Insulin glargine and exenatide in type 2 diabetes: A cost comparison in combination with oral antidiabetics

Type 2 diabetes is one of the most common chronic metabolic diseases. Around 7 million people in Germany are currently affected by it (DiabetesDE 2009). According to forecasts, the global prevalence of diabetes will double by 2030 (Wild 2004).
The key finding in type 2 diabetics is hyperglycemia due to inadequate insulin secretion by the ß-cells combined with insulin resistance in the peripheral tissue.
Type 2 diabetes is associated with typical microvascular (e.g. retinopathy, nephropathy) and macrovascular (e.g. stroke and myocardial infarction) complications. The positive effect of optimized antihyperglycemic therapy on the reduction of micro- and macrovascular late effects has been proven in long-term studies (UKPDS 1998; Holman 2008).

>> In general, type 2 diabetes is initially treated with non-drug measures, such as changing your diet, losing weight and exercising. If this does not lead to the desired result, patients receive pharmacotherapy with oral antidiabetic drugs (OAD). If metabolic control through diet and OAD is inadequate, clinical improvement can be achieved by additional administration of insulin (Holman 2007).
The German evidence-based treatment guidelines for antihyperglycemic therapy of type 2 diabetes recommend a combination therapy of metformin and a second OAD or exenatide even with an Hb1Ac value of 6.5 - 7.5%. If the Hb1Ac value is >7.5%, combination therapy of OAD with insulin is recommended (Matthaei 2009). According to the international therapy guidelines for type 2 diabetics, basal-supported oral therapy (BOT) is the most effective option for starting insulin therapy (Nathan 2006).
Exenatide (EXE) from the incretin mimetics group of drugs is increasingly being used in OAD combination therapy. The recently introduced injectable GLP-1 analogue EXE inhibits the action of glucagon during food intake and increases the amount of insulin released from the pancreas into the blood. Insulin is released from the pancreas into the blood depending on the level of blood sugar (Lilly 2010). Clinical studies show that the combination of EXE and OAD (EXE/OAD) reduces the HbA1c value with a reduced rate of hypoglycemia (De Fronzo 2005; Kendall 2005).
In the past, NPH insulin (human basal insulin) was usually used to initiate BOT. However, with BOT with NPH insulin, hypoglycemia occurs more frequently due to the intermediate, dose-dependent duration of action (12-18 hours) and an uneven effect profile. To overcome the limitations of NPH insulin, the two long-acting insulin analogues glargine (≥ 24 hours) and detemir (16 to 22 hours) were developed. Randomized controlled trials (RCT) and meta-analyses show a lower frequency of hypoglycemia with a BOT with GLA compared to a BOT with NPH insulin with comparable metabolic settings (Rosenstock 2005; Riddle 2003; Mullins 2007; Home 2010). The advantages of GLA over NPH insulin are already reflected in everyday German care. A current study shows that most type 2 diabetics in Germany now receive a BOT with GLA (Kress 2009).
In terms of clinical effectiveness (RCT), treatment with EXE/OAD is considered equivalent to therapy with GLA/OAD (BOT) (Heine 2005, Barnett 2007). However, proof of comparable effectiveness over a longer period of time and under everyday conditions is still lacking.
However, in addition to clinical effectiveness, the economic aspect is also important for the treatment decision. The direct costs incurred by diabetes for the German health system were estimated at 14.6 billion euros in 2001 (Köster 2006), and the trend is rising (Weber 2006). Therefore, this study compares the costs of a combination therapy of EXE/OAD against the costs of a BOT with GLA/OAD in type 2 diabetics using real health care data in Germany.
methods
This historical cohort study was conducted using the IMS® Disease Analyzer database. This database, which is representative of Germany, includes patient data from practice computer systems of around 3,000 practicing physicians (Becher 2009). The database offers a complete list of all relevant patient information, prescription data and diagnoses (ICD-10 classification and plain text information) as well as consumption and laboratory data for each practice. Drug consumption was determined according to the Anatomical-Therapeutic-Chemical Classification (ATC) of the European Pharmaceutical Market Research Association (EphMRA).
Type 2 diabetics who started GLA/OAD or EXE/OAD therapy between January 2007 and December 2008 and whose data were documented twelve months before and twelve months after the start of therapy were included. The patient population was also defined by the following inclusion and exclusion criteria:

• All patients: at least two prescriptions of an OAD twelve months before the start of therapy.
• GLA/OAD patients: No prescription of other insulins or EXE before starting therapy with the basal insulin GLA and during the observation period.
• EXE/OAD patients: No administration of insulin or insulin analogue before starting therapy with EXE and during the observation period.

