Cost-effectiveness of pioglitazone in patients with type 2 diabetes and a history of macrovascular disease in a Swiss setting

Objectives: To evaluate the cost-effectiveness of pioglitazone versus placebo, given in addition to existing treatment regimens, in patients with type 2 diabetes and evidence of macrovascular disease in Switzerland. Methods: Event rates corresponding to macrovascular outcomes from the PROactive (Prospective Pioglitazone Clinical Trial in Macrovascular Events) trial of pioglitazone were used to project long-term clinical outcomes as part of a modified version of the previously validated CORE Diabetes Model. Direct medical costs associated with treatment regimens, complications and patient management were accounted in 2005 values based on Swiss-specific unit costs. Time horizon was set to lifetime (35 years). Future costs and clinical benefits were discounted at 2.5% annually in line with Swiss recommendations. One-way sensitivity analyses were performed. Results: Addition of pioglitazone was associated with a reduced incidence of most diabetesrelated complications, improved life expectancy (0.258 years) and improved quality-adjusted life expectancy (0.180 QALYs) compared with placebo. Pioglitazone treatment increased direct costs by CHF 10,914 per patient over a lifetime horizon. The incremental cost-effectiveness ratio (ICER) of pioglitazone versus placebo was CHF 42,274 per life-year gained and CHF 60,596 per QALY gained. ICERs were sensitive to variation in time horizon and duration of pioglitazone treatment effects.With a willingness to pay of CHF 80,000 per QALY in the Swiss setting, there was a 62.5% chance that pioglitazone would be cost-effective. Conclusions: Addition of pioglitazone to existing therapy was projected to reduce the longterm cumulative incidence of most diabetes complications and improve quality-adjusted life expectancy. Evaluation of incremental direct medical costs associated with these clinical benefits indicated that pioglitazone is likely to be a costeffective treatment option in the Swiss setting over patient lifetimes.

Objectives: To evaluate the cost-effectiveness of pioglitazone versus placebo, given in addition to existing treatment regimens, in patients with type 2d iabetes and evidence of macrovascular disease in Switzerland.
Methods: Event rates corresponding to macrovascular outcomes from the PROactive (Prospective Pioglitazone Clinical Tr ial in Macrovascular Events) trial of pioglitazone were used to project long-term clinical outcomes as part of am odified version of the previously validated CORE Diabetes Model.Direct medical costs associated with treatment regimens, complications and patient management were accounted in 2005 values based on Swiss-specific unit costs.Time horizon was set to lifetime (35 years).Future costs and clinical benefits were discounted at 2.5% annually in line with Swiss recommendations.One-way sensitivity analyses were performed.
Results: Addition of pioglitazone was associated with ar educed incidence of most diabetesrelated complications, improved life expectancy (0.258 years) and improved quality-adjusted life expectancy (0.180 QALYs) compared with placebo.Pioglitazone treatment increased direct costs by CHF 10,914 per patient over al ifetime horizon.The incremental cost-effectiveness ratio (ICER) of pioglitazone versus placebo was CHF 42,274 per life-year gained and CHF 60,596 per QALYg ained.ICERs were sensitive to variation in time horizon and duration of pioglitazone treatment effects.With awillingness to pay of CHF 80,000 per QALYinthe Swiss setting, there was a6 2.5% chance that pioglitazone

