The Outcome of Care in People With Type 1 and Type 2 Diabetes Following Switching to Treatment With Either Insulin Glargine or Insulin Detemir in Routine General Practice in the UK: a Retrospective Database Analysis

By Currie, Craig J; Poole, Chris D; Tetlow, Tony; Holmes, Paul; McEwan, Phil

Key words: Diabetes – Effectiveness – Glycaemia – HbA^sub 1c^ – Hypoglycaemia – Outcome – Short-acting insulin – Weight

ABSTRACT

Background: Two alternative basal analogue insulin treatments are available in the UK: insulin glargine and insulin detemir. The purpose of this study was to compare the outcome of treatment with these two alternative products since their launch using routinely available, unselected data from general practice. Glargine was launched in the UK in 2002, whereas detemir was launched in 2004.

Methods: The study used proprietary data from The Health Improvement Network – THIN (around 300 GP practices) using standard statistical methods. A detailed description of each patient was available for up to 15 years that included their diagnoses, test results and treatments. The two alternative treatments were compared for 9 months following switching to one of these treatments by quarterly periods. The primary outcome measure was change in HbA^sub 1c^, whilst secondary measures were change in weight, reported hypoglycaemia and treatment discontinuation. Subjects were split by type of diabetes.

Results: Data were available for 694 people treated with detemir and 5683 people treated with glargine. The demographic and clinical profiles of the respective groups were almost identical. There was a notable improvement in HbA^sub 1c^ following treatment switching to either glargine or detemir; however, there was a trend for improved diabetes control with glargine in both type 1 and type 2 diabetes. Glargine resulted in less hypoglycaemia (relative risk = 0.75; p < 0.05). Detemir resulted in a trend for less weight gain, but this did not achieve statistical significance. There was no convincing evidence of a difference in treatment discontinuation patterns.

Conclusion: On balance, glargine showed marginal improvement over detemir in diabetes-related outcomes. There was a marked improvement in hypoglycaemia when treated with glargine over detemir.

Introduction

Achieving a physiological state of normoglycaemia in people with diabetes reduces the likelihood of progressing to diabetes-related vascular complications1. In people with type 1 diabetes, normoglycaemia can only be achieved via the administration of exogenous insulin; or, unusually, by islet transplantation. In order to achieve normoglycaemia in people with type 2 diabetes there is a typical pattern of treatment of increasingly more aggressive therapy with oral hypoglycaemic agents (OHAs) following (and sometimes in conjunction with) diet and exercise. In the UK the typical pattern of OHA treatment progression involves initial monotherapy with metformin or sulphonylureas then combination therapy with these two classes of agent. Thiazolidinediones are sometimes used in first- and second-line regimens. Once this regimen has failed and pancreatic insulin secretion is all but exhausted, combination therapy is likely to then involve the addition of thiazolidinediones or more usually, insulin2.

Insulin is a complex molecule with an equally complex mode of action3. In recent years the wild-type insulin molecule has been chemically engineered to change its pharmacokinetic properties, resulting in the development of various insulin analogue products that allow people who require insulin to better mimic the physiological properties of normal pancreatic insulin secretion15. What is termed a basal-bolus insulin regimen is now used commonly to achieve better glycaemic control, particularly in type 1 diabetes. Basal-bolus treatment involves the administration of long-acting insulin (the basal component) along with prandial administration of short-acting insulin (the bolus component).

HbA^sub 1c^ can be regarded as a measure of treatment optimisation and of treatment compliance. In clinical trials the conditions are such that it is likely that patients are more likely to achieve good diabetes control as evidenced by lower HbA^sub 1c^ values. This fully justifies the use of non-inferiority designs of clinical trials of diabetes-related products. However, following product launch, stringent trial conditions are lost and it is at this stage that it is possible to properly determine the ‘actual’ impact of any novel treatment. Randomised controlled trials (RCTs) are regarded as providing the gold standard of scientific evidence”; however, in diabetes we argue that it is necessary to generate and interpret evidence from both RCTs and from general clinical practice in order to characterise the value of a novel product because the routinely used primary end point, HbA^sub 1c^, will be directly affected by the atypical trial conditions. This is also generally true of other outcome measures. Naturally, evidence from routine clinical practice can only be collated and evaluated following product launch.

