Misdiagnosis by Design: The Story of the ADA Diagnostic Criteria
This is the little known story of how--and why--the American Diabetes Association keeps doctors from diagnosing Type 2 diabetes early.
If you wait for your doctor to give you a diabetes diagnosis, the chances are good that by the time you are diagnosed you'll already have one or more serious diabetic complications. These include retinal damage, nerve damage, and early kidney damage. It is now known that these diabetic complications only develop after years of chronic exposure to high blood sugars. But, tragically, the way that today's doctors are forced to diagnose diabetes ensures that you will get no warning that you are experiencing those chronically high blood sugars until they have reached a level so high they have already done irreversible damage.
This is not an accident. Years ago a committee of medical experts whose task was to decide how diabetes should be diagnosed decided it was better to avoid diagnosing patients with diabetes than to give them early warning that they were suffering from elevated blood sugars. As a result, these medical experts intentionally set the standards for diagnosing diabetes artificially high, so that most patients do not get diagnosed until their blood sugar has reached a level where they may soon develop the diabetic eye disease that leads to blindness.
Their reasons for doing this this made sense in the late 1970s when these diagnostic criteria were originally crafted. At that time there was no treatment that could help people with early diabetes, while delivering a diabetes diagnosis could make it impossible for their patients to get health or life insurance.
These circumstances led the experts to conclude that an early diagnosis of diabetes was more likely to harm than help their patients. So they defined diagnostic criteria that wouldn't diagnose patients with diabetes until late in the disease process, when their blood sugars were bad enough that they could safely use insulin and the early insulin stimulating drugs. These were very unsafe to use in the late 1970s unless patients had extremely high blood sugars, since patients had no access to real time blood sugar testing, as there were not yet any blood sugar meters available to patients for home use.
But though medical treatment of early diabetes has changed greatly since the late 1970s, the criteria used to make a diabetes diagnosis have not. In fact, the current criteria endorsed by the powerful American Diabetes Association make it harder, not easier for people to get an early diagnosis of diabetes. In fact, recent mainstream medical research shows that the diagnostic test the ADA promotes misses the diabetes diagnosis in up to 72% of older women and 48% of older men who would be diagnosed using a different test that is still commonly used in Europe. The ADA's recommended test also misses the diagnosis of diabetes in another vulnerable group, women of color.
To understand how this gross miscarriage of medical practice has been allowed to occur, you will need to understand, first, why Type 2 Diabetes is such a slippery syndrome to diagnose. Then you'll see what the process was where the 1978 criteria for diagnosing of diabetes were set unhealthily high and how, since then, institutional inertia has preserved this 40 year-old, out-of-date standard despite mounting evidence that much diagnosing patients when their blood sugars are at much lower, but still abnormal, levels can prevent patients from progressing to blood sugar levels that cause diabetes-related nerve damage, blindness, amputation, kidney failure and heart attack death.
Diagnosing Type 2 Diabetes is a Challenge
Type 2 Diabetes is not like Other Diseases
Diagnosing most common diseases is a straightforward process. Many diseases present with obvious symptoms like fever, cough, swelling, or persistent pain. If you show up with one of these symptoms, the doctor runs blood tests that look for specific abnormalities--like a high white blood cell count--that suggest the presence of an infection.
If blood tests don't pinpoint the problem, sophisticated imaging techniques can peer inside the body's organs and to look for characteristic patterns of damage. If imaging doesn't give a diagnosis, surgeons can take a tissue samples for pathologists to view under a microscope. Once the proper test is run, the diagnosis is clear. If the pathologist finds cancer cells in a suspicious lump, the patient has cancer. If the blood test shows antibodies to the virus that causes AIDS, the patient has HIV. If the EKG shows patterns characteristic of damaged heart muscle, the patient has suffered a heart attack.
But diabetes is different. Diabetes is only easy to diagnose looking backwards--after many years of elevated blood sugars have caused classic diabetic complications like amputation, blindness or kidney failure. People in the early stages of diabetes may have very high blood sugars for many hours of the day but still have normal blood sugars when they wake up or before they eat their dinners. More importantly, a person can walk around with blood sugars three or four times higher than normal and feel nothing. Even when your blood sugars are rising and falling by 150 mg/dl (8.3 mmol/L) or more within a few hours your only symptom may be increased hunger, weight gain, or being tired all the time, symptoms few doctors connect with blood sugar status. The rest of the symptoms that are actually caused by hours a day of exposure to elevated blood sugars, like nerve pain in the feet, recurrent urinary tract and yeast infections, impotence, and transiently blurred vision, can be easily attributed to other causes.
There's No Clear-Cut Blood Sugar Level at Which Type 2 Diabetes Begins
Defining what, exactly is a diabetic blood sugar is difficult, too, because if you randomly test hundreds of people in a general population, an expensive proposition, you have to decide when to test their blood sugars. Fasting blood sugars vary differently from post-meal sugars, and testing at different times after a meal gives you very different readings, too.
Even if you just test one measurement, like fasting blood sugar, you will find that blood sugar levels in that population do not clearly divide into normal and abnormal sugars with an abrupt step between the two. So when you chart all the fasting blood sugar test results of your population, you end up with a gentle curve that rises gradually from the 80 mg/dl (4.4 mmol/L) range where almost no people show signs of having any of the classic "diabetic" complications caused by elevated blood sugars up to the levels of 300 mg/dl (16.7 mmol/L) and more where most people will display some signs of blood sugar-related organ damage.
While this makes it easy to label people whose fasting blood sugar lies in the 80s as "normal" and those whose blood sugar is 300 mg/dl as "diabetic at least half of your population's blood sugar will fall into the intermediate range where they may or may not have mild symptoms that may or may not be related to their elevated blood sugars.
