5 Outcomes

The Diagnostics Advisory Committee (section 10) considered evidence from a number of sources (section 11). Full details of all the evidence are in the committee papers.

How outcomes were assessed

5.1 The assessment consisted of a systematic review of the clinical‑effectiveness data for the MiniMed Paradigm Veo system, the Vibe and G4 PLATINUM CGM system and comparator technologies.

5.2 In total, 54 publications reporting the results of 19 studies met the inclusion criteria. These studies included either the intervention or comparator technologies in a treatment arm. Two studies included data for the MiniMed Paradigm Veo system, 1 of which compared the system with an integrated sensor‑augmented pump therapy system without a low‑glucose suspend function (included as a clinical proxy for the Vibe and G4 PLATINUM CGM system in the absence of any data for this technology), and a further 7 studies included data for the integrated sensor‑augmented pump therapy system without a low‑glucose suspend function. The remainder of the studies reported data for capillary blood testing with continuous subcutaneous insulin and capillary blood testing with multiple daily insulin injections. No studies reported data for either continuous glucose monitoring with continuous subcutaneous insulin infusion (non‑integrated devices) or continuous glucose monitoring with multiple daily insulin injections. Of the 19 included studies:

  • 10 included adults only

  • 3 included children only

  • 3 included a mixed population (adults and children) but did not report data for each group separately

  • 2 included a mixed population and reported data for adults and children separately

  • 1 included pregnant women only.

5.3 The 1 study that included pregnant women only reported data for capillary blood testing with continuous subcutaneous insulin infusion and capillary blood testing with multiple daily insulin injections. Because no comparative data were found to assess the clinical effectiveness of the integrated sensor‑augmented pump therapy systems in pregnant women, this study was not included in the analyses.

5.4 All the studies were randomised controlled trials. The methodological quality of each study was appraised using the Cochrane risk of bias tool. Eleven of the 19 studies were rated as a high risk of bias, primarily because the patients, clinicians and assessors were not blinded to the allocation of interventions and glycated haemoglobin (HbA1c) results were interpreted with knowledge of the treatment allocation. Of the remaining 8 studies, 4 were rated as unclear risk of bias and 4 were rated as low risk of bias.

5.5 There was substantial heterogeneity in the populations included in the studies. Nine studies included people who had not used an insulin pump before, and only 4 studies reported including people who had experienced hypoglycaemia before the trial.

5.6 The results of the studies were presented as a narrative synthesis and combined into network meta‑analyses where possible. Direct head‑to‑head meta‑analyses were done using a fixed‑effect model unless significant heterogeneity was observed. Indirect meta‑analyses were done according to the method devised by Bucher et al. (1997).

Clinical effectiveness in adults

5.7 Twelve studies reported data for adults: 10 studies done solely in adults and 2 studies reported subgroup data for adults.

MiniMed Paradigm Veo system

5.8 One study (ASPIRE in‑home) compared the MiniMed Paradigm Veo system with an integrated sensor‑augmented pump therapy system (no low‑glucose suspend) at 3‑month follow‑up in adults with type 1 diabetes. This study included people who had experienced 2 or more nocturnal hypoglycaemic events during the study run‑in phase, but excluded people who had experienced more than 1 episode of severe hypoglycaemia in the 6 months before study recruitment. The study reported that hypoglycaemic events occurred less often in the MiniMed Paradigm Veo system group (3.3±2.0 weekly events per patient compared with 4.7±2.7 weekly events per patient; p<0.001), and this effect was consistent when the results were restricted to nocturnal hypoglycaemic events (1.5±1.0 weekly events per patient compared with 2.2±1.3 weekly events per patient; p<0.001). The study also reported that, for the MiniMed Paradigm Veo system, the mean hypoglycaemic area under the curve (AUC; derived from the magnitude and severity of the sensor‑measured glucose level) was significantly lower (less severe) for all hypoglycaemic events combined (p<0.001) and for nocturnal hypoglycaemia (p<0.001). There were no statistically significant differences in change in HbA1c, capillary blood glucose values, insulin use, diabetic ketoacidosis, quality of life, device‑related serious adverse events, or death.

5.9 Data from the ASPIRE in‑home study were used in a network analysis to compare the MiniMed Paradigm Veo system with:

  • the integrated sensor‑augmented pump therapy system with no low‑glucose suspend

  • capillary blood testing with continuous subcutaneous insulin infusion

  • capillary blood testing with multiple daily insulin injections.