The patient data had to be continuously documented. To check continuity, a minimum number of doctor visits before and after the start of therapy was defined. Before starting therapy, patients had to have at least two visits to the doctor, one of which was within a year before starting therapy. In the first year after starting therapy with either GLA or EXE, at least one doctor's visit was required every six months and another doctor's visit in the second year after starting therapy.
The patient characteristics were age, gender, insurance status (statutory/private), duration of diabetes, HbA1c value (if available), body mass index (BMI, if available), region (east/west) and focus of the practice (general practitioner/ specialist) documented.
Resource consumption and costs were determined for a period of 12 months for GLA/OAD versus EXE/OAD. The direct costs of antihyperglycemic therapy included GLA, EXE, OAD, blood glucose test strips, accessories (pens), consumables (needles, lancets) and the administration of glucose (iv) or glucagon to treat hypoglycemia. The direct costs of non-diabetic concomitant medication, including antihypertensives (ATC code C02, C03, C07, C08, C09), lipid regulators (ATC C10), antithrombotics (ATC B01) and cardiac therapeutics/blood clotting agents (ATC C01, B02) were recorded for both comparison groups . Furthermore, the number of general practitioner visits, specialist referrals and hospitalizations was documented.
The quantities and costs collected were adjusted for the variables age, gender, duration of diabetes, HbA1c value, BMI value, region and practice focus using a multivariate regression analysis. The significance level was set at p=0.05.
Results: Patient characteristics
A total of 1,934 patients were included in the database analysis. 1,484 patients received a BOT consisting of GLA/OAD and 450 patients received a combination therapy consisting of EXE/OAD (Fig. 2).
With regard to the patient characteristics recorded, there were significant differences between the collectives with regard to age, duration of diabetes, HbA1c value, BMI, insurance status and practice focus. The patients in the GLA/OAD group had an average age of 70.2 years compared to an average age of 58.1 years in the EXE/OAD group (p<0.0001). The average duration of diabetes was 5.4 years in GLA/OAD patients, one and a half years longer than the duration of diabetes in EXE/OAD patients. Patients in the GLA/OAD group had, on average, higher mean HbA1c and lower mean BMI values (p<0.0001). EXE/OAD patients were more often treated in practices with a diabetology focus (57.8% vs. 18.8%; p<0.0001) and were more often privately insured (11.8% vs. 7.3%; p<0.0001) as GLA/OAD patients. The two patient groups were comparable in terms of gender distribution and region of practice (Table 1).
Resource consumption:
In the unadjusted analysis, there were an average of 27% more general practitioner visits in the GLA/OAD group than in the EXE/OAD group. Likewise, hospital admissions and referrals to a specialist occurred more frequently in GLA/OAD than in EXE/OAD (Table 2).
Cost:
The mean unadjusted annual drug costs of antihyperglycemic therapy were 1,068 euros (GLA/OAD) and 1,740 euros (EXE/OAD). The costs of non-diabetic concomitant medication were higher for GLA/OAD patients (360 euros) than for EXE/OAD patients (257 euros). The costs for antihypertensive drugs accounted for the largest share (Table 3).
The unadjusted annual cost of antihyperglycemic therapy in EXE/OAD patients consisted mainly of the cost of EXE (68%), OAD (14%), and blood glucose test strips (14%). Whereas for GLA/OAD patients, blood glucose test strips accounted for the highest cost share (41%), followed by GLA (34%) and OAD (22%) (Fig. 3).
Regression analysis
In a multivariate regression analysis, the results were adjusted for age, gender, duration of diabetes, HbA1c value, BMI, practice focus and region.
It was shown that BOT with GLA led to significant annual savings in the direct costs of antihyperglycemic therapy (insulin, OAD, test strips, accessories, treatment of hypoglycemia) compared to EXE/OAD combination therapy (-640 euros; p<0, 0001).
This cost advantage for GLA was due in particular to the difference in costs for GLA and EXE (∆=809 euros; p<0.0001). In addition, a cost advantage of GLA/OAD over EXE/OAD was also found in terms of accessory costs (∆ =37 euros; p<0.0001). On the other hand, EXE/OAD patients showed lower expenditure on the adjusted costs of blood glucose test strips than GLA/OAD patients (∆=203 euros; p<0.0001). There were no significant differences between the two comparison groups for the costs of non-diabetic concomitant medication or for the OAD costs.
With regard to resource consumption (family doctor visits, specialist referrals and hospitalizations), it was shown that in the adjusted analysis the GLA/OAD patients had fewer general practitioner visits and specialist referrals, but more hospital admissions than EXE/OAD patients. However, these differences were not significant (Table 4).
discussion
The analysis carried out based on real care data shows that the adjusted annual costs of antihyperglycemic therapy for type 2 diabetics with GLA/OAD were lower than those of EXE/OAD combination therapy (-640 euros). The cost difference was particularly due to the lower drug costs of GLA compared to EXE.