Doctors used to label the upper half of that intermediate range of blood sugars "pre-diabetes" until research in the 1970s showed that some of the people who had been diagnosed with "pre-diabetes" did not go on to develop full-fledged diabetes. This made the pre-diabetes diagnosis suspect. The term went out of use in the 1990s, though doctors now are using it again.
You Can't Biopsy a Working Pancreas
If blood testing cannot give a definitive diagnosis, you might think that doctors could use advanced imaging techniques to look for abnormalities in the pancreas. After all, we know that blood sugar control is largely a function of the pancreas's insulin-producing beta cells. You might also assume that a biopsy of the pancreas might show failing beta cells long before enough of them had died to cause the patient to become diabetic. But here you run into problems that even today's sophisticated research methods have not yet solved.
The beta cells form only a very small part of the whole pancreas and even when a large proportion of them have shriveled and died, the damage cannot be seen on an MRI. Nor can you take a biopsy of the pancreas without risking severe damage to it. That's because besides producing insulin, the pancreas' other main function is to make digestive enzymes. So if you cut into a pancreas to take a tissue sample for a biopsy, these digestive enzymes spill out and do what they do best--they digest any tissue they encounter even if that tissue happens to be that of the pancreas itself.
The difficulty of studying the pancreas does not go away when the patient dies, either, which partially explains why the underlying disease process that takes place in patients with Type 2 Diabetes is still not well understood. After death those pancreatic enzymes immediately start to digest the pancreas. So unless the autopsy is conducted within an hour or two of death, it may be too late. Though there are a few researchers who have the skill to do pancreatic autopsies, there have been only a very few studied that have incorporated the findings of their research.
All this should explain why it was tough years ago for experts to agree on what actually constitutes a blood sugar level that should be labeled and treated as "Type 2 diabetes."
How the Experts Made Sure You Will Not Get an Early Diabetes Diagnosis
A Committee Defines "Diabetes" and Sets the Defining Blood Sugar Level Intentionally High
Toward the end of the 1970s, researchers had not yet come up with clear-cut medical evidence that any particular blood sugar level marked the level at which the damage characteristic of Type 2 Diabetes began. (NOTE: More recent research has very clearly determined these levels, which are discussed in detail elsewhere on this site. In the 1970s doctors could chose from any of 17 different diagnostic standards to sort out their normal patients from the diabetics. Few of these differing criteria rendered the same diagnosis for the same patient. In fact, a 1975 study of 340 people who were each evaluated using each of six different widely-used methods of diagnosing diabetes found that less than 48% the people in the study were classified the same way by any two of the six methods.(1)
This meant that whether or not you were diagnosed as being diabetic depended entirely on which criteria your doctor chose to use. If your doctor used one standard you might be diagnosed as diabetic with a 2 hour glucose Tolerance Test result of 161 mg/dl--and denied health insurance as a result. Another patient with the identical blood sugar whose doctor used a different cutoff for his glucose tolerance test--or administered a different amount of glucose when running the test would be told they were "normal"--and retain access to health insurance. To further confuse the matter, epidemiologists, people who study disease patterns in large populations, could find no no specific blood sugar level where the incidence of diabetic complications suddenly shot up.
Eventually it became clear that a definitive definition of diabetes was required, so in April of 1978, the National Institutes of Health (NIH) convened a committee to establish one common definition. The committee was called The National Diabetes Data Group (NDDG) and its members were drawn from the ranks of prominent health professionals in the English speaking world. Though it was touted as being an "international workgroup," fifteen of its nineteen members were from the United States, and six of those were government medical experts employed by NIH or the Centers for Disease Control. Of the committee's four international committee members, three were from the UK: two doctors came from Guy's Hospital Medical School in London and one from The Royal Infirmary in Edinburgh. One lone doctor represented the rest of the world, a physician from Steno Memorial Hospital in Copenhagen.
The task the committee faced was a daunting one. The cutoffs for diagnosing diabetes found in the most commonly used diagnostic standards ranged all over the place. Some diagnosed patients as diabetic when their oral Glucose Tolerance Test results were as low as 160 mg/dl (8.8 mmol/L) while others diagnosed only those whose blood sugars were higher than 240 mg/dl (13.3 mmol/L). The Fasting Plasma Glucose values these criteria identified as diabetic ranged from 110 mg/dl (6.1 mmol/L) to 130 mg/dl (7.2 mmol/L).
The reason for this variation was that when doctors started testing their patients with newly-invented blood sugar meters they discovered that some patients showed early signs of diabetic retinal damage or early diabetic kidney disease when their blood sugar levels were as low as 160 mg/dl (8.9 mmol/L) two hours after the start of a Glucose Tolerance Test while others with the same blood sugar readings did not.
Though long-term epidemiological studies of large populations showed that over time a hefty percentage of the symptom-free patients whose blood sugar had tested near 160 mg/dl (8.2 mmol/L) on a Glucose Tolerance Test or whose blood sugar was over 110 mg/dl (6.1 mmol/L) on the Fasting Plasma Glucose test did go on to develop the classic diabetic complications, some remained symptom-free. It was these "borderline" patients whose fate troubled the experts, since a diabetes diagnosis could have so many negative repercussions..
The Problem of the Diabetes "Stigma"
Remember, back in 1978 there was no drug that could prevent a pre-diabetic patient from becoming fully diabetic. Oral pills caused dangerous hypos in patients who couldn't measure their blood sugar in real time. Furthermore, the belief that dominated at the time, the eating fat caused heart disease, made it impossible for doctors to advise people with diabetes to eat the low carbohydrate diets that had been the standard way of treating Type 2 diabetes until the 1950s.