    None of the 3 additional studies incorporated into the network analysis reported whether they included people who had experienced hypoglycaemia. The network analysis included change in HbA1c and diabetic ketoacidosis at 3‑month follow‑up as outcomes. No statistically significant differences were seen in any of the comparisons.

Integrated sensor‑augmented pump therapy system (no low‑glucose suspend)

5.10 Five further studies included integrated sensor‑augmented pump therapy in a treatment arm. One study (Hirsch et al. 2008) compared an integrated sensor‑augmented pump therapy system (no low‑glucose suspend) with capillary blood testing and continuous subcutaneous insulin infusion. This study did not exclude people with hypoglycaemia unawareness. The study reported no statistically significant difference in change in HbA1c (%) between the groups at 6‑month follow‑up (−0.0364%; standard error 0.1412; p=0.80).

5.11 The remaining 4 studies (Hermanides et al. 2011; Lee et al. 2007; Peyrot et al. 2009; Bergenstal et al. 2010) compared the integrated sensor‑augmented pump therapy system (no low‑glucose suspend) with capillary blood testing combined with multiple daily insulin injections. Inclusion or exclusion criteria for hypoglycaemia were not stated in 3 of these studies. Bergenstal et al. (2010) excluded people with hypoglycaemia unawareness. The studies reported multiple outcomes at various follow‑up points.

5.12 Three‑month follow‑up (2 studies): 1 study (Lee et al. 2007) reported a statistically significant difference in the change in HbA1c (%) in favour of the integrated sensor‑augmented pump therapy system (no low‑glucose suspend; −0.97; p=0.02). This difference was not statistically significant in Peyrot et al. (2009; −0.69; p=0.071). No statistically significant differences were found for hypoglycaemic events, diabetic ketoacidosis or serious adverse events in either study.

5.13 Six‑month follow‑up (1 study): Hermanides et al. (2011) reported no statistically significant difference for hypo‑ or hyperglycaemic events. Statistically significant differences in favour of the integrated sensor‑augmented pump therapy system were found for the following outcomes:

  • change in HbA1c % (−1.1; 95% confidence interval [CI] −1.47 to −0.73)

  • number of people with HbA1c ≤7% (53 mmol/mol; 14/41 compared with 0/36; p<0.001)

  • daily insulin use (difference of −11.0 units per day; 95% CI −16.1 to −5.9; p<0.001)

  • quality of life measured by the SF‑36 (difference of 7.9; 95% CI 0.5 to 15.3; p=0.04).

5.14 Twelve‑month follow‑up (1 study, excluded people with hypoglycaemia unawareness): Bergenstal et al. (2010) reported no statistically significant differences for hypoglycaemic AUC, severe hypoglycaemia or diabetic ketoacidosis. Statistically significant differences in favour of the integrated sensor‑augmented pump therapy system (no low‑glucose suspend) were seen for the following outcomes:

  • change in HbA1c % (−0.6; 95% CI −0.8 to −0.4; p<0.001)

  • number of people with HbA1c <7% (53 mmol/mol; 57/166 compared with 19/163; p<0.001)

  • hyperglycaemic AUC (3.74 compared with 7.38; p<0.001)

  • improved quality of life measured by the SF‑36 (difference of 3; 95% CI 1.36 to 4.64)

  • fear of hypoglycaemia measured by the Hypoglycaemia Fear Survey (difference of −6.5; 95% CI −9.76 to −3.27).

5.15 All 5 studies were incorporated into several network analyses that were done to calculate effect estimates for the integrated sensor‑augmented pump therapy system (no low‑glucose suspend). The results of the analyses suggested that there was a statistically significant reduction in HbA1c % (weighted mean difference −1.10; 95% CI −1.46 to −0.74), and a statistically significant difference in the proportion of people with HbA1c <7% (53 mmol/mol; relative risk 25.55; 95% CI 1.58 to 413.59) in favour of the integrated sensor‑augmented pump therapy system (no low‑glucose suspend) when compared with capillary blood testing with multiple daily insulin injections. Quality of life (measured by the Diabetes Treatment Satisfaction Questionnaire) associated with integrated sensor‑augmented pump therapy was also significantly improved when compared with both capillary blood testing with continuous subcutaneous insulin infusion (weighted mean difference 5.90; 95% CI 2.22 to 9.58) and capillary blood testing with multiple daily insulin injections (weighted mean difference 8.60; 95% CI 6.28 to 10.92).

