International Journal of
Radiation Oncology biology
physics
www.redjournal.org
Health Care Economics
In Search of the Economic Sustainability of Hadron Therapy: The Real Cost of Setting Up and Operating a Hadron Facility Barbara Vanderstraeten, PhD,* Jan Verstraete, RN,y Roger De Croock, MScEng,z Wilfried De Neve, PhD, and Yolande Lievens, PhD* *Department of Radiotherapy, Ghent University Hospital, Gent, Belgium; yDepartment of Radiation Oncology, University Hospital Gasthuisberg, Leuven, Belgium; and zBelgian Hadron Therapy Center Foundation, Brussels, Belgium Received Sep 11, 2013, and in revised form Jan 6, 2014. Accepted for publication Jan 23, 2014.
Summary The cost of setting up and running different types of hadron therapy centers (proton only/carbon ion only/ combined) is calculated with different financing methods (public/private). We compare a business model with an activity-based costing method for calculation of the required reimbursement and the treatment costs per patient and per fraction. The complementarity of both methods is shown, as well as the impact of a delay in commissioning, a higher interest rate, higher patient throughput, and hypofractionation.
Purpose: To determine the treatment cost and required reimbursement for a new hadron therapy facility, considering different technical solutions and financing methods. Methods and Materials: The 3 technical solutions analyzed are a carbon only (COC), proton only (POC), and combined (CC) center, each operating 2 treatment rooms and assumed to function at full capacity. A business model defines the required reimbursement and analyzes the financial implications of setting up a facility over time; activitybased costing (ABC) calculates the treatment costs per type of patient for a center in a steady state of operation. Both models compare a private, full-cost approach with public sponsoring, only taking into account operational costs. Results: Yearly operational costs range between V10.0M (M Z million) for a publicly sponsored POC to V24.8M for a CC with private financing. Disregarding inflation, the average treatment cost calculated with ABC (COC: V29,450; POC: V46,342; CC: V46,443 for private financing; respectively V16,059, V28,296, and V23,956 for public sponsoring) is slightly lower than the required reimbursement based on the business model (between V51,200 in a privately funded POC and V18,400 in COC with public sponsoring). Reimbursement for privately financed centers is very sensitive to a delay in commissioning and to the interest rate. Higher throughput and hypofractionation have a positive impact on the treatment costs. Conclusions: Both calculation methods are valid and complementary. The financially most attractive option of a publicly sponsored COC should be balanced to the clinical necessities and the sociopolitical context. Ó 2014 Elsevier Inc.
Reprint requests to: Barbara Vanderstraeten, PhD, Ghent University Hospital, Department of Radiotherapy, De Pintelaan 185 0P7, 9000 Ghent, Belgium. Tel: (þ32) 9-332-19-41; E-mail: barbara.vanderstraeten@ uzgent.be This work was financially supported by the Cancer Plan Action 30 financed by the Federal Office of Health and Social Affairs in Belgium. Int J Radiation Oncol Biol Phys, Vol. 89, No. 1, pp. 152e160, 2014 0360-3016/$ - see front matter Ó 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ijrobp.2014.01.039
The foreign expert group consisted of Alejandro Mazal and Ste´phanie Bolle (Paris, France), Tadashi Kamada (Chiba, Japan), Ju¨rgen Debus and Oliver Ja¨kel (Heidelberg, Germany), Tony Lomax (Villigen, Switzerland), Philippe Lambin (Maastricht, The Netherlands), and Piero Fossati (Pavia, Italy). Conflict of interest: none.
Volume 89 Number 1 2014
Introduction Over the past decades there has been a growing concern about high and rising medical costs (1). The rapid diffusion of new technologies has been proposed as a major contributing factor in this rise (2). Radiation oncology, a highly technological and resource-demanding discipline, is obviously not immune to this general tendency (3, 4). Therefore, even if radiation therapy’s share of the total health care and oncology budget seems modest (5), particularly when considering the large proportion of cancer patients being treated with radiation (6), the radiation therapy community cannot escape from its responsibility to evaluate the financial implications of its clinical actions. Hadron therapy is an example of such a technological evolution in radiation therapy for which the financing and the value for money are heavily debated. This study was sponsored by the government of a European country where the decision to construct a hadron therapy center was pending. Before a decision, the government initiated a feasibility study to gather information regarding the scope, clinical indications, limited technical specifications, and treatment cost in view of reimbursement setting. Lack of reliable investment cost estimates makes hadron therapy projects cumbersome: “list prices” for hadron therapy equipment seem nonexistent, and quotes of vendors depend on many variables, including the country, location, scope (clinical, research, demonstration) and knowhow of the project team (conditions of collaboration for the development of new technologies). Hence, accurate cost estimates require the development of a comprehensive project plan. If the government responds positively to the feasibility study, this is the next step in the governmental process for a project of this scale. A comprehensive project plan requires knowledge of implantation site and type of center, as well as a sufficient budget (5%-7% of the investment cost) to make a business plan including detailed technical descriptions of building and equipment, requiring separate risk capital unless the decision to construct a hadron therapy center is certain. In the end, European legislation imposes the use of a European tender to organize a call for bids, finally leading to vendor selection and contract signing. For the present feasibility study, estimates of investment cost were needed for cost modeling at a stage when resources to develop a comprehensive project plan were not yet available. This seems to be a frequently encountered problem in many countries, as well as for individual project groups who need to demonstrate to potential sponsors of hadron treatments, including health insurance providers, that favorable cost-benefit ratios exist. Early-stage cost modeling has to deal with uncertainties, mainly in investment cost but also in maintenance, upgrade, and operational cost. Two cost models were evaluated. Activity-based costing (ABC), an advanced costaccounting method, has been proposed to assess the financial impact of gradual process and technology changes
The real cost of hadron therapy
153
in radiation therapy and of factors that affect patient numbers and population mixes (3, 7). Developed to compute the cost in an operational department, ABC is, however, less suited to capture the implications of setting up a new facility with carry-over investment and development costs over decades. In practice the long-term financial evaluation of setting up a new department is performed using a business model (BM). This article describes both approaches for 3 types of centers built to treat with protons only, with carbon ions only, or with protons and carbon ions. The investment cost of many centers worldwide has been financed by public resources, often being taxpayers’ money in Europe or donations of a more spontaneous kind in the United States. Less fortunate project groups may have to find private resources, typically bank loans at market conditions, to finance the investment cost. American centers are typically set up on a commercial basis and have to take the full range of capital and risks into account for their reimbursement setting (8). It is therefore not surprising that these centers are known to charge $100,000 or more per patient, whereas reimbursements of approximately V20,000 to V40,000 per patient are commonly charged for hadron treatments in European and Japanese centers. The majority of hadron projects in surrounding countries, to which a potential center in Belgium would benchmark, were started as scientific projects both in the field of clinical and technological development. Calculations of cost per treatment were made for both scenarios, public sponsoring and private financing. Amortization of the investment cost was discarded in the public sponsoring scenario. The specificity and complementarity of both cost models are evaluated: the gross financial implications of starting up a hadron facility over time and ensuing required reimbursement levels are evaluated in a BM; the treatment costs for a center in a specific year of operation are fine-tuned with ABC. Given the uncertainties in some input parameters, the presented reimbursement rates and treatment costs should be interpreted in the context of these assumptions.
Methods and Materials Type of center Three different technical solutions are considered: a carbon only center (COC), a proton only center (POC), and a combined center (CC), each operating 2 treatment rooms. The CC has 1 proton and 1 carbon ion treatment room. Proton treatments are delivered with a gantry, whereas carbon ion treatments use a fixed beam.
Input parameters Table 1 summarizes all investment costs, personnel requirements, and other operational costs. A time slot of
154
International Journal of Radiation Oncology Biology Physics
Vanderstraeten et al.