So without an available treatment, an early diagnosis was of little use to the patient.The diagnosis could hurt the patient worse than the disease itself, not only were patients diagnosed with diabetes routinely denied health and life insurance, they might also be denied employment or even the right to have a driver's license. There was no Americans with Disabilities Act yet. What was worse, this "stigma" applied whether the diagnosis was diabetes or "pre-diabetes." Any diagnosis with the "D-word" could trigger it.
Weighing these two points, the members of the NDDG committee sensibly decided to set the diagnostic criteria for diabetes as high as possible. They would intentionally avoid labeling patients with the damaging "D-word" until their condition had deteriorated so badly that they could benefit from the few anti-diabetic drugs at their disposal.
The NDDG committee concluded that it was better to leave ten people undiagnosed (and unstigmatized) than to misdiagnose one normal person as being diabetic. So the committee decided to raise the cutoff for diagnosing diabetes to a level higher than it had ever been set by any of the diagnostic criteria used during the previous decade. By setting it that high they would, they hoped, assure that none of the people they diagnosed as diabetic would ever again show a normal result on a subsequent blood sugar test.
That they did this on purpose is not a guess. This line of reasoning is set out very clearly in the NDDG committee's final published report: National Diabetes Data Group. Classification and Diagnosis of Diabetes Mellitus and Other Categories of Glucose Intolerance. Diabetes, Vol 23,(10),1979. P.1039-1057. This report is not available online and, because it is so old, it is difficult to find in most libraries. You can view a scan of a relevant page of the printed report that discuss how the NDDG committee set the criteria by clicking HERE.
Data from a Highly Atypical Population Gives the Committee a Rationale for the New, Too-High, Criteria
Once the committee decided to set the diagnostic criteria as high as ethically possible, the challenge was to locate the blood sugar level at which a patient could be counted on not to retest as "normal" ever again. Because the studies done in all European and most other populations around the world showed that featureless curve that gave no clue as to where that level might be, the committee turned instead to data that had been collected among a small, long-isolated group of native Americans, the Pima Indians. They chose them because the Pima had the distinction of being one of only two human populations ever studied in which the blood sugar test results of the people in the group, when graphed, did not rise in a smooth curve from normal to abnormal but made a sudden, step-like jump from normal to diabetic.
It could be argued, that since the diabetes that afflicted the Pima behaved differently from that of most other human populations and most likely resulted from a specific genetic mutation that had been advantageous for survival in the Pima's unique ecological situation, statistics drawn from the Pima might not be the best guide for setting the diagnostic standard for the rest of the people of the world whose diabetes did not share the same genetic basis and did not behave in the same way. But what attracted the NDDG to the Pima was that unlike other populations, when you plotted their Glucose Tolerance Test results on a graph, you could clearly see a split between those who had diabetes and those who did not--as long as you defined "diabetes" as meaning "going blind from diabetic retinopathy."
When Pima were given a 2-hour Oral Glucose Tolerance Test using 75 grams of glucose the data broke out into two distinct groups. The Pima who showed no symptoms of diabetic retinopathy had blood sugars that clustered in a range under 200 mg/dl (11.1 mmol/L). The second cluster of Pima had a very high incidence of retinopathy and their blood sugar clustered in a range that began at 240 mg/dl (13.3 mmol/L). Best of all, in view of the committee's determination that people not be misdiagnosed, if a Pima's blood sugar had been higher than 200 mg/dl at the beginning of the 12-year study which discovered this clear-cut breakpoint, the chances of that individual retesting with a lower, nondiabetic blood sugar value was almost nil. Setting the lower bound of the diagnostic cutoff for diabetes at 200 mg/dl two hours after the start of an Oral Glucose Tolerance Test would stigmatize only those who deserved that stigma and were going to be diabetic for life.
This is why, despite the fact that researchers even in 1978 knew that lots of people who were not Pima developed overt diabetic symptoms when their two hour Glucose Tolerance Test results were considerably lower than 200 mg/dl, your doctor, today, will not diagnose you as being diabetic until your blood sugar reaches the level at which a person of Pima ancestry is just about to develop irreversible retinal damage.
The Committee Chooses an Even Higher Fasting Plasma Glucose Criteria Based on No Data at All
Next the committee defined a Fasting Plasma Glucose test level which they determined was "definitely diagnostic of diabetes" setting it at 140 mg/dl (7.8 mmol/L). They did not explain the rationale for this choice though this level, too, was much higher than any level used to diagnose diabetes in the standards then in common use.
The new Fasting Plasma Glucose diagnostic criteria the NDDG chose was considerably higher than the Fasting Plasma Glucose level registered by those diabetic Pima whose data had been used to set the level for the 2-h Glucose Tolerance Test. The diabetic Pima had average Fasting Plasma Glucose readings of only 120 mg/dl (6.7 mmol/L) , not the 140 mg/dl (7.8 mmol/L) that the NDDG arbitrarily chose for their diagnostic criteria.
Most likely the committee did not expect the fasting test to be the primary test for diabetes and expected that doctors would continue to use the Glucose Tolerance Test which had been the standard for diagnosing diabetes for a decade. The very high fasting test value was probably chosen as an afterthought and set at what was known to be an extremely high level because people whose fasting blood sugar was that high already had diabetic complications and could be guaranteed not to revert back to normalcy.
The committee did note in their final report that "fasting values between 125 and 140 mg/dl probably indicate a degree of abnormal glucose tolerance that has not been assessed fully as yet."
But this information did not percolate down to the world's doctors, few of whom ever read the committee's report. Busy doctors only saw the tables of diagnostic values published by the American Diabetes Association and the World Health Organization which made it look like a diagnosis using the Fasting Plasma Glucose level of 140 mg/dl (7.8 mmol/L) was equivalent to a diagnosis made with a 2-h value of 200 mg/dl (11.1 mmol/L) on the much more expensive and time-consuming Glucose Tolerance Test. The result was that once these criteria were published most doctors began to diagnose diabetes only via the use of the cheaper Fasting Plasma Glucose test. This ensured that a decade of diabetics would be diagnosed only when their fasting blood sugar had risen to a level that almost guaranteed that they would have already developed serious complications.