Supplementary data

5.16 In addition to the studies incorporated into the network analyses, 2 observational studies and 1 randomised cross‑over study were included in a supplementary narrative analysis carried out by the External Assessment Group. The SWITCH study reported a randomised cross‑over study to assess the clinical effectiveness of an integrated sensor‑augmented pump therapy system. The study included 81 adults and 72 children, from 8 European sites, who used continuous subcutaneous insulin infusion. All patients had an integrated sensor‑augmented pump therapy system with an activated sensor, and were randomised to alternate sequences of sensor use over a 12‑month period. The study concluded that using the sensor was associated with a reduction in HbA1c (−0.43%; 95% CI −0.32 to −0.55) and a decrease in the time spent in hypoglycaemia.

5.17 Choudhary et al. (2011) reported a study designed to assess the low‑glucose suspend function of the MiniMed Paradigm Veo system over a 3‑week period in 28 adults from 6 UK centres. The study had a 2‑week run‑in period with the low‑glucose suspend function deactivated, after which patients were divided into 4 groups according to their duration of hypoglycaemic events. The study concluded that using low‑glucose suspend was associated with a significant reduction in duration of nocturnal hypoglycaemia in the group with the highest duration of hypoglycaemia during the run‑in period (mean 75.1±54 compared with 10.2±18 minutes per day; p=0.02).

5.18 Choudhary et al. (2013) reported a retrospective audit of 35 adults attending a specialist clinic, with established problematic hypoglycaemia or impaired awareness of hypoglycaemia while on optimal medical therapy. The patients were given a continuous glucose monitor in addition to either continuous subcutaneous insulin infusion or multiple daily injections. Outcomes were audited after 12 months. Of the 35 patients, 23 used the MiniMed Paradigm Veo system and 3 used the Vibe and G4 PLATINUM CGM system. The audit reported that the median rate of severe hypoglycaemia was reduced from 4.0 (interquartile range 0.75 to 7.25) episodes per patient‑year at baseline to 0.0 (interquartile range 0.0 to 1.25) episodes per patient‑year at 12 months (p<0.001). HbA1c levels were also reduced from 8.1±1.2% to 7.8±1.0% at 12 months (p=0.007). The final mean HbA1c level and median severe hypoglycaemia rate did not differ between patients who used the system with low‑glucose suspend and those who did not.

Clinical effectiveness in children

5.19 Six studies reported data for children, including 1 study (Ly et al. 2013) that reported data for the MiniMed Paradigm Veo system. Of the remaining 5 studies, 2 reported data for an integrated sensor‑augmented pump therapy system without low‑glucose suspend (used as a proxy for the Vibe and G4 PLATINUM CGM system), and 3 reported data for comparator technologies only (capillary blood testing with continuous subcutaneous insulin infusion and capillary blood testing with multiple daily insulin injections).

5.20 One study (Ly et al. 2013) reported results for the MiniMed Paradigm Veo system compared with capillary blood testing and continuous subcutaneous insulin infusion at 6‑month follow‑up in a mixed population aged 4–50 years. Seventy per cent of the patients were aged under 18 years, and people with an impaired awareness of hypoglycaemia were included. The study reported a statistically significant difference in the rate of hypoglycaemic events, with a lower rate of events in the MiniMed Paradigm Veo system group (incidence rate ratio 3.6; 95% CI 1.7 to 7.5; p<0.001). No statistically significant differences were reported for change in HbA1c, the number of people experiencing hypoglycaemic events or the hypoglycaemia unawareness score.

5.21 One study (Hirsch et al. 2008) compared an integrated sensor‑augmented pump therapy system (no low‑glucose suspend) with capillary blood testing and continuous subcutaneous insulin infusion. The study included data for the change in HbA1c (%) at 6 months and found no statistically significant difference between the technologies.

5.22 One study (Bergenstal et al. 2010) compared an integrated sensor‑augmented pump therapy system (no low‑glucose suspend) with capillary blood testing and multiple daily insulin injections, and reported multiple outcomes at 12‑month follow‑up. There was a statistically significant change in HbA1c % (−0.5; 95% CI −0.8 to −0.2; p<0.001) and a statistically significant lower hyperglycaemic AUC (>250 mg/dl; 9.2 compared with 17.64; p<0.001) in favour of the integrated sensor‑augmented insulin pump therapy. No statistically significant differences were seen for the following outcomes:

  • proportion with HbA1c ≤7% (53 mmol/mol)

  • the number of people having severe hypoglycaemic events

  • the rate of severe hypoglycaemic events

  • hypoglycaemia AUC (defined as <70 mg/dl)

  • number of patients with diabetic ketoacidosis

  • quality of life (measured by the Paediatric Quality of Life Inventory and Hypoglycaemia Fear Survey).

