Title: Insulin Pumps for Type 1 Diabetes: A Clinical and Cost Effectiveness Review

Transcription

1 Title: Insulin Pumps for Type 1 Diabetes: A Clinical and Cost Effectiveness Review Date: 14 April 2008 Context and policy issues: Diabetes mellitus (DM) is a metabolic disorder characterized by hyperglycemia (elevated blood glucose levels). DM results from a defect in insulin secretion, insulin action, or both. 1,2 It can result in severe complications, including limb amputation, blindness, kidney failure, heart disease, stroke and premature death. 3 There are two main types of DM: type 1 DM and type 2 DM. In type 1 DM, 90% of the insulin producing cells of the pancreas are destroyed, leading to severe insulin deficiency. 4 Type 2 DM is characterized by insulin resistance and relative insulin deficiency. 5 Over 2.25 million Canadians are estimated to have DM. 3 Of all the diagnosed cases, 90% are type 2 DM and 10% are type 1 DM. 3 The body needs a little insulin all the time (basal insulin, which accounts for half of the insulin production) with brief peaks of increased levels an hour or two after meals. 5 Insulin therapy remains the mainstay of treatment for Type 1 DM. The Diabetes Control and Complications trial (DCCT) has shown that intensive insulin therapy effectively delays the onset and slows the progression of diabetic complications in type 1 DM patients. 6 Insulin is administered either by injections [e.g. multiple daily injections (MDI)] or as continuous infusion using a small pump [referred to as continuous intensive insulin infusion (CSII) or insulin pump therapy]. The pump is programmed to deliver insulin and is worn by the patient externally. Insulin is delivered from the pump via insertion of a catheter into the subcutaneous tissue of the abdominal wall. 7 The insulin pump costs approximately C$6000 to C$ In addition, there is a substantial cost associated with consumables such as infusion sets which are required with pump use. At the present time, the insulin pump is not funded by all jurisdictions. An evaluation of the clinical and cost effectiveness of insulin pumps is warranted. Disclaimer: The Health Technology Inquiry Service (HTIS) is an information service for those involved in planning and providing health care in Canada. HTIS responses are based on a limited literature search and are not comprehensive, systematic reviews. The intent is to provide a list of sources and a summary of the best evidence on the topic that CADTH could identify using all reasonable efforts within the time allowed. HTIS responses should be considered along with other types of information and health care considerations. The information included in this response is not intended to replace professional medical advice, nor should it be construed as a recommendation for or against the use of a particular health technology. Readers are also cautioned that a lack of good quality evidence does not necessarily mean a lack of effectiveness particularly in the case of new and emerging health technologies, for which little information can be found, but which may in future prove to be effective. While CADTH has taken care in the preparation of the report to ensure that its contents are accurate, complete and up to date, CADTH does not make any guarantee to that effect. CADTH is not liable for any loss or damages resulting from use of the information in the report. Copyright: This report contains CADTH copyright material. It may be copied and used for non-commercial purposes, provided that attribution is given to CADTH. Links: This report may contain links to other information on available on the websites of third parties on the Internet. CADTH does not have control over the content of such sites. Use of third party sites is governed by the owners own terms and conditions.

2 Research questions: 1. What is the clinical effectiveness of insulin pumps for patients with type 1 diabetes? 2. What is the cost effectiveness of insulin pumps for patients with type 1 diabetes? 3. Is there any evidence that insulin pumps have a life cycle and should be replaced after a certain time period? Methods: A limited literature search was conducted on key health technology assessment resources, including PubMed, The Cochrane Library (Issue 1, 2008), University of York Centre for Reviews and Dissemination (CRD) databases, ECRI, EuroScan, international HTA agencies, and a focused Internet search. Results include articles published between 2003 and March 2008, and are limited to English language publications only. Filters were applied to limit the retrieval to health technology assessments (HTA), systematic reviews (SRs), meta-analyses, economic evaluations, and guidelines. Summary of findings: Two HTA reports, 5,9 four SRs, and three economic studies were identified. Findings from these reports are summarized below. Clinical effectiveness of insulin pumps Health technology assessments Two HTA reports 5,9 published in 2004 and 2005 were identified. An HTA report 5 published by Colquitt et al. in 2004, assessed the clinical and cost effectiveness of CSII compared with MDI, for the delivery of insulin in DM patients. They performed a comprehensive literature search and 20 studies satisfied their inclusion criteria. These studies included eight parallel randomized controlled trials (RCTs), nine randomized crossover trials and three non-randomized crossover studies. Of these 20 studies, 14 were in adults, four in pregnant women and two in adolescents. Characteristics of the studies are summarized in Appendix 1. No RCTs or crossover studies with children were identified. The results are presented according to the different populations studied. Adults with type 1 DM: The quality of reporting and methodology of the included 14 studies was generally poor. Allocation concealment was inadequate in four studies and not reported in the remaining 10 studies. Eight studies reported glycated haemoglobin (A1c) levels at 2.5 to 4 months of follow-up. Of these eight studies, five reported reductions (ranging between 0.3% and 2.48%), one reported no difference and two reported a slight increase with CSII as compared to MDI. Of the five studies that reported reduction, results were statistically significant in two, non-significant in two, and significance was not reported in one. Of the two studies that reported an increase, results were non-significant in one and significance was not reported in one. Five of the eight studies provided sufficient data enabling them to be combined in a meta-analysis. The corresponding weighted mean difference (WMD) was and the 95% confidence interval (CI) was (-1.28, Insulin Pumps for Type 1 Diabetes 2

