Continuous Glucose Monitoring Efficacy in Routine Use

One-half century ago, hand-held devices for home self-monitoring of blood glucose became more widely accessible. Most of the current health care providers who help people with diabetes consider self-monitoring of blood glucose a cornerstone of daily diabetes management. They likely could not imagine that highly respected diabetologists from Oxford, United Kingdom, published an article in Diabetes Care in 1980 stating that “In many patients, sufficiently good control can be obtained by this method [referring to glucose measurements limited to the outpatient clinic visits] so that it is not necessary to ask them to measure their own blood glucose concentrations or to ask them to obtain the fairly expensive meters for reading glucose oxidase strips” (1).

The introduction of real-time continuous glucose monitoring (CGM) 15 years ago received a very similar initial reception from many distinguished diabetologists. The comparison of CGM downloads with “spaghetti” and the fear of “data overload” were so universal that a randomized controlled trial (RCT) with a real-time CGM, despite substantial and clinically meaningful benefit, barely got accepted for publication (2). However, evidence-based medicine prevailed, and the landmark Juvenile Diabetes Research Foundation’s RCT clearly demonstrated a robust benefit of CGM in adults with type 1 diabetes (T1D) (3). Several RCTs followed, demonstrating the safety and efficacy of CGM in hypoglycemia (4), in the pediatric population (5), integrated with an insulin pump (6, 7), used with basal-bolus therapy (8, 9), and during pregnancy (10). Consequently, several distinguished professional associations published consensus statements and/or recommendations on CGM (11, 12). More recently, successful use of CGM was also demonstrated in people with type 2 diabetes, and many individuals with T1D and type 2 diabetes and their care teams rely on CGM as a main tool for daily diabetes management (13). Finally, time in range was proposed as a CGM-derived metric for both routine clinical management as well as clinical research (14).

CGM also enabled the rapid development of closed-loop insulin delivery with several artificial pancreas systems (15–18)—a dream of a few, one good decade ago, and a clinical reality today (19).

However, the question of the safety and efficacy of CGM in routine day-to-day home use remained largely open. Longitudinal reports from individual national diabetes centers associate improved metabolic control with increasing use of insulin pumps and CGM (20), and analyses of spontaneous uploads from CGM users demonstrated substantial benefits of CGM use (21). However, a large prospective routine-use study was missing. In a previous issue of the journal, a report from Belgium provides this extremely important piece of evidence. The prospective observational study was required by the Belgian national public insurance system upon a full reimbursement approval and included 515 adult patients with T1D for a 12-month prespecified follow-up (22). The main indication for starting CGM was hypoglycemia (56%), followed by inadequate glycemic control (26%). Baseline hemoglobin (Hb)A1c, in 417 (81%) patients who used CGM connected to an insulin pump for at least 12 months, was 7.7 ± 0.9% (61 ± 9.8 mmol/mol) and decreased to 7.4 ± 0.8% (57 ± 8.7 mmol/mol) at 12 months (P < 0.0001). Sex [female (59%)], higher education (64%), white (97%), a long history of T1D, impaired awareness of hypoglycemia (47%), and 5.7 ± 4.6 years of continuous subcutaneous insulin infusion use at baseline were the most important characteristics of this population. As expected, HbA1c, at 4, 8, and 12 months, decreased more in patients who started CGM because of inadequate glycemic control compared with patients who started because of hypoglycemia or pregnancy. Pregnant women had the lowest baseline HbA1c [7.2 ± 0.7% (55.0 ± 7.7 mmol/mol)] and decreased it even further to 6.6 ± 0.7% (49.0 ± 7.7 mmol/mol) at 4 months (P < 0.0005) and 6.9 ± 0.9% (52.0 ± 9.8 mmol/mol) at 12 months. Only the indication for starting the CGM was an independent predictor of changes in HbA1c in the multivariable analysis (P < 0.0001). The proportion of patients who achieved HbA1c < 7% (<53 mmol/mol) increased from 23% to 33% at 12 months (P = 0.001). Several other predefined, clinically important outcome measures were significantly improved: hospitalizations for severe hypoglycemia or ketoacidosis decreased from 16% to 4% (P < 0.0005), along with a decrease in admission days from 54 to 18 per 100 patient-years (P < 0.0005), representing a nationwide cost reduction of €345,509 ($430,000) during the study period. Additionally, work absenteeism decreased, and quality of life improved significantly, with a strong decrease in fear of hypoglycemia. Several CGM parameters also demonstrated benefit: the percentage of values <70 mg/dL (3.9 mM) was 5.6 ± 3.8% in the first 2 weeks compared with 4.5 ± 3.2% after 12 months (P = 0.002), independent of baseline HbA1c; glucose variability, measured by coefficient of variation (95% confidence interval), decreased from 38.7% (38.0 to 39.4) in the first 2 weeks to 37.9% (37.2 to 38.6) at 12 months (P = 0.02).

The effectiveness of CGM observed in this study may be attributable to several factors. Mean sensor use was 87.5%, which is more than in most RCTs and reported real-world use; high percentage of sensor use is repeatedly associated with higher effectiveness of CGM (3, 6, 7). Furthermore, the attrition was very low. As a result of the specific design of the study, dictated by the Belgian health authorities, clinical teams selected highly motivated patients who were able to attain a high benefit from the CGM technology. The trial, however, has all of the limitations of an observational study without a prospective control group. Specific training, education, and more intense contact with health care professionals may have contributed to the favorable outcomes. Additionally, the unavailability of blinded CGM measurements before the start of reimbursement precluded the comparison of time spent in hypoglycemia, desired range, and hyperglycemia before and after the use of CGM.

With the available amount of data from RCTs, which together include several thousand patient months, meta-analyses of these data, and judicious appraisal of studies by professional groups, the effectiveness of CGM and its clinical benefit remain beyond reasonable doubt. The Belgian experience (22) augments the evidence from good, routine clinical practice by dedicated medical teams and a unique public reimbursement system able to foster improved medical care through applied research. Despite the intense involvement of the study participants, the quality of life significantly increased. These important findings will likely benefit people with T1D in many other countries and compel various reimbursement authorities to assume a similar attitude toward CGM reimbursement. Let’s join forces and act together so that this will, in turn, stimulate scientists and corporations to develop even more accurate, smarter, smaller, less burdensome, more automated, and integrated CGM systems that will assimilate smoothly into hours, days, and years of life with diabetes, finally reducing the overwhelming burden of this chronic disease to individuals and the society.

Glossary

Abbreviations:

CGM

continuous glucose monitoring

Hb

hemoglobin

RCT

randomized controlled trial

T2D

type 2 diabetes