Table 1
Input parameters for the different technical solutions Center type Parameter
Cost/FTE/year (V)*
CC
COC
POC
20M 10M 40M 20M 4M
25M 50M 4M
15M 25M 4M
2M 1.5M
2M 1.5M
2M 1.5M
1.5M 2.5M 101.5M
2.5M 85M
1.5M 2.5M 51.5M
1 1 1 5 2 6 3 18 4 6 6 4 7
1 1 1 5 2 6 3 18 4 6 6 4 7
1 1 1 5 2 6 3 18 4 6 6 4 5
64 6.0M
64 6.0M
62 5.7M
0 0 760
50 305 0
y
Investment costs (V) Type of investment Carbon ion building Proton building Carbon ion beam equipment Proton beam equipment Imaging equipment in each room Simulation equipment Planning equipment þ record and verify Anesthesia facilities Other facilities Total investment Operational model Daily working hours (center) Working days/wk (center) Working weeks/y (center) Treatment hours/y (center) No. of working hours/wk (staff) Productivity (staff) (%) No. of working weeks/y (staff) Productive hours/FTE/y (staff) Personnel requirements General manager 180,000 Chief physicist 140,000 Chief medical doctor 180,000 Administration 55,000 Data manager 46,000 Radiation oncologists 180,000 Radiographers 56,000 Therapists 56,000 Trajectory nurses 50,000 Dosimetrists 93,000 Physicists 93,000 IT engineers 140,000 Maintenance engineers and 140,000 technicians Total no. of FTE Total cost/y (V) Exploitation costsx Maintenance cost/y (% of investment cost) Buildings Hadron beam equipment Planning equipment þ record and verify Imaging and simulation equipment, anesthesia, and other facilities Insurance cost/y (V) Energy cost/y (V) Facility maintenance cost/y (V) Medical consumable cost/patient (V) Population mixes Type of treatment Pediatric (30 fractions) Adult proton (30 fractions) Adult carbon ion (15 fractions)
07:30-20:00 5 48z 3000 38 88 45 1505
1 5 10 7 20,000 677,754 382,667 750
50 115 369
(continued on next page)
Volume 89 Number 1 2014
The real cost of hadron therapy
155
Table 1 (continued ) Center type Parameter
Cost/FTE/year (V)*
Total no. of patients/y Total no. of fractions/y
CC
COC
POC
534 10,485
760 11,400
355 10,650
Abbreviations: CC Z combined center; COC Z carbon only center; FTE Z full-time equivalent; IT Z information technology; M Z million; POC Z proton only center. * All personnel costs are assumed to be total costs, including any potential extras. See also footnote “x.” y All figures are estimates based on exploratory contacts with vendors and discussions with international experts. All investment costs are amortized over 20 years, except for IT equipment (planning systems, record and verify), which is replaced every 5 years. No costs were taken into account for the lands or for future decommissioning. z A yearly 4-week break is foreseen for maintenance and quality assurance. x All operational costs are subject to a yearly inflation of 2%, except for energy costs (2.5%). All figures are based on KCE report vol. 67A (9), but adapted to current standards based on discussions with the team of international experts and literature review (10-19).
30 minutes per fraction is allotted to adult treatments, compared with 45 minutes for pediatric treatments, assuming the latter require anesthesia (20). For ABC, a product or treatment is defined as the combination of equipment type (proton or carbon ion) and patient type (pediatric or adult). The number of patients that can be treated yearly in each center is also presented in Table 1.
Business model For the BM, treatment costs are modeled in spreadsheets using Excel (Microsoft, Redmond, WA). To analyze the financial implications of setting up a facility over time, different phases are taken into account (Fig. 1). During the 4-year preparation and commissioning phase, long-term loans with interests of 5% are allotted on the investment and personnel costs. The BM is used to determine an average reimbursement per patient for each type of center so as to generate a positive net cumulated cash flow (CCF) 16 years after start-up. This corresponds to the repayment of the first investment (in case of private financing) after 20 years, which is equal to the amortization period. To balance the negative cash outflow during the years before full operation, additional short-term loans are introduced. The reimbursement rate itself is subject to a yearly inflation rate of 2%. Sensitivity analyses (SA) are performed to investigate the effect of delays in the commissioning phase, increases in long-term interest rates and various investment or personnel costs, and a shortening of the amortization period.
Activity-based costing Activity-based costing calculates the cost per product or treatment in a center operating at full capacity. The treatment process is split into a series of activities, each of which is costed, typically on the basis of resource
consumption over a certain amount of time (3, 11). The model used for the present calculations is comparable to that described earlier for photon radiation therapy (10). All treatment-related personnel, building, and equipment costs are allocated to the products on the basis of their activity consumption. Medical consumables are directly traced to the products. Overhead cost is allocated to each product according to the number of fractions. Apart from the cost calculation for each type of treatment, ABC is used to investigate the impact of different population mixes and fractionation schedules and patient loads.
Financing methods Private financing uses a full-costing approach that assumes that all costs are covered through private financing. For the BM, this includes the investment costs, personnel costs, and interim interest on the investment capital during the commissioning phase and any cash drain after start-up until a positive cumulated net cash flow is generated after 16 years. For ABC, equivalent annual investment costs are calculated for each type of building and equipment. For public sponsoring, all investment costs, and for the BM also the personnel costs during commissioning, are taken out of the project. Hence the calculations are based purely on the operational costs.
Results Figure 2 shows the CCF calculated with the BM. The nadir takes place at 3 years after start-up because not all operational costs and capital amortization are covered by the acquired reimbursements yet. This is compensated by short-term loans to obtain the net CCF. The required reimbursement rates to obtain a positive net CCF 16 years after start-up are shown in Table 2 at baseline and for variable investment and personnel costs,
156
Vanderstraeten et al.
International Journal of Radiation Oncology Biology Physics
Fig. 1. Assumed spread of the investment costs and gradual increase of the yearly personnel costs and number of patients during the different phases of the business model. The full building costs are financed in the first year. Proton and carbon equipment investment expenditures are spread over the second (33%) and third (67%) year, whereas imaging and simulation equipment costs are spread over the third (33%) and fourth (67%) year. All other investments are made during the fourth year. During this period a fraction of the personnel is already required for preparation and commissioning. These personnel costs are modeled as a fraction of the total personnel cost of the center operating at full capacity, namely 10% (year 1), 20% (year 2), 25% (year 3), and 30% (year 4). During the first 4 years after start-up, the personnel costs are assumed to be 40% (year 1 after start-up), 65% (year 2), 90% (year 3), and 100% (year 4) of the costs at full capacity, whereas patient numbers have been assumed to increase steadily from 25% (year 1 after start-up) over 50% (year 2) and 75% (year 3) to 100% (year 4) of the total number of patients at full capacity. and various delays in commissioning and ramp-up. Moreover, for the CC with private financing every percentage increase in long-term interests adds V1757 to the required reimbursement. Using a 15-year instead of a 20-year amortization period (ie, obtaining a positive net CCF 11 years instead of 16 years after start-up), the required reimbursement rate increases by V7950 for the CC (þ15.5%), V4800 for the COC (þ14.8%), and V6550 for the POC (þ12.8%) in case of private financing. For ABC, the yearly operational costs correspond exactly to the yearly operational costs calculated with the BM for centers in full operation, not taking into account inflation. For private financing versus public sponsoring, the yearly operational costs amount to V24.8M (M Z million) versus V12.8M, respectively, for the CC, V22.4M versus V12.2M for the COC, and V16.5M versus V10.0M for the POC. The required reimbursement rates calculated with the BM are slightly higher than the average cost per patient with ABC because ABC does not take into account inflation. The costs per treatment and per fraction calculated with ABC are shown in Table 3. Halving the yearly number of patients (CC: 268; COC: 380; POC: 478) nearly doubles these costs. Table 4 summarizes the results of the SA for the publicly sponsored CC.