Introduction
This study was supported by an unrestricted grant from Ta keda Pharma AG.
pressure levels in addition to glycaemic control have ag reater impact on macrovascular events than concentrating solely on normoglycaemia [7][8][9].
The PROspective pioglitAzone Clinical Tr ial in macroVascular Events (PROactive) was one of the few large scale outcomes studies to investigate prospectively the effect of an oral glucose-lowering drug (pioglitazone) on macrovascular outcomes [10].Pioglitazone is amember of the thiazolidinedione class of antidiabetic medications and has blood glucose lowering properties.In addition to positive effects on glycaemic control, pioglitazone also has the potential to improve lipid abnormalities [11].The PROactive study enrolled 5,238 patients with type 2d iabetes and evidence of macrovascular disease [12,13].Patients were randomly allocated to receive either pioglitazone or placebo in addition to their usual treatment regimen in line with the International Diabetes Federation (IDF) European region 1999 guidelines for diabetes care [14].By the end of PROactive (36 months of follow-up), HbA1c was reduced by -0.9% (versus -0.3% with placebo), HDL-cholesterol increased by 0.54 mmol/l (versus 0.3 mmol/l) and triglycerides decreased by -0.064 mmol/l (versus an increase by 0.076 mmol/l for placebo) with pioglitazone treatment.There was an on-significant 10% relative risk reduction (RRR) with pioglitazone in the primary endpoint (P =0 .09),which was ac omposite of all cause mortality,n on-fatal myocardial infarction (NFMI) (including silent MI), stroke, acute coronary syndrome (ACS), endovascular or surgical intervention in the coronary or leg arteries, and amputation above the ankle.Pioglitazone was also associated with as ignificant RRR of 16% (P = 0.02) in the principal secondary endpoint of time to first event of death from any cause, MI (excluding silent MI) and stroke.Pioglitazone also reduced the number of patients progressing to long-term insulin therapy by approximately 50%.Ta ken together these findings demonstrate that improved glycaemic and lipid control associated with pioglitazone treatment lead to areduced incidence of macrovascular events.
An estimated 250,000 individuals in Switzerland have diagnosed type 2d iabetes [15].Recent surveys indicate that many of these patients are inadequately controlled with respect to glycaemia and lipid levels, and estimate that approximately 25% of patients have at least one macrovascular complication [15,16].In the Swiss setting the annual cost per patient with at least one macrovascular complication was reportedly three times more than that corresponding to patients without complications (CHF 5050 versus CHF 1723 per year) [15].Therefore, in the Swiss setting the addition of pioglitazone to current treatment regimens may be beneficial in clinical terms and subsequently in economic terms, due to the avoidance of future costly macrovascular events and improved quality of life.To investigate this hypothesis, ap ublished and validated model of type 2diabetes was adapted to incorporate clinical data from PROactive to estimate the impact of treatment on life expectancy,quality-adjusted life expectancy and incidence of macrovascular events, and to account for direct medical costs over patient lifetimes in aSwiss setting.

174
Cost-effectiveness of pioglitazone in patients with type 2diabetes and ahistory of macrovascular disease in aS wiss setting

PROactive
The PROactive study was aprospective, randomised, double-blind, placebo-controlled trial conducted in 19 European countries (321 centres) [10].Designed to test the hypothesis that pioglitazone used as an 'add on' therapy would lower the incidence of macrovascular events in patients with type 2diabetes and ahistory of macrovascular events, PROactive was the first adequately powered study to look at the secondary prevention of macrovascular events.In total 5238 patients were enrolled and the average length of follow-up was 36 months.
At study end there was an on-significant 10% relative risk reduction (RRR) with pioglitazone in acomposite of all cause mortality,n on-fatal MI (including silent MI), stroke, ACS, endovascular or surgical intervention in the coronary or leg arteries, and amputation above the ankle.Pioglitazone was also associated with as ignificant RRR of 16% in time to first event of death from any cause, MI (excluding silent MI) and stroke.

