes sense to me that a lot of up and downs create havoc on the body. In fact, I think I live with some of the aftermath of swinging blood sugars. Now I try to keep my standard deviation really low. And my A1c. When will we catch a break? Thanks Amy
I’m sorry, I just don’t buy into this 11% finding, not at all.
We see it every day, people with high glucose are the ones having complications. The graphs from the UKPDS and DCCT show that what kind of diabetes you have does not matter for complications. High glucose does the damage.
I shall continue to keep my glucose in the middle of the non-diabetic range (4.9 to 5.4) and SD at 14.
-Lloyd
[...] This post was mentioned on Twitter by diabetesblogs. diabetesblogs said: The Mixed-Up Role of Standard Deviation in Diabetes Care http://dd4.us/fCk6PY #diabetes #dblog [...]
My endo definitely knows about standard deviation.
I thought that standard deviation *was* known to factor into eye complications, though.
Thanks for posting about SD! Intuitively I’ve known this for a long time–I feel my best when my SD is low. But I tried to ask my kind, thoughtful, old-school endo a question about my spiky management back in the early 90s and I still remember his answer: “Don’t worry, Laura. The body only sees the A1C.” But it makes sense to me that endos like mine and Sysy’s have labeled SD “irrelevant” up till now, because they’ve had no way to measure it!
Seeing constant illustrations of my variability and grading it weekly or monthly has been the single best thing about having a CGM, and it’s what has helped me tweak my management and improve my daily health the most. What you can see, you can work on. Time for a big study on variability and complications using thousands of CGM patients’ data!
Probably the reason standard deviation is inconclusive, is there there are different ways of measuring it. You have your post-low spikes (thud-boing!) and your overguestimated carbs (whoosh-splat) types of spikes, versus those nice little gentle waves I can rarely manage to achieve.
So two people (or two days with the same person) may both have values that vary from 50 to 250, only in one of them you hit both within half an hour while with the other maybe they are low at 5am and high at 7pm and gradually drift in between.
A real study of standard dev would need to fit all participants with CGM and calculate not only the stdev, and time spent out of ideal range, but also the rate of change on an hourly basis…Makes my head hurt, and I aced statistics in college
In terms of quality-of-life, we as patients know that even if standard deviation has a relatively small influence on microvascular risk, it still spells major nuisance for people with diabetes, and the bigger question for people with type 1 diabetes (the standard deviation is, by definition, according to such endocrinologists as Dr. Zachary Bloomgarden, MD, almost always significantly higher in type 1 than it is with type 2) is how best to minimize it? There are many thoughts, for example, adding Symlin to the treatment protocol, but many patients find that can be more trouble than it’s worth and getting insurance coverage for it remains a challenge for some patients. The bottom line is that the doctor must work closely with the patient to decide how best to reduce standard deviation, and that may require more than the 15 minutes that the current U.S. payment system for healthcare reimbursement practically dictates. There is not a one-size-fits-all solution, but doing so may be worth the doctor’s investment!
Point of clarification on what “low standard deviation” is–a given data set can have one no matter what the numbers are, as long as they don’t vary between too high and low a number. Someone who’s *always* at 200-225 would have a yucky A1C but still have a low standard deviation.
As a kid growing up with diabetes it was fairly apparent that standard deviation never shows its face in A1C. It’s fairly common to have a “bad” standard deviation but an acceptable A1C and in some ways, fake a decent A1C.
If you keep your A1C the same but reduce your standard deviation; you are actually improving your control. And there is no tool on the market that shows you that.
Well, at least your endo is looking at standard deviation. But what is “acceptable” for that statistic anyway? I think someone once told me that it was about 1/3 of the mean bg, but I don’t know where I heard it or whether that rule of thumb is accurate. Anybody have ideas on this??
Seems like you need a good idea of the overall daily trend of glucose data coming from an accurate CGM system to get an accurate calculation of standard deviation. Thus, most assumptions previously made about calculating standard deviation based on several discrete daily points from self-monitoring have so much inherent error in them that the concept of calculating percentage breakdown between A1C and standard deviation as causes of complications is questionable at best and possibly misleading at worst. The real question of course is if you have at this time a sufficiently large patient population satisfied in the quality of their CGM measurements that a summary meta study from this data is actually worthwhile.
I am a type 1 Diabetic and I am also what is called a Master Black Belt for Six Sigma, a fancy way of measuring processes and using statistics to prove systemic changes.