3 0.01), indicating a non-significant reduction with CSII as compared to MDI. However, on excluding from the meta-analysis one study which used a suboptimal MDI regime, the WMD (95% CI) was (-1.19, 0.14). Six studies reported A1c levels at 6 months of follow-up. Of these six studies, five reported reductions (between 0.2% and 0.7%) with results being significant in only two; the sixth study reported an increase which was not significant. Three of the studies could be pooled and the WMD (95% CI) was (-0.57, 0.05) indicating a non-significant reduction in A1c with CSII as compared to MDI. One study reported A1c at 9 months of follow-up and found no significant difference between CSII and MDI. Four studies reported A1c levels at 12 months of follow-up. Of these four, one reported a significant reduction and three reported no difference with CSII as compared to MDI. Two of these studies could be pooled and the WMD (95% CI) was (-1.29, 0.07). Two studies reported A1c levels at 24 months of follow-up. Of these two, one reported a reduction but did not report on significance and the other reported no significant difference with CSI as compared to MDI. One study reported A1c at 41 months of follow-up and mentioned a reduction with CSII as compared to MDI but did not report on significance. In the short-term studies, insulin dose was lower for CSII than for MDI by 20% to 25% but there was little difference found in the longer term studies. Of the eight studies that reported on patient preference, four found that a greater proportion of patients preferred CSII, three found a greater proportion preferred MDI and one found no difference. One study examined the quality of life (QoL) using the Diabetes Quality of Life (DQOL) questionnaire and found no significant differences between CSII and MDI. Thirteen studies reported on hypoglycemia. Hypoglycemic events did not differ significantly between CSII and MDI in most studies, but three found fewer hypoglycemic episodes with CSII and one study found more hypoglycemia and hypoglycemic coma with CSII. The authors mentioned that observational studies showed much greater reductions in rates of severe hypoglycemia with CSII. Pregnant women with DM: The quality of reporting and methodology of the included four studies was generally poor. Allocation concealment was inadequate in two studies and not reported in two studies. There is insufficient evidence from these studies to determine if CSII or MDI is better for achieving glycemic control in pregnancy. Also, it is difficult to ascertain from the available evidence if there is any significant difference between CSII and MDI in terms of pregnancy outcomes (e.g. duration of pregnancy, delivery by caesarean section, birth weight, and death). Two studies found no significant difference between CSII and MDI in total daily insulin dose; one study found lower total daily insulin required for CSII and one study did not report on the total daily insulin dose. None of these four studies reported on patient preference or QoL. Hypoglycemic events were reported in all four studies. One study reported a severe hypoglycemic episode but did not specify the treatment group. One study found more severe hypoglycemic episodes in the MDI group than in the CSII group, but the investigators mentioned that the groups were too small to achieve statistical significance. The other two studies, however, found slightly more severe hypoglycemic episodes and undefined hypoglycemia with CSII as compared to MDI. Two studies reported the occurrence of ketoacidosis occurring in the CSII group but not in the MDI group. Insulin Pumps for Type 1 Diabetes 3

4 Adolescents with type 1 DM: The quality of reporting and methodology of the included two studies was generally poor. Allocation concealment was not reported in either study. For A1c, one study found no significant difference between CSII and MDI, whereas the other study found a benefit with CSII. None of the two studies reported on QoL. Both studies reported that there were few severe hypoglycemic episodes. Children with type 1 DM: No RCTs comparing CSII with MDI in children were identified and included in the HTA report. The authors of this HTA report did not identify any studies reporting on cost-effectiveness of CSII compared with MDI. They estimated the additional cost or cost savings associated with CSII based on the perspective of the National Health Services (NHS) in UK. The costs were given in ( 1 = C$2.43 in 30 June 2004, the year of publication of the HTA report). The cost of the pump (Diestronic H-Tron, Diestronic D-Tron, or MiniMed 508) varied between 2350 and 2562 and the annual cost of consumables required with the pump varied between and Details of these costs are shown in Appendix 2. Other additional costs associated with CSII include 74 for general patient management, and for CSII patient education. In the short term, it has been shown there is a reduction in insulin use with CSII as compared to MDI. The authors assumed a reduction in insulin cost of The evidence on the rate of hypoglycemic events with CSII as compared to MDI is conflicting. However the authors assumed a reduction in one hypoglycemic event per year with CSII as compared to MDI. The estimated additional costs (i.e. mean patient-level costs for 1, 4 and 8 years) associated with CSII versus MDI are shown in Table 1. Table 1: Estimated Additional Patient-Level Costs Associated with CSII as Compared to MDI (Figures in Parentheses Indicate the Future Costs with 6% Discounting) Time frame Assumed pump life (year) Total cost ( ) Year 1 >1 3,571 to 3,878 Years 1 to 4 4 6,058 to 7,081 (5,790 to 6,722) Years 1 to ,894 to 13,941 (10,201 to 11,871) Years 1 to ,096 to 12,178 (8,728 to 10,429) The additional cost (discounted) of CSII compared with MDI varied according to the make of the pump and the estimated life of the pump. The additional cost per annum was 1091 (with the least expensive pump and a pump life of 8 years), and 1680 (with the most expensive model and a pump life of four years). The authors were unable to perform a cost-effectiveness analysis as meaningful health outcome data which were quantifiable were unavailable. In summary, this HTA report 5 showed that most studies on adults with type 1 DM showed that compared to MDI, CSII resulted in greater improvement in A1c but no significant difference in hypoglycemic events. It was difficult to ascertain from the limited evidence on pregnant women with DM, if there was a difference between CSII and MDI therapy in terms of A1c, hypoglycemia or pregnancy outcomes. The two studies on adolescent with type 1 DM showed variable results. The authors did not conduct a cost-effectiveness analysis but provided some cost data. Assuming a pump life of eight years, for a period of eight years the additional cost associated with CSII, compared to MDI ranged between 10,096 and 12,718 (without discounting) and between 8,728 and 10,429 (with discounting). An HTA report by Côté and St-Hilaire, 9 published in 2005 evaluated CSII in comparison to MDI in patients with type 1 DM. They included in their report the findings of two HTA reports; one Insulin Pumps for Type 1 Diabetes 4