Model description and statistical approach
The important short-term clinical effects of the pioglitazone and placebo treatment regimens from PROactive were used to project long-term outcomes using a modified version of the validated and peer-reviewed CORE Diabetes Model (CDM) [17,18].The CORE Diabetes Model is ac omputer simulation model developed to determine the long-term health outcomes and economic consequences of interventions in type 1and type 2 diabetes.Disease progression is based on 15 interdependent semi-Markov sub-models that simulate progression of disease-related complications (angina, MI, congestive heart failure, stroke, peripheral vascular disease, diabetic retinopathy,macula oedema, cataract, hypoglycaemia, ketoacidosis, lactic acidosis, nephropathy and end-stage renal disease, neuropathy,f oot ulcer,a mputation and non-specific mortality).Each sub-model uses time, state and diabetes type-dependent probabilities derived from published sources.The reliability of simulated outcomes has been tested, with results extensively validated in 66 separate analyses against outcomes reported by clinical trials and epidemiological studies [18].In addition to this the CDM was presented at the Fourth Mt.Hood challenge where it compared favourably with an umber of currently available diabetes simulation models [19].
Data from PROactive was used as ab asis for longterm projections using the CORE Diabetes Model adapted to include clinical data from PROactive [20].Whilst abrief overview of the modified model used in the current analysis is provided here and in the accompanying appendix (see Appendix, fig.3), we recommend that the interested reader refer to the detailed account of the model provided in the publication of Va lentine et al. [20].In short, at otal of 15 complication-related sub-models were included in the final PROactive long-term simulation model.
For the sub-models of acute coronary syndrome, myocardial infarction, percutaneous coronary intervention, coronary artery bypass graft, bypass surgery/revascularization of the leg, hospital admission for heart failure, non-serious heart failure, oedema, transient ischaemic attack, stroke, photocoagulation and severe vision loss, event rates were taken directly from PROactive.Event rates in the placebo arm were calculated directly from the annual hazard rates observed over months 0-36 of PROactive assuming constant risk (see Appendix, table 6) [20].Hazard ratios were then applied to the event rates from the placebo arm in line with the relative risk observed for each event during PROactive to calculate event rates in the pioglitazone treatment arm.The remaining sub-models of cataract, nephropathy and endstage renal disease, neuropathy,p eripheral vascular disease, diabetic foot and amputation, were as presented in the original CDM and therefore based on transition probabilities drawn from published studies including UKPDS to simulate complication progression.
All-cause and cardiovascular mortality rates were derived from PROactive for years 1-3, with rates subsequently doubling every 10 years [21].Event rates for following years were calculated by applying relative risk adjustments [22][23][24][25][26] for each additional life-year gained (i.e. as the patient gets older,h is/her risk of experiencing an event increases).In all sub-models, the occurrence of events resulted in the accrual of event costs and, where applicable, subsequent state costs as well as assignment of the appropriate disutility values.
For the base case simulation, the clinical effects associated with the pioglitazone and placebo treatment regimens were applied as observed during PROactive for acute coronary syndrome, myocardial infarction, percutaneous coronary intervention, coronary artery bypass graft, bypass surgery/revascularisation of the leg, hospital admission for heart failure, non-serious heart failure, oedema, transient ischaemic attack, stroke, photocoagulation and severe vision loss.In those sub-models it was assumed that the treatment-related changes in HbA1c, lipids and blood pressure were reflected in the observed event rates.For those complication sub-models not based on PROactive it was necessary to apply the treatment associated effects on HbA1c, lipids and blood pressure to simulate the impact of this on future events.Tr eatment effects on HbA1c were applied separately in simulation years 1, 2a nd 3, and in subsequent years, based on the findings from PROactive, the UKPDS and Framingham (for long-term projection).Changes in HbA1c and other parameters for pioglitazone and placebo regimens were applied as summarised in table 1.The long-term progression of all of these clinical parameters subsequently followed the patterns previously described by Palmer et al. in their description of the CORE Diabetes Model [17].
The health economic analysis was performed using a non-parametric bootstrapping approach in which the progression of diabetes was simulated in 1000 patients through the model 1000 times to calculate the mean and standard deviation of costs, life expectancy and qualityadjusted life expectancy using second order Monte Carlo simulation [27].The model estimated the impact of treatment on life expectancy,quality-adjusted life expectancy (based on CODE-2 utilities, and assuming abaseline utility score of 0.814 for diabetes without complications), diabetes-related complications (cumulative event rates), direct medical costs and cost-effectiveness ratios over patient lifetimes, in line with specifications for health economic evaluations [28].Mean results from each of the 1000 iterations were used to create as catter plot, comparing the differences in costs and effects for pioglitazone and placebo treatment regimens.These values were in turn used to generate ac ost-effectiveness acceptability curve over ar ange of willingness to pay values in the Swiss setting.

