Int. J. Radiation Oncology Biol. Phys., Vol. 57, No. 2, pp. 522–535, 2003 Copyright © 2003 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/03/$–see front matter
doi:10.1016/S0360-3016(03)00579-0
CLINICAL INVESTIGATION
Radiation Oncology Practice
ACTIVITY-BASED COSTING: A PRACTICAL MODEL FOR COST CALCULATION IN RADIOTHERAPY YOLANDE LIEVENS, PH.D.,* WALTER
VAN DEN
BOGAERT, PH.D.,*
AND
KATRIEN KESTELOOT, PH.D.†
*Department of Radiation Oncology, University Hospitals Leuven, Leuven, Belgium; †University Hospitals Leuven and Center for Health Services and Nursing Research, KU Leuven, Belgium Purpose: The activity-based costing method was used to compute radiotherapy costs. This report describes the model developed, the calculated costs, and possible applications for the Leuven radiotherapy department. Methods and Materials: Activity-based costing is an advanced cost calculation technique that allocates resource costs to products based on activity consumption. In the Leuven model, a complex allocation principle with a large diversity of cost drivers was avoided by introducing an extra allocation step between activity groups and activities. A straightforward principle of time consumption, weighed by some factors of treatment complexity, was used. The model was developed in an iterative way, progressively defining the constituting components (costs, activities, products, and cost drivers). Results: Radiotherapy costs are predominantly determined by personnel and equipment cost. Treatment-related activities consume the greatest proportion of the resource costs, with treatment delivery the most important component. This translates into products that have a prolonged total or daily treatment time being the most costly. The model was also used to illustrate the impact of changes in resource costs and in practice patterns. Conclusion: The presented activity-based costing model is a practical tool to evaluate the actual cost structure of a radiotherapy department and to evaluate possible resource or practice changes. © 2003 Elsevier Inc. Radiotherapy, Cost calculation, Activity-based costing.
INTRODUCTION Radiotherapy costs Although the importance of accurate cost data has gradually become recognized by the medical profession and more specifically by radiation oncologists, literature data on the cost of radiotherapy (RT) activities and products are scarce and often contradictory (1–3). Although part of the literature analyzed the costs of specific aspects of the RT process, most efforts have focused on the cost of RT delivery—the cost per patient visit (or fraction), extrapolated to a global RT treatment cost (2– 8). Because of differences in methods, included cost components and RT activities, and health care systems, these studies yielded considerable variation in computed costs, rendering the comparison of the results unreliable (1, 3). Moreover, most of these approaches calculated the average costs; total treatment costs were obtained by multiplying the average cost per fraction by the number of fractions typically given for each type of treatment. This method may, however, not give a good representation of the consequences of fractionation changes on resource consumption (2), because it implicitly assumes a stable relationship between the cost per fraction and the
cost of other activities within the RT process, such as planning or fixation. In practice, this is usually not the case, hence the cost of the different steps should ideally be calculated separately (3). Because they are easy to obtain, charges (i.e., patient bills), whether or not adapted by cost/charge ratios (9, 10), are frequently used as proxies for resource cost. The same goes for reimbursement data of RT. Although it is theoretically possible that charges or reimbursement for a given procedure are an accurate measure of its actual resource cost, this is rarely the case in reality because of historical and political factors, government regulations, budget constraints, and market forces (11). In view of producing more accurate resource cost estimates, health care organizations have started to invest in more sophisticated cost-accounting systems capable of producing activity- or product-specific cost data (12). Activitybased costing (ABC) is a cost-accounting system that allocates the resource costs to the products using a multistep allocation procedure on the basis of activity consumption. This article describes the development and application of such an ABC system for the RT department of the Leuven University Hospital.
Reprint requests to: Yolande Lievens, Ph.D., Department of Radiotherapy, University Hospitals Leuven, Herestraat 49, Leuven 3000 Belgium. Tel: 0032 16 347600; Fax: 0032 16 347623;
E-mail:
[email protected] Received Oct 29, 2002, and in revised form Apr 17, 2003. Accepted for publication Apr 23, 2003. 522
ABC in radiotherapy
● Y. LIEVENS, W. VAN DEN BOGAERT, and K. KESTELOOT
Fig. 1. ABC schematic.
