2012 Diabetes Care Volume 45, September 2022

Population-Level Impact and Michael A. Rotondi,1 Octavia Wong,1

Michael Riddell,1 and Bruce Perkins 2
Cost-effectiveness of Continuous
Glucose Monitoring and
Intermittently Scanned
Continuous Glucose Monitoring

Downloaded from [Link] by Sandip Mitra on 27 June 2023

Technologies for Adults With
EPIDEMIOLOGY/HEALTH SERVICES RESEARCH

Type 1 Diabetes in Canada:

A Modeling Study
Diabetes Care 2022;45:2012–2019 | [Link]

OBJECTIVE
Maintaining healthy glucose levels is critical for the management of type 1 diabetes
(T1D), but the most efficacious and cost-effective approach (capillary self-monitor-
ing of blood glucose [SMBG] or continuous [CGM] or intermittently scanned
[isCGM] glucose monitoring) is not clear. We modeled the population-level impact
of these three glucose monitoring systems on diabetes-related complications, mor-
tality, and cost-effectiveness in adults with T1D in Canada.

RESEARCH DESIGN AND METHODS

We used a Markov cost-effectiveness model based on nine complication states for
adults aged 18–64 years with T1D. We performed the cost-effectiveness analysis
from a single-payer health care system perspective over a 20-year horizon, assum-
ing a willingness-to-pay threshold of CAD 50,000 per quality-adjusted life-year
(QALY). Primary outcomes were the number of complications and deaths and the
incremental cost-effectiveness ratio (ICER) of CGM and isCGM relative to SMBG.
1
RESULTS School of Kinesiology and Health Science, York
University, Toronto, Canada
An initial cohort of 180,000 with baseline HbA1c of 8.1% was used to represent all 2
Division of Endocrinology, Department of Medicine,
Canadians aged 18–64 years with T1D. Universal SMBG use was associated with University of Toronto, Toronto, Canada
 11,200 people (6.2%) living without complications and  89,400 (49.7%) deaths Corresponding author: Michael A. Rotondi,
after 20 years. Universal CGM use was associated with an additional  7,400 (4.1%) mrotondi@[Link]
people living complications free and  11,500 (6.4%) fewer deaths compared with Received 10 November 2021 and accepted 7
SMBG, while universal isCGM use was associated with  3,400 (1.9%) more people June 2022
living complications free and  4,600 (2.6%) fewer deaths. Relative to SMBG, CGM This article contains supplementary material online
at [Link]
and isCGM had ICERs of CAD 35,017/QALY and 17,488/QALY, respectively.
© 2022 by the American Diabetes Association.
CONCLUSIONS Readers may use this article as long as the
Universal use of CGM or isCGM in the Canadian T1D population is anticipated to work is properly cited, the use is educational
and not for profit, and the work is not altered.
reduce diabetes-related complications and mortality at an acceptable cost-effec- More information is available at [Link]
tiveness threshold. [Link]/journals/pages/license.
[Link]/care Rotondi and Associates 2013

Though the etiology is not known with are heterogeneous (7,8), on balance, each type of glucose monitoring (SMBG,
certainty, type 1 diabetes (T1D) appears these technologies have shown im- CGM, isCGM) to obtain the population-
to be an autoimmune disease that de- proved glycemic control (9,10), reduced level cost and QALY impact of universal
stroys insulin-producing b-cells in the hypoglycemia (11,12) and DKA (11,12) use of these technologies. Our model
pancreas resulting in lifelong depen- events, fewer hospital admissions (11), is based on a Markov cost-effective-
dence on injected insulin therapy. Self- and improved quality of life (9,11) in a ness model with nine primary states:
management of blood glucose levels is variety of real-world settings and study no complications, retinopathy, neurop-
the cornerstone of T1D care. High levels populations. In most Canadian provin- athy, nephropathy, cardiovascular dis-
of blood glucose can lead to acute ces, people with diabetes over the age ease (CVD), end-stage renal disease,
events including diabetic ketoacidosis of 65 years (and <25 years old in some lower-extremity amputation, blindness,
(DKA) and long-term micro- or macro- provinces) receive some support from and death (Supplementary Fig. A1).
vascular complications (1), while low their provincial health plan for these Moreover, at any point in the model, a
blood glucose (hypoglycemia) puts pa- technologies, but the majority of adults severe hypoglycemia (SH) or DKA event
tients at risk for immediate injury or could occur. We compared the cost-ef-

