Diagnosing diabetes is a lot like trying to sculpt warm Jell-O.
At the extremes, it's pretty easy to decide if someone has diabetes or not. For example, when I was diagnosed, I was having symptoms (constant thirst and urinating a lot), my random blood glucose (BG) level was over 300 mg/dL (to convert to mmol/L divide by 18) hours after my last meal, and my next-day fasting was 269. The glucose level in my urine was so high that the hospital recalibrated its machine to make sure the result was correct.
Clearly, I was diabetic.
At the other extreme, someone with a fasting BG level of 65 who goes up to 80 after drinking a huge glucose drink and has a hemoglobin A1c level of 4.2 obviously doesn't have diabetes.
In between the extremes, there are a lot of patterns that could or could not be considered to be diabetes.
The official guidelines for diagnosing diabetes are that you should be considered diabetic if
Your fasting BG level is 126 or greater on at least two occasions (less than 100 is considered normal) or
Your BG level is 200 or greater 2 hours after starting an oral glucose tolerance test (OGTT) with 75 grams of glucose (less than 140 is considered normal) or
A random BG level is greater than 200 and you're having symptoms.
Recently, some official diabetes groups are suggesting using the hemoglobin A1c test for diagnosis, with any result of 6.5 or greater confirmed by a second test considered diagnostic.
Some years ago, the diagnostic fasting levels were even higher, as some diabetes experts felt that a diagnosis of diabetes would cause harm because of the stigma against "diabetics" and because insurance companies would refuse to insure them. More recently, people have realized that diabetic complications occur even at BG levels below these diagnostic values, and the earlier people are diagnosed, the greater their chance of preventing complications.
However, regardless of where the cutoff points are set, none of these criteria are perfect. Anyone with diabetes knows that fasting BG levels can vary from day to day, and even testing fasting levels on two different days doesn't ensure that they represent a true value.
The same may be true of the OGTT. We all know that we can eat exactly the same thing on two different days at exactly the same time and get exactly the same amount of exercise, yet one day our postprandial BG levels will be higher than the other. Furthermore, because this test is time consuming, very few physicians use it for diagnosis.
And the A1c test is affected by a lot of things, including red blood cell lifetime, which can be genetic and is also affected by various hemolytic anemias, spleen damage, or major blood loss; and abnormal hemoglobin types. Furthermore, although most labs now claim to have standardized their A1c tests, in practice there's still variation from one lab to the other.
Hence one person might have normal BG levels all day long but have an abnormal A1c result, and another person might have elevated BG levels yet have a low A1c. I know someone who had fasting BG levels above 130 but an A1c in the 4s so her doctor refused to diagnose her until things got much worse.
To further complicate things, there are various different patterns of BG abnormalities. Some people may have normal fasting BG levels but go high after meals (this was formerly called impaired glucose tolerance). Others may have high fasting levels but not go very high after meals (this was formerly called impaired fasting glucose).
Some years ago official diabetes groups decided to merge both groups into a new category called prediabetes even though some people think their risks and outcomes differ.
These aren't the only patterns one can get. Some people may have a little impaired glucose tolerance and a little impaired fasting glucose.
Some may have normal fasting BG levels, go very high after meals, but come down again quickly, so they wouldn't satisfy the official requirement for high BG levels at 2 hours after an OGTT. This and this show the variation in BG levels after a high-carb breakfast in people considered nondiabetic.
But no one knows if such wide variations in BG levels might cause complications. Some people think wide variation is worse than sustained high BG levels. Yet the people shown in the cited graphs are considered nondiabetic. Their A1c levels are in normal ranges.
Other people may have temporary increases in BG levels because of some stress, such as surgery or emotional stress, and then revert to normal BG levels.
Diet can also affect your BG levels. Someone following a low-carb diet for weight loss might have normal fasting and postprandial BG levels and normal A1c levels as long as he or she followed the LC diet. An OGTT would show the underlying diabetic defect, but most doctors don't use that test these days, especially in someone with normal fasting BG levels. And these people would be grouped with the nondiabetics if they were included in any clinical trials.
Just fasting can affect your BG levels. Fasting is the ultimate low-carb diet, and after a long fast you'll test diabetic even if you're not on a standard carbohydrate-containing diet because when you don't need them, your body stops producing carbohydrate-processing enzymes. This is called starvation diabetes.
If you've been on a very low carb diet and you're given an OGTT, you'll probably test diabetic even if you're not, for the same reason.
Some people may have abnormal BG levels because of very high insulin resistance, which can sometimes be reversed with weight loss and exercise. They also have defective beta cells that aren't able to cope with the excess demand. But if they can reduce the insulin resistance, their beta cells can cope. Many overweight couch potatoes don't have diabetes because their beta cell mass simply expands to cover the increased need.
The same situation occurs during pregnancy. In most people, the beta cell mass expands during pregnancy to cover the increased need in late pregnancy. Some people have beta cells that are unable to do this, so they are diagnosed with gestational diabetes. After the baby is born and the demand is lowered, their BG levels revert to normal.
Others may have abnormal BG levels with a lot less insulin resistance and but even wimpier beta cells. And of course there can be all kinds of combinations of these two factors.
If diagnosing diabetes is difficult, diagnosing prediabetes is even more difficult because someone with full-blown diabetes like I had when I was diagnosed is unlikely to revert to normal no matter what they do. At that point we've lost so many beta cells that unless we figure out how to get them to regenerate, we're always going to have to be careful about our diet.
But in the prediabetes range, the probability of reverting to normal BG control is greater, especially if you're very overweight and hence are still producing a lot of insulin when you're diagnosed. Thus a diagnosis may be more vague. One month you'd qualify as prediabetic and then you'd lose some weight and you wouldn't. Then you'd regain the weight and you would.
Because of all this diagnostic vagueness, arbitrary cutoff points, and changing standards, any studies that purport to show that "diabetics" are at increased risk or decreased risk or should be taking X drug or avoiding Y practice are somewhat questionable. The older the study, the less relevant it's likely to be. In the old days they didn't even differentiate between type 1 (autoimmune; insulin requiring) and type 2.
I personally don't put a lot of trust into studies that rely on complex statistics to show some effect. You can study 10,000 patients with type 2 and show that there's a slightly better, statistically significant benefit from some treatment (usually a drug). If you're a physician interested in prescribing that drug, that suggests that the odds of success are greater if you prescribe it. (Not taking into account the biases caused by the fact that most drug studies are sponsored by drug companies that know how to manipulate data.)
But it says nothing about whether the drug will help or harm any individual patient. And as patients, that's what we want to know.
Does that mean we should simply ignore all these massive trials? I don't think so. They do tell us something; they suggest that some treatment could help or harm.
But if your doctor tells you that all "diabetics" should be taking some drug or avoiding some drug or following some other regimen and you don't think it sounds "right for you" as the TV ads are so enamored of saying, then research it carefully.
Find out if the patients in the study sound similar to you. If they were mostly elderly white men on low-fat diets and you're a young Asian woman on a low-carb diet, you might respond differently than those patients.
Diabetes comes in many flavors. Diagnosis can be arbitrary. Statements like "All diabetics should be on aspirin" are unlikely to be true, even if supported by references to some big clinical trial.