The Committee Gets Rid of "Pre-Diabetes"
After it finished defining the cutoffs for diagnosing diabetes, the committee made one more change that has had immense repercussions on your health even now. It invented new category to take the place of what had been called "borderline diabetes" or "pre-diabetes." To get rid of the "D" word--and its stigma--the committee renamed this not-normal but not-diabetic blood sugar range "Impaired glucose Tolerance" (IGT). The committee also specified that IGT should be diagnosed only after administration of a Glucose Tolerance Test in which two samples were drawn.
While remarking in the final report that IGT "may have prognostic implications and should not be ignored or taken lightly" and noting that people with IGT had "increased susceptibility to atherosclerotic disease" (i.e. heart disease) the committee went out its way to stress that only a few people with IGT would go on to develop diabetes each year and that "many will return to normal glucose tolerance spontaneously."
This redefinition of the old pre-diabetes category not only removed the stigma from the new IGT diagnosis, it also, for all practical purposes, rendered it invisible. The new, vague, euphemistic name for the condition, "impaired glucose tolerance", the assertion that IGT usually did not turn into diabetes, and the requirement that IGT be diagnosed with an expensive and time-consuming test caused most of the world's doctors to henceforth ignore it completely--even though the long-term studies cited in the NDDG's own final report showed that while it was true that the annual rate of conversion from pre-diabetes to full fledged diabetes was only 1 to 5% per year, over an eight to twelve year period from 25.9% to 45% of people with IGT went on to become severely diabetic.
Once the NDDG's new diagnostic scheme taught doctors around the world to ignore IGT, when those hudreds of thousands of people with undiagnosed IGT were eventually diagnosed with diabetes, it usually came as a complete surprise. Worse, when they were finally diagnosed, almost half of them already had one or more serious diabetic complication caused by years of exposure to the high blood sugars experienced during their years of undiagnosed IGT.
Flash Forward 20 Years -- The ADA's Inertia Makes Things Worse
The NDDG committee's chief rationale for making the decision that people with diabetes should be diagnosed as late as possible was that there were no treatments available in the 1970s that could prevent "borderline diabetes" from progressing to the real thing. But twenty years later that was no longer true. New classes of drugs made it easier to control diabetic blood sugars. New types of drugs for high blood pressure had made it possible to slow down the development of heart and kidney disease. Ophthalmologists had learned how to treat early symptoms of diabetic retinal damage with laser surgery that could prevent blindness. More importantly, while the diabetes drugs of the 1970s had not prevented people with borderline diabetes from developing the full-fledged syndrome, there was some evidence that newer drugs which had come into common use in the 1990s could.
Responding to these changes, the American Diabetes Association (ADA) convened an Expert Committee in 1995 whose specific mission was to review the scientific literature published since the NDDG diagnostic standard had been set in 1979 and to "decide if changes to the classification and diagnosis of diabetes were warranted."
The report of the ADA Committee that revised the criteria was published in this report Expert Committee on the Diagnosis and Classification of Diabetes Mellitus: Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. (The version available online is a revision of the original report published in Diabetes Care in 1997).
On the surface, the final recommendations made by this ADA committee seem minor. The press releases that appeared in the medical press made it sound like the ADA experts had merely lowered the cutoff for diagnosing diabetes which would result in more patients being diagnosed. On the surface, this made it sound like more people would be getting diagnosed earlier, allowing them to take advantage of the new drugs and preventative techniques now available.
But closer examination of the ADA's complete report discloses that what the Expert Committee did was something far more radical and that the full impact of their recommendations would be to keep millions of Americans from receiving an early diagnosis--especially those with the very early symptoms of elevated blood sugars like nerve damage, impotence and frequent fungal and bacterial infections.
That is because, though the committee's charge had been to see whether changes in the diagnosis of diabetes were warranted, it gave only a cursory review to the literature that examined what the impact had been of setting the diagnostic cutoff for diabetes as high as possible in 1978. Though it would seem obvious that the Expert Committee should have examined the question of what had happened to the health of the millions of people around the world whose diagnosis had been changed from "diabetic" to "Impaired Glucose Tolerance" by the adoption of the 1979 diagnostic criteria. It did not.
Though the Committee cited a total of 143 research articles in their final published report, only five of those articles examined the prevalence of diabetic-style complications among the people now diagnosed as having Impaired Glucose Tolerance. And though the ADA's Expert Committee cited these five articles in their report's footnotes, they did not discuss the articles' findings in the text of their report. In fact, in their discussion of Impaired Glucose Tolerance and Impaired Fasting Glucose they stated that these conditions were "not clinical entities in their own right but rather risk factors for future diabetes and cardiovascular disease"
There was no comment on the finding in one of the cited studies that the incidence of retinopathy among middle class older white older women with Impaired Glucose Tolerance was 6.4% higher than that of that of normal women in the same community, nor was there any mention of the prevalence of "diabetic" neuropathy among people with Impaired Glucose Tolerance, though this is complication that typically develops much earlier in the diabetic process than does retinopathy and results in erectile dysfunction, unremitting pain in extremities, and, eventually, failure of the nerves that control heart beat and the immune response. This omission is disturbing since research published by neurologists before the final publication of the Committee's revised report showed that this kind of "diabetic" nerve damage was frequently detectable among people whose blood sugars fell into the lower range of the Impaired Glucose Tolerance category.(2)(3)
The ADA's Expert Committee made it clear in the text of their report why they did not devote energy to evaluating whether the old diagnostic standard was still justified: They had already decided, for a reason that had nothing to do with protecting their patients health, to retain the old standard. That reason was this: "an enormous body of clinical and epidemiological data has been collected based on the 2h PG cutpoint of 200 mg/dl"and "it would be very disruptive and add little benefit to alter the well-accepted 2h PG diagnostic level of 200 mg/dl."