5 published in 2000 by the Catalonian HTA agency and the other published in 2002 by an HTA agency in the UK. In addition, they examined the literature published since 2002 and included in their report the findings from 34 clinical studies. These studies included 8 RCTs and 26 observational studies published between 2002 and Of the 8 RCTs, three (two parallel and one crossover design) were with adult patients and five (3 parallel and two crossover designs) with pediatric patients. Of the 26 observational studies, 10 (four cohort and six case-series) were with adult patients and 16 (two cohort and 14 case-series) with pediatric patients. They included results of a small survey and a Quebec-based cost analysis. They also included findings from four economic studies published between 2002 and Of these four, only one was a full published report. The RCTs showed a modest improvement in A1c levels in adult type 1 DM patients with CSII compared to MDI, but not in children with type 1 DM. However, the observational studies involving children with type 1 DM showed a greater improvement in A1c with CSII. For adult patients, CSII was as effective as MDI with insulin glargine, with respect to glycemic control. The RCTs showed no difference in severe hypoglycemic events in adults or children with CSII compared to MDI. The observational studies reported fewer hypoglycemic events with CSII compared to MDI. This could be because CSII is offered to patients who are most likely to benefit from it. The RCTs and observational studies showed no apparent improvement in QoL of type 1 DM patients with CSII. A survey involving 30 CSII users in Québec showed that patients using CSII found its use to be beneficial for several aspects of their daily life. These patients were more motivated than average, highly organized and used the pump successfully. Hence, these results cannot be generalized. The study, which was available as a full published report, showed that CSII is cost-effective if prescribed to patients who are mostly likely to benefit from it (i.e. those who experienced higher rates of hypoglycemia and required hospitalization). The Quebec based cost analysis showed that compared to MDI, the annual estimated cost of CSII is C$4,756 more per user. This estimate takes into account an average life span of five years for the pump and associated costs for training. Details of cost of supplies for CSII and MDI are shown in Appendix 3. This HTA report 9 concluded that for adult DM patients, compared to MDI with neutral protamine Hagedorn (NPH), CSII may offer a modest advantage in terms of glycemic control, with no apparent additional risk. But in children with DM, CSII does not appear to be more beneficial than MDI with NPH. In adult DM patients, CSII is as effective as MDI with insulin glargine. The evidence for the effect of the modest improvement with CSII in terms of preventing long-term complications and improving QoL are not sufficient to provide robust cost-effectiveness information. Systematic reviews and meta-analyses Four systematic reviews comparing CSII with MDI in type 1 DM patients were identified. In addition, one systematic review is in progress and at the present time the published protocol 7 is available. Weissberg-Benchell et al. 10 in 2003 published a systematic review and meta-analyses on the metabolic and psychosocial impact of CSII in adults, adolescents and children with type 1 DM. They included 52 studies comparing CSII use with other diabetes treatments (e.g. conventional therapy or MDI) either for the same individual or between groups. The studies were published Insulin Pumps for Type 1 Diabetes 5

6 between 1979 and 2001, with 22 published before 1987 and 13 published since the release of results of the DCCT in Weissberg-Benchell et al. 10 included both randomized and nonrandomized studies. These 52 studies were categorized as paired design (41) and parallel design (11) and included a total of 1,547 patients [mean sample size = 30, (range: 3 to 177)]. Of these studies, 12 included only pediatric patients, 33 included only adult patients and seven included both pediatric and adult patients. The mean age of the patients was 26 years (range: 2.3 to 49.8 years) and the average time on CSII was 53 weeks (range: 4 to 234 weeks). Not all studies reported all outcomes. Two different methods were used to analyze data, based on the study design (parallel or paired). For the parallel design the pooled WMD with 95% CI were estimated by conducting a meta-analysis. For the paired design, pre- and post-pump summary means, weighted by the inverse of the respective variances, were estimated using a repeatedmeasures model. In Table 2, results are presented categorized by study design for each of the outcome variables (i.e. A1c, blood glucose, insulin dose, and body weight) The authors pooled randomized and non-randomized studies in the same meta-analysis. CSII therapy resulted in significant improvements in glycemic control. There was variability in results reported for insulin dose requirements, change in body weight and QoL. Some studies showed a decrease in hypoglycemic events and some studies showed no change with CSII when compared with MDI therapy. Studies before 1993 showed increased risk of diabetic ketoacidosis (DKA) with CSII but in studies after 1993 the results were variable. Some studies reported infections at the catheter site. Table 2. Outcome Data from the SR by Weissberg-Benchell et al. 10 Study design Outcome variable Result* Parallel A1c CSII MDI/CT: WMD (95% CI) = 0.95 (0.8, 1.1) Paired (pre-post) A1c Pre CSII: 9.36±0.22% Post CSII: 8.96±0.11% (p=0.039 ) A1c (when average duration Pre CSII: 8.68±0.06% of pump therapy 1 year) Post CSII: 7.48±0.22% (p<0.001 ) A1c (when average duration Pre CSII: 9.4±0.23% of pump therapy < 1 year) Post CSII: 9.2±0.01% (p=0.3 ) Parallel Blood glucose CSII MDI/CT: WMD (95% CI) = (13.45, 21.16) Paired (pre-post) Blood glucose (mg/dl) Pre CSII: ±9.13 Post CSII: ±4.71 (p<0.001) Parallel Insulin dose Four studies reported a decrease, one reported an increase and one reported no Paired (pre-post) change in insulin dose with CSII therapy Insulin dose (units per day) Pre CSII: 53.69±0.11 Post CSII: 44.19±0.07 (p<0.001) Insulin dose (units per kg per day) Pre CSII: 0.74±0.04 Post CSII: 0.62±0.02 (p<0.001) Parallel Body weight Two studies reported a decrease and one reported an increase in weight with CSII therapy Paired (pre-post) Body weight (kg) Pre CSII: 68.24±0.27 Post CSII: 71.21±0.31 (p<0.001) *Values expressed as mean±sd unless otherwise stated CI = confidence interval, CSII = continuous subcutaneous insulin injection, CT = conventional insulin therapy, MDI = multiple daily injection, WMD = weighted mean difference. Insulin Pumps for Type 1 Diabetes 6