5.23 In a network analysis, the MiniMed Paradigm Veo system was compared with an integrated sensor‑augmented pump therapy system (no low‑glucose suspend) and with capillary blood testing with continuous subcutaneous insulin infusion. Data from Ly et al. (2013) and Hirsch et al. (2008) were included in the analysis. The network analysis included 1 outcome, change in HbA1c at 6 months, and showed no statistically significant difference between the technologies.

Additional clinical‑effectiveness analyses for the economic model

5.24 A full network analysis of 14 studies was done to calculate estimates of change in HbA1c and severe hypoglycaemic event rates in adults for each of the interventions and for 2 of the comparators (capillary blood testing with continuous subcutaneous insulin infusion and capillary blood testing with multiple daily injections). This analysis included 10 studies that reported data for adults only, 2 studies that reported subgroup data for adults and 2 studies that reported data for a mixed population (adults and children).

5.25 The results of the network analysis suggested that there were statistically significant changes in HbA1c in favour of both the MiniMed Paradigm Veo system (weighted mean difference −0.66; 95% CI −1.05 to −0.27) and the integrated sensor‑augmented pump therapy system without the low‑glucose suspend (weighted mean difference −0.70; 95% CI −1.05 to −0.30) when compared with capillary blood testing with multiple daily insulin injections. The results also suggested that there was a statistically significant difference in the severe hypoglycaemic event rate in favour of capillary blood testing with continuous subcutaneous insulin infusion when compared with the integrated sensor‑augmented pump therapy system without the low‑glucose suspend (weighted mean difference 3.23; 95% CI 1.10 to 9.49). This network analysis was subject to bias because data from different populations and studies with different lengths of follow‑up were pooled, which resulted in substantial heterogeneity across the studies.

Costs and cost effectiveness

Systematic review of cost‑effectiveness evidence

5.26 The External Assessment Group did a search to identify studies investigating the cost effectiveness of the MiniMed Paradigm Veo system and the Vibe and G4 PLATINUM CGM system. Two studies were included and were appraised using the Drummond et al. (1996) checklist.

5.27 Kamble et al. (2012) reported the results of a cost‑effectiveness analysis that compared an integrated sensor‑augmented pump therapy system (no low‑glucose suspend) with capillary blood testing and multiple daily insulin injections from the perspective of the US healthcare system. The study used the IMS CORE Diabetes Model with a time horizon of 60 years and included a population with an average age of 41.3 years and with inadequately controlled type 1 diabetes. The study reported that, when all health effects included in the IMS CORE Diabetes Model were taken into account, the Vibe and G4 PLATINUM CGM system was not cost effective, with incremental cost‑effectiveness ratios (ICERs) of $229,675 and $168,104 per quality‑adjusted life year (QALY) gained assuming a sensor life of 3 or 6 days respectively.

5.28 Ly et al. (2014) reported the results of a cost‑effectiveness analysis that compared the MiniMed Paradigm Veo system with capillary blood testing and continuous subcutaneous insulin from the perspective of an Australian healthcare system. The study used a de novo decision analytic model that included a population with type 1 diabetes and an impaired awareness of hypoglycaemia. The model had a time horizon of 6 months and incorporated severe hypoglycaemic events only. The study reported that the MiniMed Paradigm Veo system was cost effective, with an ICER of AU$40,803 for people aged 12 years or over.

5.29 A manuscript by Roze et al. (2015), which was unpublished and non‑peer reviewed at the time of guidance development, reported the results of a cost‑effectiveness analysis that compared the MiniMed Paradigm Veo system with capillary blood testing and continuous subcutaneous insulin infusion. The study used the IMS CORE Diabetes Model to model an adult population with inadequately controlled type 1 diabetes over a lifetime time horizon. The analysis took the perspective of the NHS and had a discount rate of 3.5% for costs and 1.5% for effects. The study reported that the MiniMed Paradigm Veo system was cost effective, with an ICER of £12,233 per QALY gained.

Economic analysis

5.30 The External Assessment Group used the IMS CORE Diabetes Model to assess the cost effectiveness of the MiniMed Paradigm Veo system and the Vibe and G4 PLATINUM CGM system in adults with type 1 diabetes who are eligible to receive an insulin pump, in accordance with NICE's technology appraisal guidance on continuous subcutaneous insulin infusion for the treatment of diabetes mellitus.

