• HER2-positive breast cancer;
  • model-based economic evaluation;
  • transferability;
  • trastuzumab


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

Introduction:  Geographic transferability of model-based cost–effectiveness results may facilitate and shorten the reimbursement process of new pharmaceuticals. This study provides a real world example of transferring a cost–effectiveness study of trastuzumab for the adjuvant treatment of HER2-positive early breast cancer from the United Kingdom to The Netherlands.

Methods:  Three successive steps were taken. Step 1: Collect available information with regard to the original model, and assess transferability using existing checklists. Step 2: Adapt transferability-limiting factors. Step 3: Obtain a country-specific estimate of cost–effectiveness.

Results:  The structure of the UK model was transferable, although some of the model inputs needed adaptation. From a health-care perspective, the Dutch estimate amounted to €5828/quality-adjusted life-year gained. From a societal perspective, the incremental cost–effectiveness ratio was dominant.

Conclusion:  Transferability of a model-based UK-study in three steps proved to be an efficient method to provide an early indication of the cost–effectiveness of trastuzumab and has led to the provisional reimbursement of the treatment.


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

Geographic transferability of model-based cost–effectiveness results from other countries can have the potential to facilitate and shorten the appraisal process regarding the reimbursement of new pharmaceuticals. Several authors have addressed the importance of the generalizability of the results of economic evaluations in health care [1,2]. In addition, checklists have been developed to assess the degree of transferability of the results of economic evaluations in health care from one geographic region to another [3–5].

Still, the actual transferability of in particular model-based cost–effectiveness results has gained less attention. This study provides a real world example of transferring a model-based economic evaluation of trastuzumab (Herceptin; Roche Netherlands, BV, Woerden, The Netherlands) for the adjuvant treatment of HER2-positive early breast cancer from the United Kingdom (UK) to the Dutch setting. Three successive steps were performed:

  • 1
    Checklists were applied to assess transferability of the UK model.
  • 2
    Adaptations to the identified transferability-limiting factors were made.
  • 3
    Cost–effectiveness of trastuzumab in the Dutch setting was estimated.


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

Step 1: Assess Transferability

The transferability decision chart by Welte et al. consists of general and specific knock out criteria of transferability. If the general knock out criteria are fulfilled (i.e., the intervention is not comparable with the one that shall be used in the decision country, or the quality of the study is not acceptable) transfer of the study would be so difficult that conducting a new study is the better option. Specific transferability criteria have been developed to identify study parts that need adaptation. The checklist by Boulenger et al. consists of 42 questions related to the methodological approach used in the published studies. Each question can be answered with 1 for “yes,” 0.5 for “partially,” and 0 for “no,” or “no information.” Because no standard is available to determine what a certain score means in terms of the level of transferability, ultimately, a transferability summary score can be calculated although interpretation is difficult. The checklist by Urdahl et al. addresses in particular four questions related to: 1) definition of target decision-maker or jurisdiction; 2) transparent reporting of model specification; 3) relevance of data inputs to target decision-maker or jurisdiction; and 4) assessment of robustness of the model to geographic variation in data inputs.

Step 2: Adapt Factors That Limit Transferability

In Table 1, a detailed overview of specific transferability criteria based on the checklist by Welte et al. [3] is given. Transferability of the study may be limited because of factors present in each of these parts. In step 2, transferability-limiting factors were linked to a specific study part and it was examined whether it was feasible to adapt this factor based on existing knowledge.

Table 1.  Transferability of the trastuzumab UK model-based cost–effectiveness analysis to The Netherlands according to the Welte checklist
Transferability factorDirect influence onEstimated relevance for transferability UK—Dutch settingEstimated correspondence between study and decision countryEstimation of CER of decision country based on CER of study country
  • *

    Detailed cost information per unit was given in the UK cost–effectiveness model. The estimated correspondence was assumed to be low, which led to the decision to replace all resource valuation with Dutch unit prices;

  • In the UK cost–effectiveness model, no loss of productivity costs was calculated. In the Dutch setting, the standard approach is the friction cost method;

  • Although detailed cost-information per unit was given in the UK cost–effectiveness model, it was not feasible to assess the difference in absolute and relative prices. Therefore, the decision was made to use UK resource use but replace UK unit costs by Dutch unit costs.