Simulation cohorts
Acohort of patients was defined with baseline demographics, complications, and important concomitant medications representative of the two treatment arms from PROactive.Long-term outcomes were calculated 1000 times in the model using asimulated population of 1000 patients to capture the effects of random variation between individual simulated patients.At baseline 66.1% of the cohort was male with amean age of 61.8 years, duration of diabetes 10 years, and am ean HbA1c level of 8.1% (see Appendix, table 7for more detail).For the purposes of this analysis patients were assumed to remain on the same treatment regimen for the duration of the simulation (35 years or death).
The proportion of patients receiving angiotensin converting enzyme inhibitors (ACE inhibitors) in the simulation cohort was set to 62.8% based on data from PROactive.This has an important influence on renal outcomes and it is relevant to note here that as urvey of Swiss patients reported as imilar rate of ACEi nhibitor use, with 59% of patients reporting treatment with this class of antihypertensive agent [16].Also in line with data from PROactive 42.9% of the simulation cohort were assigned to use of statins and 73.1% use of aspirin at baseline.Risk adjustment for the use of aspirin or statins was disabled in the analysis as the influence of these agents was already taken into account, along with the impact of ACEi nhibitors, in the cardiovascular event rates taken from PROactive.Other settings for patient management parameters (e.g.screening for renal disease and foot ulcer prevention programs) were set in line with the standard of care patients in the PROactive study were receiving, with all patients receiving regular screening and foot checkups.

Effect Mean change from baseline Pioglitazone Placebo
Change in HbA1c in year 1(%-points) -0.9 -0.2).Direct medical costs were calculated as the sum of drug acquisition costs (based on data from PROactive), patient management costs and the cost of complications.The annual costs of study medication were accounted based on am ean annual cost of CHF 1153.03 per patient for pioglitazone (taken directly from PROactive data and corresponding to daily costs of CHF 2.35, 3.12 and 3.71 for treatment with 15 mg, 30 mg and 45 mg per day) and zero for placebo.

Discounting, time horizon and perspective
Alifetime horizon of 35 years was used in the analysis to capture all relevant long-term complications, their associated costs and impact on life expectancy.Discounting was applied to costs, life expectancy and quality-adjusted life expectancy at an annual rate of 2.5% in line with current recommendations for the Swiss setting [29].The analysis was conducted from ahealthcare payer perspective in Switzerland.

Sensitivity analyses
One-way sensitivity analyses were performed to investigate the impact of varying key parameters on the base case incremental cost-effectiveness ratio (ICER).The impact of time horizon was investigated by reporting ICERs at 5, 10 and 20 years.Similarly the impact of discounting was assessed by using alternative discount rates between zero and 5%.To investigate the potential impact of pioglitazone-associated improvement in betacell function, as ensitivity analysis was performed where the HbA1c creep of 0.15% annually in years 4+ of the simulation was reduced to 0.1%.The influence of risk adjustment for age on the event rates taken from PROactive was investigated by performing an analysis where no risk adjustment for age was applied during the simulation (risk adjustment factors all set to 1).
In the base case it was assumed that the pioglitazone related treatment effects were maintained over patient lifetimes.The validity of this assumption was investigated by limiting treatment effect to aperiod of only five years, and thereafter both treatment arms followed the clinical progression based on event rates from the placebo arm of PROactive.For the base case analysis input parameters were based on the corresponding mean reported from the RCT.T oi nvestigate the impact of uncertainty surrounding input parameters probabilistic sensitivity analy- sis was performed, outcomes were projected with sampling from the distributions defined by the mean and standard deviation of patient age, HbA1c, duration of diabetes, SBP,BMI, individual lipid fractions and treatment effects observed in the RCT from which data was taken for the base case analysis.
In the base case scenario, quality-adjusted life expectancy was calculated using ap ublished formula and utility scores from the CODE-2 study [30].Whilst this approach ensures ar obust estimation of QALYs in the base case analysis (through preservation of the integrity of the CODE-2 formula), ap otential limitation of this approach is that it fails to take into account utilities associated with an umber of the endpoints reported in PROactive and captured in the present modelling analysis (e.g.heart failure, oedema and myocardial infarction).
To investigate the influence of including quality of life disutilities that were not included in the CODE-2 formula, anumber of sensitivity analyses were run to include additional quality of life disutilities.In brief, this involved repeating the base case analysis using the CORE default method of quality-adjusted life expectancy estimation [17], applying an event disutility of -0.01 for oedema and af ollow up disutility of 0( as the condition is typically short-lived), applying an event disutility to hospitalisation for heart failure of -0.121 and af ollow-up disutility of -0.181, based on the UKPDS [31], or applying an event disutility of -0.0605 for non-serious heart failure and af ollow-up disutility of 0. Utility values for oedema and non-hospitalised heart failure were based on assumptions.