The purpose of this study was therefore to compare actual clinical outcomes of two of the most recent basal insulin analogue products, insulin glargine (glargine) and insulin detemir (detemir) in routine general practice in the UK. These two products are regarded as commercially competing because they have similar pharmacokinetic properties78. Their respective manufacturers (sanofi- aventis Pty, Paris, France and Novo Nordisk A/S, Bagsvaerd, Denmark) have each made claims of superiority for their products. Clinical trials for these products were designed to demonstrate non- inferiority versus the comparative treatments – typically neutral protamine hagedorn (NPH) insulin. With regard to HbA^sub 1c^, both glargine and detemir have been shown to be non-inferior to NPH insulin. However, glargine is regarded as having a more physiological pharmacokinetic profile than detemir since it has a longer half-life and other differences7,9. On the other hand, detemir is thought to have benefit in terms of less weight gain8, insulin being an anabolic steroid that normally results in continuous weight gain. Both products have been shown, or have claimed, to reduce the likelihood of hypoglycaemia7,8 and data from only one single head-to-head trial was available for direct comparison9. From the clinical trial evidence it is difficult to favour one product over the other using existing evidence. This analysis therefore compares the outcome of these treatments following their product launch in the UK using measures of outcome that were used as the primary and secondary end points in their respective pivotal clinical trials programmes: HbA^sub 1c^, hypoglycaemia, weight gain and treatment discontinuation.

Methods

Primary and secondary end points

The primary end point was mean percent HbA1 by quarterly period up to 9 months following switching to either glargine or detemir. The secondary end points were:

* The relative risk of hypoglycaemia

* Mean change in weight

* Treatment discontinuation

Data

Data analysed in this study were derived from routine general practice in the UK. These data were sourced from a proprietary health data provider who supply a resource called The Health Improvement Network (THIN)10, and this data resource can be regarded as being akin to the General Practice Research Database (GPRD)”, a widely recognised source of UK GP data. THIN was selected since it had greater patient numbers.

THIN data are collected in a non-interventional way from the daily record keeping of primary care physicians in the UK. The records are anonymised at the collection stage so that researchers have access to an encrypted identifier for the physician’s office and the patient. They provide a longitudinal medical record for each patient.

Currently, the dataset consists of contributions from over 300 practices and data from approximately 5 million patients of whom over 2.3 million are actively registered with the practices and can be prospectively followed. The remaining patients have historical data but have either left the practice or died. There are nearly 30 million patient years of computerised data in THIN. On average, patients have full data for 6 years and may have up to 15 years of observations.

The data from THIN were provided in four linked tables that detailed the following:

* Subject demographic details

* Medical history

* Test results

* Drug treatments

Inclusion and exclusion criteria

For the purposes of this study, data were extracted that included any subject who had been treated with either glargine or detemir. Subjects were then excluded if they had been switched at any time between the two products. This was justified since there may have been some bias in that glargine was launched prior to detemir, 2002 versus 2004, respectively. Furthermore, given current scientific evidence, a switch between the two products may have resulted from non-clinical/ scientific considerations.

Type of diabetes and medical case history

The analysis was split between subjects who had either type 1 or type 2 diabetes. Diabetes diagnosis was attributed on a stepwise principal. For those few people who did not have a specific diagnosis of diabetes, a diagnosis of type 2 diabetes was attributed if the subject had received any non-insulin, diabetes-related medication; otherwise type 1 diabetes was assumed.

In order to evaluate whether the respective grou\ps were directly comparable, it was necessary to examine their medical case histories in addition to other factors. This comparison characterised diabetes- related vascular complications. THIN utilised Read codes to record diagnoses12, and these codes were grouped here under the broad headings of acute myocardial infarction (AMI), congestive heart failure (CHF), stroke, peripheral vascular disease (PVD) and a broad diagnostic category that included any codes under the heading of coronary heart disease (CHD) which included AMI and CHF as well as other appropriate CHD diagnoses.

Measurement of the primary end point: glycaemia

Long-term glucose control was measured using HbA^sub 1c^. The average HbA^sub 1c^ value was calculated by quarterly period prior to and following switching using linear interpolation of values between any two consecutive HbA^sub 1c^ observations. In order to account for any differences in baseline values, these data are presented as the difference from the mean baseline HbA^sub 1c^ value, baseline meaning the date of the first recorded glargine or detemir prescription.

