Bailey T. et al. Diabetes Technology Ther. 2015, 17(11), 1-8

predominantly due to between-sensor variance, with the linear mixed model estimating the between-subjects variance component to be 37% of the total. There were no statistically significant differences in sensor sensitivity (i.e., slope) be- tween insertion sites on either the right or left arm (P = 0.5542). Performance of the System was stable across the 14 days of wear after the first day. The percentage of readings within Consensus Error Grid Zone A (BG reference) on Days 2, 7, and 14 was 88.4%, 89.2%, and 85.2%, respectively (Fig. 3), and the MARD on the same days was 11.9%, 10.9%, and 10.8%, respectively. The mean time to first glucose results was 1 h 1 min (n = 168), and 100% of sensors were able to provide inter- stitial glucose results within 1 h 10 min after insertion. The mean lag time between the System sensor and YSI reference was 4.5 – 4.8 min. Sensor accuracy was not affected by factors such as body mass index (BMI), age, type of diabetes, clinical site, insulin administration, or hemoglobin A1c, as the percentages of readings within Consensus Error Grid Zone A were similar (Fig. 4). At Days 1 and 15, study participants completed a ques- tionnaire specific to this study and rated their experience with the System on a scale of 0 (strongly agree) to 4 (strongly disagree) for several subjective categories. On Day 1, the scorecard included seven statements about ease of use, pain compared with fingerstick, adequacy of packaging informa- tion, and pain or bleeding when applying the sensor to either arm. On Day 15, the scorecard included nine statements about the sensor relative to comfort of wear, ease of wear, whether the sensor got in the way of daily activities, pain compared with fingerstick, easier than fingerstick, and ery- thema or edema after removal of sensor from either arm. A score of 0, 1, or 2 was considered a favorable response. On Day 1, favorable ratings were reported by most respondents (‡97.2%; n = 72) for seven of seven subjective statements. Likewise, on Day 15 the majority of respondents (‡94.4%; n = 72) reported favorable ratings for nine of nine subjective statements. In total, 98.6% (142/144; both arms of 72 respondents) of responses were favorable ratings of £ 2 for the statement ‘‘The amount of pain I felt when applying the Sensor was acceptable.’’ A favorable response of £ 2 was also reported by 99.3% (143/144) of responses for statement ‘‘The amount of bleeding I experienced when applying the Sensor was acceptable.’’ There were no unexpected adverse device effects re- ported during the clinical study. One participant had a se- rious adverse event (severe hypoglycemia prior to sensor insertion) not related to the study or device. Skin issues observed in 202 site exams of 72 study participants were as follows: moderate to severe itching 0.5% of the time, moderate erythema 4.0% of the time, and 98.6% of the in- sertions had a pain rating of £ 2. Rate of mild incidences was < 9% for any individual category of skin issues men- tioned above, including edema, rash, induration, bruising, bleeding, and others. Discussion This study evaluated the performance and usability of the FreeStyle Libre system. Study results showed agreement between the System’s sensor readings and capillary BG and venous reference. The capillary BG reference provided a wider distribution of glucose results and covered up to 14 days of wear. Therefore, capillary BG was used as the pri- mary comparator for the System’s performance evaluation. Capillary BG reference provides more reference points and represents real-life accuracy during daily patient use. One strip lot was used as the capillary BG reference across all clinical sites and sensor lots so there were adequate data to determine variation between sensor lots and sites without confounding results with multiple strip lots. The System is unique among existing interstitial glucose monitoring technologies in that the wired enzyme factory- only calibrated sensor has wear time of up to 14 days without additional calibration. This lack of reliance on an external BG monitor for calibration is a potential advantage as errors in BG meters could potentially lead to glucose monitoring system errors. Sensors requiring routine user calibration several times daily can be affected by glucose instability, such as observed postprandially.15 Sensor delays or lag be- tween interstitial readings and venous or capillary readings have also been shown to vary among sensors,16 with newer- generation sensors demonstrating less lag time.15,17,18 Differences among interstitial,14,19,20 capillary, and venous readings must also be considered when comparing accuracy outcomes.9,10,21–23 Sources contributing to differences be- tween capillary BG versus venous YSI readings include the amount of blood used for testing, delays in analysis from the time of sampling, and differences in the composition of the blood samples.9,10 Collectively, these differences limit the direct comparison of accuracy outcomes among sensor technologies. Therefore, the present study was compared with reported outcomes with similar wired enzyme technology, factory-calibrated sensors, and those reporting accuracy outcomes using Consensus EGA. In a study of 55 subjects who wore a wired enzyme, FIG. 3. Stability of accuracy across 14 days of Freestyle Libre sensor wear: Consensus Error Grid Zone A (green), Zone B (blue), and Zone C (red). N is the number of paired sensor and reference data points per day of wear for all sensors used in the study. FACTORY-CALIBRATED FLASH GLUCOSE MONITORING 5

Seitenübersicht