Continuous glucose monitoring
Continuous glucose monitoring (CGM) provides information about the direction and rate of change of glucose and provides alarms to warn of impending hypoglycaemia. It is surprising that early studies of CGM failed to demonstrate a reduction in the incidence of hypoglycaemia [
30‐
32]. This may have been related to very low baseline rates of hypoglycaemia, perhaps again driven by safety concerns in regulatory studies. A patient-level data meta-analysis was more encouraging [
33] and later studies reviewed below have been more positive.
We refer here to open or real-time CGM, where the data are visible to the user and available for immediate response. Anecdotally, blinded CGM has been used to look at overnight glucose control and for educational purposes. One study failed to find any difference in hypoglycaemia detected by blinded CGM between people with and without intact awareness of hypoglycaemia [
10]; this unexpected finding may relate to the short period of data collection.
As with CSII, the importance of education cannot be underestimated. The Multicenter 2 × 2 Factorial Randomized Controlled Trial Comparing Insulin Pump with Multiple Daily Injections and Continuous with Conventional Glucose Self-monitoring (HypoCOMPaSS) was one of the largest RCTs of diabetes technologies in those with IAH. It randomised 96 people with type 1 diabetes and with IAH in a 2 × 2 fashion to receive CGM or not and CSII or not [
34]. Although the number of episodes of severe hypoglycaemia per person-year was very high at baseline and fell dramatically throughout the study, it did not differ between the four randomised groups. All participants received a short group education package and weekly support from researchers during the 24 weeks of the main study and the global benefit has been attributed to this. On the other hand, the IN CONTROL study [
35], which also recruited people with IAH, demonstrated significant reductions in the rates of severe and biochemical hypoglycaemia in participants who used CGM but who had received no intensive education. Only 35% of participants used carbohydrate counting, a marker for appropriate current diabetes education, and under half of the participants used CSII. Notably, the improvement was lost after participants crossed over to their previous therapy, thus indicating no evidence of any learning effect. Two recent studies of CGM in individuals using MDI demonstrated significant reductions in the duration of hypoglycaemia (by almost half) but rates of severe hypoglycaemia were low at baseline [
36,
37] .
Sensor-augmented pumps (SAPs) can suspend insulin delivery for actual or predicted hypoglycaemia [
38,
39]. One RCT, which compared a device providing automated suspension of insulin delivery in response to sensor hypoglycaemia vs SAP without this feature, demonstrated a 38% reduction in the number of nocturnal hypoglycaemic events [
40]. There was no episode of severe hypoglycaemia in this trial. In the same year, Ly et al [
41] reported a significant reduction in the incidence of moderate and severe hypoglycaemia using the automated threshold suspend system against CSII in 95 children and young adults.
Intermittently monitored, retrospective CGM, or ‘flash’ monitoring, provides easily accessible glucose data, including directional trends but no alarms or alerts for hypoglycaemia. Obviating the need for finger prick should improve rates of monitoring and control but this remains to be proven. An RCT in individuals with tight glucose control reported a 38% reduction in sensor-detected glucose values under 3.9 mmol/l, with no deterioration in overall glucose control, but this study excluded those with IAH or previous severe hypoglycaemia [
42]. A recent study in people with IAH found no impact on hypoglycaemia, compared with real-time CGM [
43], and further studies are needed.
Hypoglycaemia is also an issue in insulin-treated type 2 diabetes, but the use of technology to prevent hypoglycaemia in this group has not been studied in detail. An RCT of CSII vs MDI in type 2 diabetes (OPT2MISE) did show CSII to be of benefit in those who were unable to improve control despite high doses of insulin, although the incidence of hypoglycaemia was low in both groups [
44].
A common contributor to hypoglycaemia is over-correction of high glucose values, which result from repeated injections of rapid-acting insulin within a short period of time causing overlapping or ‘stacking’ of insulin doses. This is particularly important for individuals using CGM or ‘flash’ monitoring, whereby they see rapid fluctuations in glucose and/or receive alerts about glucose measurements that may be out of range. Software that can perform the calculation of insulin dose, while accounting for the ‘insulin action on board’, has been available in insulin pumps for a while. Greater use of the bolus advisor software on pumps has been associated with better control of postprandial glucose and a non-significant reduction in the frequency of postprandial hypoglycaemic events [
45]. Conclusive data are not yet available, although a meta-analysis of six small studies of bolus advisors used with insulin pumps found a non-significant reduction in the number of hypoglycaemic episodes [
46]. Incorporation of such technology into capillary plasma glucose meters and smartphone health applications has improved glucose control and reduced glucose variability, although no consistent impact on hypoglycaemia has been reported [
47‐
50]. Some studies have demonstrated reduced fear of hypoglycaemia with the use of these devices [
51] and many people with diabetes welcome their use. However, some individuals still express reservations and are reluctant to use automated advice (especially on basal insulin adjustment) without understanding how that advice was generated [
52]. There are important caveats to the use of smartphone health applications: many have not been tested as rigorously as other medical technologies and may carry risk of inappropriate dose advice [
53].
Despite these advances, we have not yet reached a Utopia wherein all have access to and are able to use the new technologies reliably to achieve optimum glucose control with minimal or no hypoglycaemia. Notably, while studies have demonstrated reduction in the number of episodes of severe hypoglycaemia with CGM, the benefit only persists while sensors are being used. None of the technologies have been shown to restore subjective awareness of hypoglycaemia or the impaired counter-regulatory responses of IAH needed for endogenous protection against severe hypoglycaemia. ‘Technological awareness’ only functions when the sensors are worn. The evidence to date thus supports a step-wise approach to the management of problematic hypoglycaemia, starting with validated structured education programmes transferring the skills of insulin dose adjustment to the users of the insulin, and progressing through use of technology, either pump or sensor, then SAP therapy with automated suspend features. If these conventional measures fail, where available, islet or pancreas transplantation may be an option [
29]. Using this algorithm, by progressing through the steps every 3–6 months and being supported throughout by an experienced multi-disciplinary healthcare team, we have been able to alleviate problems with recurrent severe hypoglycaemia in most of the individuals affected [
54]. It is unclear exactly where interventions such as flash glucose monitoring and closed-loop systems fit into this algorithm, as they have not been tested in this population; however, they are likely to provide a further level of protection. In future, a readily available fully closed-loop ‘artificial pancreas’ may change this algorithm if the device is able to maintain normoglycaemia without hypoglycaemia and proves acceptable to all users.