Measurement of secondary end points

Change in mean weight was calculated in a similar fashion to the primary end point. Hypoglycaemia was measured as a rate using records of GP diagnosis of hypoglycaemia. We were mindful that treatment discontinuation was susceptible to the issue of the differing periods of observation because of the alternative launch dates. Furthermore, people with diabetes were given ‘repeat prescriptions’, i.e., they did not necessarily need a GP appointment to collect their medications from the pharmacy. The duration of the repeat prescription could be up to a number of months. Therefore a decision that a subject had discontinued was open to interpretation, and a base case assumption had to be made in this regard that if a subject did not have another script for 12 weeks, then they had discontinued. This would have been irrelevant if the two insulin treatments had been launched simultaneously and the periods of observation had been the same. Finally, discontinuation patterns were affected by data censoring, i.e., some subjects remained on treatment at the date of transfer of the data. Importantly, the other outcome measures were not affected by these considerations. People could have discontinued treatment for a number of reasons other than switching treatment – for example, death or changing GP.

Statistical methods

A comparison of HbA^sub 1c^ was assessed using the student t- test. The relative rate of hypoglycaemia was assessed using the chi- square test. Treatment discontinuation was assessed using a Cox’s proportional hazards model (CPHM) using age and sex as covariates. An assumption was made that a subject had discontinued if they had not received a prescription for glargine or detemir for at least 6 months, and allowance was made in the CPHM for the ‘right censoring’ of the data. Statistical significance was assumed at the conventional level of p < 0.05.

Ethical approval

MREC approval was granted for this study (London MREC reference number 06/MRE02/32).

Results

Study subjects and baseline characteristics

The total number of study subjects was 6378 people. The number of subjects with type 1 and type 2 diabetes was broadly the same whereby 49% had type 1 diabetes (Table 1). Considering type 1 diabetes, 53% were male and the mean age at first recorded diabetes diagnosis was 27 years. In type 2 diabetes, 52% of subjects were male, and the mean age at onset was 49 years (Table 1).

Because of the different launch dates, fewer subjects were treated with detemir than glargine (694 vs. 5683, respectively]. The profile of subjects treated with glargine or detemir was broadly the same in both type 1 diabetes and in type 2 diabetes (Table 1); for example, the mean age at onset of type 1 diabetes was 27.5 years and 26.5 years for glargine and detemir, respectively. Further, in type 1 diabetes, the mean age at baseline was 37.9 and 37.2 years, respectively. Regarding the clinical case history, again, these were very similar; for example, in type 1 diabetes, the prevalence of AMI was 1.7% and 1.8%, respectively. In type 2 diabetes the prevalence of AMI was 7.1% and 6.1%, respectively. The only difference of note was the prevalence of PVD, where there was a lower prevalence in the glargine treated type 1 diabetes group (1.5 vs. 2.7%), and a higher prevalence in type 2 diabetes (3.7 vs. 1.7%).

Table 1. Subject baseline characteristics

Change in HbA^sub 1c^

The pattern of HbA^sub 1c^ prior to switching was similar in the comparative groups in both type 1 and type 2 diabetes (Figures 1 and 2). In the period quarter 1 (Pre-Ql) prior to switching, the mean HbA^sub 1c^ for the following groups: subjects with type 1 diabetes treated with glargine, type 1 diabetes treated with detemir, type 2 diabetes treated with glargine and type 2 diabetes treated with detemir, was as follows: 9.02%, 9.13%, 9.45% and 9.45%, respectively (mean Δ type 1 = 0.11%, Δ type 2 = 0.00%; Figures 1 and 2).

Among type 1 cases, both glargine and detemir cohorts showed an early improvement in glycaemic control (mean 90-day absolute HbA^sub 1c^ improvement -0.4%) which then levelled off. No meaningful difference emerged between the cohorts, although there was a trend for better control with glargine whereby the mean HbA^sub 1c^ was 0.05% and 0.43% at Q2 and Q3, respectively, when comparing glargine with detemir. Neither of these differences achieved statistical significance.

For type 2 patients, the same pattern was evident among detemir- treated subjects but not so for those treated with glargine who continued to show improvement throughout the analysis period. A mean improvement in HbA^sub 1c^ of 0.88% from baseline was noted for patients treated with glargine at Q3. At Ql and Q2 the relative difference in mean HbA^sub 1c^ between glargine and detemir was 0.08% and 0.05%, respectively (Figure 2). These differences did not achieve statistical significance.

The significant overlap of the 95% confidence intervals (a function of decreasing numbers in the detemir cohort) did however limit the conclusions that may be inferred from any differences.

Change in weight

For type 2 cases those treated with detemir appeared to show almost no weight gain on average in the first 6 months of treatment whereas those treated with glargine gained 0.5 kg on average (Figures 3 and 4). These differences did not achieve statistical significance (p = 0.78).

