Clinical Briefs

By Louis Kuritzky, MD, Clinical Assistant Professor, University of Florida, Gainesville. Dr. Kuritzky is a consultant for GlaxoSmithKline and is on the speaker's bureau of GlaxoSmithKline, 3M, Wyeth-Ayerst, Pfizer, Novartis, Bristol-Myers Squibb, AstraZeneca, Jones Pharma, and Boehringer Ingelheim.

A1c as a Screening Tool for Diabetes

The most recent 2007 position paper from the ADA suggest that all adults over age 45 receive screening for diabetes by a fasting blood glucose (FBG). Measurement of FBG is not always convenient for patients. The National Health and Nutrition Examination Survey (NHANES) population provides a window of observation into the potential utility of A1c, as opposed to FBG for screening purposes.

Based upon the NHANES subset of 6,012 individuals who had both FBG and A1c measured, an A1c of 5.8% had a sensitivity of 86% and a specificity of 92% for the confirmed diagnosis of diabetes.

The ADA has to date not supported use of A1c for screening because:

1) Although there is a nationally standardized method for measuring A1c, not all laboratories use the same method, hence A1c can vary from lab to lab;

2) At the earliest stages of diabetes, the A1c may be less sensitive than either the FBG or postprandial glucose.

Since 50% of persons with newly diagnosed diabetes already possess one or more of the complications of diabetes at the time of diagnosis, it is obvious we are not diagnosing early enough. Since A1c can be obtained at any time of day, patients do not have to arrive at any particular time and may or may not be fasting. The authors suggest that (using the A1c method of the DCCT) an A1c < 6.0 is normal, 6.0-6.9% is prediabetic, and > 7.0% is diabetic. Finally, they offer the opinion that an A1c > 5.8% could stimulate further investigation or closer observation.

Buell C, et al. Diabetes Care. 2007;30(9):2233-2235.

Effects of Cinnamon on Glucose and Lipids

The concept that cinnamon could affect glucose control was first addressed in 1990, and subsequently supported by animal studies demonstrating improved insulin signal transduction, lowered glucose, and favorable effects upon lipids when cinnamon was administered. In 2003, a clinical trial in Pakistan gave further credence to the cinnamon-glucose relationship by showing that 40 days of cinnamon ingestion reduced fasting glucose, triglycerides, and LDL. Blevins, et al performed the first study of cinnamon treatment in diabetics in the United States.

Sixty diabetic subjects were randomized to receive either 500 mg/d cinnamon or placebo, both administered in identical capsules each night with the evening meal. Study subjects were followed for 3 months. Fasting glucose, total cholesterol, LDL, HDL, triglycerides, insulin levels, and A1c were measured at baseline and monthly thereafter.

At study end, there were no discernible effects of cinnamon upon any of the variables. There were meaningful differences between this study population and that of the Pakistan trial, in addition to ethnicity: for instance, the fasting glucose and triglyceride levels in the American population were much lower than the Pakistani study subjects. On the other hand, the dose of cinnamon used in the Pakistan trial was much higher: 1-6g daily. A beneficial metabolic impact of cinnamon in type 2 diabetic Americans has not been confirmed.

Blevins SM, et al. Diabetes Care. 2007;30(9):2236-2237.

Do Residents Have a Strong Enough Grasp of Statistics?

Best utilization of the currently available literature requires skillful interpretation of the statistical methods used to design and implement a study, as well as critical appraisal skills for the outcomes of the trial. The majority of currently practicing clinicians received little or no specific training in biostatistics, and although evidence-based medicine curricula in medical schools around the country are growing in number, they are not yet universal. Hence, it should come as no surprise that resident physicians may not be as well equipped to address the statistical aspects of the literature as we would like.

To assess knowledge of residents in reference to statistics, a 20-question test was completed by 277 residents from 11 different residency programs in the United States. The questions addressed commonly encountered issues in clinical trials.

Overall, the resident knowledge score was approximately 41% (out of 100%). Residents with previous epidemiology training performed somewhat better, as did residents who most recently entered training.

Several recent clinical trials have failed to achieve statistical significance for their primary endpoint, thus rendering the secondary endpoints uncertain, yet clinicians who are unaware of this statistical "boundary" may give credence to those secondary endpoints, which in essence, remain unproven. More attention to instruction in pertinent biostatistics during clinical years may be required to remedy some of these deficits.

Windish DM, et al. JAMA. 2007;298(9):1010-1012.