Discussion Activity-based costing was benchmarked against the BM by comparison of the yearly operational costs. Both calculation methods are valid alternatives, each providing a
different but complementary perspective on the cost of hadron therapy. For both models a COC provides the lowest cost per treatment, despite the high investments costs, owing to a combination of treatment time and productivity. The treatment costs calculated with ABC (Table 3) demonstrate how the lower number of fractions for carbon ion treatments compensates for the higher investment costs. Indeed, because of the higher annual patient throughput related to lower fraction numbers, carbon ion treatments are less expensive than proton treatments in adults. Similarly, pediatric treatments with treatment time slots of 45 minutes translate into a higher cost per fraction and per treatment. In the literature, even longer treatment times have been reported for pediatric patients (22). It is well recognized that the overall machine time per treatment (ie, the daily treatment time multiplied by the number of fractions) is the most important determinant of radiation therapy treatment cost (3). Analogous to this, Perrier et al (11) investigated the impact of the duration of different steps within a single fraction (eg, time of immobilization, beam-up time) and found similar relationships between time and cost. In line with the literature (13), it is shown that hadron treatments delivered in a CC are more expensive than in centers dedicated to protons or carbon ions, especially in case of private financing. This is owing to economies of scale, with lower incremental investment costs required for a second treatment room similar to the first. This impact is less pronounced for public sponsoring because the investment costs are neglected. The costs per fraction presented in Table 3 seem to be relatively high compared with literature data, especially for carbon ion treatments. For CC, an average cost of V1128
Volume 89 Number 1 2014
157
The real cost of hadron therapy
Fig. 2. Cumulated cash flow (top) and net cumulated cash flow (bottom) for each technical solution and both financing methods in millions of euros as a function of time (in years). Black lines: combined center; dark grey lines: carbon only center; light grey lines: proton only center. Solid lines: private financing; dotted lines: public sponsoring. per fraction, or V20,304 per patient, has been reported (13). The observed differences can be explained by the different assumptions used: shorter treatment time slots (18 minutes) and/or higher utilization rates of the treatment facility (14 working hours per day), resulting in annual patient numbers of approximately 600 patients per treatment room. These numbers seem overly optimistic, from an operational point of view as well as from the context of patient referral. Table 2
For COC, fraction costs of V1028 and V1130, and corresponding treatment costs of V20,560 and V22,590, have been calculated (11, 23). Sensitivity analyses were performed to analyze the impact of population mix and fractionation in a publicly sponsored CC (Table 4). The impact of the change in population mix is negligible, because the potentially higher number of pediatric patients in Belgium is limited in
Required reimbursement rates calculated with the BM for the different technical solutions and both financing methods Required reimbursement rate (V/patient)
Type of technical solution and financing CC Private financing Public sponsoring COC Private financing Public sponsoring POC Private financing Public sponsoring
SA total investment cost
SA total personnel cost
SA delay in commissioning and ramp-up
Baseline
25%
þ25%
þ50%
þ75%
þ100%
30%
1y
2y
3y
51,150 27,550
42,800 24,750
59,500 30,350
67,850 33,150
76,200 35,950
84,550 38,750
47,791 24,191
55,650 29,950
60,900 32,750
67,200 36,000
32,400 18,400
27,400 16,750
37,400 20,050
42,400 21,700
47,400 23,350
52,400 25,000
30,040 16,040
35,250 19,950
38,600 21,800
42,500 23,950
51,200 32,300
44,400 30,000
58,000 34,600
64,800 36,900
71,600 39,200
78,400 41,500
46,384 22,484
55,750 35,150
61,100 38,450
67,300 42,350
Abbreviations: BM Z business model; SA Z sensitivity analysis. Other abbreviations as in Table 1.
158
International Journal of Radiation Oncology Biology Physics
Vanderstraeten et al.
Table 3 Average costs per patient, per treatment, and per fraction (in V) calculated with ABC for the different technical solutions and financing methods CC Cost Average cost per patient Average cost per fraction Cost per treatment Pediatric proton Adult proton Adult carbon ion Cost per fraction Pediatric proton Adult proton Adult carbon ion
COC
POC
Private financing
Public sponsoring
Private financing
Public sponsoring
Private financing
Public sponsoring
46,443 2365
23,956 1220
29,450 1963
16,059 1071
46,342 1545
28,296 943
69,701 48,185 42,749
35,579 26,791 21,507
29,450
16,059
61,591 43,842 -
34,878 27,217 -
2323 1606 2850
1186 892 1434
1963
1071
2053 1461 -
1163 907 -
Abbreviation: ABC Z activity-based costing. Other abbreviations as in Table 1.
respect to the total numbers. The impact of fractionation on the treatment cost is highlighted in the SA on the number of fractions per carbon ion treatment. Because the cost of treatment preparation is independent of the number of fractions, its relative impact on the cost per fraction increases nonlinearly in more hypofractionated treatment schedules. This is in contrast to calculations performed with non-ABC cost-accounting models, which mostly rely on the poor approximation that the treatment cost scales linearly with the number of fractions (13). However, the gradual increase in cost per fraction does not preclude the positive effect of lower fraction numbers on the total treatment cost. This supports the advantages of hypofractionation, in terms of patient convenience as well as from a financial perspective. Similar conclusions have been drawn in other reports (11, 13). Resources should be optimally used, especially when their costs are high. For the BM (Table 2) the required reimbursement is highest for POC and lowest for COC, in line with the respective annual patient throughput. That this is related to operational as well as investment costs is shown by the fact that similar conclusions can be drawn for public and private centers. Obviously, these conclusions only hold true in the situation of optimal patient referral and full occupancy. In our BM analysis, the reimbursement rates required to make a center sustainable are almost twice as high for private financing as for public sponsoring (Table 2). It is clear that the commonly accepted charges in European and Japanese centers are insufficient to cover the full costs, including capital investment. In addition, the budgetary implications and risks of setting up a center in a commercial environment cannot be overestimated. In line with the highest capital investments, the cumulated cash drain at 3 years is most pronounced in CC in a private setting. Because of the omission of investment costs, this drain is by definition smaller in case of public sponsoring. The risks are demonstrated by the very high sensitivity of the required reimbursement to delays in
commissioning and ramp-up. For the privately financed CC, losing a full year of productivity roughly corresponds to V27M cash flow loss, which must be compensated by short-term bank loans and results in an increase in reimbursement of V4500. For private financing, investment costs evidently have an important influence on the treatment cost. Sensitivity analyses studying the effect of a 25% decrease up to a 100% increase of investment costs were performed (Table 2) to illustrate the large variations in calculated reimbursement rate due to the differences and uncertainties surrounding investment cost, which are a reality. Another uncertainty relates to the personnel requirements (Table 1), based on the Belgian Health Care Knowledge Centre report (9) and adjusted to the number and type of treatment machines in each center. In reality, however, personnel requirements also depend on patient numbers. Moreover, wage costs and the type of personnel involved in different activities in the radiation therapy process vary considerably among countries. Because the international expert team considered our estimates to be at the high end, SA were performed studying the effect of a 30% decrease in personnel costs. This has more impact on public sponsoring than private financing, because of the relatively larger share of personnel costs in the total treatment cost. Similarly, the effect is also highest for the POC. Both SA on investment and personnel costs did not alter the conclusions from the baseline calculations. Because equipment and personnel costs together constitute approximately 80% of the total cost of radiation therapy, it was less important to analyze potential variations in other operational expenses. Finally, as mentioned throughout the text, all calculations assume a center operating at full capacity after the ramp-up period. It is clear that costs will be higher if this cannot be ascertained in reality, for example by means of mandatory referral of certain indications by existing photon radiation therapy facilities, with common