Downloaded from [Link] by Sandip Mitra on 27 June 2023

require private insurance plans or sti-
death (2) and possibly a greater risk for pends from public organizations like Di- fectiveness of SMBG, CGM, and isCGM
developing cognitive impairment (3) abetes Canada or incur these costs as in this T1D cohort over a 20-year pe-
and earlier cardiovascular mortality (4). personal expenses. Decision makers riod using the T1D complication rates,
To monitor glycemia and adjust insulin must determine whether these new relative effects of CGM and isCGM,
doses, most people with T1D use self- technologies are appropriate for large- costs, and QALYs in Appendix A.
monitoring of blood glucose (SMBG) scale use and provide reasonable value
techniques, which requires lancing the in the context of limited health care re- Estimates of Efficacy of isCGM and
fingertip to obtain a drop of capillary sources. The purpose of this study is to CGM
blood that is applied to a test strip and estimate the impact on diabetes-related As a preliminary step, we required an es-
inserted into a glucose meter device. complications, mortality, and relative timate of clinical efficacy of isCGM and
Recent development of innovative tech- cost-effectiveness of SMBG, CGM, and CGM vs. SMBG to estimate the clinical
nological devices for T1D management isCGM for adults with T1D in Canada. improvements that can be attributed to
has led to new strategies for monitoring the national use of these advanced mon-
glucose levels in interstitial fluid, including itoring technologies. With this informa-
RESEARCH DESIGN AND METHODS
continuous glucose monitoring (CGM), tion and relevant epidemiological and
Model Overview and Initial Cohort economical information, we can evaluate
such as the Dexcom G6 or Medtronic
Guardian Connect, and intermittently Our model is adapted from the Ontario our primary outcomes of the absolute
scanned continuous glucose monitoring Health (OH) (formerly Health Quality number of complications and deaths
(isCGM) (previously known as flash glu- Ontario) (13) report and previous that may be prevented and the cost-ef-
cose monitoring), such as the Abbott work by Garc
ıa-Lorenzo et al. (14) and fectiveness of these interventions.
FreeStyle Libre systems. CGM systems McQueen et al. (15). The OH (13) study Identifying a summary measure for
provide measurement of glycemic lev- was a governmental report with investi- efficacy in clinical trials of glucose moni-
els throughout the day and night, gation of the cost-effectiveness of pub- toring strategies and technologies pre-
which reduces the risk of undetected licly funding CGM for all residents of sents a challenge. Some trials have
hypo- and hyperglycemic events, espe- the province of Ontario with T1D. Their been focused on recruitment of subjects
cially nocturnal hypoglycemic events modeling study assumed a baseline av- with elevated HbA1c, for which the pri-
(5). CGM systems are also equipped erage HbA1c of 8.8% for the population mary outcome of interest is improve-
with alarms that alert users if glucose and an SMBG testing regimen of an av- ment in average glycemia (e.g., 16,17).
levels exceed or decline preset limits, erage of six tests per day. Primarily due In other trials investigators have re-
enabling the user to take insulin or car- to higher device costs at the time, the cruited subjects with target HbA1c and
bohydrates to avoid a dangerously high report did not support public funding of evaluated the outcome of reduction in
or low blood glucose. By comparison, CGM at the willingness-to-pay threshold risk of hypoglycemia (e.g., 18,19). While
while isCGM systems continuously moni- of CAD 50,000 per quality-adjusted life- the effect of glucose monitoring ap-
tor interstitial glucose levels, they only year (QALY). Our model improves on proaches on HbA1c levels remains an
display glycemic levels when the sen- these approaches, as it includes the key important predictor for complications of
sor is scanned, and initial generations outcome of DKA events, focuses both T1D (18), there is a demonstrated need
of these systems do not have alarms. on CGM and on isCGM, and includes to move beyond this metric, as it does
CGM and isCGM may allow for better the most recent costs of these devices. not fully capture glycemic variability or
profiling of glucose management then Our initial cohort was age weighted to risk of hypoglycemia (20). To this end,
SMBG or HbA1c levels alone, since isCGM represent the 180,000 Canadians with there has been an increased focus on
and CGM can more easily track the in- T1D between the ages of 18 and 64 time in range (TIR) as an alternative
dividual’s percent time below, above, years. This value was obtained by apply- measure of glycemic control, particu-
and within the recommended glycemic ing the overall estimates of rates of dia- larly for CGM systems. The International
targets (6). betes in Canada to recent Canadian Consensus on Time in Range defined
While randomized trials examining census data (see Appendix A1). We as- TIR as the time spent in the optimal gly-
the efficacy of these new technologies sume that all 180,000 Canadians use cemic control range of 3.9–10.0 mmol/L
2014 Cost-effectiveness of CGM and isCGM in Canada Diabetes Care Volume 45, September 2022

