The Nuts and Bolts of Achieving End Points With Real-Time Continuous Glucose Monitoring

Currently available CGM devices measure interstitial glucose. The physiologic lag for equilibration of glucose between the blood and interstitial fluid compartment has important implications for sensor calibration and clinical decision-making during times when the glucose level is changing.

Current generations of CGM devices are calibrated by the patient using fingerstick blood glucose measurements. To optimize CGM accuracy, it is important that the patient calibrate the device during steady-state conditions. After meals, the glucose level will often increase by >3 mg · dl⁻¹ · min⁻¹, and this, in conjunction with the physiologic lag in equilibration of the blood and interstitial glucose that is often in the range of 10–15 min, can lead to differences between glucose levels in the blood and interstitium of as much as 30–45 mg/dl (3). If the CGM is calibrated with blood glucose measurement postprandially, this will lead to upward setting of the sensor and compromise the accuracy of the device in detecting hypoglycemia.

The physiologic lag can have important implications with regard to detection and treatment of hypoglycemia. When the glucose is falling rapidly, the interstitial glucose generally lags behind blood, and in this situation, the actual blood glucose could be quite low, even when the interstitial/sensor glucose is normal (4,5). The practical implication is that in situations when the glucose is falling (such as after exercise), even if the sensor glucose is normal, the patient will need to perform fingerstick glucose measurement to clarify whether to treat; this is especially important before driving. During the recovery from hypoglycemia, the increase in the interstitial/sensor glucose will often lag behind the blood glucose (6). At times when blood glucose has already normalized, the sensor/interstitial glucose will still be low, and some patients who rely on RT-CGM to assess response to treatment will mistakenly assume they need to consume more carbohydrates, resulting in overtreatment of hypoglycemia. The practical implication is that the patient should be instructed of the need to perform fingerstick glucose measurements to assess recovery from hypoglycemia. With the adoption of RT-CGM, the current guidelines for treatment of hypoglycemia (i.e., the rule of 15: take 15 g carbohydrate and recheck 15 min later) need to be modified.

Patients need to be aware that because of physiologic lag when the glucose is either rising or falling rapidly, there will be a marked difference between the sensor reading and fingerstick measurements and that these discrepancies do not necessarily indicate that the sensor is inaccurate. Understanding the basis for these differences can be important in ensuring that patients do not lose confidence in the technology.

While RT-CGM alarms have obvious potential utility in preventing and minimizing hypoglycemia, clinical experience indicates that with some patients, this benefit is counterbalanced by an increased frequency of hypoglycemia due to excessive postprandial bolusing. This tendency for some patients to overreact to postprandial excursions and take excessive insulin is illustrated in Fig. 1.

This tendency for excessive postprandial bolusing is a common problem with RT-CGM. A major focus of education and follow-up care of the patient using RT-CGM will often need to be addressed at reducing this risk. Factors that patients need to consider before taking extra insulin to treat postprandial hyperglycemia include the following: 1) Residual insulin “on board” from premeal bolus, 2) direction of “trend” arrow on glucose sensor, and 3) type of carbohydrate eaten.

In the STAR 1 trial involving sensor-augmented insulin pumps, subjects followed guidelines to modify food and correction boluses based on the rate of change of the glucose levels detected by the real-time sensor (7). It was recommended that if the glucose level is rising at 1–2 mg · dl⁻¹ · min⁻¹, the calculated food/correction bolus should be increased by 10%, and if the level is rising >2 mg · dl⁻¹ · min⁻¹, the calculated bolus should be increased by 20%. Conversely, if the glucose level is declining at 1–2 mg · dl⁻¹ · min⁻¹, the calculated food/correction bolus should be decreased by 10%, and if the level is declining by >2 mg · dl⁻¹ · min⁻¹, the calculated bolus should be decreased by 20%.

As illustrated in Fig. 2, after eating high–glycemic index carbohydrate foods, there is a rapid spike in glucose level (8). The insulin pharmacodynamic curve (9) indicates that at the time when glucose is peaking (∼1–2 h after the meal), there is a substantial amount of residual insulin “on board” from the premeal bolus, and if the patient were to take an extra bolus, there would be a considerable risk for hypoglycemia from dose stacking.

In contrast, as illustrated in Fig. 3, after a low–glycemic index meal, glucose absorption tends to be prolonged, and if the glucose is elevated ≥2 h after the meal, an additional bolus may be needed to bring the level down to target.

Figure 4 illustrates the tendency of some patients to respond to declining glucose levels (shown in the upper panel) by suspending basal delivery of insulin by their pump (shown in the lower panel). The end result is an increase in glycemic variability.

This practice (which is in part an attempt to minimize carbohydrate/calorie intake for treatment of hypoglycemia) has been noted in ∼30% of pump patients using RT-CGM. Patients need to be aware that suspending basal insulin is often not an effective way to treat hypoglycemia and should receive careful guidance to prevent and minimize this cause for glycemic instability.

CGM can be of potential benefit to patients with type 1 diabetes using intensive insulin therapy. To use continuous glucose data safely and effectively, patients need to be skilled in diabetes self-management, and the widespread adoption of RT-CGM into diabetes care will need to be coupled with comprehensive self-management education.