Activity-based costing In an era of rapidly increasing production complexity, product size, and volume diversification, ABC was developed as a response to the shortcomings of traditional cost accounting methods. Whereas the latter focus on the products and directly assign resource costs (e.g., wages, equipment) to the products on the basis of the assumption that each product consumes a certain amount of these resources, ABC (Fig. 1) focuses on the activities (13–15). More specifically, the assignment of costs through ABC occurs in two stages: cost objects (i.e., services or products) consume activities, activities consume (resource) costs. In practice, this means that the indirect costs are allocated to the products on the basis of the activities consumed within the production process (e.g., medical wage costs are allocated to different activities such as simulation or follow-up) using “first-stage cost drivers.” The further allocation of the costs of activities to the products is performed in a second step using “secondstage cost drivers.” Traditional approaches use only a few (mostly volume-based) cost drivers. In ABC, a large number of diverse cost drivers may be used, reflecting the relation between the resource costs and activities and between the activities and products. The practical scheme that underlies this stepwise allocation, and thus the constituting components (i.e., costs, activities, products, and cost drivers), is, however, unique for each specific situation. The major advantage of ABC lies in a more accurate cost computation, especially in situations in which product diversity is important and in which the indirect costs, not directly traceable to the products, represent an important proportion of the total costs. Additionally, ABC also allows an in-depth product analysis by explaining the relationship between the products and
523
activities. The improved insights in the (cost) structure of products and of departments supports continuous process improvements, referred to as activity-based management (13). The potential drawback of ABC systems lies in the time and resource consumption associated with the development and management of such complex cost-accounting techniques. Because the precision of an ABC model depends critically on the level of detail of its constituting components, an appropriate choice of this level of detail is crucial. If the process analysis or the product definition is not detailed enough (aggregation), the obtained cost per product will not be considered relevant because of a lack of specificity. Conversely, if too many (sub)activities and products are defined (disaggregation), the whole calculation process becomes too complex and difficult to perform, which in turn translates into a greater workload and associated cost in the development phase, as well as during routine use, of the program. The difficulty in designing a good cost system thus lies in achieving a model that is easy and economic to maintain yet does not introduce excessive distortions (16). We describe the way in which this was approached in the Leuven radiotherapy department. METHODS AND MATERIALS Development of an ABC system for the Leuven radiotherapy department The described cost-accounting model was developed for the Leuven radiotherapy department using the product and activity data of 2000. In 2000, the department delivered 1769 external beam radiotherapy (EBRT) courses, operated with four treatment machines (linear accelerators, of which one was a multileaf collimator [MLC]) and two simulators. The personnel working in the department consisted of 11 full-time equivalent (FTE) radiation oncologists (5.5 FTE staff members, 5.5 FTE residents), 29.13 FTE paramedical staff, and 3.47 FTE administrative staff. The ABC program of the Leuven radiotherapy department was developed in an iterative way by progressively defining its components. Because the main incentive of this exercise was to gain insight into the cost structure of the Leuven radiotherapy department and to obtain accurate cost estimates of the delivered radiation treatments for additional use in management applications and for economic evaluations, a detailed approach was necessary. The different allocation steps defined within the developed model are schematically shown in Fig. 2. Specific to this model is that a distinction was made between the different levels of activities: (1) activity groups were defined as the global amount of a certain activity (e.g., simulation) and included all the specific activities incorporated into this group; and (2) specific activities represent activities linked to a specific product (e.g., simulation of a tangential breast irradiation). The definition of two activity levels translated into the
524
I. J. Radiation Oncology
● Biology ● Physics
Volume 57, Number 2, 2003
Fig. 2. ABC in Leuven radiotherapy department: allocation steps.
