March 01, 2010
4 min read
Save

Examining the relationship between HbA1c, mortality

You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact customerservice@slackinc.com.

In a thought-provoking study published in the Lancet in February, Currie et al analyzed the relationship between HbA1c and survival in patients with diabetes, and reported associations between lower HbA1c and mortality. Interest in their report was undoubtedly piqued by the researchers’ suggestion that the findings extended those of the ACCORD trial, in which intensive treatment to a HbA1c goal of 6% was associated with increased mortality, and findings from the VADT, in which severe hypoglycemia was associated with markedly increased mortality.

What are we to make of this?

The large U.K. General Practice Research Database was used to obtain clinical information on nearly 28,000 patients with diabetes whose treatment had been intensified from oral monotherapy to combination metformin/sulfonylurea therapy, and more than 20,000 patients with diabetes who had received oral monotherapy and were then given insulin with or without additional oral hypoglycemic medications. The mean follow-up was 4.5 years and 5.2 years in the two groups.

In the combination oral therapy group, mortality was highest at 9% among those in the lowest (≤6.72%) and highest (>9.85%) HbA1c deciles; however, mortality was flat, at 7%, for HbA1c values ranging from 6.73% to 9.11%, with an increase to 8% in the penultimate decile of HbA1c ranging 9.12% to 9.84%. In the insulin-receiving group, using the same HbA1c deciles, 18% of those in the lowest and 17% of those in the highest deciles died during follow-up, with the lowest mortality at 11% for those in the fifth decile with HbA1c 7.69% to 7.96%. Statistical adjustment for age, sex, cigarette use, cholesterol, baseline BMI, cardiovascular disease, number of general practitioner visits and a comorbidity index including a variety of ailments (in particular, CVD and renal insufficiency) gave similar conclusions. Furthermore, progression to a first CVD event was 1.6-fold among those in the lowest HbA1c decile compared with those in the fourth decile, centered at an HbA1c of 7.5%, and then increased progressively as the HbA1c increased from 8% to 10.5%.

Zachary T. Bloomgarden, MD
Zachary T. Bloomgarden
Yehuda Handelsman, MD
Yehuda Handelsman

There are flaws in the assumption that the association of lower HbA1c with higher mortality reflects causality. For example, a relatively high serum creatinine cutoff of 1.7 mg/dL was reached by 16% of the lowest decile of insulin-treated patients, 14% of the second decile, 13% of the third and fourth, and 9% to 10% of those in the seventh to tenth deciles. Therefore, the inclusion of renal disease in the comorbidity index may not have altogether eliminated this cause of both lower glucose and of higher mortality.

A related and important caveat is the recognition that not all patients glycated hemoglobin to the same degree for a given level of glycemia, with lower HbA1c seen in a variety of illnesses and, hence, potentially in itself is somewhat misleading as a marker of mortality. This may particularly be an issue with the use of many different HbA1c assays in a non-standardized fashion in this epidemiologic study. Furthermore, one may wonder whether the inclusion in the study of only those patients who required additional treatment, for unspecified reasons, and the exclusion of the cohort of persons whose glycemic control was considered to be stable, might in some fashion have biased the outcome to lead to the finding of greater mortality and, perhaps, of more CVD in those with lower HbA1c levels. It is intriguing to note that the researchers of the present study excluded patients aged younger than 50 years, and that in the metformin plus sulfonylurea and insulin cohorts age increased progressively with lower HbA1c levels. So, although statistical adjustment was used, there may have been a dual bias, first from increasing mortality of greater age, and, second and more subtly, from the inclusion of progressively more patients with more advanced disease in the lower HbA1c groups. We do not have information as to whether lipid and blood pressure-lowering and anti-platelet treatments were given preferentially to younger rather than older patients, which could further confound interpretation of the findings.

Turning point

Nevertheless, we may be at a turning point in the “less is better for everyone” approach to glycemia, employing a more nuanced approach to recognize that “tight control” might not be optimal for all patients with diabetes. Those with less CVD, for example, appeared in the VADT to have greater benefit of glycemic control. And, intriguingly, different approaches of achieving similar degrees of glucose-lowering may not lead to equal outcomes, with a comparison of different insulin treatment regimens showing that patients with type 2 diabetes started on basal insulin with subsequent addition of preprandial boluses had significantly fewer adverse events and less CVD than those started on prandial or biphasic insulin.

An issue which could not be addressed in the current study is whether it is good glycemic control that somehow is bad or whether it is hypoglycemia, not documented in the existing dataset, that led to adverse outcome, or whether greater age, other morbidities such as renal insufficiency from malignancy associated with both low HbA1c and with adverse outcome, are acting as confounders. If the latter is the case, then we need to simply carefully evaluate our patients. If it is truly good control that has adverse effects, then all of our current concepts must be reassessed. But if hypoglycemia is the mediator, we must be careful not to conflate this common adverse effect of glycemic treatment with the notion that glycemic treatment is itself bad. Whether incretin-based treatments and future approaches are similarly less likely to cause hypoglycemia will lead to better outcome than the metformin plus sulfonylurea or (presumably) NPH insulin-based treatments used for patients in the Currie study, and still recommended as “first line” in the American Diabetes Association treatment algorithm, will be the subject of intensive investigation in coming years.

Zachary T. Bloomgarden, MD, is Clinical Professor in the Department of Medicine at Mount Sinai School of Medicine, Endocrine Today Editorial Board member, and Editor of the Journal of Diabetes. Yehuda Handelsman, MD, is Medical Director of the Metabolic Institute of America.

For more information:

  • Currie CJ. Lancet. 2010;375:481-489.
  • Duckworth W. N Engl J Med. 2009;360:129-139.
  • Gallagher EJ. Journal of Diabetes. 2009;1:9-17.
  • Gerstein HC. N Engl J Med. 2008;358:2545-2559.
  • Holman RR. N Engl J Med. 2009;361:1736-1747.
  • Nathan DM. Diabetes Care. 2009;32:193-203.
  • Reaven PD. Diabetes. 2009;58:2642-2648.
  • Shrom D. Journal of Diabetes. 2010;doi:10.1111/j.1753-0407.2010.00066.x.