  • CER, cost-effectiveness ratio.

Methodological characteristics    
 PerspectiveCosts and effectsHighHighUnbiased
 Discount rateCosts and effectsHighMediumUnbiased
 Medical cost approachDirect medical costsHighLow*Biased
 Productivity cost approachProductivity costsNot calculated
Health-care system characteristics    
 Absolute and relative prices in healthcareDirect medical costsHighLowUncertain
 Practice variationCosts and effectsHighHighUnbiased
 Technology availabilityCostsHighHighUnbiased
Population characteristics    
 Disease incidenceCosts and effectsHighHighUnbiased
 Case—mixCosts and effectsHighHighUnbiased
 Life expectancyCosts and effectsHighHighUnbiased
 Health—status preferencesEffectsHighLowBiased
 Acceptance, compliance, incentives to patientsCosts and effectsn.a.
 Productivity and work loss timeProductivity costsNot calculated
 Disease spreadCosts and effectsNot relevant 

Step 3: Estimate Country Specific Cost–Effectiveness

Based on the adapted study parts, a country-specific estimate of cost–effectiveness was calculated. Also, uncertainty was addressed in probabilistic sensitivity analysis and one-way or n-way sensitivity analyses.

The Case: Model-Based Cost–Effectiveness Analysis of Trastuzumab in the Early Setting from the UK

In the UK study, a Markov cohort model was used to calculate the long-term costs, effects, and cost–effectiveness of adjuvant trastuzumab for 1 year compared with observations (no adjuvant treatment with trastuzumab in the early breast cancer setting, although trastuzumab is given to patients who progress to metastatic disease). The study population consisted of patients aged 50 years and older with HER2-positive early breast cancer. In Figure 1, all model transitions between the health states are presented.


Figure 1. Model transitions between health states (Wardley 2006) [6]. All patients start in the disease-free survival state and in subsequent cycles they may move to the other health states: loco-regional/contralateral recurrence, metastatic disease, and death. The model distinguishes between “severe cardiac adverse events,” and “other than severe cardiac adverse events” that can occur during treatment with trastuzumab. Patients who develop a cardiac side effect during the treatment period are assumed to stop treatment, and from that moment on have the same risk of developing breast cancer recurrence and metastasis as nontreated patients. After the occurrence of one of the cardiac side effects, patients may continue to experience “chronic cardiac adverse events” for the rest of their lives. *With or without cardiac event.

Download figure to PowerPoint

Outcomes considered in the model are life-years, quality adjusted life-years (QALYs), and health-care costs. Cycle length in the model was 1 year, and the time horizon was 45 years. For utility and cost calculations, a half-cycle correction was applied. A probabilistic sensitivity analysis was performed. All model inputs are presented in Table 2.

Table 2.  Summary of model inputs
 United KingdomThe Netherlands
Discount ratesRate (%)Rate (%)
 Mean* (€)Lower (€)Upper (€)Mean (€)Lower (€)Upper (€)
  • *

    Exchange rate: £1 =€1.4;

  • Tengs and Wallace, 2000 [8];

  • Hayman, 1998 [10];

  • §

    Carter, 1998 [11];

  • Van Hanswijck de Jonge, 2006 [12];

  • **

    Wardley, 2006 [6];

  • ††

    Lidgren, 2007 [13];

  • ‡‡

    HERA trial (Piccart–Gebhart, 2005 [14]);

  • §§

    After 5 years of simulation in both treatment arms these transition probabilities are adjusted for time in the disease-free state using a stepwise function: 6 to10 years: 0.58, 11 years to end: 0.37 (EBCTCG, 2006 [15]);

  • ¶¶

    Year 1–5 HERA trial (Piccart–Gebhart, 2005), from year six assumptions;

  • ***

    Trastuzumab in the metastatic stage is only used in the ‘no trastuzumab’ in the adjuvant setting’ treatment arm.

