Camparisan of one-year costs of Type 2 diabetes ...

International Journal of C!inical Pharrnacology and Therapeutics, Val. 48- No. 2/2010 (129-137)

Camparisan of one-year costs of Type 2 diabetes treatment with insulin glargine or insulin detemir in a basal supported oral therapy (BOT) in Germany S. Pscherer\ E.S. Dietrich 2 , F.-W. Dippel 3 and A.R. Neilson 2 1Kiinikum

Traunstein, Germany, 2 HealthEcon AG, Basel, Switzerland, and Deutschland GmbH, Berfin, Germany

3 Sanofi-Aventis

Key words Basal insulin- insulin glargine- insulin detemir- Type 2 diabe tes - cost analysis

Received May 8, 2009; accepted October 21, 2009 Correspondence to A.R. Neilson HealthEcon AG, Steinentorstr. 19, Postfach 1510, 4001 Basel, Switzerland aneilson@ healthecon.com

Abstract. Objective: A one-year cost ana lysis comparing basal insulin analogues glar gine (IG, Lantus®) versus detemir (ID, Leve mir®) in combination with oral antidiabetic drugs (basal supported oral therapy; BOT) in insulinnaive Type 2 diabetes patients in Ger many based on the results of a randomized controlled clinical trial (RCT). The trial dem onstrated equivalent treatment efficacy. Mate rials and methods: Total direct diabetes treat ment costs were estimated from the perspec tive of the German statutory health insurance (SHI) for the time horizon of one-year. Simu lated resources included medication (insulin, oral antidiabetic drugs) and consumable items (needles, blood glucose test strips and lancets). Initial and fmal insulin doses per kg body weight and proportion of patients with once/twice daily insulin injection were taken from the above mentioned RCT. Unit costs were taken from official Gennan price lists and sources. Deterministic-(DTA) and probabilis tic sensitivity analyses (PSA) on resource use and unit costs were performed to test robust ness of the results. Results: Average annual treatment costs per patient (base case) were € 849 for glargine and € 1,334 for detemir re sulting in cost savings of€ 486 per patient per year (36% ). Costs of insulins were € 469 (IG) and € 746 (ID). Costs of consumable items amounted at€ 380 (IG) and € 588 (ID) respec tively. Sensitivity analyses confmned the find ings in favor of insulin glargine. PSA results found cost savings ranging from € 429 to € 608 (5th/95th percentiles). Conclusions: The cur rent model estimated that insulin glargine was associated with lower annual treatrnent costs of € 486 (36%) compared to the use of insulin detemirwhile the same glycemic control is ex pected to be achieved.

lntroduction Cunent American and European diabetes consensus statements [16] and Gennan-spe

cific treatrnent guidelines [ 14] recommend the initiation ofbasal or prandial insulin injections in Type 2 diabetes after failure oflifestyle inter vention in combination with metformin ther apy. A recently published 52-week multinatio nal, randomized, open-labe!, parallel-group non-inferiority trial [20] comparing directly the addition of basal insulin analogues glar gine or detemir to existing therapy with oral antidiabetic drugs (OAD) in Type 2 diabetic patients. The trial demonstrated that glargine reaches the same efficacy target as detemir with only one daily administration and lower insulin doses. The primary endpoint was baseline ad justed HbAlc at the end of treatment. Sec ondary endpoints included fasting plasma glucose (FPG), within-participant variation in ten point self-measured plasma glucose profiles (SMPG), proportion of participants achieving HbAlc:::; 7.0% with and without hypoglycemia, incidence of hypoglycemia, adverse events and standard safetyparameters. Insulin-naive adults (n = 582, HbAlc 7.5-10%, BMI:::; 40.0 kg/m2) were random ized to receive either insulin detemir or glar gine once daily (evening) actively titrated to FPG target:::; 6.0 nunol/1. An additional morn ing insulin detemir dose was permitted if pre dinnerplasma glucose (PG) was> 7.0 nunol/1 after achieving FPG < 7.0 tmnol/1. Mean baseline HbA 1c decreased from 8.6 to 7.2 and 7.1% (ns) with detemir and glargine, respective1y. FPG improved from 10.8 to 7.1 and 7.0 mmol/1 (ns), respectively. In both treatment groups 52% achieved the target HbA1c:::; 7.0%; 35% (90/259) glargine patients and 33% (82/248) detemir patients without hypog1ycemia. Within-participant