Volume 89 Number 1 2014
The real cost of hadron therapy
159
Table 4 Sensitivity analyses on average cost per patient, per treatment, and per fraction (in V) calculated with ABC for the publicly sponsored combined center Parameter No. of patients per year Pediatric proton (30 fx) Adult proton (30 fx) Adult carbon ion (15 fx) Adult carbon ion (20 fx) Adult carbon ion (16 fx) Adult carbon ion (12 fx) Adult carbon ion (6 fx) Adult carbon ion (4 fx) Total Cost per treatment (V) Pediatric proton (30 fx) Adult proton (30 fx) Adult carbon ion (15 fx) Adult carbon ion (20 fx) Adult carbon ion (16 fx) Adult carbon ion (12 fx) Adult carbon ion (6 fx) Adult carbon ion (4 fx) Average cost per patient (V) Cost per fraction (V) Pediatric proton (30 fx) Adult proton (30 fx) Adult carbon ion (15 fx) Adult carbon ion (20 fx) Adult carbon ion (16 fx) Adult carbon ion (12 fx) Adult carbon ion (6 fx) Adult carbon ion (4 fx) Average cost per fraction (V) Total cost per year (V)
Baseline
Pediatric: 60 min per fraction
Pediatric: 100 patients
Carbon ion fractionation mix
50 115 369 534
50 90 369 509
100 40 369 509
50 115 20 100 100 30 450 865
35,579 26,791 21,507 23,956
42,373 27,683 22,113 25,088
35,496 27,627 22,073 25,088
33,036 24,304 23,733 20,066 16,399 11,026 9320 15,136
1186 892 1434 1220 12.8M
1412 923 1474 1312 12.8M
1173 921 1472 1312 12.8M
1101 810 1187 1254 1367 1838 2330 1292 13.1M
Abbreviation: fx Z fractions. Other abbreviations as in Tables 1 and 3. The sensitivity analyses include: (1) 60-minute instead of 45-minute treatment time slots for pediatric patients (at the expense of the number of adult patients that can be treated on the proton facility); (2) 100 instead of 50 pediatric patients (at the expense of the number of adult patients that can be treated on the proton facility); and (3) mixed number of fractions per treatment for the carbon ion patient population, based on the 2008 Cancer Registry data on cancer incidence and delivered radiation therapy treatments (21).
agreements regarding which center will be responsible for the follow-up of the patients and whether or not a fee for referral can be included. In conclusion, the BM and ABC are equally valid models to calculate the cost of hadron therapy. Each method can provide important data from a different perspective. The choice of financing model has a very large impact on the feasibility of the project. Taking into account the commonly charged treatment costs in existing European centers, a private approach does not seem sustainable. A POC would be today’s technically most safe solution. However, the number of pediatric patients being limited, investment in a POC may not be justified. A large proportion of adult patients eligible for hadron therapy may indeed experience more clinical benefits from the higher biological effect obtained with carbon ion treatments. For
this reason, although a CC may keep open all perspectives, a COC may be a promising alternative. Another argument in favor of the COC is that further shortening of the fractionation schedules, as investigated in ongoing trials, will result in a higher throughput, hence in even more competitive treatment costs. The decision to invest in a hadron therapy center in Belgium will obviously not be based solely on financial and budgetary grounds, but will and should also depend on clinical needs, including a vision regarding intercenter collaboration and patient referral.
References 1. Organisation for Economic Co-Operation and Development. Key table: Total expenditure on health as a percentage of gross domestic
160
2. 3.
4. 5. 6.
7.
8.
9.
10.
11.
12.
Vanderstraeten et al.
product. http://www.oecd-ilibrary.org/social-issues-migration-health/ total-expenditure-on-health_20758480-table1. Accessed June 27, 2013. Bodenheimer T. High and rising health care costs. Part 2: Technologic innovation. Ann Intern Med 2005;142:932-937. Lievens Y, van den Bogaert W, Kesteloot K. Activity-based costing: A practical model for cost calculation in radiotherapy. Int J Radiat Oncol Biol Phys 2003;57:522-535. Sullivan R, Peppercorn J, Sikora K, et al. Delivering affordable cancer care in high-income countries. Lancet Oncol 2011;12:933-980. Norlund A. Costs of radiotherapy. Acta Oncol 2003;42:411-415. Delaney GP, Jacob S, Featherstone C, et al. The role of radiotherapy in cancer care: Estimating optimal utilization from a review of evidencebased clinical guidelines. Cancer 2005;104:1129-1137. van Loon J, Grutters J, Macbeth F. Evaluation of novel radiotherapy technologies: What evidence is needed to assess their clinical and cost effectiveness, and how should we get it? Lancet Oncol 2012;13:169-177. Johnstone PAS, Kerstiens J, Helsper R. Proton facility economics: The importance of “simple” treatments. J Am Coll Radiol 2012;9: 560-563. Huybrechts M, Obyn C, Gailly J, et al. Hadrontherapy. Health Technology Assessment. KCE Report 67A (D/2007/10.273/50). Belgian Health Care Knowledge Centre (KCE); 2007. Van de Werf E, Verstraete J, Lievens Y. The cost of radiotherapy in a decade of technology evolution. Radiother Oncol 2012;102: 148-153. Perrier L, Combs SE, Auberger T, et al. A decision-making tool for a costly innovative technology. J d’Economie Me´dicale 2007;25:367380. Grutters J, Pijls-Johannesma M, Ruysscher D, et al. The costeffectiveness of particle therapy in non-small cell lung cancer: Exploring decision uncertainty and areas for future research. Cancer Treat Rev 2010;36:468-476.
International Journal of Radiation Oncology Biology Physics 13. Peeters A, Grutters JPC, Pijls-Johannesma M, et al. How costly is particle therapy? Cost analysis of external beam radiotherapy with carbon-ions, protons and photons. Radiother Oncol 2010;95: 45-53. 14. Nakagawa Y, Yoshihara H, Kageji T, et al. Cost analysis of radiotherapy, carbon ion therapy, proton therapy and BNCT in Japan. Appl Radiat Isot 2009;67:S80-S83. 15. Goitein M, Jermann M. The relative costs of proton and X-ray radiation therapy. Clin Oncol 2003;15:S37-S50. 16. Peckham A. Canadian Radiation Oncologists Manpower Study Report. RA9812. Canadian Association of Radiation Oncologists (Markham, Ontario, Canada) November 22, 1999. 17. Thomas SJ, Vinall A, Poynter A, et al. A multicentre timing study of intensity-modulated radiotherapy planning and delivery. Clin Oncol 2010;22:658-665. 18. The ABT Study of Medical Physicist Work Values for Radiation Oncology Physics Services: Round III. Final Report. Abt Associates, Inc. (Cambridge, Massachusetts, USA) March 2008. 19. Weber DC, Poortmans PMP, Hurkmans CW, et al. Quality assurance for prospective EORTC radiation oncology trials: The challenges of advanced technology in a multicenter international setting. Radiother Oncol 2011;100:150-156. 20. Johnstone PAS, McMullen KP, Buchsbaum JC, et al. Pediatric CSI: Are protons the only ethical approach? Int J Radiat Oncol Biol Phys 2013;87:228-230. 21. Belgian Cancer Registry. Tables 2008. http://www.kankerregister.org. 22. Singhal M, Vincent A, Simoneaux V, et al. Overcoming the learning curve in supine pediatric proton craniospinal irradiation. J Am Coll Radiol 2012;9:285-287. 23. Ja¨kel O, Land B, Combs S, et al. On the cost-effectiveness of carbon ion radiation therapy for skull base chordoma. Radiother Oncol 2007; 83:133-138.