  • HERA, Herceptin Adjuvant; DFS, disease-free survival; EBCTG, Early Breast Cancer Trialists' Collaborative Group.

Health state costs      
 One year in disease-free state2.4392555.0881.2761332.663
 First year in local- or contralateral recurrence state18.94844134.6789.37095543.238
 Cost of a recurrence event4.5913.8255.2471.6051.3911.901
 First year in metastatic state31.37516.86953.94620.1923.91852.889
 One year in the metastatic state after first year17.5388.21127.08711.2662.82522.768
Cardiac-related costs      
 Severe cardiac adverse events10.1427.49112.79214.2267.11328.452
 Other cardiac adverse events1.9521.7572.1467.1133.55714.226
 Chronic cardiac adverse events4423994877.1133.55714.226
 Heart monitoring per year1.6148082.4221.2757051.394
Trastuzumab-related costs      
 HER2 test715288114109420
 Trastuzumab in the adjuvant setting34.18822.98845.97641.54137.45749.942
 Administration in the adjuvant setting3.9823.5844.3793.9063.4724.340
 Trastuzumab in the metastatic setting24.52516.49132.98329.82624.0003.600
 Administration in the metastatic setting7.9637.1678.7609.1148.6809.548
Productivity costs   Mean  
 Days lost in recurrence and metastatic state   65 daysfixed 
 Costs per day lost   €308fixed 
Utilities for health statesMeanLowerUpperMeanLowerUpper
 Disease-free survival first year0.7490.7030.7950.696††0.6340.747
 Disease-free survival after first year0.8470.8070.8860.779††0.7450.811
 Metastasis disease first year0.4840.4260.5420.685††0.6200.735
 Metastasis disease after first year0.484§0.4260.5420.685††0.6200.735
 Local recurrence0.8100.7600.8700.779††0.7000.849
Disutility for events      
 Local recurrence event0.2400.1920.288as in UK
 Contralateral breast cancer0.2400.1920.288
 Cardiac event0.300**0.2500.350
 DFS to contralateral breast cancer§§0.0070.0020.009as in UK
 DFS to metastasis§§0.0790.0390.106
 DFS to local recurrence§§0.0290.0090.035
 Local recurrence to metastasis0.0780.0390.106
 Cardiac adverse events0.0050.0030.010
 Cardiac event to local recurrence0.0290.0090.035
 Cardiac event to metastasis0.0780.0390.106
 Probability of mild cardiac event0.0350.0300.040
 Metastasis state to death0.2950.2000.700
 HER2 positive rate20%fixed 
 Incidence cardiac adverse events1.2%fixed 
Relative risks trastuzumab¶¶Years    
 Disease-free survival to local recurrence1–50.5100.3201.000as in UK
11 to end0.8620.3201.000
 Disease-free survival to metastasis1–50.5200.4001.000
11 to end0.8790.4001.000
 Local recurrence to metastasis1–50.4000.1501.080
11 to end0.6760.1501.080
Metastasis to death***0.800fixed 
Background mortalityUK life tablesDutch life tables


