Description:
<jats:sec><jats:title>Background:</jats:title><jats:p> Despite much discussion regarding the clinical relevance of glycemic variation (GV), little discourse has addressed the properties of the data set from which it is derived. We aimed to assess the minimum duration of data required using continuous glucose monitoring (CGM) that most closely approximates to a gold standard 90-day measure. </jats:p></jats:sec><jats:sec><jats:title>Methods:</jats:title><jats:p> Data from 20 children and adolescents with type 1 diabetes were examined. All participants had CGM data sets of 90 days duration, from which standard deviation (SD), coefficient of variation (CV), mean amplitude of glycemic action (MAGE), and continuous overlapping net glycemic action (CONGA<jats:sub>1-8</jats:sub>) were calculated for the overall period and then investigational periods of 2, 4, 6, 12, 18, 24, and 30 days. The percentage difference between each measure and the overall measure per time period was assessed. </jats:p></jats:sec><jats:sec><jats:title>Results:</jats:title><jats:p> As the duration of the CGM data set increased, the percentage error continued to decrease, giving a metric approximating more closely toward the overall measure. Median SD and CV differed from the overall measure by <10% at 12 days duration. The frequency of interruptions to the CGM trace rendered MAGE and CONGA unreliable, hence SD and CV were reported. </jats:p></jats:sec><jats:sec><jats:title>Conclusions:</jats:title><jats:p> We suggest that data sets used to infer GV should be of a minimum duration of 12 days. MAGE and CONGA exhibit poor performance in the setting of frequent trace interruption. </jats:p></jats:sec>