130

Pscherer, Dietrich, Dippeland Neilson

variability for self-monitored SMPG and pre-dinner PG did not differ by insulin treat ment. There were no differences in standard safety parameters between treatments. Ad verse events recorded as serious tended to be of a wide ranging disparate nature, with no clear pattern of between-treatment differ ences[20]. At the end ofthe trial, 45% ofthe patients treated with detemir required an injection once daily, while 55% had to inject an additional morning dose. The majority of patients admin istering insulin detemir twice daily (n = 103) were transferred to this regime within 12 weeks oftreatment. All glargine patients were treated with once daily injections. In both groups, basal insulinwas initiated once dai1y in the evening at a dose of 12 U/kg and titrated according to a structured algo rithm to a fasting plasma g1ucose target Ievel :::; 6 mmol/1 in the absence of hypoglycemia. After 52 weeks detemir patients required an average 77% higher dose than glargine pa tients. Doses were even higher for patients with once dai1y administration: detemir: 0.78 U/kg (0.52 with once daily dosing, 1.00 U/kg with twice daily dosing); glargine: 0.44 U/kg. The RCT [20] demonstrated that insulin glargine provides equiva1ent efficacy with enhanced patient convenience of insulin ap p1ication (once daily) compared to insulin detemir (once or twice daily), in patients with Type 2 diabetes. The comparative annual costs ofboth regimes have already been reported for severa1 countries [5, 7, 8, 11, 17], but no eva1 uation has yet been conducted for Gerrnany. This was the purpose ofthe present analysis. A cost analysis which compares only the costs ofthe alternative treatment strategies is a justified economic eva1uation method in the current study since evidence from the RCT [20] demonstrated that the alternative strate gies are at least clinically equivalent (non inferiority trial). Such economic evaluation approaches are useful to support decision making on the financing of insulin initiation therapies for patients with Type 2 diabetes in different healthcare settings. Moreover, since transferability from one country to another is usually restricted, country-specific evalua tions are required that take into account coun try-specific features such as treatment guide 1ines, epidemiology of diabetes, patterns of

health service resource use, unit costs and reimbursement regulations [4].

Methods The present study conducted a direct cost comparison between the two different insulin analogue treatment strategies in combination with OADs based on the findings ofthe RCT [20] for the German healthcare setting. A number of assumptions made in the modeling analysis have a certain degree ofuncertainty, but in general the va1ues chosen for calcu1a tion adopted a conservative approach, gener ally biased agairrst the treatment arm with insulin glargine. The present evaluationwas perforrned as a cost minimisation analysis ac cording to the Gennan reconnnendations on health economic evaluation [23].

Perspective and time horizon The economic analysis involved an as sessment of direct healthcare costs only and therefore takes the costs perspective of the German statutory health insurance (SHI), in an open care setting. The time period is the first year after initiating insulin treatment.

Cast determinants Identification, measurement and valu ation of the key cost deterrninants comprise the main activities in conducting a cost analy sis. In the present study, all unit costs were taken from official price lists and sources for the current year, 2008. The cost deterrninants that were included, resource utilization in each case, and the unit costs applied, are pre sented in detail below. The assumptions char acterize the base case ofthe cost ana1ysis and were further investigated in sensitivity analy ses. All key model assumptions used in the analyses are smmnarized in Table I.

Basal insu/in analogues In the RCT [20], the evening starting dos age for both insulin glargine and insulin detemir was 12 U/kg bodyweight. Glargine

131

Treatment costs of inadequately controlled T2 diabetes

Table 1.

Key parameters used in the cost analysis.

Parameter

Values used in the base case analysis

Assumptions and values used in probabil istic sensitivity analysis

Source/comments

Final dose per kg for once daily injection- glargine

0.44 U/kg

Gamma, approximated using a mean = 0.44; so= 0.045

Actual SO was not reported in the RCT [20] Range approximated assuming 10% variation of mean final dose

Final dose per kg for once daily injection - detemir

0.52 U/kg

Gamma, approximated using a mean = 0.52; so= 0.053

As above

Final dose per kg for twice daily injections - detemir

1.00 U/kg

Gamma, approximated using a mean = 1.00; S0=0.102

As above

55%

Min = 0.485 Max = 0.613

127/231 patients in detemir treatment group in the RCT [20].