7 The systematic review and meta-analysis by Weissberg-Benchell et al. 10 included both randomized and non-randomized studies and showed that CSII therapy was associated with improved glycemic control compared with conventional therapy or MDI. There appeared to be a decrease in hypoglycemic events and increase in DKA with CSII therapy but results varied. There was no conclusive evidence for improvement in QoL. Retnakaran et al. 11 in 2004 published a systematic review and meta-analysis comparing CSII with MDI in Type 1 DM patients, where both intensive regimens used short acting insulin analogues (insulin lispro or insulin aspart). They included three RCTS [design: parallel (2), crossover (1); with a total of 147 type 1 DM patients; range: 27 to 79 patients] published between 2000 and Mean patient age ranged from 36 to 44 years, and treatment duration ranged between four and nine months. The studies differed significantly in mean baseline A1c levels (7.95%, 8.2% and 9.27%). Reductions in outcome variables were expressed as WMD (95% CI). Reductions obtained with CSII compared to MDI were 0.34% (-0.09%, 0.77%), 9.7% (-11.3%, 35.8%), and 12.2 (6.4, 18.0) for A1c, rate of hypoglycemic episodes, and insulin dose (units/day) respectively. There was one incidence of diabetic ketoacidosis (DKA) in each of the two treatment groups. The authors also showed that for A1c, the relative benefit of CSII over MDI increased with higher baseline A1c. In summary, Retnakaran et al. 11 showed a modest improvement in A1c with CSII as compared to MDI, but the difference was statistically non-significant. However, for patients with higher baseline A1c, greater benefits were achieved with CSII. Barnard et al. 12 in 2007 published a systematic review on the quality of life (QoL) associated with insulin pump use in Type 1 DM. They consulted with professional groups and searched electronic databases and published references up to July They included 17 studies [RCT (5), non-randomized controlled study (3), comparative observational study (2), observational (pre-post) study (7)] published between 1988 and Five RCTs were included. One RCT with adult type 1 DM patients (13 on CSII and 14 on MDI) showed that there were no statistically significant differences between the treatment groups for any of the Diabetes Quality of Life (DQOL) subscales used. One RCT with 79 adult type 1 DM patients treated with CSII or MDI showed moderate improvements in general health status and some aspects of health-related QoL using the SF-36 questionnaire. Two RCTs (crossover design) with type 1 DM pediatric patients (23 patients in one RCT and 19 in the other) using DQOL showed there was no significant difference in QoL between the CSII and MDI groups. One RCT with 26 type 1 DM pediatric patients, using DQOL and Parent Rated QoL questionnaires, showed that there were no differences in parental QoL. After the completion of the review, the authors identified one RCT, with 272 DM patients, which showed that overall DQOL score was significantly higher with CSII compared to MDI. Three non-randomized controlled studies were included. One study with adult type 1 DM patients with well-matched groups (49 on CSII and 79 on MDI) showed no significant differences between the two groups with respect to QoL, well-being or satisfaction with treatment, using DQOL questionnaire, Diabetes Treatment and Satisfaction Questionnaire and World Health Organization well-being questionnaire. One study with adult type 1 DM patients (12 on CSII and 16 on MDI) showed that QoL was better with CSII than with MDI in four subscales of an unvalidated Insulin Therapy Related Quality of Life questionnaire. One study with adolescent type 1 DM patients showed improvement in self-efficacy, depression and QoL in both CSII and MDI users. Insulin Pumps for Type 1 Diabetes 7