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

Step 1: Transferability Check

Information regarding the UK study included a poster [6], a technology appraisal by the National Institute for Health and Clinical Excellence (NICE) [7], a mathematical model, technical documentation, clinical data, and a costing study (provided by Roche). The main reason for using the UK model was that its validity had been established by NICE. In addition, the UK study passed the general knock out criteria from Welte et al. [3]. Hence, the model structure was considered of good quality. As presented in Table 3, the following input parameters needed adaptation: price level, discount rate, background mortality, utilities, resource use, unit prices, and productivity loss. Three questions in the checklist by Boulenger et al. [4] were answered with “no.” First, no conversion rate was given; second, no analyses were conducted to explore geographic variability; and third, caveats regarding the generalizability of the results were not discussed. We interpreted the resulting score of 94% as an adequate description of the model inputs. The transferability questions, as formulated by Urdahl et al., could also be answered. It is common procedure in the UK that cost–effectiveness analyses are commissioned by NICE. The manufacturer (i.e., Roche) provides the evidence for submission. Transition probabilities and health utilities were adequately described in the technical documentation. The UK resource use was based on expert opinion (N = 7) and the clinical data input were derived from the Herceptin Adjuvant (HERA)-trial [14]. An assessment of the robustness of the model to geographic variation in data inputs was not included in the UK analysis.

Table 3.  Overview of the results of step 1 (transferability check) and step 2 (adaptations) in the transferability process
Study partStep 1Step 2Explanation
Transferability limiting factorAdapted
  • *

    Dutch Central Bureau of Statistics,;

  • Pharmacoeconomic guidelines, Health Care Insurance Board;

  • Oostenbrink, Dutch manual for costing research;

  • §

    § Cost prices of the Health Care Insurance Board;

  • Cost prices of the Maastricht University Medical Centre;

  • **

    **Dutch Pharmaceutical Compass;

  • ††

    Koopmanschap et al. [16].

Model structure   
 Health statesNoTransferable
General inputs   
 Price levelYesYesPrice index figures*
 Discount rate costsYesYesPhamacoeconomic guidelines
 Discount rates effectsYesYesPhamacoeconomic guidelines
Transition probabilities   
 Transition probabilitiesNoTransferable
 Risk reductionsNoTransferable
 Background mortalityYesYesDutch life tables*
Utilities assigned to health states/events/paths   
 Health state description?PartlyNo Dutch utilities available. Used Swedish study (Lidgren, 2007) and expert opinion instead
 Health state valuationYesPartly
Costs assigned to health states/events/paths   
 Health-care resource useYesPartlyNo Dutch data available
 Health-care unit pricesYesYesDutch standard prices‡,§, Dutch Hospital prices, Dutch Pharmaceutical Compass**
 Patient and family resource useNoNot included in the study
 Patient and family unit pricesNoNot included in the study
 Resource use in other sectorsNoNot included in the study
 Unit prices in other sectorsNoNot included in the study
 Productivity costs approachYesYesFriction cost method‡,††
 Productivity lossYesYesExpert opinion
 Valuation of lost productivityYesYesDutch standard prices
Model analysis   
 Type of model analysisNoCohort simulation, transferable
 Probabilistic sensitivity analysisNoTransferable
 One-way, N-way sensitivity analysesYesYesSensitivity analyses on input parameters (HER2 positivity rate) and adaptations added

The overall conclusion of the transferability check was that the structure of the model was transferable to the Dutch setting.

Step 2: Adaptations Made to the UK Model-Based Cost–Effectiveness Study

Table 3 presents an overview of the adaptations to transferability-limiting study parts. For the adaptation of the health-care resource use, we could only partly apply the Dutch data.The reason was that in The Netherlands, no detailed information was available about the economic consequences of breast cancer regarding the various health states. Instead, we used the UK data because these were readily available, and subsequently assessed their impact on the incremental cost–effectiveness ratio (ICER) in sensitivity analyses. More information about the adaptations can be found in the appendix at:

Step 3: Cost–Effectiveness Estimate for the Dutch Situation

Results (Table 4) indicate that, from a health-care perspective, treatment with trastuzumab leads to a gain of 2.25 QALYs (discounted). In terms of cost–effectiveness, the use of trastuzumab for a 50 years old patient amounts to €5.828 per QALY gained (discounted). From a societal perspective (including indirect costs), the ICER is dominant. The acceptability curves (Fig. 2) show that for the base case analysis (50 years, health-care perspective), the probability that the net benefit is positive, is 1 for thresholds of €26,000/QALY and higher. One-way sensitivity analyses demonstrated that results were only sensitive to the use of trastuzumab in the metastatic phase for the comparator group (Fig. 3) and not, for example, to a 50% decrease in the costs of the metastatic phase.