Beta, approximated using a = 104; ß = 127

Assumed range considered to reflect the 95% confidence interval

Resource consumption

Proportion(%) of insulin detemir patients with twice daily injections

Unit costs glargine (Lantus®, 100 U/ml Cartridge, Sanofi-Aventis, 2700 U: 9 x 3 ml)

0.049 €/U

Min = 0.044 €/U Max = 0.054 €/U

10% variation of base case price. Conserva live assumption to include range of different prices according to pack size

detemir (Levemir®, Penfill Cartridge, Novo Nordisk 3,000 U: 10 x 3 ml)

0.049 €/U

Min = 0.044 €/U Max = 0.054 €/U

As above

Needles

0.269 €/ needle

Min = € 0.242 Max = € 0.296

10% variation of base case price. Assumed uniform

Blood glucoselest strips

0.658 €/ lest strip

Min = € 0.592 Max = € 0.724


Lancets

0.113 €/ lancet

Min = € 0.102 Max = € 0.124


was administered once daily at bedtime as per study protocol following a structured dose ti tration algorithm. Insulin detemir was also given once daily at bedtime with an option of adding a second dose if pre-dinner PG was > 7.0 rmnol/1, but only if pre-breakfast plasma glucose (FPG) was< 7.0 Illi11111 or nocturnal hypoglycemia precluded achievement ofFPG target. The actual quantities of insulin glargine consumed based on the RCT [20] were used for the base case analysis. F ollowing a similar approach taken in the Argentinean cost analy sis [17], we assumed that the average final dose for each insulinwas the initial one with a linear titration over the 52 weeks and calcula tions of mean doses were based on patient's final weights: Initial dose per kg + (final dose per kg- initial dose per kg)/2) x final weight.

In the trial insulin injection was con ducted using the refill pen injector OptiPen Pro I® (Sanofi-Aventis, Paris, France) and FlexPen® (Novo Nordisk, Bagsvaerd, Den mark) for glargine and detemir respectively. Each cartridge contains 3 ml solution (1 ml contains 100 U). Insulin glargine is available on the Gennan market as Lantus® (Sanofi-Aventis). The fonnulation for insulin detemir used in the trialwas Levemir® (Novo Nordisk). For the analysis, we considered only the insulin and not the devices or pens as these are usually given to the patients free of charge as a sample. In addition the price dif ferences between the compared devices are very small. Each pen can be filled with car tridges with 3 ml solution (1 ml contains 100 U). For insulin prices, see Table I. For both drugs we assumed the price of the most economical pack size, i.e.: insulin glargine

132

Pscherer, Dietrich, Dippel and Neilson

(Lantus® cartridge: 2,700 IU, 9 x 3 ml); insu lin detemir (Levemir® cartridge: 3000 IU, 10 x 3 ml). Prices were taken as the pharmacy sales price according to Lauer Taxe including VAT as reimbursed by SHI [12]. Manufac turer rebates (6%) and pharmacy rebates (2.30 € per prescription) were of no impor tance in this comparison.

Oral antidiabetic drugs metformin, insu/in secretagogues, a-glucosidase inhibitors On the basis of the RCT study protocol [20], the use ofOADs (active ingredient and dosage) was recomrnended to remain stable during the study (i.e., no actual data reported in the article to the contrary). Therefore we held these items of resource utilization con stant in both groups.