(70–180 mg/dL) in a 24-h period (6). It Epidemiologic Complications, using a combination of manufacturer re-
has been further suggested that a 10% Rates, Quality of Life, and Cost sources and publicly available informa-
increase in TIR corresponds to approxi- Parameters tion on 12 August 2021. As the focus is
mately a 0.5%–0.8% reduction in HbA1c, Epidemiologic parameters were identi- on the cost-effectiveness of glucose
depending on the baseline HbA1c levels fied based on commonly reported T1D monitoring and complications, costs of
and population characteristics (6). Given complications in literature and those insulin delivery and treatment (e.g.,
the importance of TIR in CGM and used in the OH report (13). All partici- pumps or daily injections) were not in-
isCGM systems, we selected TIR as our pants began in the no complications cluded. Note that our goal is not to com-
primary measure of efficacy. state. After the first year, participants pare individual monitoring technologies.
To estimate the efficacy of CGM and can transition to a complications state,
Thus, to ensure an impartial evaluation
isCGM, we focused on randomized con- remain in the no complications state, or
of all competing monitoring systems, we
die. Once in a complications state, partic-
trolled trials (RCTs). A recent meta-analy- apply an average cost of relevant tech-
ipants can transition to a more severe
sis of RCTs (8) was published comparing nologies across models, and differences
complications state (e.g., from retinopa-

Downloaded from [Link] by Sandip Mitra on 27 June 2023

the efficacy of CGM, including isCGM, in sensor efficacy and reliability are not
thy to blindness), remain there, or die
with SMBG. An updated literature included. We assumed an annual cost of
due to the increased risk of death from
search in June 2021 did not reveal any CAD 2,019, 3,930, and 2,540 for SMBG
their complication or other causes. Up to
additional randomized trials that were (average of six tests per day), CGM
two complications could also be com-
relevant for this analysis. To ensure gen- (based on the average annual cost of the
bined and at any point in the model, an
eralizability of other study results to our Dexcom G6 and Medtronic Guardian
acute SH or DKA event could occur
study, we excluded studies based on Connect), and isCGM (based on the
(Supplementary Fig. A1). Of particular
isCGM, sensor-augmented pumps, pedi- FreeStyle Libre), respectively. Full details
note, while previous analyses included
atric patients, and patients with type 2 can be found in Appendix A3. Complica-
SH events, this cost-effectiveness study is
diabetes to estimate the efficacy of tion costs were based on the sources
the first with incorporation of the cost
CGM versus SMBG (Supplementary presented by OH (13) in 2018 and in-
and QALY impact of DKA events. Costs
Table A2). This yields an estimated im-
for SH or DKA events are added to the flated by 4.29% (24) to account for infla-
provement in TIR of 1.48 h/day (95% CI tion in 2021 Canadian dollars (CAD)
individual’s annual cost along with a cor-
0.77–2.20) based on five studies. As ex- responding reduction in QALY for that (Supplementary Table A12).
pected, due to the variety of patient specific year. For simplicity, the occur-
characteristics, interventions, and study rence of both an SH and DKA event in a Outcomes
timelines, heterogeneity was moderate single year was excluded from the We examined two primary outcomes: 1)
to substantial, with an I2 (21) of 69% model. Given the novel inclusion of DKA the modeled absolute number of com-
(Supplementary Fig. A2). Regarding events in our model, a detailed literature plications and deaths that may be pre-
isCGM, only one RCT (18) examining search was performed to determine the vented by national use of CGM and
adults in Europe has been performed. most appropriate estimate of the rate of isCGM for adults with T1D and 2) the
Other isCGM studies were included in DKA events, costs, and reductions of cost-effectiveness of CGM and isCGM
their review (8) but included people with QALY in our population. The baseline relative to SMBG. For cost-effectiveness,
type 2 diabetes and samples overlapping DKA and SH event rates were obtained we calculated the incremental cost-ef-
with Bolinder et al. (18) and are thus ex- from Pettus et al. (22), as their study fectiveness ratio (ICER) of CGM versus
cluded. Bolinder et al. (18) show an in- was based on a large sample of 31,430 SMBG and isCGM versus SMBG. We
crease in TIR of 1.00 h/day (0.43–1.57) adults with T1D in the U.S., with 85% used the traditional willingness-to-pay
(Supplementary Table A3). To ensure that in our target age range of 18–64 years. threshold of CAD 50,000/QALY as our
our cost-effectiveness analysis is conserva- Full details for both SH and DKA events cost-effectiveness threshold.
tive, we used the lower bounds of the can be found in Appendix A4. Baseline
95% CIs for TIR (i.e., an improvement in annual transition probabilities of these Statistical Analysis
TIR of 0.77 and 0.43 h/day for CGM and parameters, and associated references All analyses were performed with the
isCGM, respectively) to estimate the im- for the SMBG group, are included in heemod package (25) in R (26) and are
pact of CGM and isCGM on our baseline Supplementary Table A7. Age-dependent from the perspective of a single-payer
complication rates. We note that long- CVD, CVD mortality, and other mortality health care system. Note that cost esti-
term studies examining the relation- rates are presented in Supplementary mates include direct publicly funded
ships between glycemic control and Tables A8–A10, respectively. health diabetes monitoring and compli-
micro- and macrovascular complications QALYs were obtained from the most cation costs and do not include societal
have traditionally relied on HbA1c. Where recent available resources (Supple- costs (e.g., lost income or productivity).
required, we adopt the assumption mentary Table A11) and are consistent Consistent with the parameters of the
that a 10% increase in TIR is approxi- with values in the most recent cost- OH (13) report, we assumed a baseline
mately equal to a 0.8% absolute re- effectiveness analysis of CGM for T1D average HbA1c of 8.1% (27) and an an-
duction in HbA1c for our model to in the U.K. (23). Note that when an indi- nual discount rate of 1.5% for both
estimate the impact of CGM and vidual experiences multiple acute or costs and QALYs. These discount rates
isCGM on reducing these complication chronic complications disutilities are ad- are based on the Canadian standards
rates (Supplementary Tables A4–A6). ditive. We obtained intervention costs for cost-effectiveness models, which
[Link]/care Rotondi and Associates 2015