Model structure

5.31 The structure of the IMS CORE Diabetes Model is a simulation model designed to predict the long‑term health outcomes and costs associated with the management of both type 1 and type 2 diabetes. The model structure comprises 17 interdependent Markov sub‑models that represent the most common diabetes‑related complications. This includes stroke, peripheral vascular disease, diabetic retinopathy, hypoglycaemia and ketoacidosis.

5.32 The model was adapted to reflect the NHS and personal social services perspective, the population and interventions included in the assessment, and parameters were inflated to 2015 values where necessary. The model was run as a cohort simulation with a time horizon of 80 years.

Model inputs

5.33 The model was populated with data from the clinical‑effectiveness review, published literature and routine sources of cost and prevalence data. Where appropriate, parameter estimates were taken from the draft NICE guideline on type 1 diabetes. If published data were unavailable, the External Assessment Group used expert opinion to derive estimates to populate the model. A discount rate of 3.5% was applied to both costs and effects.

5.34 One of the comparators listed in the final scope, continuous glucose monitoring with multiple daily insulin injections, was excluded from the analysis because no data were found for this comparator in the clinical‑effectiveness review. In addition, the clinical effectiveness of the comparator, non‑integrated continuous glucose monitoring and continuous subcutaneous insulin therapy, was assumed to be equivalent to that of the Vibe and G4 PLATINUM CGM system (as derived from data for an integrated sensor‑augmented pump therapy system with no low‑glucose suspend function) because no data were found for this comparator.

5.35 The clinical‑effectiveness outcomes included in the model were reduction in HbA1c from baseline and number of severe hypoglycaemic events. A baseline HbA1c value of 7.26% was applied and estimates of mean HbA1c change from baseline were derived from the clinical‑effectiveness review. The values showed that HbA1c increased from baseline for capillary blood testing with continuous subcutaneous insulin infusion (0.05) and capillary blood testing with multiple daily insulin injections (0.64), but decreased for the MiniMed Paradigm Veo system (−0.02), Vibe and G4 PLATINUM CGM system (−0.06) and non‑integrated continuous glucose monitoring with continuous subcutaneous insulin infusion (−0.06).

5.36 In the IMS CORE Diabetes Model, the change in HbA1c level was assumed to happen within the first 12 months, thereafter annual progression occurred (that is, a 0.045% increase in HbA1c each year). The value for annual progression was chosen to correspond with the assumptions made in the economic model for the draft NICE guideline on type 1 diabetes, and was taken from the Diabetes Control and Complications Trial.

5.37 Severe hypoglycaemic event rates for the interventions and the comparators were estimated from the clinical‑effectiveness review. The rate of severe hypoglycaemic events was lowest for the MiniMed Paradigm Veo system (1.9584 per 100 patient‑years) and highest for capillary blood testing with multiple daily insulin injections (19.584 per 100 patient‑years). No baseline event rates were needed for this parameter because the model assumed that these values were treatment specific.

Costs

5.38 Costs included in the model were associated with the primary prevention of diabetes‑related complications, managing diabetes‑related complications, treating diabetes (including the costs of the interventions), and related hospital costs. NHS costs were taken from routinely available data and from related NICE clinical guidelines. The costs of the MiniMed Paradigm Veo system and the Vibe and G4 PLATINUM CGM system were £2961.62 and £3195.48 respectively. A total cost per year for the technologies was calculated assuming a 4‑year lifespan for the insulin pumps, which takes into account the need to replace consumables such as insulin cannulas and reservoirs, glucose monitoring sensors and transmitters, and batteries. The total cost per year for the MiniMed Paradigm Veo system was £4862.10 and for the Vibe and G4 PLATINUM CGM system was £5298.65. The costs of comparator technologies were taken from the draft NICE guideline on type 1 diabetes and from the published literature. In the base‑case analysis, comparator technology costs were weighted by UK market share.

Health‑related quality of life

5.39 The utility values applied to each health state were derived from the published literature. A disutility of −0.012 was applied to a severe hypoglycaemic event, which also incorporated the disutility associated with fear of hypoglycaemia.

Base‑case analysis

5.40 For the purposes of decision‑making, the ICERs per QALY gained or lost were considered. The following key assumptions were applied in the base‑case analysis:

  • The population has a mean age of 41.6 years, has had diabetes for a mean duration of 27.1 years, and has a mean HbA1c of 7.26%.