vast majority of the costs (wage, equipment, and space, all indirect costs) being allocated using a three-step procedure. As is common practice in ABC, the first-stage cost drivers allocated the resource costs to the activity groups. The second step, using second-stage cost drivers, allocated the costs of these various activity groups to the specific activities. It was only in the third step that the costs were allocated to the products. In this final step toward the calculation of the (treatment-related) product costs, the activity costs per product were added. Moreover, multiplication factors (or third-stage cost drivers), accounting for the complexity of the treatments (i.e., the number of fractions and number of fields), were introduced in this step. A very complex allocation principle with a large diversity of cost drivers could be avoided by introducing an extra allocation step between activity groups and activities. The straightforward principle of time consumption, weighed by some factors of treatment complexity, was used instead. A few cost components were not allocated to the products following this stepwise allocation procedure. Part of the material costs were defined as direct costs and therefore were directly assigned to the products. The overhead costs were treated in varying ways. Some of the so-called departmental support costs were related to machine maintenance and were therefore reallocated to the equipment costs. The remainder of the departmental support costs (e.g., research and teaching) and the hospital support costs (e.g., general management) were allocated to the products using a parameter that mimics product complexity. These overhead costs were not allocated through the intermediary of activity consumption but by using a complexity parameter, for which the number of fractions was chosen.
After having performed all described allocations, the global cost of the product equals the sum of the wage, equipment, and space (treatment-related) activity costs, the specific material cost, and the allocated proportion of the overhead costs. Within the development of the ABC model, the costs, activities, products, and cost drivers were progressively defined. This resulted in a large data set. To link these data— using the defined allocation principles and cost drivers and with the aim of calculating the RT product costs—a computer program was developed using Microsoft Excel software. The consecutive Excel files, mirroring the ABC structure, are presented in Table 1. Costs All cost inputs (Table 2) used in the ABC program were obtained from the Leuven University Hospital in 2000 and expressed in Euro (€). The actual costs were used for all resources; for the durable inputs, equivalent annual costs were calculated on the basis of the actual price or the current replacement value. Five cost categories were defined: wage, equipment, space, material, and overhead costs. On the basis of the way they are assigned, costs can be divided into direct costs (directly traced to the cost objects) and indirect costs (not directly traceable to the cost objects, but indirectly allocated using cost drivers). The only direct costs within a RT process are some specific material costs, including the costs of fixation masks, X-ray films, simulation and portal films, and the materials to produce shielding blocks. Because the use of masks is standard for certain types of treatments, and because the
ABC in radiotherapy
● Y. LIEVENS, W. VAN DEN BOGAERT, and K. KESTELOOT
525
Table 1. Files of activity-based costing program File group Definition files Activity group files
Activities file Products file
File Definition costs Definition activities Definition products Time activity groups Wage activity groups Space activity groups Equipment activity groups Total activity groups Cost activities Cost products
Function Definition of center-specific parameters Interim calculations based on defined parameters
Calculation of treatment-related unit costs Calculation of product costs
number of films used and blocks produced per type of RT is equally defined, their cost can be directly and unequivocally traced to that specific RT treatment. The vast majority of the cost inputs of a RT process (i.e., costs of personnel, equipment, and buildings, remainder of the material costs, and overhead costs) are consumed during the production of a wide range of different RT products. It is therefore impossible to trace directly (a part of) these costs to a specific RT treatment. Hence, these costs were considered indirect costs and were allocated to the RT products using cost drivers. For wage, equipment, and space costs, a stepwise allocation using the ABC method was used; the nonspecific material costs and overhead costs were allocated on the basis of the number of fractions.