A B

C D

Downloaded from [Link] by Sandip Mitra on 27 June 2023

Figure 1—Number of people over a 20-year horizon with no complications (A), who died (B), with minor complications (C) or with major complica-
tions (D).

assume cost discounts equal to the ap- CGM and isCGM by 10% and 25% to living without any complications of
proximate long-term borrowing costs represent potential bulk purchase agree- T1D (Fig. 1A) and prevents approximately
for the Canadian government (28) and ments. Additional details regarding 4,600 deaths (Fig. 1D). Figure 1B and
equal discount rates for QALYs. sensitivity analyses are included in C must be interpreted cautiously, as
Appendix B. we predict a larger number of individ-
Sensitivity Analyses uals living with minor complications
To ensure the robustness of our model, RESULTS with CGM and isCGM as individuals
we performed a variety of one-way sen- Modeled Number of Complications remain alive and live with fewer severe
sitivity analyses. They include variation and Deaths complications throughout the study pe-
in individual model parameters accord- Figure 1 presents the general trends of riod. Additional figures (Supplementary
ing to their plausible range (Supple- no complications, minor complications Figs. C1–C3) presenting the number of
mentary Table A7), cost and QALY dis- (neuropathy, nephropathy, retinopathy), people in each individual complication
counts of 0%–5%, incorporating a more major complications (CVD, end-stage re- state can be found in Appendix C.
liberal estimate of the efficacy of CGM nal disease, lower-extremity amputation,
and isCGM with use of the point esti- blindness), and death. After 20 years, if Cost-effectiveness of CGM and
mates and upper bounds of the 95% CIs the entire Canadian population uses CGM, isCGM
for TIR instead of our conservative esti- 7,400 more people are living with- Table 1 contains the cost, number of
mate with use of the lower bound, out complications compared with use QALYs, and ICERs for each of the three
comparing CGM and isCGM with a re- of SMBG (Fig. 1A). Similarly, the num- interventions. These results provide evi-
vised recommended SMBG regimen of ber of deaths is reduced by 11,500 dence that both CGM and isCGM are
8 or 10 tests per day (vs. 6 tests per (Fig. 1D). In comparison with SMBG, cost-effective interventions at the tradi-
day), and reducing the annual cost of isCGM keeps 3,400 more Canadians tional willingness-to-pay threshold of
CAD 50,000/QALY. We note that while
the costs of funding CGM and isCGM
Table 1—Cost and QALY projection and ICERs for an initial cohort of 180,000
are higher, they generate significant
adult Canadians using SMBG, CGM, and isCGM over a 20-year horizon cost savings due to lower costs of com-
Total cost Complications Interventions
plications of T1D. Finally, note that the
Strategy (CAD) cost (CAD) cost (CAD) QALY ICERa government cost for these technologies
SMBG 12,166,922,680 7,142,676,195 5,024,246,485 2,062,023 — in Canada would be much lower as
some individuals already benefit from
CGM 16,080,940,460 5,862,493,277 10,218,447,183 2,173,798 35,017
employer-sponsored health plans and
isCGM 12,981,653,727 6,548,934,955 6,432,718,772 2,108,612 17,488 national funding for these technologies
a
CAD/QALY (relative to SMBG). would likely reduce costs due to bulk
purchase agreements.
2016 Cost-effectiveness of CGM and isCGM in Canada Diabetes Care Volume 45, September 2022

Sensitivity Analyses SMBG over a 20-year horizon while pro- in plausible parameters, common cost
The impact of departures from various ducing a higher total number of QALY. and QALY discounts, and measures of
model assumptions is presented in Figs. Similarly, if Canadians are assumed to efficacy. These results support use of
2 and 3. Variation in the rate of DKA be using a larger number of SMBG tests public health resources to increase uni-
events has the largest impact on our per day (i.e., 8 or 10), isCGM is cost-ef- versal access to CGM and isCGM for the
observed ICERs, but this is due to the fective at any willingness-to-pay thresh- adult T1D population in Canada and
comparatively large degree of uncer- old, as the increased costs of the emphasize the need for greater equity
tainty in this parameter. Moreover, even additional daily SMBG tests exceed the in technology adoption.
considering the extreme case, the low- annual cost of isCGM and yield a lower In the Canadian context, an earlier
est rates of DKA events produce an total QALY. Based on these results, our cost-effectiveness analysis in Ontario,
ICER of approximately CAD 53,021/ model is reasonably robust against de- that included data up to January 2017,
QALY for CGM, which nears our thresh-
partures from our initial assumptions. did not find sufficient evidence for the
old of CAD 50,000/QALY (Fig. 2). Simi-
public funding of CGM for people with
larly, a QALY discount of 5% also