To the authors' knowledge, this is the first study that compares the costs of type 2 diabetics undergoing GLA/OAD and EXE/OAD in Germany based on a representative database. However, some fundamental limitations must be taken into account when interpreting the results of this database analysis. Since this is a historical cohort study, randomization of patients could not be carried out. This is reflected in structural differences between patient groups. The patients in the EXE/OAD group were younger, had a shorter duration of diabetes, a lower HbA1c value, a higher BMI, were more likely to have private insurance and were more often treated in a diabetology practice than GLA/OAD patients. The differences in the basic characteristics mentioned can only partially be explained by the different prescription habits. According to the German evidence-based treatment guidelines, EXE/OAD combination therapy is recommended for HbA1c values >6.5%, while BOT should only be used for HbA1c values >7.5% (Matthaei 2009). The higher BMI value in the EXE group could be due to the expected weight-reducing effect of EXE. It is reasonable to assume that a doctor would be more likely to prescribe EXE rather than GLA to an obese patient with a borderline HbA1c level. These partially understandable and explainable differences in the basic characteristics could also cause the different levels of resource consumption in the two groups. In fact, doctor visits, specialist referrals and hospital admissions were documented more frequently in the GLA/OAD group. This could be due in particular to the age difference (GLA patients were on average 70.2 years old versus 58.1 years in the EXE group), as it can be assumed that older patients have more comorbidities and therefore have more doctor visits and hospital admissions. Furthermore, the consumption of blood sugar test strips was different in both groups, as GLA patients self-monitor their blood sugar levels more often than patients who are only treated with OAD or EXE/OAD. GLA patients spent 436 euros on blood sugar test strips, while EXE patients only used test strips for a total of 243 euros. Despite the above-mentioned structural differences between the comparison groups, the unadjusted analysis already showed a cost advantage for GLA over EXE, which remained even after the multivariate regression analysis, corrected for the above-mentioned differences.
Two cost-effectiveness studies, which compare the costs and benefits of GLA/OAD and EXE/OAD in the treatment of type 2 diabetics based on a validated diabetes model, also come to the conclusion that the costs of a BOT GLA are below those of EXE/OAD combination therapy (Mittendorf 2009; Brändle 2009). The German study examined GLA versus EXE in type 2 diabetics over a 10-year period and concluded that the 10-year costs with EXE were 3,854 euros per patient higher than the costs of BOT with GLA, with a comparable quality of life and survival time (Mittendorf 2009). The study within the Swiss healthcare system was modeled over a time horizon of 35 years. EXE was also more expensive than GLA (CHF 8,378 per patient in 35 years) with the same effectiveness and survival time (Brändle 2009). In principle, it must be taken into account in modeling studies that data from different sources (epidemiological data, data from clinical studies and cost data) are incorporated into the simulation process. In comparison to the present real-life study, which depicts the actual reality of care, modeling can only be used to achieve the best possible approximation of real everyday care.
The results of the present study indicate a savings potential for the German healthcare system through the use of GLA in BOT compared to EXE/OAD combination therapy in type 2 diabetics. Despite the limitations of this study discussed, database analyzes fundamentally provide important insights that contribute to understanding real health care processes and can represent a valuable addition to randomized controlled trials for decision-makers in the health system (Silverman 2009).
conclusion
BOT with GLA can be viewed as at least equivalent to EXE/OAD combination therapy in terms of clinical effectiveness (Heine 2005; Barnett 2007). From a health economics perspective, based on this database analysis, BOT with GLA can be considered superior to EXE/OAD combination therapy with the same effectiveness and lower costs. The results of this study favor BOT with GLA, which could lead to savings in the treatment of type 2 diabetes in Germany compared to EXE/OAD combination therapy. <<

glossary
Basal-supported oral therapy: Combination of a long-acting insulin (basal insulin) and an oral antidiabetic drug (tablet). Human insulin: Insulin that is completely similar in structure and composition to the human insulin hormone. Today it is produced biosynthetically. NPH insulin: intermediate-acting human insulin (12-18 hours). Insulin analogues: human insulin modified by changing the number and order of amino acids with new properties. There are short-acting and long-acting variants (e.g. insulin glargine).
Legend: SD: standard deviation; MW: mean, GLA= insulin glargine, BOT= basal-supported oral therapy, EXE= exenatide, OAD= oral antidiabetics.