Radiation Oncology biology
physics
www.redjournal.org
Health Care Economics
In Search of the Economic Sustainability of Hadron Therapy: The Real Cost of Setting Up and Operating a Hadron Facility Barbara Vanderstraeten, PhD,* Jan Verstraete, RN,y Roger De Croock, MScEng,z Wilfried De Neve, PhD, and Yolande Lievens, PhD* *Department of Radiotherapy, Ghent University Hospital, Gent, Belgium; yDepartment of Radiation Oncology, University Hospital Gasthuisberg, Leuven, Belgium; and zBelgian Hadron Therapy Center Foundation, Brussels, Belgium Received Sep 11, 2013, and in revised form Jan 6, 2014. Accepted for publication Jan 23, 2014.
Summary The cost of setting up and running different types of hadron therapy centers (proton only/carbon ion only/ combined) is calculated with different financing methods (public/private). We compare a business model with an activity-based costing method for calculation of the required reimbursement and the treatment costs per patient and per fraction. The complementarity of both methods is shown, as well as the impact of a delay in commissioning, a higher interest rate, higher patient throughput, and hypofractionation.
Purpose: To determine the treatment cost and required reimbursement for a new hadron therapy facility, considering different technical solutions and financing methods. Methods and Materials: The 3 technical solutions analyzed are a carbon only (COC), proton only (POC), and combined (CC) center, each operating 2 treatment rooms and assumed to function at full capacity. A business model defines the required reimbursement and analyzes the financial implications of setting up a facility over time; activitybased costing (ABC) calculates the treatment costs per type of patient for a center in a steady state of operation. Both models compare a private, full-cost approach with public sponsoring, only taking into account operational costs. Results: Yearly operational costs range between V10.0M (M Z million) for a publicly sponsored POC to V24.8M for a CC with private financing. Disregarding inflation, the average treatment cost calculated with ABC (COC: V29,450; POC: V46,342; CC: V46,443 for private financing; respectively V16,059, V28,296, and V23,956 for public sponsoring) is slightly lower than the required reimbursement based on the business model (between V51,200 in a privately funded POC and V18,400 in COC with public sponsoring). Reimbursement for privately financed centers is very sensitive to a delay in commissioning and to the interest rate. Higher throughput and hypofractionation have a positive impact on the treatment costs. Conclusions: Both calculation methods are valid and complementary. The financially most attractive option of a publicly sponsored COC should be balanced to the clinical necessities and the sociopolitical context. Ó 2014 Elsevier Inc.
Reprint requests to: Barbara Vanderstraeten, PhD, Ghent University Hospital, Department of Radiotherapy, De Pintelaan 185 0P7, 9000 Ghent, Belgium. Tel: (þ32) 9-332-19-41; E-mail: barbara.vanderstraeten@ uzgent.be This work was financially supported by the Cancer Plan Action 30 financed by the Federal Office of Health and Social Affairs in Belgium. Int J Radiation Oncol Biol Phys, Vol. 89, No. 1, pp. 152e160, 2014 0360-3016/$ - see front matter Ó 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ijrobp.2014.01.039
The foreign expert group consisted of Alejandro Mazal and Ste´phanie Bolle (Paris, France), Tadashi Kamada (Chiba, Japan), Ju¨rgen Debus and Oliver Ja¨kel (Heidelberg, Germany), Tony Lomax (Villigen, Switzerland), Philippe Lambin (Maastricht, The Netherlands), and Piero Fossati (Pavia, Italy). Conflict of interest: none.
Volume 89 Number 1 2014
Introduction Over the past decades there has been a growing concern about high and rising medical costs (1). The rapid diffusion of new technologies has been proposed as a major contributing factor in this rise (2). Radiation oncology, a highly technological and resource-demanding discipline, is obviously not immune to this general tendency (3, 4). Therefore, even if radiation therapy’s share of the total health care and oncology budget seems modest (5), particularly when considering the large proportion of cancer patients being treated with radiation (6), the radiation therapy community cannot escape from its responsibility to evaluate the financial implications of its clinical actions. Hadron therapy is an example of such a technological evolution in radiation therapy for which the financing and the value for money are heavily debated. This study was sponsored by the government of a European country where the decision to construct a hadron therapy center was pending. Before a decision, the government initiated a feasibility study to gather information regarding the scope, clinical indications, limited technical specifications, and treatment cost in view of reimbursement setting. Lack of reliable investment cost estimates makes hadron therapy projects cumbersome: “list prices” for hadron therapy equipment seem nonexistent, and quotes of vendors depend on many variables, including the country, location, scope (clinical, research, demonstration) and knowhow of the project team (conditions of collaboration for the development of new technologies). Hence, accurate cost estimates require the development of a comprehensive project plan. If the government responds positively to the feasibility study, this is the next step in the governmental process for a project of this scale. A comprehensive project plan requires knowledge of implantation site and type of center, as well as a sufficient budget (5%-7% of the investment cost) to make a business plan including detailed technical descriptions of building and equipment, requiring separate risk capital unless the decision to construct a hadron therapy center is certain. In the end, European legislation imposes the use of a European tender to organize a call for bids, finally leading to vendor selection and contract signing. For the present feasibility study, estimates of investment cost were needed for cost modeling at a stage when resources to develop a comprehensive project plan were not yet available. This seems to be a frequently encountered problem in many countries, as well as for individual project groups who need to demonstrate to potential sponsors of hadron treatments, including health insurance providers, that favorable cost-benefit ratios exist. Early-stage cost modeling has to deal with uncertainties, mainly in investment cost but also in maintenance, upgrade, and operational cost. Two cost models were evaluated. Activity-based costing (ABC), an advanced costaccounting method, has been proposed to assess the financial impact of gradual process and technology changes
The real cost of hadron therapy
153
in radiation therapy and of factors that affect patient numbers and population mixes (3, 7). Developed to compute the cost in an operational department, ABC is, however, less suited to capture the implications of setting up a new facility with carry-over investment and development costs over decades. In practice the long-term financial evaluation of setting up a new department is performed using a business model (BM). This article describes both approaches for 3 types of centers built to treat with protons only, with carbon ions only, or with protons and carbon ions. The investment cost of many centers worldwide has been financed by public resources, often being taxpayers’ money in Europe or donations of a more spontaneous kind in the United States. Less fortunate project groups may have to find private resources, typically bank loans at market conditions, to finance the investment cost. American centers are typically set up on a commercial basis and have to take the full range of capital and risks into account for their reimbursement setting (8). It is therefore not surprising that these centers are known to charge $100,000 or more per patient, whereas reimbursements of approximately V20,000 to V40,000 per patient are commonly charged for hadron treatments in European and Japanese centers. The majority of hadron projects in surrounding countries, to which a potential center in Belgium would benchmark, were started as scientific projects both in the field of clinical and technological development. Calculations of cost per treatment were made for both scenarios, public sponsoring and private financing. Amortization of the investment cost was discarded in the public sponsoring scenario. The specificity and complementarity of both cost models are evaluated: the gross financial implications of starting up a hadron facility over time and ensuing required reimbursement levels are evaluated in a BM; the treatment costs for a center in a specific year of operation are fine-tuned with ABC. Given the uncertainties in some input parameters, the presented reimbursement rates and treatment costs should be interpreted in the context of these assumptions.
Methods and Materials Type of center Three different technical solutions are considered: a carbon only center (COC), a proton only center (POC), and a combined center (CC), each operating 2 treatment rooms. The CC has 1 proton and 1 carbon ion treatment room. Proton treatments are delivered with a gantry, whereas carbon ion treatments use a fixed beam.
Input parameters Table 1 summarizes all investment costs, personnel requirements, and other operational costs. A time slot of
154
International Journal of Radiation Oncology Biology Physics
Vanderstraeten et al.