Downloaded from [Link] by Sandip Mitra on 27 June 2023

CONCLUSIONS T1D (13). However, this analysis was
produces a slightly elevated ICER of
CAD 51,086 compared with our cost-ef- Our results show that compared with based on earlier estimates of efficacy
fectiveness threshold for CGM. In all SMBG, universal adoption of CGM or and higher costs, prior to the reporting
considered cases, isCGM meets our isCGM is predicted to reduce diabetes- of updated evidence of effectiveness
cost-effectiveness threshold (Fig. 3). Of related complications and mortality and the reduction in the absolute costs
particular note, if bulk government pur- over a 20-year time horizon, and they of these technologies. In a recent cost-
chases decrease the cost of isCGM by are cost-effective strategies for monitor- effectiveness study (29) where investi-
25%, this technology is cost-effective at ing glycemia in Canadian adults living gators compared the Dexcom G6 with
any willingness-to-pay threshold, as the with T1D. Results are consistent across SMBG in Canada using the IQVIA Core
total cost of isCGM is lower than that of multiple assumptions including variation Diabetes Model (IQVIA CDM) (30),

Figure 2—Sensitivity analyses examining the impact of various assumptions on ICERs for CGM. Baseline model of cost-effectiveness ICER of CAD
35,017/QALY and willingness-to-pay threshold of CAD 50,000/QALY are indicated. ESRD, end-stage renal disease; LEA, lower extremity
amputation.
[Link]/care Rotondi and Associates 2017

Downloaded from [Link] by Sandip Mitra on 27 June 2023

Figure 3—Sensitivity analyses examining the impact of various assumptions on ICERs for isCGM. Baseline model of cost-effectiveness ICER of CAD
17,488/QALY and willingness-to-pay threshold of CAD 50,000/QALY are indicated. ESRD, end-stage renal disease; LEA, lower extremity amputa-
tion. ***Model is cost-effective at any willingness-to-pay threshold.

results show that the Dexcom G6 is a relative to SMBG at any willingness-to- in July 2021 the U.S. relaxed its qualifi-
cost-effective glucose monitoring sys- pay threshold. In addition, we note cation rules for CGM to encompass
tem with an ICER of CAD 16,931/QALY. that the costs for these devices for the nearly all adults with T1D receiving
However, this study did not include health payer in Canada would be mark- Medicare benefits. Our results may also
consideration of isCGM and was finan- edly lower, as many individuals already be of interest to private health insurers
cially supported by Dexcom, raising have isCGM or CGM devices funded by in the U.S. and other jurisdictions, as
concerns of potential conflicts of inter- their employer-sponsored private ben- we have shown that isCGM is in fact a
est. Similarly, while moderate cost-ef- efits plans. lower-cost glucose monitoring regimen
fectiveness has been shown for isCGM Internationally, a 2018 study on the compared with SMBG when individuals
in Ontario (31) and Quebec (32), our cost-effectiveness of CGM in Spain did are averaging 8 or 10 tests per day,
updated results support their coverage not show sufficient cost-effectiveness while improving health outcomes. How-
across Canada. Canadian health care given the higher technology costs at the ever, it should also be stressed that only
spending decisions are independently time (14). However, our findings are CGM technology, and not isCGM, can
determined by the 10 provincial and 3 consistent with the most recent studies be used to inform sensor-augmented
territorial governments; however, our of CGM and isCGM use in the U.K (23), and hybrid closed loop insulin pump
results support cooperative and coordi- France (33), and the U.S (34), while systems for more customized insulin de-
nated national approaches to health isCGM alone has been shown to be a livery to help minimize both hyper- and
care spending, which could generate cost-effective intervention in a study hypoglycemia in T1D (38).
further cost savings due to bulk gov- based in Sweden (35). These more re- Unique strengths of our analysis are
ernment purchases. This is best illus- cent findings have led to the inclusion the consideration of current device costs,
trated in our sensitivity analysis, where of these technologies in public health the incorporation of DKA events in the
we showed that a 25% reduction in plans across the world, including Spain Markov model, use of TIR as a measure
the cost of isCGM is cost-effective (36), France (33), Italy (37), and notably, of efficacy, and our examination of both
2018 Cost-effectiveness of CGM and isCGM in Canada Diabetes Care Volume 45, September 2022