Clinical outcomes
Based on clinical findings for PROactive, long-term projections with am odified version of the CORE Diabetes Model indicated that treatment with pioglitazone was associated with improvements in life expectancy and qualityadjusted life expectancy (expressed in qualityadjusted life years, QALYs) compared to placebo.Mean life expectancy increased by 0.258 years with pioglitazone and after adjustment for quality of life an improvement of 0.180 QALYs was projected versus placebo (table 3).When the discount rate was set to zero (no discounting), mean life expectancy in the pioglitazone treatment arm was 0.406 years longer than in the placebo arm.
Estimation of long-term complication rates demonstrated that, over patient lifetimes, the pioglitazone treatment regimen was associated with ar educed number of events versus placebo for most diabetes-related outcomes, including MI, TIA, stroke, PCI, ACSand CABG.Exceptions to this were projected for heart failure, oedema and leg revascularisation where pioglitazone was associated with increased cumulative events versus placebo.

Lifetime costs and cost-effectiveness
Over patient lifetimes, treatment with pioglitazone was associated with higher direct medical costs than the placebo regimen (table 3).Direct costs increased by CHF 10 914 with pioglitazone compared to placebo.This increase was largely due to increased drug acquisition costs (CHF 63 813 versus CHF 55 633).Complication-related costs were slightly higher in the pioglitazone arm (difference CHF 2734) due mainly to the increase in hospitalisation for heart failure (difference CHF 5680) and longer life expectancy (resulting in patients being exposed to the risk of complications for al onger period).Tr eatment with pioglitazone was associated with ar educed cost for stroke events by CHF 2953 per patient, for MI events (CHF 901), ACS( CHF 956) and PCI and CABG (CHF 1226) compared to placebo.
Estimation of incremental cost-effectiveness ratios (ICER) for pioglitazone versus placebo treatment produced values of CHF 42 274 per life year gained and, taking quality of life into account, CHF 60 596 per QALYg ained (table 3).The values from the 1000 means (each from 1000 patients) of incremental costs and incremental effectiveness (in terms of quality-adjusted life ex-

Ta ble 3
Summary of base case results for pioglitazone versus placebo.
pectancy) were used to generate as catter plot on the cost-effectiveness plane (fig.1).This analysis shows that the majority of points were in the upper right quadrant of the plane, indicating increased effectiveness and increased costs associated with pioglitazone treatment over placebo.These values were then used to create ac ost-effectiveness acceptability curve, by assessing what proportion of values fell below set willingness to pay values (fig.2).The analysis demonstrated that, with awillingness to pay of CHF 80 000 per QALYi nt he Swiss setting, there was a6 2.5% chance that pioglitazone would be cost-effective.Assuming aw illingness to pay of CHF 61,000 (€ 40 000) per QALY, in line with the threshold applied in ar ecently published analysis of drugeluting stents in the Swiss setting (2004 costs), there was a5 4% chance that pioglitazone would be cost-effective.

