Real-world patients with chronic diseases, including diabetes or hypertension, frequently differ from those in more perfect clinical trial settings. Cummings et al 1 describe an innovative approach to measuring glucose control over time by using statistical calculations to measure the area under the curve (AUC) for fluctuations in glycosylated hemoglobin (HbA1c) from the ideal Յ7% (see Figure 1). Though an observational study such as this over 6 to 10 years is less than an ideal design, it observes patients in their environment, including times of varying adherence to lifestyle changes and medication, and removes the Hawthorne effect (patients do better while being observed during a clinical trial). It may also indirectly measure clinical inertia and the response of the health care system to elevated values.The authors take 10 different perspectives of the HbA1c, ranging from the AUC above HbA1c Ͼ7% over time, to sophisticated statistical analysis including "root mean square." The end point was renal function over time using estimated glomerular filtration rate (eGFR) calculated by the Modification of Diet in Renal Disease equation.A strength of the study is the large number of African-American patients. The more rapid decline in this ethnic group irrespective of sex compared with white patients was striking (Figure 2). Yet the determining factors accounting for 27.3% of the variance in the decrease in renal function over time were age, mean systolic blood pressure, initial HbA1c, initial eGFR, and number of HbA1c measurements, not race or sex. Although various HbA1c measurements accounted for 4.9% of the variance in decline of eGFR, the following 5 measures accounted for 4.5% of the variance: AUC above 7%, AUC approximately 7%, HbA1c decrease, HbA1c standard deviation, and HbA1c root mean square approximately 7%.Although this study has important methodologic limitations, it also has important methodologic value: What do we really know about the importance of fluctuations in HbA1c in routine patient care? and What are the best methods for characterizing this variability? Although other studies have looked at single-point or mean HbA1c measurements, this study has taken numerous statistical approaches to characterize HbA1c variation in routine practice over time and has attempted to relate that variability to an important diabetes endpoint. Though this study suggests a relatively smaller role of HbA1c variability compared with other predictors, it was stimulating to follow the sophisticated thought processes.