In short, because research studies had standardized on this definition of diabetes, it would make it hard to align old research with newer research using a better diagnostic standard. The needs of researchers were deemed far more important than those of patients given very late diagnoses How little real substance there was to this argument was shown when the Committee went on to change the Fasting Plasma Glucose test diagnostic cutoff, which was also heavily used in research.
The Committee Again Intentionally Sets the New Cutoff for the Fasting Plasma Glucose Test Too High
Though the committee claimed they were only adjusting the cutoff used on the Fasting Plasma Glucose test so that it diagnosed the same people who were diagnosed with a two hour Glucose Tolerance Test result of 200 mg/dl (11.1 mmol/L) or higher, this is not, in fact, what they did.
As you'll remember, that 200 mg/dl two hour Glucose Tolerance Test cutoff had been set to match the level at which Pima Indians began to show a steep rise in the incidence of diabetic retinopathy. But, for reasons that are not fully explained in their report, the Expert Committee chose not to use the Fasting Plasma Glucose test data collected in the same Pima studies which had been used to set the original 2-hour Glucose Tolerance Test cutoff. Had they done so, they would have had to set the new Fasting Plasma Glucose cutoff to be used for diagnosing diabetes at 120 mg/dl (7.0 mmol/L) since that is the level that corresponds to the 200 mg/dl 2-h test in the Pima population.
Nor did the committee choose to set the new Fasting Plasma Glucose test cutoff using data collected in a well-crafted large-scale study of the United States population. Data from that study, The National Health and Nutrition Examination Survey (NHANES III), showed that for Americans, the fasting blood glucose level that corresponds to the two hour Glucose Tolerance Test value of 200 mg/dl is 121 mg/dl (6.7 mmol/L).
Instead, the committee decided to set the cutoff for diagnosing diabetes on the Fasting Plasma Glucose test at 126 mg/dl (7.0 mmol/L), basing their choice on the fasting blood sugar results found in a study of 13 Pacific Native populations. The reason they gave for doing this was that this was the highest value found in any published study, anywhere.
In the words of the committee,
We chose to set a cutpoint at the upper end [of all values found in studies]. "This value is slightly higher than most of the estimated cutpoints that would give the same prevalence of diabetes as the criterion of 2h PG, 200 mg/dl. That is, slightly fewer people will be diagnosed with diabetes if the new PFG criterion is used alone than if either the FPG or the OGTT is used and interpreted by the previous WHO and NDDG criteria.
Once again, as had been in the case in 1979, the experts had come down firmly on the side of not diagnosing diabetes and had rejected data drawn from studies of heterogeneous populations around the world in favor of that drawn from isolated and probably genetically unique isolated ethnic group.
The Committee Tells Doctors Not To Use the Glucose Tolerance Test
It could be argued--and later it was argued when the Expert Committee found itself with a need to defend what they had done--that by dropping the Fasting Plasma Glucose cutoff to 126 mg/dl the ADA's Experts had improved the diagnostic capabilities of the Fasting Plasma Glucose test, since setting it at 126 mg/dl diagnosed more people than the 140 mg/dl cutoff set in 1979. But any benefit this change might have conferred on patients was wiped out by the two other major changes the Expert Committee made in the diagnostic criteria.
Because, once it finished changing the Fasting Plasma Glucose test cutoff so that it supposedly matched the 2-h Glucose Tolerance Test cutoff--though the committee admitted it really did not--the ADA's Expert Committee recommended that doctors stop using the 2 hour Glucose Tolerance Test.
The justification the Committee gave for this decision was that a recent study had shown that when the Glucose Tolerance Test was administered to the same people twice in a 26 week period it yielded results that varied by 16.7% while the Fasting Plasma Glucose test's results varied only by 6.4%.
The committee interpreted this to mean that the Glucose Tolerance Test was not reliable, though it had been the gold standard for blood sugar testing for fifty years. Though the committee still stated elsewhere in their report that the Glucose Tolerance Test was "an invaluable tool in research."
The the report makes it clear later on that its real reason for abolishing the use of the Glucose Tolerance test was because it was "more costly and time consuming." In the Brave New World of American health care in the 1990s, cost-cutting trumped patient care.
It gets worse. Not only did the committee announce that the Glucose Tolerance Test was "not recommended for routine use," it also recommended that as a further cost-cutting measure, doctors should not test for diabetes even with the cheaper fasting test in "presumably healthy individuals." The only patients who should be offered screening for high blood sugar were those who were overweight, particularly those over 40.
Too bad for the one out of five people with Type 2 Diabetes who NIH statistics show are notoverweight.
The ADA Replaces Impaired Glucose Tolerance with Impaired Fasting Glucose--though Research Shows They Are Not The Same Syndrome
Having told doctors not to use the Glucose Tolerance Test, the Committee was left with the problem of what to do about "Impaired Glucose Tolerance" the old "pre-diabetes" diagnostic category that, though it had been shunted to the sidelines, still identified patients, some of whom had early symptoms of diabetes and many of whom would go on to develop the full fledged syndrome.
Since Impaired Glucose Tolerance could only be diagnosed using the discarded Glucose Tolerance Test, the committee needed to come up with a new way to designate the people whose blood sugar fell into the very large range that lay between normal and diabetic. So the Expert Committee invented a new category, "impaired fasting glucose" which it defined as being any Fasting Plasma Glucose value that fell between 110 mg/dl and 125 mg/dl (6.1 and 6.9 mmol/L).