  • The insulin pumps used in the integrated systems and as stand‑alone devices have a lifetime of 4 years.

  • Four capillary blood tests are needed each day for monitoring blood glucose with either continuous glucose monitoring or capillary blood testing.

  • Forty‑eight units of short‑acting insulin are used each day for continuous subcutaneous insulin infusion.

  • Forty‑eight units of insulin are used each day for multiple daily insulin injections, with twice‑daily insulin detemir (long‑acting; 24 units in total) and 3 boluses of short‑acting insulin at meal times (24 units in total).

  • Three HbA1c tests are needed each year.

  • Treatment effects (HbA1c) are estimated as the mean reduction from the baseline value derived from the clinical‑effectiveness review. This reduction occurs over the first year, then annual progression (0.045%) occurs.

  • The clinical effectiveness of the Vibe and G4 PLATINUM CGM system and the continuous glucose monitoring with continuous subcutaneous insulin infusion (non‑integrated) system are equivalent. The clinical‑effectiveness data for these technologies were derived from the integrated sensor‑augmented pump therapy system with no low‑glucose suspend function.

  • The probability of death from a severe hypoglycaemic event is 0%.

5.41 The results of the probabilistic base‑case analysis suggested that the MiniMed Paradigm Veo system was not cost effective when compared with capillary blood testing with multiple daily insulin injections (£123,375 per QALY gained) and capillary blood testing with continuous subcutaneous insulin infusion (£730,501 per QALY gained). The Vibe and G4 PLATINUM CGM system was not cost effective when compared with capillary blood testing with multiple daily insulin injections (£133,323 per QALY gained) and capillary blood testing with continuous subcutaneous insulin infusion (£668,789 per QALY gained).

5.42 When the MiniMed Paradigm Veo system was compared with continuous glucose monitoring with subcutaneous insulin infusion (non‑integrated), it was associated with an incremental QALY loss (−0.0192) and had an ICER of £422,849 saved per QALY lost. This was driven by the non‑integrated system having the highest decrease in HbA1c from baseline in the clinical‑effectiveness review. This decrease in HbA1c led to a decrease in the number of lifetime diabetes‑related complications, which compensated for the higher number of hypoglycaemic events observed with the non‑integrated system, despite the MiniMed Paradigm Veo system having the lowest number of lifetime hypoglycaemic events (0.622 severe hypoglycaemic events per person). Compared with the non‑integrated continuous glucose monitoring and continuous subcutaneous insulin infusion, the cost effectiveness of the Vibe and G4 PLATINUM CGM system was driven by the cost of the comparator, which resulted in the Vibe and G4 PLATINUM CGM system having an incremental cost of £674.

5.43 The cost‑effectiveness plane for the base‑case probabilistic sensitivity analysis showed a positive correlation between costs and QALYs, with the treatments that included continuous glucose monitoring associated with both increased cost and increased QALYs. The results of the probabilistic sensitivity analysis were also plotted on a cost‑effectiveness acceptability curve, which showed that the probability of technologies that contain continuous glucose monitoring being cost effective was 0% for all maximum acceptable ICERs included in the analysis. This was because the cost of the technologies was too large to be offset by the additional QALYs gained when compared with capillary blood testing.

5.44 Two alternative base‑case scenarios were also run. The first scenario excluded multiple daily insulin injections and assumed all insulin therapy was delivered by continuous subcutaneous insulin infusion. This was intended to reflect the recommendation made in NICE's technology appraisal guidance on continuous subcutaneous insulin infusion for the treatment of diabetes mellitus, which supports using continuous subcutaneous insulin infusion as an option when multiple daily insulin injections are not considered appropriate. The results of this full incremental analysis showed that when capillary blood testing with multiple daily insulin injections was excluded from the analysis, the MiniMed Paradigm Veo system and Vibe and G4 Platinum CGM system were extendedly dominated (that is, dominated by a combination of 2 alternatives) and dominated (has higher costs and worst outcomes) respectively, by the non‑integrated system. The cost‑effectiveness acceptability curve for this analysis showed that capillary blood testing with continuous subcutaneous insulin infusion was the strategy with the greatest probability of being cost effective.