CGM and isCGM independently, without should focus on the evaluation of the model interpretation, framing the results
sponsorship from device manufacturers. cost-effectiveness of these technologies within the literature and public health con-
text, and revised the work for critical context
However, there are potential limitations. for children and youth, given their in- and clarity. All authors contributed to critical
First, cost estimates for the nine compli- creased risk for poor glycemic control revision of the first draft and all subsequent
cation states may be underestimated, as leading to acute SH or DKA events. In ad- drafts of the manuscript. M.A.R. is the guar-
health care costs have increased beyond dition, in our review here we highlight antor of this work and, as such, had full ac-
typical rates of inflation (39). This may cess to all the data in the study and takes
the comparatively small body of evi-
responsibility for the integrity of the data and
lead to an underestimation of compara- dence from RCTs for isCGM relative to the accuracy of the data analysis.
tive cost-effectiveness of CGM and the number of studies for CGM. More
isCGM. Secondly, cost estimates have broadly, long-term clinical studies evalu- References
been primarily based on Ontario data, ating the risk of micro- and macrovascu- 1. Fowler MJ. Microvascular and macrovascular
and absolute costs may not generalize to lar complications based on TIR would complications of diabetes. Clin Diabetes 2008;
all Canadian provinces. However, we ex- also be of use to formally quantify the 26:77–82
2. Seaquist ER, Anderson J, Childs B, et al.
pect that the estimation of relative cost- relationship between TIR and these ad-

Downloaded from [Link] by Sandip Mitra on 27 June 2023

Hypoglycemia and diabetes: a report of a
effectiveness between technologies will verse outcomes. workgroup of the American Diabetes Association
remain unbiased. Third, in implementing In conclusion, both CGM and isCGM and the Endocrine Society. Diabetes Care 2013;
estimates of efficacy from randomized are cost-effective tools for the manage- 36:1384–1395
clinical trials, we assume that estimates 3. Rehni AK, Dave KR. Impact of hypoglycemia
ment of T1D in the Canadian adult pop-
on brain metabolism during diabetes. Mol
are durable over the 20-year horizon. ulation. In addition to their ease of use Neurobiol 2018;55:9075–9088
Fourth, we examined a hypothetical and ability to improve day-to-day living 4. Lu CL, Shen HN, Hu SC, Wang JD, Li CY. A
baseline population without consider- with T1D, their widespread adoption is population-based study of all-cause mortality
ation of population growth or new-onset anticipated to lead to a significant re- and cardiovascular disease in association with
T1D during the 20-year examination pe- prior history of hypoglycemia among patients
duction in risk of developing SH, DKA,
with type 1 diabetes. Diabetes Care 2016;39:
riod. Fifth, we assumed that the efficacy and micro- or macrovascular complica- 1571–1578
of a new isCGM technology that includes tions and acceptable cost-effectiveness 5. Price D, Walker T. The rationale for continuous
alarms to indicate hyper- and hypoglyce- compared with SMBG. As these technol- glucose monitoring-based diabetes treatment
mia without a cost increase (FreeStyle Li- ogies further improve and costs con- decisions and non-adjunctive continuous glucose
bre 2; Abbott) is comparable with that monitoring use. Eur Endocrinol 2016;12:24–30
tinue to decrease, these results suggest 6. Battelino T, Danne T, Bergenstal RM, et al.
of its earlier version. As a result, our results that individuals with T1D should be Clinical targets for continuous glucose monitor-
may underestimate cost-effectiveness of strongly supported by public health pro- ing data interpretation: Recommendations from
newer isCGM technology. Additionally, grams and insurers to choose and im- the International Consensus on Time in Range.
we emphasize that our results are appli- plement either technology. Diabetes Care 2019;42:1593–1603
7. Evans M, Welsh Z, Ells S, Seibold A. The
cable to adults with T1D aged 18–64
impact of flash glucose monitoring on glycaemic
years and that we cannot extrapolate control as measured by HbA1c: A meta-analysis
these estimates of cost-effectiveness to Funding. This study was funded through a of clinical trials and real-world observational
children or older adults. Due to lack of fee-for-service consulting agreement between studies. Diabetes Ther 2020;11:83–95
population-level data, we could not in- M.A.R. and JDRF Canada. 8. Maiorino MI, Signoriello S, Maio A, et al. Effects
The funders had a role in the study design of continuous glucose monitoring on metrics of
clude other clinical factors in our model, and provided feedback on the manuscript glycemic control in diabetes: a systematic review
such as age of onset of T1D, which may draft but had no role in data collection or with meta-analysis of randomized controlled trials.
affect risk of complications. Similarly, lack data analysis or interpretation. Diabetes Care 2020;43:1146–1156
of long-term clinical data limits formal Duality of Interest. M.C.R. has received 9. Rouhard S, Buysschaert M, Alexopoulou O,
validation of our Markov cost-effective- speaker honoraria from Medtronic Diabetes, Preumont V. Impact of flash glucose monitoring
Insulet Corporation, Ascensia Diabetes Care, on glycaemic control and quality of life in
ness model, but our detailed sensitivity Novo Nordisk (through JDRF T1D Performance patients with type 1 diabetes: a 18-month follow-
analyses show that our conclusions re- in Exercise and Knowledge [PEAK] program), up in real life. Diabetes Metab Syndr 2020;
main robust against various departures Xeris Pharmaceuticals, Lilly Diabetes, and Lilly 14:65–69
from model assumptions. Moreover, our Innovation. B.P. has received speaker hono-