The Committee gave no reason for why it had selected the 110 mg/dl cutpoint that defined the bottom of this range, observing that, in fact, like every other criteria they and their fellow experts had introduced,"this choice is somewhat arbitrary."
The World Rejected the ADA Criteria though US Patients Were Stuck with Them
So radical were these recommendations that when the ADA Expert Committee's report was circulated to diabetes professionals around the world, the international diabetes community refused to adopt the ADA committee's recommendation to use only the Fasting Plasma Glucose test to screen for diabetes. No other private or governmental health authority besides the American Diabetes Association endorsed the decision to scrap the Glucose Tolerance Test, and the international health community also rejected the idea that the ADA's the newly designated category "Impaired Fasting Glucose" was equivalent to Impaired Glucose Tolerance.
A Large Body of Research Determined that Impaired Fasting Glucose is not the Same as Impaired Glucose Tolerance
European researchers responded by publishing studies that revealed that the people diagnosed as diabetic by the Fasting Plasma Glucose test were not the same people diagnosed by the Glucose Tolerance Test.
A British study of 26,190 European men and women in 13 different European populations found that the application of the American criteria to European populations induced changes in the prevalence of diabetes ranging from an increase of 13.2% to a reduction of 4.0%. An Italian study concluded,
The sensitivity of F[asting] P[lasma] G[lucose] using the new threshold … appears to be markedly different in ethnically different populations. The category of I[mpaired] F[asting] G[lucose] identifies patients who are different from those affected by I[mpaired] G[lucose] T[olerance], regardless of the population studied, and a clear distinction between the two categories should be maintained.
A German study showed that the Fasting Plasma Glucose test misdiagnosed a full 7% of the German diabetic population. Another British study's title says it all: "Impaired Glucose Tolerance And Fasting Hyperglycaemia Have Different Characteristics." A Japanese study showed that "F[asting] P[lasma] G[lucose] criteria alone would overlook many subjects with 'diabetic type' in Japan." A team of Spanish researchers, who studied the blood sugar profiles of 1,034 randomly sampled residents of Asturias, Spain found that "using fasting glucose only (ADA1997) diagnoses 36.3% of those with diabetes." (4)
Lest it seem that these results only applied to European and Asian populations, an American study team headed by James Meigs of Harvard Medical School found the same effect in the Baltimore Longitudinal Study of Aging. They started with a group of 815 men and women 60% of whom started out as normal, and 40% of whom had IGT and administered blood tests to them every two years for at least ten years. (5)
Over this period 42% of the normal subjects who developed abnormal Fasting Plasma Glucose did not develop abnormal Glucose Tolerance. This led them to conclude that those who displayed abnormal blood sugars on the fasting test alone were suffering from a different disorder than those with abnormal 2 hour Glucose Tolerance Test results.
Not only that, but in the population they studied, "the more common pathway appeared to be the development of abnormal 2hPG levels [on Glucose Tolerance Testing] with normal F[asting] P[lasma] G[lucose] levels." This, of course, means that the majority of the people who became diabetic in this group would not have been identified using the ADA diagnostic criteria until very late in the development of their diabetes--if they were diagnosed at all--since not all of them ever displayed abnormal Fasting Plasma Glucose values.
Meigs' research report spelled out the implications of this finding very clearly. It stated,
We now know that Type 2 Diabetes is preventable when subjects with impaired glycemia [blood sugar] are detected and treated. Detection programs relying solely on elevated F[asting] P[lasma] G[lucose] levels using current diagnostic thresholds may only detect the more uncommon I[mpaired] F[asting] G[lucose] phenotype, and miss a substantial number of subjects at risk for diabetes on the basis of abnormal 2hPG levels.
Research Pinpoints Who is Not Getting Diagnosed: Older Women and Women of Color
But while all the preceding studies came to the inescapable conclusion that a significant percentage of people with diabetes were being ignored by the ADA diagnostic criteria, they did not ask the all important question of who these people might be. It was another group of researchers, one from the Department of Epidemiology and Population Health at the London School of Hygiene lead by Joceline Pomerleau who discovered exactly who it was that the ADA diagnostic criteria were missing. The answer they found was women and lots of them--especially women of color.
When the London School of Hygiene researchers applied the new ADA diagnostic criterion to a large multiracial population they found that though the total percentage of people diagnosed as diabetic did go up slightly (1%) when the researchers applied the ADA standard to diagnose diabetes, the increase was found to be entirely due to the diagnosing of more men. The new ADA standard diagnosed significantly fewer women than did the Glucose Tolerance Test recommended by WHO. As the researchers explained in their report,
Compared with men, women had significantly lower Fasting Plasma Glucose levels despite higher 2h post load glucose levels. The observed differences persisted after adjustments of age and ethnicity as well as BMI.
Not only that, but when they broke out their data by ethnicity, "Fasting glucose was consistently lower in women than in men in each ethnic group."
The researchers suggested that
Physiologic sex differences in metabolic adaptation to fasting could partly explain the differences observed between men and women. It has been consistently reported that women have lower Fasting Plasma Glucose levels than men and that women display a more rapid fall in plasma glucose during prolonged fasting compared with men.
