Continuous subcutaneous insulin infusion for the treatment of diabetes mellitus

From the evidence presented in the NICE technology appraisal guidance on the use of long‑acting insulin analogues for the treatment of diabetes and a search for evidence that has emerged since, the assessment group concluded that multiple daily injection (MDI) therapy based on long-acting insulin analogues is more efficacious than MDI therapy based on older insulins, such as neutral protamine Hagedorn (NPH, Isophane), in type 1 diabetes mellitus. Therefore, in type 1 diabetes mellitus, analogue-based MDI therapy was used as the comparator for CSII therapy. For type 2 diabetes mellitus, a Cochrane review from 2007 concluded that there was no benefit of long-acting insulin analogues over NPH. Taking NICE's technology appraisal guidance 53 into account, which limited the use of glargine in type 2 diabetes mellitus, the assessment group concluded that in type 2 diabetes mellitus, the advantage of long-acting insulin analogues over NPH was not proven and both remained valid comparators for CSII therapy in this appraisal.

The assessment group found 4 randomised controlled trials (RCTs) in type 1 diabetes mellitus that compared CSII with analogue MDI therapy and 4 RCTs in type 2 diabetes mellitus that compared CSII with MDI therapy (3 NPH-based and 1 glargine-based). One RCT conducted in children and adolescents with type 1 diabetes mellitus reported a statistically significant greater reduction in HbA1c levels following 16 weeks of CSII therapy compared with analogue-based MDI therapy (1% decrease versus no change). In the 3 further studies in type 1 diabetes mellitus and the 4 studies in people with type 2 diabetes mellitus, there was little evidence of a statistically significant difference between CSII and MDI therapy in terms of HbA1c reduction. There were statistically significant fewer episodes of severe hypoglycaemia with CSII therapy compared with MDI therapy in the study in children and adolescents with type 1 diabetes mellitus. The other studies had too few hypoglycaemic episodes to comment or showed no difference. There was no evidence of a statistically significant difference in insulin dose being administered, mean blood glucose levels, glucose variability or weight between the 2 groups. There was no statistically significant difference in the quality of life for people with type 1 diabetes mellitus between CSII and MDI therapy in the 2 studies that reported this outcome. In type 2 diabetes mellitus, 1 study found no difference in the quality of life and another reported a statistically significant difference in treatment satisfaction favouring CSII therapy. Many of these studies were of brief duration, and improvements in glycaemic control may be gained as the person's expertise with the use of the pump increases with time. Conversely, any improvements in HbA1c levels observed in the short term cannot be assumed to persist in the longer term.

The assessment group reported a total of 48 observational studies. The majority of the patients in these studies had type 1 diabetes mellitus; only 4 studies included a small number of people (3% to 7%) with type 2 diabetes mellitus. It noted that, although these studies carried a greater risk of bias than the RCTs, they were much larger, of longer duration and more representative of people likely to be considered for CSII therapy in routine clinical practice than the populations in the RCTs available. A total of 46 studies reported HbA1c levels before and after CSII therapy. Of these, all 18 studies in the adult/mixed age group showed a statistically significant decrease in HbA1c levels (in the range 0.2% to 1.4%) after initiation of CSII therapy. Of the 23 studies in the children/adolescent age group, 20 studies showed a decrease in HbA1c levels after starting CSII therapy (in the range 0.2% to 1.2%) with the difference reaching statistical significance in 13 studies. Three studies showed an increase in HbA1c levels after starting CSII therapy; however, in 2 of these studies, the increase was not statistically significant, and the third study did not report whether the HbA1c increase of 0.6% was statistically significant. In 5 studies in young children, the HbA1c decrease after starting CSII therapy was in the range 0.2% to 1.6%, with the change being statistically significant in 4 of these studies. Twenty-six observational studies compared the rate of severe hypoglycaemic episodes in people receiving CSII and MDI therapy. Of the 10 studies in the adult/mixed age groups, 8 reported statistically significant decreases in the rate of severe hypoglycaemic episodes after starting CSII therapy. The rate ratios were in the range 0.07 to 0.40. In the remaining 2 studies, there were no episodes before or after beginning pump use. Of the 11 studies in children and adolescents, 1 had no severe hypoglycaemic episodes and the other 10 reported decreases in their frequency following initiation of CSII therapy, with rate ratios in the range 0.12 to 0.80. In 4 of these studies, the reduction was statistically significant, 3 did not report significance, and 3 did not show a statistically significant decrease. In 5 studies in young children, the rate ratios were in the range 0 to 0.81, with the difference being statistically significant in 3. Observational studies also showed no statistically significant increase in the incidence of diabetic ketoacidosis, some minor weight gain and, in some studies, an overall reduction in daily insulin dose, after initiation of CSII therapy. Of the 22 studies that reported continuation rates, these were in the range of 74% to 100% at 1 to 5 years.