Consumable items: needles, blood glucose test strips and /ancets As mentioned above, utilization of refill pen injectors were assumed in the present analysis. With respect to the number of nee dles, test strips and lancets needed per patient we estimated as one per each insulin injec tion. However, findings from a recent Euro pean diabetes patient survey reported that 93% ofmen with diabetes in Gennany used the same needle several times, on average 9.2 inj ections with the same needle [ 19]. ClickFine® (Ypsomed, Sulzbach, Germany) needles were assumed for both regimes as these can be fitted in all makes of pen. The price of needles were based on official mail order prices [6, 26]. One lancet and one glu cose test strip is required for each blood glu cose measurement. Thus, in the base case, a single daily measurement was assumed for glargine and either one or two times daily for detemir. Following the results in the underly ing RCT [20] we assumed that 55% of pa tients received twice daily injections with detemir. These assumptions were varied in the sensitivity analyses. It was assumed that Softclix® (Roche Diagnostics, Basel, Switzerland) lancets were used. As a wide

range ofblood glucose test strips are available at almost identical prices, a uniform price [6] could be established so that cost was inde pendent of any particular measuring system (Table 1).

Further resource uti/ization In the RCT [20], the incidence of hypo glycemia was comparable between treatrnent regimes: 5.8 vs. 6.2 episodes per patient-year with insulin detemir and insulin glargine, re spectively. Major hypoglycemic events (and also those likely to have greatest resource use implications requiring medical assistance) were rare with both insulins and could not be statistically analyzed. We therefore assumed no difference in these related costs in our analysis between the two therapeutic groups.

Data management and ca/culations Data on resource use and unit costs were entered into Microsoft Excel 2003 spread sheets. All calculations as well as presenta tion of results were done using prices (Euro) to two decimal places. Calculations, tables and graphs were generated using Microsoft Excel2003.

Al/owance for parameter uncertainty: sensitivity analysis We perfonned a number ofsensitivity ana lyses to explore the robustness of the study re sults to changes in the value of key cost pa~ rameter estimates. First, the assumptions made in the base case analysis were varied in simple one-way sensitivity analyses in order to test the robust ness ofthe base case results to alternative as sumptions in price, resource use, uncertainty in other assumptions and possible deviations from the underlying RCT results for routine clinical care. In the first part ofthe detennin istic sensitivity analyses (DSA), the most im portant cost detem1inants were altered in dependently of one another by ± 25%. The sensitivity analyses are summarized in a tor

133


Table 2.

Base case cost analysis results.

Resource use

Unit costs (€)

Total units consumed/year

Costs, €/year (%)*

L'l. in costs, € (%)**

glargine

detemir

glargine

detemir

glargine

detemir

9,564

15,207

0.049 €/U (Lantus®, 100 U/ml Cartridge, Sanofi-Aventis, 2700 U: 9 x 3 ml)

0.049 € /U (Levemir®, Penfill Cartridge, Nova Nordisk 3,000 U: 10 x 3 ml)

469.14 (55.2)

745.95 (55.9)

276.82 (37.1)

Needles

365 in each case

566 in each case

0.269 €/needle

98.22 (11.5)

152.24 (11.4)

-54.02 (35.5)

Blood glucose test strips

As above

0.658 €/lest stri p

240.17 (28.3)

372.26 (27.9)

-132.09 (35.5)

Lancets

As above

0.113 €/lancet

41.15 (4.8)

63.79 (4.8)

-22.63 (35.9)

848.68 (100)

1,334.25 (100)

-485.57 (36.4)

Insulin (U)

Total costs, €/ year

*The % values in the glargine and determir columns represent the % that each cost component contributes to total costs, **The % values in the final column represent the % difference in each cost component between glargine and detemir groups.

nado plot showing the cost drivers in de scending order of importance. To address shortcomings in performing only univariate sensitivity analysis, we also perfonned pro babilistic sensitivity analysis (PSA) taking the form of a Monte Carlo simulation (MCS) in which key model parameters were simulta neously varied by replacing parameter esti mates with an appropriate distribution [2]. For example, parameters with respect to the proportion of once daily, twice daily injec tions are bounded on the interval 0- 100, so it wou1d be inappropriate to apply a distribution that gave a non-negligible value outside of that range. In this case, a betadistributionwas specified to reflect the normal distribution and restriction to values between zero and 100 (%). With respect to insulin consump tion, variation in resource use values (final dose per kg) was notreported in the trial itself. A range of values was therefore approxi mated assuming a 10% variation of the mean final dose. For MCS, a gamma distribution was fitted by estimating the standard devia tion. F or unit costs a 10% variation of base case prices were assumed to reflect the mini mum and maximum negligible values. Out put from this MCS (1 ,000 iterations, using RiskAmp Add-In for Excel® 2007, Struc tured Data, LLC, New York, NY, USA. www.riskamp.com) generate estimates ofthe likelihood of different Ievels of annual treat ment cost-differences between the two groups

given the uncertainty in model parameter val ues. Finally, a number of modified seenarios were carried out to test specifically variations more applicable to routine care in Germany. These variations included: only one needle per ten insulin injections in both groups, all determir patients received once-daily injec tions (low dosage); replacing detemir with NPH (with dosage, price and distribution of once: twice daily injections based on a re cently published observational trial in Ger many [9]); number ofblood glucose measure ments different in routine care.