taking place in an organization, which activities are associated with which part of the production process, and how these activities are linked to the consumption of resources and the generation of revenues. In our model, a first global distinction was made between two groups of activities: treatment-related activities, directly linked to the delivery of RT; and supporting activities, supporting the production process and the department but not linked to specific RT treatments. Figure 3 graphically shows the further subdivision of these categories into subcategories. Treatment-related activities. Thirty treatment-related activities (Table 3), performed in the process of delivering RT to patients, were defined and aggregated into four treatmentrelated activity subcategories:
Activities Activities are defined as repetitive actions performed in the production process of an organization. Because they are the central components of an ABC system, through which the indirect resource costs are allocated to the products, the first prerequisite in the development of an ABC program is to perform an activity analysis. This identifies all activities
General activities: performed in each type of RT product Product-related activities: related to certain types of RT products Fractionation-related activities: vary in frequency as a function of the number of fractions delivered Field-related activities: vary in frequency as a function of the number of fields delivered
Table 2. Cost categories by cost type and allocation basis Cost category
Cost input
Cost type
Wage costs Equipment costs Space costs Material costs
Reference wage costs Equivalent annual costs ⫹ external maintenance contracts ⫹ CRS Equivalent annual costs Consumed materials
Indirect costs Indirect costs
Stepwise ABC Stepwise ABC
Overhead costs
Hospital* support Departmental† support
Indirect costs Direct costs Indirect costs Indirect costs Indirect costs
Stepwise ABC Direct assignment Fraction number Fraction number Fraction number
Total
Allocation basis
Amount (€) 2,445,318€ 1,444,944€ 227,453€ 55,863€ 52,465€ 604,268€ Part of rows 1–3, 5 4,830,313€
Abbreviations: ABC ⫽ activity-based costing; CRS ⫽ care-related support (see definition activities); IT ⫽ information technology; FTE ⫽ full-time equivalent. * Contains overhead costs assigned by hospital to radiotherapy department; costs of nonassigned buildings and equipment, financial costs, general costs, maintenance costs, and costs of energy were allocated on basis of number of square meters of department; cost of general coordination and administration, personnel accommodation, the IT department, and industrial medicine were allocated by number of FTE staff working in department. † Contains costs related to departmental support activities (care and non– care-related costs of wage, equipment, space and material); they are defined within additional steps of ABC program.
526
I. J. Radiation Oncology
● Biology ● Physics
Volume 57, Number 2, 2003
Fig. 3. Activity categories of RT department.
The definition of the fractionation- and field-related activities was judged crucial because of the relation of these components to the treatment complexity and the resulting cost. Supporting activities. Supporting activities included departmental-supporting activities and hospital-supporting activities. Departmental-supporting activities were those ac-
tivities performed by personnel of the RT department, specific to the department but not product specific. They were further subdivided into care-related support (e.g., oncologic case discussions) and non– care-related support (e.g., research and teaching). Hospital-supporting activities were those activities per-
Table 3. Treatment-related activities, aggregated into four categories and further subcategories Category General activities
Subcategory First contact Appointment planning Permanence—file management Morning discussion Simulation Planning
Product-related activities
Fractionation-related activities
Field-related activities
Quality control start Social work Discharge Trial discussion Mask—immobilization Preparation therapy
Dietary advice Psychological advice Radiotherapy delivery Weekly follow-up Weekly chart round Weekly quality control Blocks—bolus In vivo dosimetry Portal imaging
Details Status and/or multidisciplinary consultation and/or consultation in other department
Basic planning and 1D planning or 2D planning or 3D planning and supervision planning
Localization and CT in RT position and drawing target volume and discussion of radiotherapy plan
ABC in radiotherapy
● Y. LIEVENS, W. VAN DEN BOGAERT, and K. KESTELOOT
formed for many or all other departments in the hospital by personnel not belonging to the RT department. Examples of such activities include general management, heating and maintenance, and information technology support. Following the practice in the Leuven hospital, the model allocated part of these global hospital overhead costs to the RT department on the basis of the number of FTEs and number of square meters of space. Non–EBRT activities. Some activities (i.e., consultations, brachytherapy, and services provided to other RT departments) were defined as non–EBRT activities. Because these were irrelevant to the cost of EBRT, they were excluded from the calculation. However, because these activities partly consume the same resources as EBRT, their time (and related resource) consumption had to be defined before being separated from the rest of the cost calculation. Products In this model, a product was defined as any type of RT. Because the aim was to analyze the resource cost of the different radiation products delivered in the Leuven