The assessment group also accessed data from the Insulin Pump Clinical Database. This information is collected from a group of centres with considerable experience of using pumps and reflects results with routine care in centres with expertise.

The assessment group identified 6 observational studies of CSII therapy in pregnant women with type 1 diabetes mellitus. One study showed statistically significant lower HbA1c levels in women on CSII compared with MDI therapy, although the assessment group felt the results should be disregarded because there appeared to be selection bias in the study. Overall, there were no statistically significant differences in either maternal or fetal outcomes of pregnancy between CSII and MDI therapy.

The clinical-effectiveness evidence from the manufacturers was based on a so-far unpublished meta-analysis of RCTs and observational studies of people with type 1 diabetes mellitus, comparing CSII therapy with isophane-based or glargine-based MDI therapy. The meta-analysis provided data on the baseline rate of, and the reduction in, severe hypoglycaemic events in the group of people receiving CSII therapy. The greatest reduction occurred in people who had the highest initial hypoglycaemia frequency. Glycaemic control was also significantly better for those people receiving CSII therapy; the greatest reduction was achieved in people who had the highest baseline HbA1c levels when receiving MDI therapy. Detailed results of this study were designated as 'academic in confidence'.

The clinical specialists emphasised the robustness of the evidence that allowed the decrease in HbA1c levels to be used as a reliable proxy for avoidance of the long‑term complications of diabetes mellitus. The relationship between complications and HbA1c levels is such that, for any particular decrease in HbA1c, the benefits in terms of complications avoided are greater for a higher baseline HbA1c. In addition, it is possible that the variability in blood glucose level may also have a role to play in the development of long-term complications, and there is evidence that CSII therapy can reduce this variability. The clinical specialists also reported that, regardless of the mode of treatment, the higher the baseline HbA1c the greater the value of any improvement would be. The paediatric clinical specialist emphasised the difficulty in controlling blood glucose in very young children because of their sensitivity to insulin, small size and irregular lifestyle, and noted the special relevance of insulin pumps to this age group. The time of puberty was also identified as a difficult time to control diabetes because of hormonal and psychosocial reasons. Children also have a greater lifetime risk of complications because these depend on the duration of the disease and an early onset makes for a potentially longer time lived with diabetes. The patient expert described the stress that being the parent of a child with diabetes entails and the effects of this on a person's social and professional life.

In summary, there is little evidence from the RCTs of a significant difference between CSII and MDI therapy in terms of a decrease in HbA1c levels or in the rate of severe hypoglycaemic episodes in people with diabetes mellitus. Observational studies show a much greater improvement (decrease) in HbA1c levels with CSII therapy, as well as statistically significant decreases in the rate of severe hypoglycaemia episodes. There is no clear evidence of any greater benefit of CSII over MDI therapy in pregnancy.

A systematic review of the cost-effectiveness literature for insulin pumps conducted by the assessment group found 11 publications. Except for 1 study which developed a relatively simple Markov model and another in which the model was not reported, all publications used the Centre for Outcomes Research (CORE) model. Three studies that were performed in the UK and took the health service perspective reported incremental cost‑effectiveness ratios (ICERs) for CSII therapy compared with MDI therapy of £11,500, £26,300 and £32,800 per quality‑adjusted life year (QALY) gained. In the cost-effectiveness studies, the most common assumed improvement in HbA1c levels with CSII therapy compared with MDI therapy was 1.2%.