Table 1
Input parameters for the different technical solutions Center type Parameter
Cost/FTE/year (V)*
CC
COC
POC
20M 10M 40M 20M 4M
25M 50M 4M
15M 25M 4M
2M 1.5M
2M 1.5M
2M 1.5M
1.5M 2.5M 101.5M
2.5M 85M
1.5M 2.5M 51.5M
1 1 1 5 2 6 3 18 4 6 6 4 7
1 1 1 5 2 6 3 18 4 6 6 4 7
1 1 1 5 2 6 3 18 4 6 6 4 5
64 6.0M
64 6.0M
62 5.7M
0 0 760
50 305 0
y
Investment costs (V) Type of investment Carbon ion building Proton building Carbon ion beam equipment Proton beam equipment Imaging equipment in each room Simulation equipment Planning equipment þ record and verify Anesthesia facilities Other facilities Total investment Operational model Daily working hours (center) Working days/wk (center) Working weeks/y (center) Treatment hours/y (center) No. of working hours/wk (staff) Productivity (staff) (%) No. of working weeks/y (staff) Productive hours/FTE/y (staff) Personnel requirements General manager 180,000 Chief physicist 140,000 Chief medical doctor 180,000 Administration 55,000 Data manager 46,000 Radiation oncologists 180,000 Radiographers 56,000 Therapists 56,000 Trajectory nurses 50,000 Dosimetrists 93,000 Physicists 93,000 IT engineers 140,000 Maintenance engineers and 140,000 technicians Total no. of FTE Total cost/y (V) Exploitation costsx Maintenance cost/y (% of investment cost) Buildings Hadron beam equipment Planning equipment þ record and verify Imaging and simulation equipment, anesthesia, and other facilities Insurance cost/y (V) Energy cost/y (V) Facility maintenance cost/y (V) Medical consumable cost/patient (V) Population mixes Type of treatment Pediatric (30 fractions) Adult proton (30 fractions) Adult carbon ion (15 fractions)
07:30-20:00 5 48z 3000 38 88 45 1505
1 5 10 7 20,000 677,754 382,667 750
50 115 369
(continued on next page)
Volume 89 Number 1 2014
The real cost of hadron therapy
155
Table 1 (continued ) Center type Parameter
Cost/FTE/year (V)*
Total no. of patients/y Total no. of fractions/y
CC
COC
POC
534 10,485
760 11,400
355 10,650
Abbreviations: CC Z combined center; COC Z carbon only center; FTE Z full-time equivalent; IT Z information technology; M Z million; POC Z proton only center. * All personnel costs are assumed to be total costs, including any potential extras. See also footnote “x.” y All figures are estimates based on exploratory contacts with vendors and discussions with international experts. All investment costs are amortized over 20 years, except for IT equipment (planning systems, record and verify), which is replaced every 5 years. No costs were taken into account for the lands or for future decommissioning. z A yearly 4-week break is foreseen for maintenance and quality assurance. x All operational costs are subject to a yearly inflation of 2%, except for energy costs (2.5%). All figures are based on KCE report vol. 67A (9), but adapted to current standards based on discussions with the team of international experts and literature review (10-19).
30 minutes per fraction is allotted to adult treatments, compared with 45 minutes for pediatric treatments, assuming the latter require anesthesia (20). For ABC, a product or treatment is defined as the combination of equipment type (proton or carbon ion) and patient type (pediatric or adult). The number of patients that can be treated yearly in each center is also presented in Table 1.
Business model For the BM, treatment costs are modeled in spreadsheets using Excel (Microsoft, Redmond, WA). To analyze the financial implications of setting up a facility over time, different phases are taken into account (Fig. 1). During the 4-year preparation and commissioning phase, long-term loans with interests of 5% are allotted on the investment and personnel costs. The BM is used to determine an average reimbursement per patient for each type of center so as to generate a positive net cumulated cash flow (CCF) 16 years after start-up. This corresponds to the repayment of the first investment (in case of private financing) after 20 years, which is equal to the amortization period. To balance the negative cash outflow during the years before full operation, additional short-term loans are introduced. The reimbursement rate itself is subject to a yearly inflation rate of 2%. Sensitivity analyses (SA) are performed to investigate the effect of delays in the commissioning phase, increases in long-term interest rates and various investment or personnel costs, and a shortening of the amortization period.
Activity-based costing Activity-based costing calculates the cost per product or treatment in a center operating at full capacity. The treatment process is split into a series of activities, each of which is costed, typically on the basis of resource
consumption over a certain amount of time (3, 11). The model used for the present calculations is comparable to that described earlier for photon radiation therapy (10). All treatment-related personnel, building, and equipment costs are allocated to the products on the basis of their activity consumption. Medical consumables are directly traced to the products. Overhead cost is allocated to each product according to the number of fractions. Apart from the cost calculation for each type of treatment, ABC is used to investigate the impact of different population mixes and fractionation schedules and patient loads.
Financing methods Private financing uses a full-costing approach that assumes that all costs are covered through private financing. For the BM, this includes the investment costs, personnel costs, and interim interest on the investment capital during the commissioning phase and any cash drain after start-up until a positive cumulated net cash flow is generated after 16 years. For ABC, equivalent annual investment costs are calculated for each type of building and equipment. For public sponsoring, all investment costs, and for the BM also the personnel costs during commissioning, are taken out of the project. Hence the calculations are based purely on the operational costs.
Results Figure 2 shows the CCF calculated with the BM. The nadir takes place at 3 years after start-up because not all operational costs and capital amortization are covered by the acquired reimbursements yet. This is compensated by short-term loans to obtain the net CCF. The required reimbursement rates to obtain a positive net CCF 16 years after start-up are shown in Table 2 at baseline and for variable investment and personnel costs,
156
Vanderstraeten et al.
International Journal of Radiation Oncology Biology Physics
Fig. 1. Assumed spread of the investment costs and gradual increase of the yearly personnel costs and number of patients during the different phases of the business model. The full building costs are financed in the first year. Proton and carbon equipment investment expenditures are spread over the second (33%) and third (67%) year, whereas imaging and simulation equipment costs are spread over the third (33%) and fourth (67%) year. All other investments are made during the fourth year. During this period a fraction of the personnel is already required for preparation and commissioning. These personnel costs are modeled as a fraction of the total personnel cost of the center operating at full capacity, namely 10% (year 1), 20% (year 2), 25% (year 3), and 30% (year 4). During the first 4 years after start-up, the personnel costs are assumed to be 40% (year 1 after start-up), 65% (year 2), 90% (year 3), and 100% (year 4) of the costs at full capacity, whereas patient numbers have been assumed to increase steadily from 25% (year 1 after start-up) over 50% (year 2) and 75% (year 3) to 100% (year 4) of the total number of patients at full capacity. and various delays in commissioning and ramp-up. Moreover, for the CC with private financing every percentage increase in long-term interests adds V1757 to the required reimbursement. Using a 15-year instead of a 20-year amortization period (ie, obtaining a positive net CCF 11 years instead of 16 years after start-up), the required reimbursement rate increases by V7950 for the CC (þ15.5%), V4800 for the COC (þ14.8%), and V6550 for the POC (þ12.8%) in case of private financing. For ABC, the yearly operational costs correspond exactly to the yearly operational costs calculated with the BM for centers in full operation, not taking into account inflation. For private financing versus public sponsoring, the yearly operational costs amount to V24.8M (M Z million) versus V12.8M, respectively, for the CC, V22.4M versus V12.2M for the COC, and V16.5M versus V10.0M for the POC. The required reimbursement rates calculated with the BM are slightly higher than the average cost per patient with ABC because ABC does not take into account inflation. The costs per treatment and per fraction calculated with ABC are shown in Table 3. Halving the yearly number of patients (CC: 268; COC: 380; POC: 478) nearly doubles these costs. Table 4 summarizes the results of the SA for the publicly sponsored CC.