10. Lind M, Olafsd
ottir AF, Hirsch IB, et al.
results are broadly consistent with the raria from Abbott, Medtronic, Insulet, and Sustained intensive treatment and long-term
Novo Nordisk and research support to his re- effects on HbA1c reduction (SILVER study) by
results of Roze et al. (29), who used the search institute from Boehringer Ingleheim, CGM in people with type 1 diabetes treated with
externally validated IQVIA CDM (40), pro- Novo Nordisk, and the Bank of Montreal. No MDI. Diabetes Care 2021;44:141–149
viding additional confidence in our ap- other potential conflicts of interest relevant 11. Charleer S, De Block C, Van Huffel L, et al.
proach. Finally, we acknowledge that the to this article were reported. Quality of life and glucose control after 1 year of
evidence base of trials examining the im- Author Contributions. M.A.R. developed nationwide reimbursement of intermittently
the Markov cost-effectiveness model, per- scanned continuous glucose monitoring in adults
pact of CGM do not include consideration formed analyses, and wrote the initial draft living with type 1 diabetes (FUTURE): a pro-
of the additional benefits conferred by of the manuscript. O.W. supported the analy- spective observational real-world cohort study.
use of CGM in automated insulin delivery sis methods, including model coding; per- Diabetes Care 2020;43:389–397
systems. We expect that the global im- formed preliminary literature searching for 12. Parkin CG, Graham C, Smolskis J. Continuous
pact of CGM will be amplified beyond cost and effectiveness data; and wrote por- glucose monitoring use in type 1 diabetes:
tions of RESEARCH DESIGN AND METHODS and appen- longitudinal analysis demonstrates meaningful
the cost-effectiveness estimates reported dices. M.R. provided substantial contributions improvements in HbA1c and reductions in health
here with greater adoption of automated to study design and revised the draft for criti- care utilization. J Diabetes Sci Technol 2017;
insulin delivery systems. Future studies cal content. B.P. provided valuable insight into 11:522–528
[Link]/care Rotondi and Associates 2019

13. Health Quality Ontario. Continuous moni- 22. Pettus JH, Zhou FL, Shepherd L, et al. 32. Jobin N, Arbour S. Systeme flash de survei-
toring of glucose for type 1 diabetes: a health Incidences of severe hypoglycemia and diabetic llance du glucose (FreeStyle Libre). Quebec,
technology assessment. Ont Health Technol ketoacidosis and prevalence of microvascular Canada, Abbott, 2018, p. 34
Assess Ser 2018;18:1–160 complications stratified by age and glycemic 33. Roze S, Isitt JJ, Smith-Palmer J, et al. Long-
14. Garc
ıa-Lorenzo B, Rivero-Santana A, Vallejo- control in U.S. adult patients with type 1 term cost-effectiveness of Dexcom G6 real-time
Torres L, et al. Cost-effectiveness analysis of real- diabetes: a real-world study. Diabetes Care 2019; continuous glucose monitoring system compared
time continuous monitoring glucose compared 42:2220–2227 with self-monitoring of blood glucose in people
to self-monitoring of blood glucose for diabetes 23. Roze S, Isitt J, Smith-Palmer J, Javanbakht M, with type 1 diabetes in France. Diabetes Ther
mellitus in Spain. J Eval Clin Pract 2018;24: Lynch P. Long-term cost-effectiveness of Dexcom 2021;12:235–246
772–781 G6 real-time continuous glucose monitoring 34. Wan W, Skandari MR, Minc A, et al. Cost-
15. McQueen RB, Breton MD, Ott M, Koa H, versus self-monitoring of blood glucose in effectiveness of continuous glucose monitoring for
Beamer B, Campbell JD. Economic value of patients with type 1 diabetes in the U.K. Diabetes
adults with type 1 diabetes compared with self-
improved accuracy for self-monitoring of blood Care 2020;43:2411–2417
monitoring of blood glucose: the DIAMOND
glucose devices for type 1 diabetes in Canada. J 24. Bank of Canada. Inflation calculator, 2020.
randomized trial. Diabetes Care 2018;41:1227–1234
Diabetes Sci Technol 2015;10:366–377 Accessed 19 July 2021. Available from https:/www.
35. Bilir SP, Hellmund R, Wehler B, Li H, Munakata
16. Haak T, Hanaire H, Ajjan R, Hermanns N, [Link]/rates/related/inflationcalculator/
J, Lamotte M. Cost-effectiveness analysis of a flash