Hypoglycaemia

The risk of hypoglycaemia in patients ultimately switched to either glargine or detemir was the same in the year prior to switching (relative rate (RR) = 0.98, p > 0.05). There was a 9% reduction in reported hypoglycaemia following switching to the detemir group (RR = 0.91, p > 0.05) and a 30% reduction in the rate of reported hypoglycaemia following switching to glargine (RR = 0.70, p < 0.05; Table 2).

Figure 1. Quarterly change in HbA^sub 1c^ from baseline (switching) in type 1 diabetes after initiation of basal insulin

Figure 2. Quarterly change in mean HbA^sub 1c^ from baseline (switching) in type 2 diabetes after initiation of basal insulin

Figure 3. Quarterly change in mean weight from baseline in type 1 diabetes after initiation of basal insulin

Figure 4. Quanerly change in mean weight from baseline in type 2 diabetes after initiation of basal insulin

Treatment discontinuation

There was no convincing evidence of any difference in treatment discontinuation (Figure 5], although sensitivity analysis on the assumptions favoured glargine. Using the CPHM, overall 3%, of people discontinued treatment at 6 months and 6% at 1 year.

Table 2. Rate of reported hypoglycaemia prior to and following initiation of basal insulin

Figure 5. Cox proportional hazards model to determine treatment discontinuation patterns with glargine and detemir

Discussion

This study has shown that, following introduction into routine clinical practice in the UK, glargine and detemir had differing clinical impact. Regarding the primary end point for this study, diabetes control as measured by HbA^sub 1c^ subjects treated with glargine had evidence of marginally improved diabetes control following switching treatments although this did not achieve statistical significance due to the small numbers of patients using detemir. However, detemir showed benefits in terms of weight gain whereby those patients who switched to detemir had on average no evidence of any weight gain in the period following switching treatment. Finally, there was no convincing evidence of a difference in treatment discontinuation patterns.

This study had a number of strengths and limitations. These data were from a large number of subjects, particularly in the glargine group. Furthermore, there were no real inclusion or exclusion criteria imposed on the GPs when they made a decision to treat with one of the alternative insulin therapies, unlike a clinical trial. A further advantage of these data is that the observations were available for an extended period prior to the data of switching, or at what would normally be termed baseline, in a clinical trial. On the other hand, the fact that these data were from general practice introduced heterogeneity into the observations. For example, this may have affected the accuracy of diagnoses and increased variability across-the-board. However, these problems were common to both treatments, and the added variability in these data may have resulted in there being less likelihood of them achieving statistical significance. An important consideration when interpreting this evaluation was the differing time period of these observations. Thus, there may be a ‘familiarity’ effect with regard to insulin glargine since it was launched earlier, and the license for insulin detemir did not include management of type 2 diabetes except as part of a basal-bolus regimen. Nevertheless, there is \good evidence that the clinical profile of subjects treated with the alternative treatments is generally indistinguishable.

Placing these findings in a wider context, there have been a number of other studies using observational data13-16. In general, though, the results using observational data were generally stronger than those seen in the randomised clinical trials. Two main conclusions are evident from these trials. Firstly, the observational studies generally supported a reduction in hypoglycaemic event rates for patients shifting from a previous insulin regimen to one including insulin glargine. secondly, the studies mirrored the randomised clinical trial results with the greatest reductions being seen in the most severe events.

These findings potentially have policy implications. Should it be proved conclusively that glargine results in improved diabetes control over detemir, as suggested here, this will impact on the relative cost effectiveness of the two alternative treatments. In terms of their cost-effectiveness, these data certainly support the decision by the National Institute of Clinical Excellence to allow use of glargine at the products launch”, especially in type 1 diabetes. Importantly, these data would infer that it is also vital to collate intelligence post-launch about new drugs and treatments to complement the intelligence derived from clinical trials. Finally, the sister analysis designed to evaluate the relative cost of treatment found that treatment with glargine was also notably less expensive18.

This study showed potentially important differences between these two alternative treatments, however there is a need for further research to corroborate these findings. It would be of value to replicate this study using an alternative source(s) or to repeat the study at a later date when increased longitudinal data are available for detemir. In a wider sense, some proper evaluation needs to take place in order to better understand the relative virtues of clinical trial versus real-life data when assessing the value of new products. Finally, there should be a comprehensive review of the cost effectiveness of the two products based on these findings.

Conclusion

Treatment with insulin glargine in both type 1 and type 2 diabetes resulted in a trend for improved diabetes control and a reduction in hypoglycaemia when compared to treatment with insulin detemir. Insulin detemir showed a small improvement in weight change.