The manufacturers provided a joint economic evaluation that was also based on the CORE model. The patient cohort characteristics were based on a UK database of 3,000 adults with type 1 diabetes mellitus followed up over a period of 9 years. The average age was 37.8 years. Three values for the decrease in HbA1c levels with CSII therapy were considered: the lower value for the decrease in HbA1c levels was based on the results of the confidential meta‑analysis (see section 4.1.6); the upper value for the decrease in HbA1c levels was 1.29%, reflecting the higher baseline HbA1c seen in the UK population; the third value was an intermediate decrease in HbA1c levels, midway between the upper and lower values. The decreases in HbA1c levels were applied to a baseline of 9.4% in each analysis. The rate of severe hypoglycaemic episodes was assumed to be reduced by approximately 75% during treatment with CSII therapy compared with MDI therapy. Health-state utilities were those reported from the United Kingdom Prospective Diabetes Study (UKPDS), which was conducted in people with type 2 diabetes mellitus.

All 3 analyses showed a gain in QALYs for people receiving CSII compared with MDI therapy at an increased cost. The ICER for CSII compared with MDI therapy was £34,330 per QALY gained for the lower value of decrease in HbA1c levels, £16,842 per QALY gained for the upper value of decrease in HbA1c levels, and £22,897 per QALY gained for the intermediate decrease in HbA1c levels.

The assessment group commented that the industry submission assumed that the cost of a severe hypoglycaemic episode was £413, which included a hospital stay. As in clinical practice, only a minority of people with severe hypoglycaemia are hospitalised. This would overestimate the average cost of managing an episode of severe hypoglycaemia and underestimate the ICER of CSII compared with MDI therapy. The assessment group also noted that the manufacturers' economic model did not consider the benefits associated with the avoidance of the fear of hypoglycaemia, the reduced incidence of depression, or the avoidance of cognitive impairment in children, which could result from improved glycaemic control. The effect of excluding such considerations would be an overestimation of the calculated ICERs.

The assessment group's economic evaluation was also based on the CORE model. It simulated a cohort of people with type 1 diabetes mellitus with an average age of 40 years. The base-case analysis assumed a baseline HbA1c level of 8.8% before CSII therapy, and a reduction of 0.9% while receiving CSII therapy, based on preliminary data from the Insulin Pump Clinical Database for people aged 20 to 39 years. Sensitivity analyses for the effect of CSII therapy on the reduction in HbA1c levels from a baseline of 9.0% were also undertaken. A baseline rate for severe hypoglycaemic events of 18.7 per 100 person years was assumed for the base case, as in previous modelling carried out for the development of the NICE technology appraisal on the use of long‑acting insulin analogues for the treatment of diabetes; the effect of CSII therapy on this rate was investigated using reductions of 50% and 75%. Sensitivity analysis assumed a higher baseline rate of severe hypoglycaemia equal to that in the manufacturers' submission (designated as academic in confidence; see section 4.1.6). Lastly, the assessment group modelled a cohort of people who were assumed to have good control with an HbA1c level of 7.5% (in whom further reductions would not occur) but who had a very high rate of severe hypoglycaemia (134 episodes per 100 person years), with an investigation of reductions of 50% and 75% with CSII therapy.

The costs of the pumps were annualised, and it was assumed that the associated infusion equipment was changed every 3 days, as recommended. For CSII therapy, the daily requirement of insulin was 0.6 IU per kg, which was less than the 0.7 IU per kg required for MDI therapy. CSII therapy was assumed to incur a one-off training cost of £240 when a person starts to use the pump. Both CSII and MDI therapy were assumed to incur the same costs for blood glucose testing. The average cost of an episode of severe hypoglycaemia was assumed to be £65.