Downloaded from [Link] by Sandip Mitra on 27 June 2023

Riveline JP, Rayman G. Use of flash glucose- 25. Filipovi
c-Pierucci A, Zarca K, Durand-Zaleski I.
Markov models for health economic evaluation: glucose monitoring system for patients with type 1
sensing technology for 12 months as a diabetes receiving intensive insulin treatment in
replacement for blood glucose monitoring in the R package heemod. 25 April 2017 [preprint].
arXiv:1702.03252v2 Sweden. Eur Endocrinol 2018;14:73–79
insulin-treated type 2 diabetes. Diabetes Ther 36. La Moncloa. Official site for the President of
26. R Core Team. R: a language and environment
2017;8:573–586 the Government of Spain and the Council of
for statistical computing. Vienna, Austria, 2020.
17. Battelino T, Conget I, Olsen B, et al.;
Accessed 8 February 2021. Available from Ministers. The Interterritorial Council agrees to
SWITCH Study Group. The use and efficacy of
[Link] the creation of the Digital Health Commission,
continuous glucose monitoring in type 1
27. Aronson R, Brown RE, Abitbol A, et al. The which will contribute to co-governance with the
diabetes treated with insulin pump therapy: a
Canadian LMC Diabetes Registry: a profile of the Regional Governments and interoperability of all
randomised controlled trial. Diabetologia 2012;
demographics, management, and outcomes of projects in this area. June 20, 2021. Accessed 21
55:3155–3162
individuals with type 1 diabetes. Diabetes September 2021. Available from [Link]
18. Bolinder J, Antuna R, Geelhoed-Duijvestijn P,
Technol Ther 2021;23:31–40 [Link]/lang/en/gobierno/news/
Kr€oger J, Weitgasser R. Novel glucose-sensing
28. Attema AE, Brouwer WBF, Claxton K. Paginas/2021/20210630_digital-[Link]
technology and hypoglycaemia in type 1 Discounting in economic evaluations. Pharmaco-
diabetes: a multicentre, non-masked, randomised 37. Seidel D, Boggio Mesnil F, Caruso A. Reim-
Economics 2018;36:745–758 bursement pathways for new diabetes techno-
controlled trial. Lancet 2016;388:2254–2263 29. Roze S, Isitt JJ, Smith-Palmer J, Lynch P.
19. Heinemann L, Freckmann G, Ehrmann D, logies in Europe: top-down versus bottom-up. J
Evaluation of the long-term cost-effectiveness of Diabetes Sci Technol 2019;13:118–122
et al. Real-time continuous glucose monitoring in the Dexcom G6 continuous glucose monitor
adults with type 1 diabetes and impaired hypo- 38. Weisman A, Bai J-W, Cardinez M, Kramer CK,
versus self-monitoring of blood glucose in people
glycaemia awareness or severe hypoglycaemia Perkins BA. Effect of artificial pancreas systems on
with type 1 diabetes in Canada. Clinicoecon
treated with multiple daily insulin injections glycaemic control in patients with type 1
Outcomes Res 2021;13:717–725
(HypoDE): a multicentre, randomised controlled diabetes: a systematic review and meta-analysis
30. Palmer AJ, Roze S, Valentine WJ, et al. The
trial. Lancet 2018;391:1367–1377 CORE Diabetes Model: projecting long-term of outpatient randomised controlled trials. Lancet
20. Beck RW, Bergenstal RM, Riddlesworth TD, clinical outcomes, costs and cost-effectiveness of Diabetes Endocrinol 2017;5:501–512
et al. Validation of time in range as an outcome interventions in diabetes mellitus (types 1 and 2) 39. Aon Public Limited Company. 2021 Global
measure for diabetes clinical trials. Diabetes Care to support clinical and reimbursement decision- medical trend rates report. Accessed 19 October
2019;42:400–405 making. Curr Med Res Opin 2004;20(Suppl. 1): 2021. Available from [Link]
21. Deeks JJ, Higgins JPT, Altman D. Analysing S5–S26 [Link]/research/2021-global-medical-
data and undertaking meta-analyses: In Cochrane 31. Ontario Health (Quality). Flash glucose trend-rates-report
Handbook for Systematic Reviews of Interventions. monitoring system for people with type 1 or 40. Palmer AJ, Roze S, Valentine WJ, et al.
Higgins JPT, Thomas J, Chandler J, et al., Eds. type 2 diabetes: a health technology assess- Validation of the CORE Diabetes Model against
Chichester, U.K., John Wiley & Sons, 2019, pp. ment. Ont Health Technol Assess Ser 2019; epidemiological and clinical studies. Curr Med
241–284 19:1–108 Res Opin 2004;20(Supp. 1):S27–S40