The simple truth is that only 11% of the test results of the A1C correlated to complications. I have had a great A1C and I am a cyclist who exercises, yet I have had the worse year ever with all kinds of complications. I am underweight and not because of highs but because soo much has gone wrong, I have had a hard time with digestion. I do not have celiac nor any other contributing factors, except some pretty high variations with BS due to overeating and keeping sugars as tight as possible to achieve A1C of 6.0 I even had an A1C of 5.9 at one point.
Most people say I look like the epitome of health and all I have to say is Thks, but if you only knew the year I had. I felt like I had been through chemo and nauseated all the time. it has taken me since May to keep food down. I finally have been able to eat almost normally in the last 3 weeks, although I still get hit every now and then. I am still losing some weight, but it has definitely slowed down. I have had 5 specialists looking into my situation and test after test. While they continue to look I am convinced I need to have better control and even if it means a higher A1C.
Well, at least your endo is looking at standard deviation. But what is “acceptable” for that statistic anyway? I think someone once told me that it was about 1/3 of the mean bg, but I don’t know where I heard it or whether that rule of thumb is accurate. Anybody have ideas on this??
_____________________________________________________________
A type 2 on a pump can do better than 1/3 the mean bg, I do better than 1/6 the mean bg, though it took a few months to get that good at adjusting a pump.
-Lloyd
The simple truth is that only 11% of the test results of the A1C correlated to complications.
_________________________________________________________
Until this is confirmed by pier review, and confirmed my other studies, this is not so. That is the scientific method.
-Lloyd
Great article Amy, but certainly humbling and troubling to hear that microvascular complications are only roughly correlated with a1c’s!
I too am one of those patients who did NOT have horrible a1c’s, except for a brief period pre-pump way back in college. I too suffered terribly with gastroparesis and slow digestion to the point of malnutrition and feeding tubes. One thing I can definitely say though is don’t give up. After a few years of better blood sugars, my digestion seems almost normal. To me, that is proof positive that good numbers improve nerve-system complications even if they are not as strongly correlated with microvascular complications.
It seems I have autonomic neuropathy as manifested through digestive woes; I have however no kidney damage and no eye damage. It does seem that something else plays a role in damage.
My endo said the score was irrelevant too…huh now you have me thinking I gotta look further into this….thanks amy
Well all I can say is after Getting D – Adult Onset at age 45 and it being in my Whole Family ( Mother & Siblings)?
-I strove to get as low a BG’s as I could,striving for 80′s ave.
-Had alot of hypo’s , ave 2–3 per day
-had alot of Rebounds ( ave 200′s) too
-But had 6% A1c’s !
-Also got Retinopathy – 3 Eye Hemmorages over 2 yrs..
- I thne stopped being so Aggressive and set the Bar Higher.. 100′s
-I still Ave 6 -6.5% A1c’s and have alot less Lows and Highs and been 3 yrs and no new problems (yet)
And yes, I think at least 50% of these ‘Complications” are out of our Control, We have a Auto Immune Disease the wants to Destroy every cell and organ in our body and we don’t have the Drugs to stop that process yet..
Btwn Inaccurate Test meters, CGMS and Unstable Insulin Pumps and Not so Fast insulins, we lack the Right tools to do much better..
Can have 100 BG’s 10x for the day and just 2 , 150-200 can throw the 12x ave, 24 hrs ave out of wack
ave 120′s 8x and have 2, 175′s and your 130′s
It’s so Frustrating getting a 170 after 2 hrs, after having such a “perfect day” going, isn’t it?
Btwn Have to Know Exactly the Right Tot Carbs, Have to have 100 B4 eating and have to Kow the right Correction Bolus for 1,2 and 3 hrs to take if High..
It just gets overwhelming and seeing as what’s the % of T1′s that are lucky enough to Get a Pump? 25%? what about the other 75%, they’re SOL, right?
And oyu wonder why These Endo’s Are so Laid Back and don’t want us to Be so Aggressive?
They know, Once you ave 7% or less A1c’s, the rest is upto That other 50-75% that we have no control over..
And it’s In Medical Science and God’s hands..
I still think, Getting Stem Cell Injections is the answer, even getting them once a Month is better than Nothing..
If it cures our D? at least gives us Level 80-100 bg’s all the time? Count me in.. At least I won’t have to Inject myself 8-10x a day anymore and figur out all that complicated stuff and Fustrations..