The report concluded by pointing out that not only did the new ADA standard show bias against women in general, that bias was particularly pronounced in the case of women of Afro-Caribbean ancestry who were significantly underdiagnosed by the ADA criterion. The researchers found this particularly troubling since this was a group that other studies had shown to have a higher mortality rate from diabetes than males of the same population.(6)
The findings of Dr. Pomerleau's team were reinforced by another study which examined 5,388 adults in Mauritius who had not been previously diagnosed with diabetes. That study concluded,
In Mauritius, the distribution of impaired glucose metabolism differs by sex. The observation that I[mpaired] F[asting] G[lucose] is more prevalent in men and I[mpaired] G[lucose] T[olerance] more prevalent in women raises important questions about . . .the ability of the current glucose thresholds to equally identify men and women at high-risk of developing diabetes.(7)
The ADA Expert Committee--Under Pressure--Makes a Token Change in the Diagnostic Critiera
As the criticism of the new standard grew in scholarly circles, The Committee of Experts responded in a testy "Follow-up Report on the Diagnosis of Diabetes Mellitus" published in November of 2003. You can read it at
Follow-up Report on the Diagnosis of Diabetes Mellitus The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Diabetes Care Diabetes Care 26:3160-3167, 2003
In this new report the committee strenuously defended their original decisions, while offering only tepid reasoning and no supportive research citations to back their decisions.
Though the Follow-up Report did mention some of the European studies that found that the ADA's recommended FPG test identified a different population as diabetics than the Glucose Tolerance Test recommended by WHO, The Follow-up Report made no reference at all to the issue of gender and ethnic differences so many of the researchers had noted.
The committee explained away the few differences found in research they admitted were real by saying
... the data are too scant to subclassify Type 2 Diabetes formally into two distinct diseases, based on the use of F[asting ]P[lasma ]G[lucose] or 2h P[lasma] G[lucose on oral Glucose Tolerance Testing].
Then the Committee called into question the accuracy of the European data stating,
all these epidemiologic studies are based on a single glucose measurement, while the full criteria for the diagnosis of diabetes require a confirmatory test in asymptomatic subjects. Differences in prevalence by the two different criteria could therefore also result from the large day-to-day variability in the tests.
The only concession the committee made in the Follow-up Report in response to the growing volume of criticism of its new criteria was to lower the upper bound of the "normal" category on the Fasting Plasma Glucose test from 110 to 100 mg/dl. (6.1 to 5.6 mmol/l) However, while the committee did admit that the fasting glucose test number they had previously set to distinguishing normal people from people with slightly abnormal blood sugar was wrong, it did not revisit the far more damaging inadequacy of the far too high fasting test cutoff they had set for diagnosing diabetes.
The impact of this cannot be underestimated. The committee had previously advised the American medical profession that they should no longer administer Glucose Tolerance Tests, labeling them inaccurate and giving insurance companies the go-ahead to refuse to pay for them. Doctors all over America were now telling their patients that if their fasting blood sugar numbers were under 126 mg/dl they were not diabetic though the research made it clear that a significant number of women, especially women color, could have fasting blood sugars well under this cutoff and still have diabetic post-meal blood sugars high enough to cause retinal damage and other diabetic complications.
But though the Expert Committee admitted in the body of its report that
several major studies have now documented the ability to prevent or delay the onset of diabetes in individuals with IGT, only identifiable by definition using an O[ral[ G[lucose] T[olerance] T[est].
it did still did not recommend that doctors use the oral Glucose Tolerance Test to identify these people whose diabetes could be prevented or delayed.
As a result, the summaries of the Follow-up Report which appeared in medical newsletters and in press releases that went to newspapers only alerted doctors to the new, lower diagnostic cutoff for the "normal" category and the only message that got through to America's busy doctors was that the borderline for diagnosing people as having Impaired Fasting Glucose (a category most doctors completely ignored) had been lowered, not that the 126 mg/dl cut off for diagnosing diabetes missed a huge population, many of them female and people of color, whose post-meal blood sugar control was known--even by the ADA's experts--to deteriorate long before their fasting blood sugar did.
In Search of More Cost Savings the ADA Now Recommends the Use of an Even Less Accurate Screening Test--the A1c
Things have not improved since 2003. Indeed, it could be argued they have gotten worse. In response to the need to cut costs, in 2010 the ADA, which had previously described the cheap A1c test as not reliable for diagnosing diabetes, reversed itself and announced that the A1c could now be used as a screening and diagnostic tool.
The problem with this is that while the A1c is a useful test for studying large populations, where only averages are of interest to researchers, the A1c test values individuals receive often do not reflect their true blood sugar performance. The research documenting this is discussed at length HERE and HERE.
The studies showing how inaccurate the A1c test can be when administered to individuals were evaluating A1c tests processed by labs. But once the A1c was adopted as a screening tool, doctors stopped referring their patients to labs for this test. Instead most family practice doctors now use kits they can use to test patients in their offices. This allows them to bill insurers for an additional procedure, increasing the amount they are paid for routine visits. But these kit A1c in-house tests are even less accurate than the lab A1c tests. (Details of the research establishing this can be found HERE.)
The reliance on the A1c as a screening test means that people with anemia, the sickle cell trait, a thallassemia gene, or any other abnormality of the red blood cells, and anyone whose red blood cells live longer or shorter lives than average are likely to get an extremely inaccurate A1c test result which may result in many of them being told they have normal blood sugar when, in fact, their blood sugars are high enough to cause diabetic complications. I have heard of people with post meal blood sugars as high as 300 mg/dl receiving A1c test results in the 4.7% range that is the very low end of normal.
Finally, as if all this wasn't bad enough, doctors have decided, based on research discussed HERE, that the diagnostic cutoff for diabetes is 6.5%, supposedly correlating to an average blood glucose of 140 mg/dl (7.8 mmol/L), but most doctors have been taught they don't need to treat diabetes until the A1c has reached 7.0%, which supposedly correlates to an average blood sugar of 155 mg/dl (8.7 mmol/L). Since these are average blood sugars, people achieving these averages are almost always experiencing post-meal blood sugars well over the 200 mg/dl level that even the Pima-based diagnostic criteria considered diabetic. In fact, there is good evidence, based on heart disease research that any A1c over 6.0% should be treated as diabetic. I have been told by my endocrinologist that 5.7% is often indicative of diabetes, too.