8 Two comparative observational studies were included. One study with 197 adult type 1 and type 2 DM patients (140 on CSII and 57 on MDI) used the Insulin Delivery System Rating Questionnaire. CSII users scored lower than MDI users on daily activity interference, diabetes worry and social burden and higher on treatment satisfaction and psychological well-being. The other study compared QoL in 183 patients who had self-selected CSII, conventional therapy or intensive conventional therapy. Those using CSII and remaining on treatment for one year reported greater improvements in treatment satisfaction, as measured by the diabetes-specific treatment satisfaction scale, than those choosing either form of conventional therapy. Seven uncontrolled observational studies (pre-post type), including a total of 387 DM patients assessed QoL after changing to CSII therapy from either conventional therapy or intensified conventional therapy. The main assessment tool was the DQOL Questionnaire. Four studies were in adults and three in children and adolescents. Of the four studies in adults one showed no difference and three showed an improvement in QoL with CSII. The three studies with children and adolescents mentioned improvement in QoL with CSII, but one of these three studies provided no evidence. Barnard et al. 12 concluded that there was conflicting evidence reported in the various studies on QoL. Many of the included studies were small and of low quality. There is uncertainty concerning QoL benefits of CSII. Ludvigsson and Samuelsson 13 in 2007 published a critical review and meta-analysis on CSII versus MDI. Their main focus was on children and adolescents with type 1 DM but they included in addition results for adult patients. They included 18 studies (with a total of 935 children and adolescents with type 1 DM) comparing CSII with conventional MDI or TDI (three times daily injection). The studies were published between 2001 and Of these 18 studies, four were RCTs (with a total of 107 patients) and the remaining 14 were observational studies (six were retrospective studies). In addition to these 18 studies, they also included 13 more studies with a total of 1180 DM patients, comparing CSII with MDI using rapid-acting insulin analogues together with insulin glargine (MDI-G). The studies were published between 2003 and Only four of these studies were with children and adolescents, and of these, three were RCTs; the remaining nine studies were with adults. In Table 3, the results are presented categorized by the patient population [(children and adolescents) and adults] and by the interventions compared [(CSII versus MDI or TDI) and (CSII versus MDI-G)]. Compared to MDI or TDI or MDI-G, CSII resulted in lower A1c levels and lower daily insulin doses in children and adolescents (Table 3). Some studies showed that CSII, compared to the other modalities, resulted in an increase in body mass index (BMI), whereas other studies showed no difference or a decrease in BMI. Some of the studies showed that CSII resulted in more diabetic ketoacidosis but fewer severe hypoglycemic episodes compared to the other modalities, however a few studies showed no difference. A few studies reported infection at the catheter site. One RCT showed that QoL was better with CSII therapy than with MDI, while another RCT showed no difference. Two RCTs (with children) showed that QoL was the same with CSII and MDI-G. However in one of these two RCTs, 14/16 patients continued with CSII and 12/16 patients in the MDI-G group switched to CSII. Insulin Pumps for Type 1 Diabetes 8

9 Table 3: Outcome Data in the Review by Ludvigsson and Samuelsson 13 DM patients Children and adolescents (age range: 3 to 17 years) Children and adolescents (age range: 4 to 15 years) Adults (age range: 31 to 43 years; no mention of age in 3 studies) Interventions * Study design CSII versus MDI or TDI CSII versus MDI-G CSII versus MDI-G Number of DM patients per study All (18) 16 to (-0.64,-0.46); (18) Only RCTs [(4) includes two crossover] Difference (CSII other) in A1c (%) Insulin dose (units/kg/day) 16 to (-0.51, -0.11); (4) All (4) 19 to (-0.77, -0.09); (4) Only RCTs (2) 19 to (-0.64, -0.28) (2) All (9) 32 to (-0.19, -0.11); (7) Only RCTs [(3), includes 1 crossover] 38 to (-0.21, 0.07); (2) (-0.21, -0.13); (8) (-0.23, -0.05); (2) (-0.18, -0.10); (4) (-0.33, 0.03) (2) -6.3 (-7.2, -5.4) (3) -5.2 ** (-5.91, -4.49); (1) * All = RCTs + observational studies; the single number in parenthesis indicates the number of studies in the particular group or number of studies used for the particular analysis; difference expressed as mean with 95% confidence interval; ** units/24 hour. CSII = continuous subcutaneous insulin injection, MDI = multiple daily injection, MDI-G = multiple daily injection using in addition insulin glargine, TDI = three times daily injection In the systematic review by Ludvigsson and Samuelsson 13 there appears to be a slightly greater reduction in A1c with CSII compared to the other modalities. Some of the studies showed that CSII compared to the other modalities resulted in more diabetic ketoacidosis but fewer severe hypoglycemic episodes. QoL data is sparse and it is uncertain whether CSII significantly improves QoL. A Cochrane systematic review by Misso et al. is being conducted to assess the effects of CSII compared to MDI in type 1 DM patients; at the present time only the protocol 7 is available. Cost-effectiveness of insulin pumps Economic evaluations Three economic studies published in 2003, 2005 and 2007 were identified. Scuffham and Carr 14 used a Markov model to estimate the cost-effectiveness of CSII compared to MDI for patients with type 1 DM. This study was partly funded by Medtronic (manufacturer of an insulin pump) and was published in Monte Carlo simulations were undertaken for Insulin Pumps for Type 1 Diabetes 9