5.45 The second scenario excluded comparators that included capillary blood testing and assumed all glucose monitoring was done using continuous glucose monitoring. This was intended to show the impact of the low‑glucose suspend function of the MiniMed Paradigm Veo system. The results of this full incremental analysis showed that the MiniMed Paradigm Veo system was the least expensive strategy. The Vibe and G4 PLATINUM CGM system remained dominated by the non‑integrated system, largely because the technologies were assumed to be equally effective but the Vibe and G4 PLATINUM CGM system was more expensive. The cost‑effectiveness acceptability curve for this analysis showed that the MiniMed Paradigm Veo system was the strategy with the highest probability of being cost effective at maximum acceptable ICERs of £20,000 and £30,000 per QALY gained.

Analysis of alternative scenarios

5.46 Several scenario analyses were done to assess the impact of the assumptions made in the base‑case analysis. The following assumptions were assessed:

  • applying the baseline population characteristics from the draft NICE guideline on type 1 diabetes

  • frequency of daily capillary blood tests

  • amount of insulin used a day

  • no progression of HbA1c after 1 year

  • no HbA1c change in year 1

  • rates of severe hypoglycaemic events

  • mortality of 4.9% for severe hypoglycaemia

  • method of estimating QALYs

  • 4‑year time horizon

  • addition of a utility increment for fear of hypoglycaemia

  • average annual cost for non‑integrated systems without market share weighting.

5.47 The ICERs changed substantially under the following assumptions:

  • no change in HbA1c in year 1

  • mortality rate of 4.9% for severe hypoglycaemia

  • utility increment of 0.0329 for fear of hypoglycaemia.

5.48 The ICERs did not change substantially in the remaining scenarios modelled, including when a relative risk for severe hypoglycaemic events of 0.125 was applied to the MiniMed Paradigm Veo system.

5.49 The ICERs changed substantially when it was assumed that there was no change in HbA1c in the first year. In this scenario, the Vibe and G4 PLATINUM CGM system was dominated when compared with all 3 comparators included in the analysis, the MiniMed Paradigm Veo system dominated in the comparison with the non‑integrated system and had ICERs of £3,344,672 and £4,871,356 per QALY gained when compared with capillary blood testing with multiple daily insulin injections and capillary blood testing with continuous subcutaneous insulin infusion respectively.

5.50 The ICERs changed substantially when it was assumed that severe hypoglycaemia had a mortality rate of 4.9%. In this scenario, the Vibe and G4 PLATINUM CGM system was dominated when compared with capillary blood testing with continuous subcutaneous insulin infusion, had an ICER of £126,689 per QALY gained when compared with capillary blood testing with multiple daily insulin injection and an incremental cost of £657 when compared with the non‑integrated system. The MiniMed Paradigm Veo system dominated in comparison with the non‑integrated system and had ICERs of £87,818 and £374,626 per QALY gained when compared with capillary blood testing and multiple daily insulin injections and capillary blood testing with continuous subcutaneous insulin infusion respectively.

5.51 The ICERs also changed substantially when a utility increment of 0.0329 was applied to represent a reduction in fear of hypoglycaemia. This utility increment was applied only to the MiniMed Paradigm Veo system and the Vibe and G4 PLATINUM CGM system. In this scenario, the MiniMed Paradigm Veo system dominated in comparison with the non‑integrated system and had ICERs of £64,012 and £74,088 per QALY gained when compared with capillary blood testing with multiple daily insulin injections and capillary blood testing with continuous subcutaneous insulin infusion respectively. The Vibe and G4 PLATINUM CGM system had ICERs of £70,103 and £74,089 per QALY gained when compared with capillary blood testing with multiple daily insulin injections and capillary blood testing with continuous subcutaneous insulin infusion respectively. When compared with the non‑integrated system, the addition of the utility increments resulted in the Vibe and G4 PLATINUM CGM system having 0.5824 incremental QALYs and an ICER of £1157 per QALY gained.

Analysis of population subgroups

5.52 The External Assessment Group also produced supplementary analyses for 2 population subgroups, adults who have difficulty maintaining target HbA1c and adults who often have hypoglycaemic events. Supplementary literature searches were done to inform the analyses and additional evidence networks were constructed to calculate treatment effects. These analyses included a utility decrement of −0.064 and a mortality rate of 0.01596 for severe hypoglycaemic events.

5.53 For the first population subgroup, adults who had difficulty maintaining target HbA1c, the clinical evidence was restricted to studies that included patients with a baseline HbA1c of 8.5% or more and supplementary evidence was included to complete the network. HbA1c treatment effects were taken from Pickup et al. (2011) and the Eurythmics study. Treatment effects were assumed to be equivalent for the MiniMed Paradigm Veo system, the Vibe and G4 PLATINUM CGM system and non‑integrated continuous glucose monitoring with continuous subcutaneous insulin infusion (mean reduction of −0.23 from baseline). In the base‑case analysis for this population, it was assumed that the baseline HbA1c was 8.6% and the rate of hypoglycaemia was 0.