The base-case analysis with a reduction of HbA1c levels of 0.9% and a severe hypoglycaemic event rate of 18.7 episodes per 100 person years reduced by 50%, over a time horizon of 50 years, produced an ICER of £37,712 per QALY gained for CSII compared with MDI therapy. Changing the reduction in the rate of severe hypoglycaemia events to 0% or 75% did not change the ICER significantly. With the higher baseline rate of severe hypoglycaemia assumed in the manufacturers' submission (designated as academic in confidence; see section 4.1.6), a 50% reduction, and baseline HbA1c levels reduced to 7.9% from a baseline of 8.8%, the ICER was £36,587 per QALY gained. When a greater reduction in HbA1c levels of 1.4% was assumed, with no reduction in severe hypoglycaemic event rates, the ICER was £24,720 per QALY gained. In the cohort with good glycaemic control, when there was assumed to be no improvement in HbA1c levels but the severe hypoglycaemic event rate was 134 per 100 person years, the ICER was £273,992 per QALY gained for a 50% reduction and £152,058 per QALY gained for a 75% reduction in severe hypoglycaemia rate. Avoidance of severe hypoglycaemic events can lead to quality of life gains by avoiding the disutility of the event itself and because of the reduced fear of such events. In the scenario with a 0.9% decrease in HbA1c from a baseline of 8.8% and a 50% reduction in the rate of severe hypoglycaemia events from that in the manufacturers' submission, which was associated with an ICER of £36,587, an assumed annual 0.01 quality of life increment in the CSII arm decreased the ICER to £29,300 per QALY gained. When the assumed quality of life increment was 0.03, the ICER decreased to £21,000 per QALY gained. In the cohort with good glycaemic control, when there was assumed to be no improvement in HbA1c levels, the severe hypoglycaemic event rate was 134 per 100 person years, an annual quality of life increment of 0.05 was assumed and a reduction in the rate of severe hypoglycaemia events by 50%, the ICER was £28,600 per QALY gained. For the same cohort but with a 75% reduction in severe hypoglycaemia events, and an annual quality of life increment of 0.04 the ICER was approximately £31,300 per QALY gained.

The appraisal committee reviewed the data available on the clinical and cost effectiveness of CSII for the treatment of diabetes mellitus. It discussed the important outcomes of treatment, including reduced HbA1c levels, reductions in both hypoglycaemia and the fear of recurrent severe hypoglycaemia, and other aspects of quality of life for both adults and children. It weighed up the RCT evidence, the observational evidence and the views expressed by the clinical specialists and patient experts. It considered evidence on the nature of the condition and the value placed on the benefits of CSII by people with diabetes mellitus and those who represent them. It was also mindful of the need to take account of the effective use of NHS resources.

The committee first considered the evidence from the RCTs. The committee was persuaded that these few, small trials of relatively short duration could not be relied on alone to capture the benefits of CSII therapy. The committee noted that the larger quantity of evidence from observational trials showed significant and larger benefits from the initiation of CSII therapy. It was aware of the biases operating in such studies and the need to interpret these results with caution. However, it agreed that people enrolled in observational studies would more closely resemble the population in routine clinical practice that would be considered as likely candidates for CSII therapy. In addition, these studies included a larger number of people and ran over longer periods of time. The committee concluded that it was appropriate to use evidence from such studies as well as from the RCTs to inform its decisions.

The committee considered the effect of CSII therapy on quality of life. It was persuaded by the clinical specialists and patient experts that the use of insulin pumps yielded quality of life benefits, such as flexibility, autonomy, and improved sleep and socialisation. In particular, the committee agreed that decreasing the rate of severe hypoglycaemic episodes would improve the quality of life for those people who experience frequent and disabling episodes and who consequently live with fear of such episodes.

The committee therefore concluded that CSII therapy had a valuable effect on blood glucose control. HbA1c levels were reduced, particularly when these levels were high at baseline (approximately 9.0%). The committee noted that the decrease in HbA1c levels after beginning CSII therapy was clearly related to the initial HbA1c level, with more pronounced decreases in HbA1c levels seen in people with higher baseline levels. The committee also heard that complications increased at more than a linear rate as HbA1c levels increased, leading to the conclusion that decreases in HbA1c from a higher baseline level would result in larger reductions in complications. It was also persuaded that CSII therapy could reduce the rate of hypoglycaemic episodes, and it heard from the patient experts that when hypoglycaemia occurs in people using CSII therapy, it does so gradually and with sufficient time for the pump user to take remedial action.

The committee considered the results of both the manufacturers' and the assessment group's economic modelling. In both, ICERs for CSII therapy were sensitive to the assumed decrease in HbA1c levels. The committee noted that only pronounced decreases in HbA1c levels, contingent on a baseline level well above 7.5%, brought the ICERs down to a level usually considered to be acceptable.