10 10,000 hypothetical patients. The authors used the perspective of the National Health Service (NHS) of UK and a time horizon of eight years (the estimated life of an insulin pump). The model was populated using data on procedures and costs for England and Wales. The main outcome modeled was Quality Adjusted Life Years (QALYs), derived from one study on QoL. Overall QoL with MDI was considered to be 5.3% worse compared with CSII. Summary of costs used in the analysis are shown in Appendix 4. Results for the base case are shown in Table 4. Table 4: Base Case Estimates * with an Eight Year Time Horizon (Scuffmann And Carr 14 ) Item Treatment type Difference between the two treatments CSII MDI CSII - MDI Cost per patient ( ) 9,514±1,337 4,052±1,792 5,462±897 QALY 7.32± ± ±0.20 ICER [Incremental cost ( )per QALY] NA NA 11,461±3,656 * Values are presented as reported by the authors and are expressed as mean ± standard deviation, costs are for 2001 UK pounds ( 1 = C$2.14 in 29 June ) CSII = continuous subcutaneous insulin infusion, ICER = Incremental cost-effectiveness ratio, MDI = multiple daily injection, NA = not applicable, QALY = Quality Adjusted Life Year A sensitivity analysis was undertaken to determine factors that affect the results. Results were most sensitive to the number of hypoglycemic episodes per patient and to the utility weights used to estimate QALYs. If the utility gained from CSII was reduced to 2% from the 5.3% used for the base case analysis the ICER would be doubled. A two-way sensitivity analysis showed that ICER was 33,000 per QALY when both the utility weight and hypoglycemia rate were low and 380 per QALY when both these factors were high. The discount rate had a relatively small effect on the results. This study 14 estimated an ICER of 11,461±3,656 per QALY gained. CSII was most costeffective in patients who had more than two severe hypoglycemic events per year and who required hospitalization at least once a year. CSII is less beneficial for those with well controlled DM and few severe hypoglycemic episodes. Roze et al. 15 in 2005 published a study on the long-term costs and outcomes with CSII compared to MDI in type 1 DM patients in the UK. This study was funded by Medtronic. The analysis was done using the CORE Diabetes model (which uses standard Markov/Monte Carlo simulation techniques). Disease progression was simulated in a cohort with baseline characteristics taken from published UK studies (mean age 26 years, duration of diabetes 12 years, mean A1c 8.68%). All costs were expressed in 2003 values ( 1 = C$2.24, 30 June ). The authors used the perspective of the National Health Service (NHS) of UK and a life-time horizon (of 60 years). A discounting rate of 3.0% per annum was applied to both costs and clinical outcomes. Results for the base case are shown in Table 5. A sensitivity analysis was undertaken to test the robustness of the results. The results were most sensitive to variations in hypoglycemia rates and in A1c levels with CSII compared to MDI. Insulin Pumps for Type 1 Diabetes 10

11 Table 5: Base Case Estimates * with a Life-Time Horizon (Roze Et Al. 15 ) Item Treatment type Difference between the two treatments CSII MDI CSII - MDI Direct cost ( ) 80,511±1,257 61,104±1,249 19,407±1,727 QALY 12.03± ± ±0.19 ICER [Incremental cost ( ) per QALY] NA NA 25,648 * Values are presented as reported by the authors, and costs are for 2003 UK pounds ( 1 = C$2.24, 30 June ) CSII = continuous subcutaneous insulin infusion, ICER = Incremental cost-effectiveness ratio, MDI = multiple daily injection, NA = not applicable, QALY = Quality Adjusted Life Year This model-based analysis 15 comparing CSII with MDI in Type 1 DM patients in the UK demonstrated that improvements in glycemic control associated with CSII, resulted in reduced diabetes related complications, leading to improvements in life expectancy and QALYs. Compared to MDI, CSII was associated with an ICER of 25,648 per QALY, representing good value for money by UK standards. Cohen et al. 16 in 2007 published a study on the long-term costs and outcomes with CSII (using insulin lispro or insulin aspart) compared to MDI [using NPH plus insulin lispro or insulin aspart] in type 1 DM patients in Australia. This study was funded by Medtronic Australasia. The analysis was done using the CORE Diabetes model (which uses standard Markov/Monte Carlo simulation techniques). Disease progression was simulated in adult patients (mean age 43 years, duration of DM 17 years, mean A1c 8.2%) and adolescent patients (mean age 17 years, duration of DM 6 years, mean A1c 8.9%) with baseline characteristics taken predominantly from Australasian National Diabetes Information Audit and Benchmarking. All costs were expressed in 2006 A$ (A$1 = C$0.83, 30 June ). Only direct medical costs were considered; indirect and non-medical costs were excluded. Some cost details are provided in Appendix 5. The authors used the perspective of the Australian single-payer health care system and a life-time horizon (of 60 years). A discounting rate of 5.0% per annum was applied to both costs and clinical outcomes. For the base case, the ICER was A$74,147 per QALY for adult patients. Results for the base case (using a pump life of 8 years) are shown in Table 6. Sensitivity analyses showed that these findings in Australian adults and adolescents with type 1 DM were robust for variations in a range of assumptions, but considerably sensitive to changes in A1c levels. Altering the reduction in A1c from 1.2% (base case) to 0.51% resulted in an ICER of A$124,201 per QALY in adult patients. Reducing the life span of the pump from eight years (base case) to six years resulted in an ICER of A$84,371 per QALY for adult patients. The sensitivity analyses showed that compared to MDI, CSII was associated with ICERs ranging between A$53,022 and A$259,646 per QALY gained, with ICERs for most scenarios being below the threshold value (A$76,000 per life year gained), hence representing good value for money in the Australian setting. Insulin Pumps for Type 1 Diabetes 11

