Diabetes Discoveries & Practice Blog

Interpreting A1C: Variability Across Populations

A doctor and a patient

A1C is a useful but imperfect tool for diagnosis of diabetes, because test responses vary considerably across ethnic and racial populations.

K. M. Venkat Narayan, MD, MSc, MBA, is a professor of medicine & epidemiology at Emory University who has done substantial research exploring variability in type 2 diabetes across populations. In the final post of our Interpreting A1C blog series, he explains why diagnosing diabetes with the hemoglobin A1C test can sometimes be problematic and offers suggestions for using the test in patient care and research.

Q: Research has found that the hemoglobin A1C test potentially misclassifies some people. Can you describe the problem?

A: We need to step back and understand what A1C is. Put simply, it captures hemoglobin glycation over the past 120 days, which is the normal lifespan of the red blood cell. Over the past 10 to 15 years, A1C has become popular as a test for management of diabetes and, increasingly, as a test for diagnosis of diabetes. However, in that same timeframe, a body of research has emerged showing that there is considerable variation in A1C results across populations.

For example, A1Cs may be .25 –1.0 percentage point higher in African Americans compared to Caucasians in the United States, and this is not necessarily reflecting differences in glucose levels. This difference has also been found in Asian American and Hispanic populations, and increasingly we're hearing reports of disparate results, compared to U.S. Caucasians, from low- and middle-income countries. My colleague Unjali Gujral looked at a cohort called the MASALA cohort (Mediators of Atherosclerosis in South Asians Living in America), which is Indian Americans living in California and Chicago, age 40 and over, and another cohort in India called the CARRS cohort (Centre for cArdiometabolic Risk Reduction in South-Asia Surveillance Study), also age 40 and over. She found that 19% of the study participants in Chennai, India, 27% in Delhi, India, and 11% in the United States had diabetes based on the A1C definition of around 6.5%, but not based on abnormal fasting glucose or the 2-hour glucose tests. Evidence of this problem has also come from African American and Hispanic populations who were part of the National Health and Nutrition Examination Survey (NHANES) and other surveys.

A1C is a simple test, but when we use A1C alone for diagnosis, we are overestimating or underestimating the amount of diabetes in the population. It is a considerable problem, especially in non-Caucasian populations in the United States and, also across the lower- and middle-income countries of the world, which have the largest burden of diabetes.

Q: What causes this variability in response to the A1C test?

A: As I said earlier, A1C captures hemoglobin glycation over a 120-day period in the red blood cells. So that assumes a normal red blood cell span of 120 days. Anything that affects red blood cell survival can affect the A1C levels, such as acute blood loss, mutations of various amino acid sequences, sickle cell anemia, thalassemia, or iron deficiency anemia. Other conditions, like advanced liver or kidney disease, also can impact A1C measurements.

Some populations, for example the African American population in the U.S., the African population in their own countries, and Caribbean populations, all have high levels of sickle cell disease and thalassemia. For them, the measurement becomes unreliable and you get artificially high levels of diabetes diagnosis.

Also, in individuals without diabetes, just a third of the variance of A1C is explained by glucose levels, age, and body mass index. This suggests that there might be many other factors involved in individual variance such as genetics. Results also can differ based on the kind of assays that are used.

Q: Is underdiagnosis also a concern?

A: Overdiagnosis is a bigger concern. But some have argued that people with anemia might be underdiagnosed, so, yes, both over- and under-diagnosis are problems to be concerned about.

Q: What are the risks from diagnosing someone with diabetes who doesn’t actually have the disease?

A: Overdiagnosis can suggest a higher burden of diabetes than truly exists, so that's a problem at the epidemiological level. At the patient level, overdiagnosis is a false positive, and the risk of that is two-fold. Number one, a diagnosis of diabetes is serious. It can have implications for insurance or driver's licenses in certain places, and also there is the risk of overtreatment. Some might argue that the fact that a person’s A1C is high means that they have prediabetes and it is appropriate to treat them with lifestyle interventions. That may be right, but I still think that mislabeling is a risk given the factors that go along with it. Especially in vulnerable populations, misdiagnosis becomes a very serious problem.

Q: Is interpreting A1C results also a challenge for ongoing patient management?

A: When it comes to clinical management, there are two factors that might make A1C less of a concern. Number one is, A1C is seldom used alone. Clinicians use A1C alongside measurements of glucose as part of routine practice. And secondly, the A1C for an individual patient becomes his or her own control. For example, if you measured my A1C, even if it’s higher, my next value at the end of three months is obtained using the same method, so there is an internal validity.

Q: What guidance can you offer for clinicians using the A1C test diagnostically?

A: When it comes to diagnosis, the convenience of the A1C is that a person doesn't need to be fasting, and you don't have to give the person a two-hour glucose load, so you can do it any time of the day. That convenience needs to be balanced with this problem of unreliability or overestimation in some populations. So be aware of the populations with which the diagnostic test may be less reliable or may be overestimating or underestimating.

And secondly, diagnose using a rule-in, rule-out approach. What that means is, if the A1C level is below 5.5, you don't need to worry about misdiagnosis, and if it's a level above 7, you don't need to worry. But for the intermediate group, those with A1C levels within 5.5 to 7, you might rule them in to do additional glucose testing. To do the A1C and then glucose testing on everybody would be cumbersome, costly, and unnecessary. But if you take a rule-in, rule-out approach and do additional glucose testing only on patients with A1C of 5.5 to 7, you eliminate three fourths of the extra testing.

The other important principle here is, never make a diagnosis of diabetes based on one test. Always make it on two tests. It could be two A1Cs or an A1C and a fasting or two-hour blood glucose test.

Q: Are health care professionals aware of the problems with A1C?

A: Awareness may not be as widespread as we want it to be. Many endocrinologists would be aware, I would imagine, but primary care providers, particularly in underserved areas or in the poorer countries of the world, may not be aware. On the other hand, in the poorer countries of the world, at least for now, the A1C test is very expensive and is not used as often as you might think.

Q: What research is being conducted to improve the accuracy and utility of this test?

A: We need more data on A1C standardization in populations that may have these other conditions such as sickle cell anemia, thalassemia, regular anemia, or in populations with high levels of liver disease or kidney disease. We make recommendations on A1C diagnostic levels based on high-income country research, and people from the poorer countries attend our conferences and take away messages that might not apply to their populations. So, I think there is a certain responsibility here. For example, I have colleagues in Trinidad and Jamaica, and we are urging them to measure A1C in a population sample and also get 2-hour glucoses and fasting glucoses. Then we can analyze the data and come up with local standards.

Comments

Blog Tools

Share

About

Diabetes Discoveries and Practice Blog
Dialogue with thought leaders on emerging trends in diabetes care

Tags

Archive

Disclaimer

We welcome comments; all comments must follow our comment policy.

Blog posts written by individuals from outside the government may be owned by the writer and graphics may be owned by their creator. In such cases, it is necessary to contact the writer, artist, or publisher to obtain permission for reuse.