What Does This Mean for You?
If you have any symptoms that suggest you might be diabetic or if you have a family history of diabetes, don't settle for your doctor's assurance that your blood sugar is "normal" if that assurance is based only on the results of a Fasting Plasma Glucose Test or an A1c test. These test may call you normal or only slightly impaired when you are, in fact, experiencing chronic elevated blood sugars high enough to damage your nerves, eyes, and kidneys.
To determine if your blood sugar is abnormal, ideally you would ask your doctor test your blood sugar after a meal or give you a Glucose Tolerance Test. Unfortunately, few doctors will conduct these kinds of tests unless they believe there is a very good reason for them. To provide that reason, you should test your blood sugar at home using the technique described HERE. If the home testing shows you are prediabetic or diabetic, take action no matter what your doctor tells you. You can do a lot with dietary changes alone, using the strategy described HERE. If that isn't enough to return you to normal blood sugars, find a new doctor who will support your quest for better health.
1 Valleron AJ, Eschwege E, Papoz L, Rosselin GE. Agreement and discrepancy in the evaluation of normal and diabetic oral glucose Tolerance Test. Diabetes 14: 585-593, 1975. http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=1140514
2 Below are the five studies of IGT cited in the Expert Committee report:
-- Fujimoto WY et. al. Prevalence of complications among second-generation Japanese-American men with diabetes, impaired glucose tolerance or normal glucose tolerance,. Diabetes 36:730-739, 1987 http://diabetes.diabetesjournals.org/cgi/content/abstract/36/6/730
-- Fuller JH et. al. Coronary-heart disease risk and impaired glucose tolerance: the Whitehall Study. Lancet i:1373-1376, 1980 http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=6104171
-- Klein R, et. al. Visual imairment and retinopahthy in people with normal glucose tolerance, impaired glucose tolerance and newly diagnosed NIDDM. Diabetes Care 14:914-918, 1991. http://care.diabetesjournals.org/cgi/content/abstract/14/10/914
-- McCartney P, et. a. The Bedford Survey: observations on retina and lens of subjects with impaired glucose toerance and in controls with normal glucose tolerance. Diabetes Metab 9:303-306, 1983
-- Beks PJ et. al, Peripheral arterial disease in relationship to glycemic level in elderly Caucasian population: the Hoorn Study. Diabetologia 38:86-96, 1995. http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=7744233
3 The studies linking neuropathy to impaired glucose tolerance are summarized HERE
4 The cited studies challenging the equivalence of Impaired Fasting Glucose with Impaired Glucose Tolerance are as follows:
-- DECODE Study Group, on behalf of the European Diabetes Epidemiology Study Group. Will new diagnostic criteria for diabetes mellitus change phenotype of patients with diabetes? Reanalysis of European epidemiological data.. BMJ. 1998 August 8; 317 (7155): 371-375 http://bmj.bmjjournals.com/cgi/content/full/317/7155/371
-- Edoardo Mannucci, Alessandra De Bellis, Anna Maria Cernigoi, Carla Tortul, Carlo M. Rotella, Mario Velussi. Further Data on the Comparison Between World Health Organization and American Diabetes Association Diagnostic Criteria. Diabetes Care, 22: 1755-1757, 1999 http://care.diabetesjournals.org/cgi/reprint/22/10/1755
-- Kohler C, Temelkova-Kurktschiev T, Schaper F, et al.Prevalence of newly diagnosed Type 2 Diabetes, impaired glucose tolerance and abnormal fasting glucose in a high risk population. Data from the RIAD study using new diagnostic criteria for diabetes. Dtsch Med Wochenschr, Sep 17 1999, 124(37) p1057-61
-- Davies MJ; Raymond NT; Day JL; Hales CN; Burden AC; Impaired glucose tolerance and fasting hyperglycaemia have different characteristics. Diabet Med 2000 Jun;17(6):433-40 https://www.ncbi.nlm.nih.gov/pubmed/10975211
-- Kuzuya T, Nakagawa S, Satoh J, Kanazawa Y, Iwamoto Y, Kobayashi M, Nanjo K, Sasaki A, Seino Y, Ito C, Shima K, Nonaka K, Kadowaki T; Committee of the Japan Diabetes Society on the diagnostic criteria of diabetes mellitus. Report of the Committee on the classification and diagnostic criteria of diabetes mellitus. Diabetes Res Clin Pract. 2002 Jan;55(1):65-85 http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=11755481
--Botas, P, Delgad E, Castano G, Diaz de Grenu C, Prieto J, and DiazCadorniga, FJ. Comparison of the diagnostic criteria for diabetes mellitus, Who-1985, ADA 1997 and WHO-1999 in the adult population of Asturias (Spain). Diabetic Medicine 20 (11) p. 904 2003 http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=14632715
5 James B. Meigs, Denis C. Muller, David M. Nathan, Deirdre R. Blake, and Reubin Andres The Natural History of Progression From Normal Glucose Tolerance to Type 2 Diabetes in the Baltimore Longitudinal Study of Aging. Diabetes, Vol. 52, June 2003 1475-1484 http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=12765960
6 J Pomerleau, PM McKeigue and N Chaturvedi, Relationships of fasting and postload glucose levels to sex and alcohol consumption. Are American Diabetes Association criteria biased against detection of diabetes in women? Diabetes Care, Vol 22, Issue 3 430-433 http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=10097924
7 Williams JW, Zimmet PZ, Shaw JE, de Courten MP, Cameron AJ, Chitson P, Tuomilehto J and Alberti KGMM; Gender differences in the prevalence of impaired fasting glycaemia and impaired glucose tolerance in Mauritius. Does sex matter? Diabetic Medicine 20 (11) 915 2003 http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=14632717