Sensitivity analyses
Sensitivity analysis demonstrated the results were most sensitive to variation in the time horizon and assumptions on the duration of the benefits of pioglitazone treatment seen in the trial (table 4).Shorter time horizons were associated with increased ICERs compared to the base case because development of many long-term complications was not captured.Many of the clinical and subsequent economic benefits, such as reduced rates and costs of nephropathy and macrovascular complications, associated with improved HbA1c levels in patients on pioglitazone, are more likely to occur at later stages beyond the 10-year time horizon.
Where the effect of pioglitazone treatment on CVD risk was applied for the first five years of the simulation only,l ife expectancy and qualityadjusted life expectancy were reduced for pioglitazone treatment in the sensitivity analysis compared to the base case.To tal costs were largely unchanged which produced an ICER of CHF 439 313 per QALYgained for pioglitazone versus placebo.Addition of various disutility values not included in the base case analysis and probabilistic sensitivity analysis to investigate the impact of uncertainty surrounding patient-and treatment effect-related input variables had little effect on the overall findings of the analysis.
Reducing the annual HbA1c creep from 0.15% to av alue of 0.1% to simulate ap ioglitazone associated delay in β-cell deterioration improved projected QALE and reduced direct costs resulting in an ICER of CHF 33 845 per QALY gained compared to the base case value of CHF 60 596 per QALYgained.
Cost-effectiveness acceptability curve for pioglitazone versus placebo.