Discussion Activity-based costing was benchmarked against the BM by comparison of the yearly operational costs. Both calculation methods are valid alternatives, each providing a
different but complementary perspective on the cost of hadron therapy. For both models a COC provides the lowest cost per treatment, despite the high investments costs, owing to a combination of treatment time and productivity. The treatment costs calculated with ABC (Table 3) demonstrate how the lower number of fractions for carbon ion treatments compensates for the higher investment costs. Indeed, because of the higher annual patient throughput related to lower fraction numbers, carbon ion treatments are less expensive than proton treatments in adults. Similarly, pediatric treatments with treatment time slots of 45 minutes translate into a higher cost per fraction and per treatment. In the literature, even longer treatment times have been reported for pediatric patients (22). It is well recognized that the overall machine time per treatment (ie, the daily treatment time multiplied by the number of fractions) is the most important determinant of radiation therapy treatment cost (3). Analogous to this, Perrier et al (11) investigated the impact of the duration of different steps within a single fraction (eg, time of immobilization, beam-up time) and found similar relationships between time and cost. In line with the literature (13), it is shown that hadron treatments delivered in a CC are more expensive than in centers dedicated to protons or carbon ions, especially in case of private financing. This is owing to economies of scale, with lower incremental investment costs required for a second treatment room similar to the first. This impact is less pronounced for public sponsoring because the investment costs are neglected. The costs per fraction presented in Table 3 seem to be relatively high compared with literature data, especially for carbon ion treatments. For CC, an average cost of V1128
Volume 89 Number 1 2014
157
The real cost of hadron therapy
Fig. 2. Cumulated cash flow (top) and net cumulated cash flow (bottom) for each technical solution and both financing methods in millions of euros as a function of time (in years). Black lines: combined center; dark grey lines: carbon only center; light grey lines: proton only center. Solid lines: private financing; dotted lines: public sponsoring. per fraction, or V20,304 per patient, has been reported (13). The observed differences can be explained by the different assumptions used: shorter treatment time slots (18 minutes) and/or higher utilization rates of the treatment facility (14 working hours per day), resulting in annual patient numbers of approximately 600 patients per treatment room. These numbers seem overly optimistic, from an operational point of view as well as from the context of patient referral. Table 2
For COC, fraction costs of V1028 and V1130, and corresponding treatment costs of V20,560 and V22,590, have been calculated (11, 23). Sensitivity analyses were performed to analyze the impact of population mix and fractionation in a publicly sponsored CC (Table 4). The impact of the change in population mix is negligible, because the potentially higher number of pediatric patients in Belgium is limited in
Required reimbursement rates calculated with the BM for the different technical solutions and both financing methods Required reimbursement rate (V/patient)
Type of technical solution and financing CC Private financing Public sponsoring COC Private financing Public sponsoring POC Private financing Public sponsoring
SA total investment cost
SA total personnel cost
SA delay in commissioning and ramp-up
Baseline
25%
þ25%
þ50%
þ75%
þ100%
30%
1y
2y
3y
51,150 27,550
42,800 24,750
59,500 30,350
67,850 33,150
76,200 35,950
84,550 38,750
47,791 24,191
55,650 29,950
60,900 32,750
67,200 36,000
32,400 18,400
27,400 16,750
37,400 20,050
42,400 21,700
47,400 23,350
52,400 25,000
30,040 16,040
35,250 19,950
38,600 21,800
42,500 23,950
51,200 32,300
44,400 30,000
58,000 34,600
64,800 36,900
71,600 39,200
78,400 41,500
46,384 22,484
55,750 35,150
61,100 38,450
67,300 42,350
Abbreviations: BM Z business model; SA Z sensitivity analysis. Other abbreviations as in Table 1.
158
International Journal of Radiation Oncology Biology Physics
Vanderstraeten et al.
Table 3 Average costs per patient, per treatment, and per fraction (in V) calculated with ABC for the different technical solutions and financing methods CC Cost Average cost per patient Average cost per fraction Cost per treatment Pediatric proton Adult proton Adult carbon ion Cost per fraction Pediatric proton Adult proton Adult carbon ion
COC
POC
Private financing
Public sponsoring
Private financing
Public sponsoring
Private financing
Public sponsoring
46,443 2365
23,956 1220
29,450 1963
16,059 1071
46,342 1545
28,296 943
69,701 48,185 42,749
35,579 26,791 21,507
29,450
16,059
61,591 43,842 -
34,878 27,217 -
2323 1606 2850
1186 892 1434
1963
1071
2053 1461 -
1163 907 -
Abbreviation: ABC Z activity-based costing. Other abbreviations as in Table 1.
respect to the total numbers. The impact of fractionation on the treatment cost is highlighted in the SA on the number of fractions per carbon ion treatment. Because the cost of treatment preparation is independent of the number of fractions, its relative impact on the cost per fraction increases nonlinearly in more hypofractionated treatment schedules. This is in contrast to calculations performed with non-ABC cost-accounting models, which mostly rely on the poor approximation that the treatment cost scales linearly with the number of fractions (13). However, the gradual increase in cost per fraction does not preclude the positive effect of lower fraction numbers on the total treatment cost. This supports the advantages of hypofractionation, in terms of patient convenience as well as from a financial perspective. Similar conclusions have been drawn in other reports (11, 13). Resources should be optimally used, especially when their costs are high. For the BM (Table 2) the required reimbursement is highest for POC and lowest for COC, in line with the respective annual patient throughput. That this is related to operational as well as investment costs is shown by the fact that similar conclusions can be drawn for public and private centers. Obviously, these conclusions only hold true in the situation of optimal patient referral and full occupancy. In our BM analysis, the reimbursement rates required to make a center sustainable are almost twice as high for private financing as for public sponsoring (Table 2). It is clear that the commonly accepted charges in European and Japanese centers are insufficient to cover the full costs, including capital investment. In addition, the budgetary implications and risks of setting up a center in a commercial environment cannot be overestimated. In line with the highest capital investments, the cumulated cash drain at 3 years is most pronounced in CC in a private setting. Because of the omission of investment costs, this drain is by definition smaller in case of public sponsoring. The risks are demonstrated by the very high sensitivity of the required reimbursement to delays in
commissioning and ramp-up. For the privately financed CC, losing a full year of productivity roughly corresponds to V27M cash flow loss, which must be compensated by short-term bank loans and results in an increase in reimbursement of V4500. For private financing, investment costs evidently have an important influence on the treatment cost. Sensitivity analyses studying the effect of a 25% decrease up to a 100% increase of investment costs were performed (Table 2) to illustrate the large variations in calculated reimbursement rate due to the differences and uncertainties surrounding investment cost, which are a reality. Another uncertainty relates to the personnel requirements (Table 1), based on the Belgian Health Care Knowledge Centre report (9) and adjusted to the number and type of treatment machines in each center. In reality, however, personnel requirements also depend on patient numbers. Moreover, wage costs and the type of personnel involved in different activities in the radiation therapy process vary considerably among countries. Because the international expert team considered our estimates to be at the high end, SA were performed studying the effect of a 30% decrease in personnel costs. This has more impact on public sponsoring than private financing, because of the relatively larger share of personnel costs in the total treatment cost. Similarly, the effect is also highest for the POC. Both SA on investment and personnel costs did not alter the conclusions from the baseline calculations. Because equipment and personnel costs together constitute approximately 80% of the total cost of radiation therapy, it was less important to analyze potential variations in other operational expenses. Finally, as mentioned throughout the text, all calculations assume a center operating at full capacity after the ramp-up period. It is clear that costs will be higher if this cannot be ascertained in reality, for example by means of mandatory referral of certain indications by existing photon radiation therapy facilities, with common