Table 4.  Outcomes deterministic cost–effectiveness analysis base case analysis (50 years, 20% HER2 positive)
 United KingdomDutch estimate based on UK model based analysis
TrastuzumabNo treatmentIncrementalTrastuzumabNo treatmentIncremental
Cost (€)QALYsCost (€)QALYsCost (€)QALYsICER (€)Cost (€)QALYsCost (€)QALYsCost (€)QALYsICER (€)
Health-care perspective              
Societal perspective              
 Discounted       123.72712.98132.18210.72−8.4552.25Dominant
 Undiscounted       154.29515.73165.26612.71−10.9713.02Dominant
 Cost (€)LYCost (€)LYCost (€)LYGICER (€)Cost (€)LYCost (€)LYCostLYGICER (€)
  1. QALY, quality adjusted life-years; ICER, incremental cost–effectiveness ratio; LY, life-year; LYG, life years gained.

Health-care perspective              
Societal perspective              
 Discounted       123.72717.11132.18214.32−8.4552.79Dominant
 Undiscounted       154.29520.70165.26616.96−10.9713.74Dominant

Figure 2. Cost–effectiveness acceptability curves for the UK and The Netherlands (health-care and societal perspective). QALY, quality adjusted life-years.

Download figure to PowerPoint


Figure 3. Cost–effectiveness acceptability curves for sensitivity analyses on the data inputs for the Dutch health-care perspective. QALY, quality adjusted life-years.

Download figure to PowerPoint

In addition, replacing UK background mortality with Dutch figures had virtually no influence. Using discount rates as recommended in the Netherlands slightly increased the probability of a positive net benefit although mainly adaptations in the health-care costs reduced the probability that net benefit was positive compared with the UK estimate.

As shown in Figure 2, based on cost–effectiveness thresholds of €14,000 (Dutch societal perspective) and €30,000 (Dutch health-care perspective), the probability of a positive net benefit is very close to 1. As a result, for cost–effectiveness thresholds of €30,000 and higher, the expected value of information is zero. Because both cost–effectiveness thresholds are considerably lower than the informal Dutch threshold of €80,000 [9], it is not meaningful to perform expected value of perfect information calculations.


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References

In this study, we obtained a Dutch estimate of the cost–effectiveness of trastuzumab for the early-stage breast cancer setting based on a model-based cost–effectiveness study developed for the UK setting, initiated by Roche.

The structure of the UK model was transferable, although some of the model inputs needed adaptation. Because of the absence of Dutch data, it was not feasible to fully adapt utilities and health-care resource use. While the epidemiology of cancer is very well documented in The Netherlands, there is little information about the economic consequences of breast cancer. As a result, it turned out to be difficult to collect Dutch resource use on a detailed level. We assumed a high correspondence in practice variation between the UK and The Netherlands regarding the various health states (disease-free survival, recurrence, and metastatic phase) for the group of HER2-positive early breast cancer patients. Therefore, we decided to use resource volume data regarding the different health states of the UK study as a proxy for the Dutch resource use, while we replaced nearly all unit costs with Dutch estimates. A limitation of the current model adaptation is that the extent to which the use of UK resource volume data has resulted in an upward or downward bias of the Dutch cost–effectiveness estimate is unknown. Nevertheless, sensitivity analysis showed that the ICER was only sensitive to the use of trastuzumab in the no comparator group and not, for example, to a 50% decrease in the costs of the metastatic phase.

A challenge for the geographical transferability of model-based cost–effectiveness results is the level of transparency in reporting the model study. This requires not only transparent reporting of the methods and results, but also access to the underlying mathematical model. At this moment, this is far from common. A recommendation to editorial boards and researchers would be to examine possibilities to make the mathematical model an integral part of a scientific publication of a model-based cost–effectiveness analysis. Another challenge regarding the transferability of economic evaluations is the availability of health-care resource data. More attention should be given to a reliable registration of resource consumption related to the different health states of diseases like, for example, breast cancer.