Discussion
In Switzerland it is estimated that approximately 25% of patients with type 2diabetes have a history of macrovascular events, and therefore in common with patients from PROactive these patients are at an increased risk for future cardiovascular events compared to most Swiss patients with diabetes.The findings of this long-term analysis of the cost-effectiveness of pioglitazone versus placebo treatment arms from PROactive indicate that the pioglitazone regimen would be associated with an ICER of approximately CHF 42 274 per life year gained and CHF 60 596 per QALY gained over patients' lifetimes for those with ahistory of macrovascular events.In the base case analysis, treatment with pioglitazone was associated with improvements in life expectancy of 0.258 years and quality-adjusted life expectancy of 0.180 QALYs, and higher direct medical costs (CHF 10 914) over patient lifetimes.Cost-effectiveness acceptability curve analysis indicated that there would be a6 2.5% likelihood that pioglitazone would be cost-effective with awillingness to pay threshold of CHF 80 000 per QALYg ained.In line with ar ecent Swiss based analysis of the cost-effectiveness of drug-eluting stents, aw illingness to pay threshold of CHF 61 000 (€ 40 000) per QALYw as also examined, and under this assumption the likelihood that pioglitazone would be cost-effective was approximately 54%.Whilst we acknowledge that these represent arbitrary willingness to pay thresholds, to our knowledge ad efined threshold has not been an-incremental quality-adjusted life expectancy (QAlys) incremental costs (CHF) Scatter plot of incremental costs and incremental effectiveness for pioglitazone versus placebo.
The red ellipse indicates a95% confidence interval.
nounced for the Swiss setting.Indeed, in many countries the definition of willingness-to-pay thresholds has become ac ontroversial issue, with some rejecting or limiting the use of this in decision-making, whilst in others such as the UK there has been ac all for both increases and decreases of the currently defined threshold by different sectors of the health care industry [33].
Recently as imilar analysis of pioglitazone in the UK setting was published where the ICER of pioglitazone versus placebo was £5 396 (approximately CHF 13 274) per QALYg ained [20].The difference in results between these settings was largely due to differences in cost structure, for example the daily cost of treatment with 30 mg pioglitazone was CHF 3.12 in the Swiss setting versus CHF 2.85 (£ 1. 19) in the UK, whilst the cost of hospitalisation for MI was CHF 14 174.93 versus CHF 15 027.85 (£ 6219.03) and for haemodialysis CHF 81 226.65 versus CHF 63 003.60 (£ 26 073) in the Swiss and UK settings respectively.T hese differences in costs and projected outcomes emphasise the im-portance of conducting country-specific health economic evaluations that take these differences into consideration [34][35][36].
There have been two major cost-effectiveness analyses in the treatment of patients with type 2 diabetes published in recent years.In 2008, Gaede et al. published al ong-term cost-effectiveness of intensive multifactorial intervention versus conventional therapy based on data from the STENO Diabetes Centre in Denmark [37].In this population with inadequate glycaemic control and microalbuminuria at baseline, long-term projections with ab espoke, trial-based model indicated that intensive therapy was associated with aa nI CER of €2 538 per QALYg ained from a Danish healthcare payer perspective.In 2001, the UKPDS group reported an analysis designed to estimate the cost-effectiveness of intensive bloodglucose control with metformin compared with conventional therapy (primarily diet) in newly-diagnosed overweight patients with type 2d iabetes [38].This analysis showed that the metformin intervention was dominant to conventional therapy (improved life expectancy and reduced costs due to complications avoided).Whilst it is difficult to make any form of comparison between cost-effectiveness studies in different populations of diabetes patients in different cost settings, these data serve to suggest that there is potential to have highly cost-effective or even cost-saving interventions in the right target population.
Even comparisons between studies conducted in the same country-specific setting can be difficult due to differences in the approach taken and reporting of health economic outcomes according to different measures such as cost per quality adjusted life year gained versus per life year saved and cost per event avoided.Nevertheless areview of the published literature identified an umber of studies conducted in the Swiss setting that serve to place the results presented here into ac ontext relevant to the healthcare payer (table 5).It should be noted here that some of the listed cost-effectiveness studies report outcomes in terms of life years gained, and accordingly these should only be compared directly to the current study ICER of CHF 42 274 per life year gained.In contrast, the remaining studies listed in table 5, report ICERs ranging between CHF 10 700 and CHF 61 550 (€ 40 467) per QALYg ained which can be compared to the cost-effectiveness of pioglitazone reported here at CHF 60 596 per QALYgained over patients' lifetimes.
Ap otential weakness of this analysis was the conservative approach taken to the estimation of quality-adjusted life expectancy using only data from CODE-2.This estimation did not capture changes in quality of life associated with several macrovascular endpoints (MI, ACS, PCI, CABG, TIA, stroke, oedema or revascularisation of the leg).It is possible that this methodology may underestimate the improvements in quality-adjusted life expectancy as the formula does not capture some of the benefits of pioglitazone treatment (reduced rates of MI ACS, PCI and CABG, stroke) or certain disadvantages such as oedema (although this was partially captured by the inclusion of BMI disutility data), hospitalisation for heart failure and revascularisation of the leg.This was addressed in the sensitivity analysis by including quality of life disutilities related to these endpoints and resulted in ICERs that were higher than projected in the base case because of ar educed between-treatment group difference in quality-adjusted life expectancy (ICER range CHF 63 531-94 654 per QALYgained versus base case CHF 60 596 per QALYgained).Of particular note, when disutilities for all CVD events captured by the model were included the difference in QALE, was 0.135 QALYs versus 0.180 QALYs in the base case, and the corresponding ICER was CHF 80 612 per QALYgained.
As econd potential criticism of the current study was the simulation of ap an European cohort as opposed to aSwiss cohort.The reason for this was that we were unable to identify sufficiently detailed and published reports of Swiss diabetes patients with ah istory of macrovascular events.Given that such patients comprise only approximately 25% of the Swiss diabetes population, it was considered inappropriate to use the profile of "typical" Swiss patients.However, PROactive was conducted exclusively in European countries and 0.8% of the 5238 patients were recruited from Swiss centres.Therefore, until suitable cohort data specific for Switzerland is published, we believe that this approach was the most appropriate for the current analysis.We also acknowledge that given the association of pioglitazone with both benefits (improved HbA1c and lipids) and side effects (increased HF and oedema) the provision of information regarding the lowest numbers needed to treat and the highest number needed to harm would be of value to health care providers.Unfortunately this is not currently possible with the PROactive model but would be an important consideration for future development.
Given that declining β-cell function is a major contributor to deterioration in glucose tolerance, it is reasoned that the potential for pioglitazone-related beta-cell preservation, as suggested by studies such as the COM06, might serve to delay the progression of type 2d iabetes [39].The association between TZDs and stabilisation of β-cell function has not been reported with other oral antidiabetic agents such as metformin and sulfonylureas, and clearly warrants further investigation.Sensitivity analysis that accounted for pioglitazone-associated stabilisation of β-cell function showed improvement in the projected QALE and reduced complication costs leading to an ICER of CHF 33 845 per life year gained.As new information becomes available regarding the longer-term impact of TZDs on pancreatic islet function it will be of interest to investigate the potential impact on health and economic outcomes.
The PROactive cohort was recruited based on ad iagnosis of type 2d iabetes and evidence of macrovascular events, with approximately 19% and 46% of patients reporting ahistory of stroke or AMI respectively.T he positive impact of pioglitazone on outcomes observed in PROactive is likely to be ac onsequence of both improved glycaemic control and improved lipid levels.Surveys of Swiss type 2diabetes patients indicate that approximately 50% of patients have dyslipidemia, 60% are hypertensive and approximately 25% of patients have at least one macrovascular complication [15,16,40,41].Hence the predominately European cohort included in PROactive and the outcomes reported from PROactive are highly relevant to at least 25% of type 2diabetes patients currently cared for in the Swiss setting.Based on outcomes from PROactive, the health-economic analysis presented here has shown that for type 2 diabetes patients with ah istory of macrovascular complications the addition of pioglitazone to current treatment regimens would represent an acceptable treatment option both in clinical and economic terms in the Swiss setting.
Prof. EErdmann is amember of the PROactive Executive Committee.We would like to acknowledge the collaboration and commitment of the PROactive investigators and their staff, without whom the present study would not have been possible.The PROactive simulation model of diabetes is amodified version of the CORe diabetes Model.As shown here in diagrammatic form, the PROactive simulation model is composed of sub-models developed using data observed in the PROactive trial and submodels based on published literature as occursinthe original CORe diabetes Model.CdM =CORe diabetes Model; SBP =S ystolic blood pressure; PVd =P eripheral vascular disease; AMi =Acute myocardial infarction; ACS=Acute coronary syndrome; HF =Heartfailure; TiA =Transient ischemic attack; SVl=S evere vision loss