12 Table 6: Base Case Estimates * with a Life-Time Horizon (Cohen Et Al. 15 ) Population Adults with type 1 DM Adolescents with type 1 DM Item Treatment type Difference between the two treatments CSII MDI CSII - MDI Direct cost (A$) 123,402±2,113 88,760±2,055 34,642 QALY 7.95± ± ICER [Incremental cost NA NA 74,147 (A$) per QALY] Direct cost (A$) 148,918±2, ,139±2,320 41,779 QALY 9.64± ± ICER [Incremental cost NA NA 74,661 (A$) per QALY] * Values are presented as reported by the authors and are expressed as mean ± standard deviation, and costs are for 2006 Australian dollars (A$1 = C$0.83, 30 June ) CSII = continuous subcutaneous insulin infusion, ICER = Incremental cost-effectiveness ratio, MDI = multiple daily injection, NA = not applicable, QALY = Quality Adjusted Life Year This model-based analysis 16 comparing CSII with MDI, in Type 1 DM patients in Australia showed that ICERs were A$74,147 and A$74,661 per QALY gained for adults and adolescent type 1 DM patients respectively, representing good value for money by Australian standards. The results were sensitive to changes in A1c levels. Life cycle of insulin pumps No HTA reports, systematic reviews or studies assessing the life cycle of insulin pumps were identified. Manufacturer s of insulin pumps provide a minimum four year warranty. 18 In a costeffectiveness analysis, 14 a pump life of eight years was used for the base case scenario. In a cost analysis 9 a pump life span of five years was used. Limitations The studies included in the HTA report by Colquitt et al. 5 were fairly old and generally of low quality. These studies were done with older and bulkier pumps hence their findings may not be relevant with respect to newer pumps. Further progress has also been made with MDI. The HTA report by Côté and St-Hilaire 9 included more recent studies (between 2002 and 2005) but many of those were case-series studies and prone to bias. Many of the studies included in the SRs had a small number of patients. Of the four systematic reviews, three 10,12,13 included both RCTs and observational studies. Observational studies provide results from a real-world setting but have potential limitations which need to be taken into consideration in interpreting results. Since the treatment is not allocated through randomization, there is a possibility of confounding due to imbalances in unmeasured and unmeasurable risk factors. Also two of systematic reviews included many older studies, results of which may no longer be relevant in the present setting as many improvements in both CSII and MDI have occurred over the years. One systematic review 11 included only RCTs but there were only three included and the majority were small trials. Insulin Pumps for Type 1 Diabetes 12

13 QoL data were sparse. The studies used different tools to assess QoL, making comparisons between studies difficult. The three economic studies were model-based and involved a number of assumptions. However, sensitivity analyses were conducted to determine which factors impacted the results and to what extent. Only direct medical costs were considered in the analyses. Non-medical costs such as lost productivity and transportation costs were not considered. The economic studies relate to settings in UK and Australia so may not be directly applicable to the Canadian setting. No evidence on the life cycle of insulin pumps was identified. Conclusions and implications for decision or policy making: Two HTA reports, 5,9 four systematic reviews, and three economic studies comparing the clinical and cost-effectiveness of CSII with MDI in type 1 DM patients were identified. No evidence on the life cycle of insulin pumps was identified. Overall there appeared to be a modest improvement in A1c levels with CSII. Data on rates of hypoglycemic events and effect on QoL with CSII as compared to MDI were conflicting. CSII appeared to be beneficial in patients with high baseline A1c values. A Cochrane systematic review is being conducted 7 and may provide more information concerning CSII therapy. All three economic studies (two conducted in UK and one in Australia) estimated ICER and demonstrated good value for money from the perspective of the health care system of the particular country. The ICERs were sensitive to variations in A1c levels, hypoglycemic event rates and utility weights associated with QoL. All the three studies were funded by the manufacturer of the pump. Careful patient selection is widely recognized as an important element in the judicious use of CSII therapy. 11 Screening of patients for suitability of CSII may be useful. 14 CSII may be an option for patients who are unable to achieve good glycemic control with MDI. Patients who have higher rates of severe hypoglycemic events and require hospitalization may benefit with CSII therapy. A high rate of hypoglycemia is, however, not sufficient criteria for CSII therapy. Patients need to be motivated and be able to use the pump and the risk of discontinuation needs to be small. 14 CSII appears to be a good investment when targeted to those who might benefit most. 14 Prepared by: Srabani Banerjee, MSc, PhD, Research Officer Kelly Farrah, MLIS, Information Specialist Health Technology Inquiry Service htis@cadth.ca Tel: Insulin Pumps for Type 1 Diabetes 13

14 References: 1. Canadian Diabetes Association clinical practice guidelines. Toronto: The Association; Available: (accessed 2008 Apr 14). 2. Powers AC. Diabetes mellitus. In: Kasper DL, Fauci AS, Longo DL, Braunwald E, Hauser SL, Jameson JL, editors. Harrison's principles of internal medicine. 16th ed. New York: McGraw-Hill; p Diabetes: facts and figures. [website]. Ottawa: Public Health Agency of Canada; Available: (accessed 2008 Mar 10). 4. Disorders of carbohydrate metabolism. In: Beers MH, Berkow R, editors. The Merck manual of diagnosis and therapy. 17th ed. Whitehouse Station (NJ): Merck Research Laboratories; p Colquitt JL, Green C, Sidhu MK, Hartwell D, Waugh N. Clinical and cost-effectiveness of continuous subcutaneous insulin infusion for diabetes. Health Technol Assess (Winch Eng) 2004;8(43):iii-95. Available: (accessed 2008 Apr 14). 6. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group. N Engl J Med 1993;329(14): Misso ML, O'Connor DA, Egberts KJ, Shaw J. Continuous subcutaneous insulin infusion (CSII) versus multiple insulin injections for type 1 diabetes mellitus [Cochrane protocol]. Cochrane Database of Systematic Reviews: Protocols 2005 Issue 1 John Wiley & Sons, Ltd Chichester, 2005;(1). 8. Answers to common questions for members of the Canadian Diabetes Association. Toronto: Ingle International; Available: (accessed 2008 Apr 4). 9. Côté B, St-Hilaire C. Comparison of the insulin pump and multiple daily insulin injections in intensive therapy for type 1 diabetes. Montreal: Agence d'évaluation des technologies et des modes d'intervention en santé; Available: (accessed 2008 Apr 9). 10. Weissberg-Benchell J, Antisdel-Lomaglio J, Seshadri R. Insulin pump therapy: a metaanalysis. Diabetes Care 2003;26(4): Available: (accessed 2008 Apr 14). 11. Retnakaran R, Hochman J, DeVries JH, Hanaire-Broutin H, Heine RJ, Melki V, et al. Continuous subcutaneous insulin infusion versus multiple daily injections: the impact of baseline A1c. Diabetes Care 2004;27(11): Available: (accessed 2008 Mar 25). 12. Barnard KD, Lloyd CE, Skinner TC. Systematic literature review: quality of life associated with insulin pump use in Type 1 diabetes. Diabet Med 2007;24(6): Insulin Pumps for Type 1 Diabetes 14