Volume 89 Number 1 2014
The real cost of hadron therapy
159
Table 4 Sensitivity analyses on average cost per patient, per treatment, and per fraction (in V) calculated with ABC for the publicly sponsored combined center Parameter No. of patients per year Pediatric proton (30 fx) Adult proton (30 fx) Adult carbon ion (15 fx) Adult carbon ion (20 fx) Adult carbon ion (16 fx) Adult carbon ion (12 fx) Adult carbon ion (6 fx) Adult carbon ion (4 fx) Total Cost per treatment (V) Pediatric proton (30 fx) Adult proton (30 fx) Adult carbon ion (15 fx) Adult carbon ion (20 fx) Adult carbon ion (16 fx) Adult carbon ion (12 fx) Adult carbon ion (6 fx) Adult carbon ion (4 fx) Average cost per patient (V) Cost per fraction (V) Pediatric proton (30 fx) Adult proton (30 fx) Adult carbon ion (15 fx) Adult carbon ion (20 fx) Adult carbon ion (16 fx) Adult carbon ion (12 fx) Adult carbon ion (6 fx) Adult carbon ion (4 fx) Average cost per fraction (V) Total cost per year (V)
Baseline
Pediatric: 60 min per fraction
Pediatric: 100 patients
Carbon ion fractionation mix
50 115 369 534
50 90 369 509
100 40 369 509
50 115 20 100 100 30 450 865
35,579 26,791 21,507 23,956
42,373 27,683 22,113 25,088
35,496 27,627 22,073 25,088
33,036 24,304 23,733 20,066 16,399 11,026 9320 15,136
1186 892 1434 1220 12.8M
1412 923 1474 1312 12.8M
1173 921 1472 1312 12.8M
1101 810 1187 1254 1367 1838 2330 1292 13.1M
Abbreviation: fx Z fractions. Other abbreviations as in Tables 1 and 3. The sensitivity analyses include: (1) 60-minute instead of 45-minute treatment time slots for pediatric patients (at the expense of the number of adult patients that can be treated on the proton facility); (2) 100 instead of 50 pediatric patients (at the expense of the number of adult patients that can be treated on the proton facility); and (3) mixed number of fractions per treatment for the carbon ion patient population, based on the 2008 Cancer Registry data on cancer incidence and delivered radiation therapy treatments (21).
agreements regarding which center will be responsible for the follow-up of the patients and whether or not a fee for referral can be included. In conclusion, the BM and ABC are equally valid models to calculate the cost of hadron therapy. Each method can provide important data from a different perspective. The choice of financing model has a very large impact on the feasibility of the project. Taking into account the commonly charged treatment costs in existing European centers, a private approach does not seem sustainable. A POC would be today’s technically most safe solution. However, the number of pediatric patients being limited, investment in a POC may not be justified. A large proportion of adult patients eligible for hadron therapy may indeed experience more clinical benefits from the higher biological effect obtained with carbon ion treatments. For
this reason, although a CC may keep open all perspectives, a COC may be a promising alternative. Another argument in favor of the COC is that further shortening of the fractionation schedules, as investigated in ongoing trials, will result in a higher throughput, hence in even more competitive treatment costs. The decision to invest in a hadron therapy center in Belgium will obviously not be based solely on financial and budgetary grounds, but will and should also depend on clinical needs, including a vision regarding intercenter collaboration and patient referral.
References 1. Organisation for Economic Co-Operation and Development. Key table: Total expenditure on health as a percentage of gross domestic
160
2. 3.
4. 5. 6.
7.
8.
9.
10.
11.
12.
Vanderstraeten et al.
product. http://www.oecd-ilibrary.org/social-issues-migration-health/ total-expenditure-on-health_20758480-table1. Accessed June 27, 2013. Bodenheimer T. High and rising health care costs. Part 2: Technologic innovation. Ann Intern Med 2005;142:932-937. Lievens Y, van den Bogaert W, Kesteloot K. Activity-based costing: A practical model for cost calculation in radiotherapy. Int J Radiat Oncol Biol Phys 2003;57:522-535. Sullivan R, Peppercorn J, Sikora K, et al. Delivering affordable cancer care in high-income countries. Lancet Oncol 2011;12:933-980. Norlund A. Costs of radiotherapy. Acta Oncol 2003;42:411-415. Delaney GP, Jacob S, Featherstone C, et al. The role of radiotherapy in cancer care: Estimating optimal utilization from a review of evidencebased clinical guidelines. Cancer 2005;104:1129-1137. van Loon J, Grutters J, Macbeth F. Evaluation of novel radiotherapy technologies: What evidence is needed to assess their clinical and cost effectiveness, and how should we get it? Lancet Oncol 2012;13:169-177. Johnstone PAS, Kerstiens J, Helsper R. Proton facility economics: The importance of “simple” treatments. J Am Coll Radiol 2012;9: 560-563. Huybrechts M, Obyn C, Gailly J, et al. Hadrontherapy. Health Technology Assessment. KCE Report 67A (D/2007/10.273/50). Belgian Health Care Knowledge Centre (KCE); 2007. Van de Werf E, Verstraete J, Lievens Y. The cost of radiotherapy in a decade of technology evolution. Radiother Oncol 2012;102: 148-153. Perrier L, Combs SE, Auberger T, et al. A decision-making tool for a costly innovative technology. J d’Economie Me´dicale 2007;25:367380. Grutters J, Pijls-Johannesma M, Ruysscher D, et al. The costeffectiveness of particle therapy in non-small cell lung cancer: Exploring decision uncertainty and areas for future research. Cancer Treat Rev 2010;36:468-476.
International Journal of Radiation Oncology Biology Physics 13. Peeters A, Grutters JPC, Pijls-Johannesma M, et al. How costly is particle therapy? Cost analysis of external beam radiotherapy with carbon-ions, protons and photons. Radiother Oncol 2010;95: 45-53. 14. Nakagawa Y, Yoshihara H, Kageji T, et al. Cost analysis of radiotherapy, carbon ion therapy, proton therapy and BNCT in Japan. Appl Radiat Isot 2009;67:S80-S83. 15. Goitein M, Jermann M. The relative costs of proton and X-ray radiation therapy. Clin Oncol 2003;15:S37-S50. 16. Peckham A. Canadian Radiation Oncologists Manpower Study Report. RA9812. Canadian Association of Radiation Oncologists (Markham, Ontario, Canada) November 22, 1999. 17. Thomas SJ, Vinall A, Poynter A, et al. A multicentre timing study of intensity-modulated radiotherapy planning and delivery. Clin Oncol 2010;22:658-665. 18. The ABT Study of Medical Physicist Work Values for Radiation Oncology Physics Services: Round III. Final Report. Abt Associates, Inc. (Cambridge, Massachusetts, USA) March 2008. 19. Weber DC, Poortmans PMP, Hurkmans CW, et al. Quality assurance for prospective EORTC radiation oncology trials: The challenges of advanced technology in a multicenter international setting. Radiother Oncol 2011;100:150-156. 20. Johnstone PAS, McMullen KP, Buchsbaum JC, et al. Pediatric CSI: Are protons the only ethical approach? Int J Radiat Oncol Biol Phys 2013;87:228-230. 21. Belgian Cancer Registry. Tables 2008. http://www.kankerregister.org. 22. Singhal M, Vincent A, Simoneaux V, et al. Overcoming the learning curve in supine pediatric proton craniospinal irradiation. J Am Coll Radiol 2012;9:285-287. 23. Ja¨kel O, Land B, Combs S, et al. On the cost-effectiveness of carbon ion radiation therapy for skull base chordoma. Radiother Oncol 2007; 83:133-138.