Ta ble 6
Summary of events, event rates and hazard rates from PROactive Study.

178
Cost-effectiveness of pioglitazone in patients with type 2diabetes and ahistory of macrovascular disease in aS wiss setting Figure 3Overview of the PROactive simulation model.
CostsDirect medical costs were expressed in 2005 Swiss Francs (CHF).Swiss unit costs were retrieved from published sources (table 2) and those not expressed in 2005 CHF were inflated using indices from the Swiss Statistical Ye arbook 2006, published by the Swiss Federal Statistical Office.Where Swiss costs could not be identified no costs were accounted (table [42]C =high density lipoprotein cholesterol; LDL-C =low density lipoprotein cholesterol; BMI =b ody mass index.Long-term progression (beyond 4y ears) follows patterns described in Scuffham and Chaplin[42]

176 Cost-effectiveness of pioglitazone in patients with type 2diabetes and ahistory of macrovascular disease in aS wiss setting Event cost (CHF) Followupcost (CHF) Reference(s)
*Where Swiss costs could not be identified no costs were accounted

effectiveness of pioglitazone in patients with type 2diabetes and ahistory of macrovascular disease in aS wiss setting StudyC ost Intervention Currency ICER year
(11)(12)d Wkly2 009;139(11)(12):173-184 •w ww.smw.ch Values shown are means with standard deviation in parentheses.ICER =incremental cost-effectiveness ratio; QALY=quality-adjusted life years.ICER =Incremental cost-effectiveness ratio; QALY=Quality-adjusted life years; LY G=Life years gained *based on exchange rates corresponding to the year of study as supplied by www.xe.com