15 13. Ludvigsson J, Samuelsson U. Continuous insulin infusion (CSII) or modern type of multiple daily injections (MDI) in diabetic children and adolescents a critical review on a controversial issue. Pediatr Endocrinol Rev 2007;5(2): Scuffham P, Carr L. The cost-effectiveness of continuous subcutaneous insulin infusion compared with multiple daily injections for the management of diabetes. Diabet Med 2003;20(7): Roze S, Valentine WJ, Zakrzewska KE, Palmer AJ. Health-economic comparison of continuous subcutaneous insulin infusion with multiple daily injection for the treatment of Type 1 diabetes in the UK. Diabet Med 2005;22(9): Cohen N, Minshall ME, Sharon-Nash L, Zakrzewska K, Valentine WJ, Palmer AJ. Continuous subcutaneous insulin infusion versus multiple daily injections of insulin: economic comparison in adult and adolescent type 1 diabetes mellitus in Australia. Pharmacoeconomics 2007;25(10): year currency converter. Ottawa: Bank of Canada; Available: (accessed 2008 Apr 8). 18. ADP: insulin pumps and supplies. In: Ministry of Health and Long-Term Care [website]. Toronto: The Ministry; Available: (accessed 2008 Apr 4). Insulin Pumps for Type 1 Diabetes 15

16 Appendices: Appendix 1: Characteristic of Studies Included in HTA Report by Colquitt et al. 5 Population Adults with type 1 DM Pregnant women (3RCTs in type 1 DM and 1 RCT in both type 1 and type 2 DM) Number of studies 14 (4 parallel RCTs, 7 crossover RCTs and 3 nonrandomized crossover studies) 4 (all RCTs) Adolescents 2 (all crossover RCTs) Publication year 1982 to 2001 (only two studies in 2000 to 2001) 1986 to to 1989 * Mean value unless indicated otherwise Study country Canada (3), Denmark (3), France (1), Germany (1), Italy (1), Norway (2), UK (1), and USA (2) Germany (1), Italy (2), and USA (1) Canada (1) and USA (1) Number of patients Mean age * (year) Duration (month) 5 to to to to to to to 20 (range) 4 to 6 Appendix 2: Cost Of Maintenance And Consumables Required With The Pump (Colquitt et al., ) Item Annual Cost ( ) * Pump maintenance to Insulin cartridges/reservoirs to Adaptors for insulin cartridges/reservoirs Infusion sets to Batteries to * Cost of pump (Diestronic H-Tron, Diestronic D-Tron, or MiniMed 508) varied between 2350 and The exchange rate for 2004 was 1 = C$2.3 Insulin Pumps for Type 1 Diabetes 16

17 Appendix 3: Estimates of Costs for CSII and MDI (Côté and St. Hilaire, ) Item Annual Cost (C$) for Year 1 Annual Cost Difference (C$) for Year 1 Annual Cost (C$) for subsequent years Annual Cost Difference (C$) for subsequent years CSII MDI CSII-MDI CSII MDI CSII-MDI Lancets (blood glucose monitor) Test strips (blood glucose 2,310 1, ,190 1, monitor) Urine ketone test strips Antiseptic swaps Transparent adhesive dressings Costs are per patient and in 2004 C$, CSII = continuous subcutaneous insulin infusion, MDI = multiple daily injection. Appendix 4: Summary of costs used in the analysis by Scuffham and Carr, Item Cost ( ) * Cost due to hypoglycemic event: - hospital treatment cost 779±187 - consultation cost 61±14 -paramedic cost 218±74 Cost due to ketoacidosis (in-patient treatment cost) 821±203 Cost of CSII pump 2,180±180 Service charge for CSII pump 325±50 Monthly cost of CSII consumables 63±6 Monthly cost of MDI consumables 10±3 Monthly cost of insulin 29±10 * Cost are in 2001 UK pounds ( 1 = C$2.14 in 2001) cost presented as mean±standard deviation, assuming one service in month 1 of year 6. CSII = continuous subcutaneous insulin infusion, MDI = multiple daily injection. Insulin Pumps for Type 1 Diabetes 17

18 Appendix 5: Annual Costs of CSII and MDI Used in the Analysis by Cohen et al., Item Cost (A$) * Adult type 1 DM patients using Adolescent type 1 DM patients using CSII MDI CSII MDI Pump NA NA Insulin 1, , , , Consumables (pump supplies, syringes and needles) Self monitoring blood glucose (lancets and glucose strips) 2, , , , , , Medical professional 1, , assistance Total for year 1 7, , , , Total for year 2 and beyond 6, , , , * Cost are in 2006 A$ (A$1 = C$0.83, 30 June ) Annual costs were calculated based on a life span of 8 years for the pump. CSII = continuous subcutaneous insulin infusion, MDI = multiple daily injection. Insulin Pumps for Type 1 Diabetes 18