The analyses in this paper indicated that treatment with trastuzumab results in an average gain of 2.79 life-years and 2.25 QALYs. From a health-care perspective, the cost–effectiveness of trastuzumab for a patient aged 50 amounts to €5828 per QALY saved. From a societal perspective, the ICER is dominant. The findings of this transferability study support the use of trastuzamab as a cost–effective adjuvant treatment for patients with HER2-positive early stage breast cancer in The Netherlands. As such, our study provides a real-world example in which three successive steps were undertaken to transfer a foreign model-based economic evaluation. This proved to be an efficient method in order to obtain an early indication of the cost–effectiveness of adjuvant trastuzumab and has led to the provisional reimbursement of this treatment in The Netherlands.

Source of financial support: Roche.


  1. Top of page
  3. Introduction
  4. Methods
  5. Results
  6. Conclusions
  7. References
  • 1
    Goeree R, Burke N, O'Reilly D, et al. Transferability of economic evaluations: approaches and factors to consider when using results from one geographic area for another. Review. Curr Med Res Opin 2007;23:67182.
  • 2
    Sculpher MJ, Pang FS, Manca A, et al. Generalisability in economic evaluation studies in health care: a review and case studies. Health Technol Assess 2004;49:1192.
  • 3
    Welte R, Feenstra T, Jager H, Leidl R. A decision chart for assessing and improving the transferability of economic evaluation results between countries. Pharmacoeconomics 2004;22:85776.
  • 4
    Boulenger S, Nixon J, Drummond M, et al. Can economic evaluations be made more transferable? Eur J Health Econ 2005;6:33446.
  • 5
    Urdahl H, Manca A, Sculpher M. Assessing generalisability in model-based economic evaluation studies. A structured review in osteoporosis. Pharmacoeconomics 2006;24:118197.
  • 6
    Wardley AM, Cameron DA, Bell R, et al. Cost–effectiveness analysis of adjuvant therapy with trastuzumab (Herceptin) for early breast cancer. Ann Oncol 2006;17:ix95.
  • 7
    Nice Appraisal guidance 107. Trastuzumab for the adjuvant treatment of early-stage HER2-positive breast cancer. August 2006.
  • 8
    Tengs TO, Wallace A. One thousand health-related quality-of-life estimates. Med Care 2000;38:583637.
  • 9
    Health Insurance Council. Fair and Sustainable Care. Amstelveen, 2006 (in Dutch).
  • 10
    Hayman JA, Hillner BE, Harris JR, Weeks JC. Cost–effectiveness of routine radiation therapy following conservative surgery for early-stage breast cancer. J Clin Oncol 1998;16:10229.
  • 11
    Carter KJ, Ritchey NP, Castro F, et al. Treatment of early-stage breast cancer in the elderly: a health–outcome-based approach. Med Decis Making 1998;18:2139.
  • 12
    Van Hanswijck de Jonge P, Doyle S, Farina C, Walker M. Poster Presentation 190 P: Elicitation of UK Health utilities in primary, recurrent and metastatic breast cancer. 31st European Society Medical Oncology (ESMO), Istanbul, Turkey, 2006.
  • 13
    Lidgren M, Wilking N, Jonsson B, Rehnberg C. Health related quality of life in different states of breast cancer. Qual Life Res 2007107381.
  • 14
    Piccart-Gebhart MJ, Procter M, Leyland-Jones B, et al. Herceptin Adjuvant (HERA) Trial Study Team. Trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer. N Engl J Med 2005;353:165972.
  • 15
    Early Breast Cancer Trialists' Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet 2005;365:1687717.
  • 16
    Koopmanschap MA, Rutten FF, Van Ineveld BM, Van Roijen L. The friction cost method for measuring indirect costs of disease. J Health Econ 1995;14:17189.