5.54 The results of the base‑case analysis for adults who had difficulty maintaining target HbA1c suggested that the MiniMed Paradigm Veo system was not cost effective compared with capillary blood testing with multiple daily injections and capillary blood testing with subcutaneous insulin infusion, which had ICERs of £86,334 and £79,281 per QALY gained respectively. Similarly, the Vibe and G4 PLATINUM CGM system was not cost effective compared with capillary blood testing with multiple daily injections and capillary blood testing with continuous subcutaneous insulin infusion, which had ICERs of £95,017 and £92,674 per QALY gained respectively. In the comparison with the non‑integrated continuous glucose monitoring and continuous subcutaneous insulin infusion system, the MiniMed Paradigm Veo system dominated and the Vibe and G4 PLATINUM CGM system was dominated, although these comparisons were driven solely by cost.

5.55 The assumptions around treatment effects, incidence of severe hypoglycaemia and the impact of reduced duration of sensor use were investigated in 4 scenario analyses, which largely resulted in less favourable ICERs than those reported for the base case. Lower ICERs were obtained for the MiniMed Paradigm Veo system when it was assumed that the low‑glucose suspend function resulted in a lower incidence of hypoglycaemia when compared with capillary blood testing with multiple daily injections (£81,255 per QALY gained) and capillary blood testing with continuous subcutaneous insulin infusion (£62,025 per QALY gained).

5.56 For the second population subgroup, adults who experienced frequent hypoglycaemic events, the clinical evidence was restricted to studies that included patients with a baseline HbA1c of 8.5% or less and supplementary evidence was included to complete the network. It was assumed that the population had HbA1c (7.26%) that remained stable, but they had frequent hypoglycaemic events. Rates of severe hypoglycaemia were calculated using incidence rate ratios reported in the ASPIRE and STAR‑3 studies, and pooled estimates from Hirsch et al. (2008), the SWITCH study and Battelino et al. (2011). Incidence rates for severe hypoglycaemia ranged from 4.33 per 100 patient‑years for the MiniMed Paradigm Veo system to 38.36 per 100 patient‑years for capillary blood testing with multiple daily insulin injections. An incidence rate of 33.33 per 100 patient‑years was applied to both the Vibe and G4 PLATINUM CGM system and non‑integrated continuous glucose monitoring with continuous subcutaneous insulin infusion.

5.57 The results of the base‑case analysis for adults who had frequent hypoglycaemic events suggested that the MiniMed Paradigm Veo system was not cost effective compared with capillary blood testing with multiple daily injections and capillary blood testing with continuous subcutaneous insulin infusion, which had ICERs of £188,124 and £189,326 per QALY gained respectively. When compared with non‑integrated continuous glucose monitoring and continuous subcutaneous insulin infusion, the MiniMed Paradigm Veo system dominated. The Vibe and G4 PLATINUM CGM system was dominated when compared with both capillary blood testing with continuous subcutaneous insulin infusion and non‑integrated continuous glucose monitoring with continuous subcutaneous insulin infusion. When compared with capillary blood testing with multiple daily injections, the Vibe and G4 PLATINUM CGM system had an ICER of £1,538,493 per QALY gained.

5.58 The assumptions around HbA1c treatment effects and impact of fear of hypoglycaemia were investigated in 3 scenario analyses, which resulted in more favourable ICERs than those reported for the base case. The lowest ICERs were produced when a utility increment of 0.0329 associated with a reduced fear of hypoglycaemia was applied to both the MiniMed Paradigm Veo system and the Vibe and G4 PLATINUM CGM system. This resulted in the MiniMed Paradigm Veo system dominating non‑integrated continuous glucose monitoring with continuous subcutaneous insulin infusion and having ICERs of £80,692 and £57,857 per QALY gained when compared with capillary blood testing with multiple daily injections and capillary blood testing with continuous subcutaneous insulin infusion respectively. The Vibe and G4 PLATINUM CGM system had an ICER of £1161 per QALY gained compared with non‑integrated continuous glucose monitoring with continuous subcutaneous insulin infusion, but ICERs of £141,953 and £124,144 per QALY gained when compared with capillary blood testing with multiple daily injections and capillary blood testing with continuous subcutaneous insulin infusion respectively.

  • National Institute for Health and Care Excellence (NICE)