The committee was aware that guidelines for people with diabetes, including those of NICE, recommend a target HbA1c level of 7.5%, or lower in the presence of vascular complications. The committee endorses these previously recommended target levels for HbA1c. However, the committee noted that there were several means to this end, including the use of structured education, and it considered that the contribution to be made by insulin pumps had to be restricted to circumstances where the pumps were a cost-effective option. The committee agreed that at very high baseline HbA1c levels the decrease expected with CSII could make CSII therapy cost effective because of the avoidance of long-term complications. However, at baseline levels of less than 9.0%, CSII would only be cost effective if an additional quality of life benefit was assumed. This benefit could be derived from the avoidance of the fear of hypoglycaemia as well as from other quality of life improvements associated with the use of insulin pumps themselves which were not captured in the base-case economic modelling. The committee judged that when a plausible small quality of life benefit is assumed, CSII would be cost effective at a baseline HbA1c level of 8.5% or above, and therefore concluded that CSII therapy is recommended as a treatment option for adults with type 1 diabetes mellitus whose HbA1c levels have remained high (that is, at 8.5% or above) on MDI therapy (including, if appropriate, the use of long-acting insulin analogues) despite a high level of care.

The committee was aware that lower HbA1c levels are associated with a greater risk of episodes of hypoglycaemia and that attempts to achieve target haemoglobin levels with MDI can result in a greater risk of hypoglycaemic episodes. The committee considered the economic modelling for people who reach a target HbA1c level of 7.5% on MDI, but who experience the occurrence of disabling hypoglycaemic episodes. The committee accepted that such episodes can be significantly decreased with CSII therapy. Although this effect did not have a pronounced impact on the ICERs in the assessment group's base case, the committee considered that there would be a greater quality of life benefit due to the avoidance of the fear of hypoglycaemia by the use of CSII. Taking into account the fact that the model excluded any impact of mild and moderate hypoglycaemia episodes on the quality of life, the committee agreed that CSII therapy could be considered an appropriate use of resources in the NHS for adults with baseline HbA1c levels below 8.5% but who experience disabling hypoglycaemia with further intensification of MDI therapy aimed at decreasing the HbA1c level. Therefore, the committee concluded that CSII therapy is recommended as a treatment option for adults with type 1 diabetes mellitus when attempts to achieve target HbA1c levels, typically 7.5%, with MDI therapy result in the person experiencing disabling hypoglycaemia. Disabling hypoglycaemia is, for the purpose of this guidance, defined as repeated and unpredictable occurrence of hypoglycaemia that results in persistent anxiety about recurrence and is associated with a significant adverse effect on quality of life.

The committee paid special attention to the use of CSII therapy in children. It heard from the clinical specialists about the importance of this treatment in very young children. It was also aware of the difficulties and delay that a trial of MDI therapy, to prove that such therapy was ineffective, would entail in this young patient group. The committee was concerned that the CORE model was not validated in children and the data from adults used in the model could not be extrapolated to children. However, the committee considered that all the factors relevant to favouring CSII therapy for adults in whom MDI therapy had failed to achieve an acceptable HbA1c level, or who experienced disabling hypoglycaemia, applied as least as much to children. In addition, the committee heard from the clinical specialists that managing the delivery of small insulin doses in very young children and delivering midday doses of insulin to young school children are difficult or impractical, leading to ineffective glycaemic control. The committee therefore decided that, on balance, CSII therapy should be recommended for children younger than 12 years with type 1 diabetes mellitus as a treatment option without an intensive trial of MDI therapy if such a trial was felt clinically inappropriate or impractical. This recommendation is consistent with NICE guideline on the diagnosis and management of type 1 diabetes in children, young people and adults.

In developing the recommendations for children, the committee considered that children 12 years and older would normally be competent to self-inject an afternoon dose of insulin at school, which would allow for a proper trial of MDI therapy. The committee also agreed that because MDI therapy based on long-acting analogues is more efficacious for type 1 diabetes mellitus than MDI therapy based on older insulins, MDI therapy should be judged to be unable to attain the required HbA1c levels only if it is based, if clinically appropriate, on long-acting insulin analogues. The committee considered its approach for children on insulin pumps who reach the age of 12 years and who, following this guidance, would not have had a trial of MDI. The committee considered that the continuation of CSII could not equitably be supported without a trial of MDI after that stage. The committee was mindful that making such a recommendation too strict could mean a change in insulin regimen for children who had achieved satisfactory control of their blood glucose level. The committee discussed the implications of undergoing such a trial, especially during a period where children experience many developmental, social and educational changes. It concluded that such a trial of MDI would normally need to be undertaken sometime before a child reached adulthood at the age of 18 years.