Results Base case analysis The identification, measurement and val uation ofthe relevant cost items are presented in Table 2. Based on the assumptions made for the base case, once-daily insulin glargine in combination with OADs generated aver age ammal cost savings amounting to € 486 (approx. 36% reduction) per patient lower than those with insulin detemir combination therapy. The most important impact factors in descending order explaining this overall dif ference were due to the lower insulin costs of glargine (€ 277; L1 = 37%), followed by lower

134

Pscherer, Dietrich, Dippel and Neilson Parameter va ried Mea n final d ailyd ose(detemir)

PrCe of flsultndetemir Mean fina!dailydose (glargine}

Price ofinsu!ing!argine

Proportion 1x-d ailyinjections withd eiemir

Prk:e ofesl strlps

price of needles and lancets

€200

€250

€300

€350

€400

€450

€500

€550

€600

€650

€700

Range ot cost sav !ngs ~ €299 • fß 72

Figure 1. One-way sensitivity analysis: Cast savings of glargine versus detemir per patient per year*. *The central verticalline represents the base case cost savings. The bar eilher side ofthe verticalline repre sent the cost savings when each parameter is varied ± 25% of the values applied in the base case (Table 1, 2nd column).

Table 3.

Sensitivity analysis: results with modified scenarios. Cast savings with glargine (€/patient/year)

!:> costs from base case

All detemir patients (100%) receive one injection per day (0.52 U/kg)

-57.97

reduced by 428.60

Detemir replaced by NPH-insulin (dosage, price and distribution based on LIVE-OE [9])

-92.30

reduced by 394.27

One blood glucose measurement per day in both groups

-330.84

reduced by 155.73

One needle per ten injections in both groups

-436.95

reduced by 49.62

Scenario

costs ofblood glucose measurements includ ing test strips and lancets (€ 155; LI,= 36%) and then insulin administration costs in terms ofneedles used (€ 54; LI,= 36%).

Sensitivity analysis Univariate sensitivity analysis are sum marized in Figure 1 as a tornado diagram showing the cost savings of insulin glargine resulting from changes in different cost deter minants. The length of the horizontal bars conesponds to the difference in average costs between glargine and detemir groups over the specified variables of interest depicted on the y-axis. The vertical line transecting the bars represents the cost difference between the groups in the point-estimate (non-probabilis tic) base case. The base case with annual cost savings of € 486 per patient is represented by the central axis. The tornado diagram ranks the cost parameters based on the magnitude of their impact on the cost differences between the t\vo treatment groups. The results show

clearly that the insulin consumptions and in sulin price have the highest impact. For all these variations, positive cost advantages and thus real cost savings were seen with insulin glargine. Modified seenarios to investigate assump tions more applicable to routine care con firmed the robustness of the base case results (Table 3). However, as already mentioned in the methods section, the deterministic sensi tivity analysis does not take into account how realistic or likely these variations and corre sponding results are. Uncertainty issues in this context are accounted for based on point values and typically provide a set of ("static") worst-case vs. best case scenarios. To investigate this issue further, we ex plored further parameter uncertainty by per fomüng a probabilistic sensitivity analysis using Monte Carlo simulation. The parame ters and their distributions are shown in Table 1. This process involving probabilistic sensi tivity analysis (in which model parameters were varied simultaneously) across assumed distributions demonstrated that overall total


Figure 2a. Monte Carlo simulation results: Fre quency (%) of specific cost savings*. *Based on 1,000 cost model iterations including variables on resource use: insulinfinal daily doses; % of twice daily injections with determir and unit costs: insulin prices, prices of consumable items. ~