Acknowledgements

This study was funded by sanofi-aventis UK.

References

1. Stevens RJ, Kothari V, Adler AI, et al. The UKPDS risk engine: a model for the risk of coronary heart disease in Type II diabetes (UKPDS 56). Clin Science 2002; 102:679-9

2. Bagust A, Evans M, Beale S, et al. A model of long-term metabolic progression of type 2 diabetes mellitus for evaluating treatment strategies Pharmacoeconomics 2006;24(Suppl 1):5-19

3. Conlon JM. Evolution of the insulin molecule: insights into structure-activity and phylogenetic relationships. Peptides 2001;22:1183-93

4. Garber AJ. Premixed insulin analogues for the treatment of diabetes mellitus. Drugs 2006;66:31-49

5. Plank J, Siebenhofer A, Berghold A, et al. Systematic review and meta-analysis of short-acting insulin analogues in patients with diabetes mellitus. Arch Int Med 2005;165:1337-44

6. Barton S. Which clinical studies provide the best evidence? BMI 2000;321:255-6

7. Wang F, Carabino JM, Vergara CM. Insulin glargine: a systematic review of a long-acting insulin analogue. Clin Ther 2003:25:1541-77

8. Soran H, Younis N. Insulin detemir: a new basal insulin analogue. Diabetes Obes Metab 2006;8:26-30

9. Pieber TR, Eugene-Jolchine I, Derobert E. Efficacy and safety of HOE901 versus NPH insulin patients with type 1 diabetes. Diabetes Care 2000:23:157-62

10. THIN [http://www.epic-uk.org/thin.htm] (accessed 21st April 2006]

11. GRRD [http://www.gprd.com/home/] (accessed 21st April 2006)

12. Brown PJB, Warmington V, Laurence M, et al. Randomised crossover trial comparing the performance of Clinical Terms Version 3 and Read Codes 5 byte set coding schemes in general practice. BMJ 2003;326:1127-30

13. Gallen I, Carter C. Prospective audit of the introduction of insulin glargine (Lantus) into clinical practice in type 1 diabetes. Practical Diabetes lnt 2004;21:110-14

14. Fischer JS, McLaughlin T, Loza L, et al. The impact of insulin glargine on clinical and humanistic outcomes in patients uncontrolled on other insulin and oral agents: an office-based naturalistic study. Curr Med Res Opin 2004:20:1703-10

15. Bullano MF, Al-Zakwani IS, Fisher MD, et al. Differences in hypoglycemia event rates and associated cost-consequence in patients initiated on long-acting and intermediate-acting insulin products. Curr Med Res Opin 2005;21:291-8

16. Ruhnau KJ, Schneider K, Schweitzer MA. Introduction of insulin glargine to basal-bolus therapy improves metabolic control in patients with type 2 diabetes in everyday clinical practice. Diabetologia 2005;48(Suppl 1):A303

17. Diabetes (types 1 and 2) – long acting insulin analogues (No. 53). London, National Institute for Health and Clinical Excellence [http://www.nice.org.uk/page.aspx?o=TA053] (accessed 21st April 2006)

18. Poole CD, Tetlow T, McEwan P, et al. The prescription cost of managing people with type 1 and type 2 diabetes following initiation of treatment with either insulin glargine or insulin detemir in routine general practice in the UK: a retrospective database analysis. Curr Med Res Opin 2007;23(Suppl l):S41-48

CrossRef links are available in the online published version of this paper: http://www.cmrojournal.com

Paper CMRO-3634_4, Accepted for publication: 11 January 2007

Published Online: 07 February 2007

doi: 10.1185/030079906X167534

Craig J. Currie(a), Chris D. Poole(b), Tony Tetlow(b), Paul Holmes(c) and Phil McEwan(d)

a Department of Medicine, School of Medicine, Cardiff University, Cardiff, UK

b Cardiff Research Consortium, Heath Park, Cardiff, UK

c sanofi-aventis UK, Guildford, Surrey, UK

d School of Mathematics, Cardiff University, Cardiff, UK

Address for correspondence: Dr Craig Currie, Pharma Research Centre, 3 Pant yr Wyn, Cardiff,

CF23 5HS, UK. Tel.: +44 (O) 29 2048 1184; Fax: +44 (O) 29 2048 1184; email: currie@cardiff.ac.uk

Copyright Librapharm 2007

(c) 2007 Current Medical Research and Opinion. Provided by ProQuest Information and Learning. All rights Reserved.