The committee considered the use of CSII therapy in women with type 1 diabetes mellitus who were pregnant or planning a pregnancy. It agreed that such women benefited from the use of insulin pumps but that there was no evidence that the criteria for the use of pumps should differ from those applied to other adults.

The committee considered the use of CSII therapy in people with type 2 diabetes. There was no evidence placed before the committee that showed that CSII therapy improved outcomes in this group of individuals. Furthermore, the economic modelling was limited to cohorts with type 1 diabetes mellitus. The committee was aware that type 2 diabetes mellitus occurred in a much more heterogeneous group of people, including some severely insulin‑resistant individuals. The committee also heard that there would be subgroups of the type 2 diabetic population, such as those with a body mass index below a certain level or those with low levels of C peptides, who would possibly stand to benefit from insulin pumps, but was aware that there was no evidence to support such a recommendation. In conclusion, the committee decided that, in the absence of evidence of improved outcomes, the use of CSII therapy in people with type 2 diabetes mellitus could not be recommended as a cost‑effective use of NHS resources.

The committee noted that adherence to any insulin regimen required a high degree of motivation, commitment and competence from patients and carers to ensure that it was both safe and effective. Insulin therapy requires attention to maintaining personal hygiene, blood glucose testing several times a day, estimating carbohydrate and calorie consumption throughout every day and self-injection. The committee was aware that structured education of people with diabetes and their carers in flexible insulin therapy can, of itself, improve glycaemic control and improve chances of successful CSII therapy. The committee agreed that structured education is very important in maintaining and improving glycaemic control and an important adjunct to CSII. The use of CSII should not be considered to replace the need for education.

The use of effective insulin pump therapy would require resetting of the cannula every 2 to 3 days and programming the pump. The committee was aware that the use of a pump was only likely to be cost effective when it was used appropriately. The committee therefore considered that all patients potentially eligible for this treatment, and their carers if appropriate, should have the opportunity for an informed discussion with the responsible clinician about the advantages and disadvantages of insulin pumps, including the requirements for using a pump effectively. Insulin pump therapy should only begin where it is agreed that the patient is likely to benefit from the intervention. However, the committee agreed that most of these requirements were equally applicable to injection therapy and patients who were eligible would have already shown a high level of care of their diabetes, if necessary with the aid of carers. The elements of a high level of care are detailed in NICE's guideline on the management of type 1 diabetes in adults and children.

The recommendations made by the committee were based on people who switched from MDI to CSII therapy experiencing a benefit in terms of either improved glycaemic control and decreased HbA1c levels, or a decrease in the rate of hypoglycaemic episodes and the fear resulting from these, and a resulting improvement in their quality of life. However, the committee was mindful that not all people started on CSII therapy would gain these benefits and the continued use of an expensive therapy in the absence of demonstrable benefits would be an inappropriate use of resources. The committee therefore felt the need to specify that if such benefits were not gained within a reasonable time period, CSII therapy should be withdrawn. The appropriate targets in terms of decrease in HbA1c levels or avoidance of hypoglycaemia, as well as the length of the trial period and the maintenance of improvements beyond that period, would need to be set on an individual basis by the responsible clinician in consultation with the pump user, relevant carers and other healthcare professionals. The committee also agreed that before withdrawing CSII, there should be further efforts at providing support and care for the pump user, including further structured education, where necessary.

The committee was aware that the delivery of insulin was just one aspect of the management of diabetes mellitus. The care provided should also include advice on diet, lifestyle, exercise and education. The management of a person's diabetes mellitus when using CSII should involve interaction with a multidisciplinary team specialised in the treatment of diabetes mellitus. Such specialist teams are especially important during the initial period when a new pump user is developing expertise. The committee was unable to make firm recommendations for the exact composition of specialist teams and considered this a decision for the individual centre. However, the committee expected that the specialist team would normally include a physician with a specialist interest in insulin pump therapy, a diabetes specialist nurse and a dietitian, among others. The committee agreed that structured education was a vital component of CSII therapy and would need to be ongoing and intensified before withdrawal of CSII. In addition, education was an important aspect of optimal MDI therapy and an inability to achieve targets on MDI would normally require further patient education before switching to CSII.