radiotherapy department and to use these cost data in economic evaluation studies, an aggregate product definition would have been insufficient. A high level of detail (i.e., products defined as specific treatments per tumor type) was therefore used. The products were organized in large categories per organ system (e.g., cranial RT, breast RT). Within each organ system, further distinction was made between the different tumor types, closely related to the specific treatment setups. If within one tumor type, separate target volumes (e.g., primary tumor and locoregional lymph nodes, whole organ and boost volume) can be both irradiated concomitantly and consecutively, these different target volumes were defined as separate products. Moreover, because the resource consumption (and cost) of a target volume (e.g., locoregional lymph nodes in breast cancer), irradiated separately or in addition to another volume can differ quite substantially, two different products were defined in these situations. Finally, several variables influencing the treatment complexity and therefore potentially also the cost, were taken into account: number of fractions, number of fields, and whether a planning CT scan and three-dimensional (3D) planning was performed. Cost drivers Cost drivers are used to allocate indirect costs to products. As mentioned before, our model used three, instead of two, levels of cost drivers, thus replacing a large diversity of cost drivers by a straightforward principle of time consumption, weighed by some factors of treatment complexity. It follows that the definition of the consumed time is the cornerstone of the calculation model. The time spent per type of personnel for each activity and per specific product (e.g., the technologists’ time spent on a simulation of a tangential breast irradiation) was defined on the basis of information from time slots (especially for the radiation
527
technologists) and interviews and time calculations (for the other personnel categories). First-stage cost drivers: allocation of resource costs to activity groups. First, wage costs were allocated to the different activity groups on the basis of the percentage of time spent per type of personnel to that activity (e.g., percentage of yearly technologists’ time spent on simulation). The defined time estimates per type of personnel, per activity, and per product type were used to calculate this percentage (or proportion) of time per activity group and per type of personnel, on the basis of the product-mix of the year under investigation. As an example, the calculation for the technologists is demonstrated (Table 4): The time per activity per specific product was multiplied by the total number of these products delivered in the year under investigation, resulting in the summed time per activity group All the times per activity group of the technologists were added and augmented with the yearly technologists’ time spent on care and non– care-related supporting activities and with the yearly time spent on activities not related to EBRT The global summation thus amounts in a total technologists’ time per year, equal to 100% With the “rule of three,” the proportion of the technologists’ time spent to the different activity groups was then calculated Knowing the global technologists’ wage costs, the cost for each activity group, treatment related and supporting and non-EBRT activities, could then be defined The same approach was then repeated for the calculation of the proportional time, and correlated costs, for the different activity groups and various personnel categories. The summation of all the personnel costs equals the wage cost per activity group. This approach is especially useful for calculating the time consumption of those personnel categories that have an irregular time schedule and/or accumulate several activities within the same time slot. More importantly, it eliminates the potential problem related to the use of standard time schedules in which the total time consumption per activity group may be overestimated (in the case of incomplete occupation of the personnel) or underestimated (in the case of a higher workload than predicted on the basis of the available staff). Without correction, the “idle” time related to incomplete occupation would be neglected, thus underestimating the real cost. In the converse situation, an overestimation of the real cost would occur. Because the actual time consumption of the investigated year was used as the calculation basis, the calculated cost was corrected for the productivity of the department during that year. Second, space costs were allocated to the activities as a function of the activities taking place in the different rooms. Because most rooms share their space among different activities (e.g., the simulation room is used for EBRT and
528
I. J. Radiation Oncology
● Biology ● Physics
Volume 57, Number 2, 2003
Table 4. Detailed example of different cost calculation steps performed in allocation of wage costs (time and cost data used were those for radiation technologists) Level Activity group Time Simulation Total Total yearly time
Treatment
Specific time
Tangential breast Bone metastases (APPA, 5 fractions) … Treatment-related activities Simulation RT delivery … CRS Non–CRS Non-EBRT
80 min ⫻ 319 ⫽ 25.520 min 70 min ⫻ 177 ⫽ 12.390 min
Total Proportional time Simulation
— 2,404 h 21,170 h — 3.163 h 2.588 h 67 h 2.404 h/34.920 h
Total time/cost
144,240 min ⫽ 2,404 h
34,920 h ⫽ 100% 6.88%
Total yearly wage cost 829,994€ Cost Simulation Activity Proportional time Breast simulation Cost Breast simulation
829,994€ ⫻ 6.88%
57,139€
80 min/144.240 min
0.555 ⫻ 10⫺3
0.555 ⫻ 10⫺3 ⫻ 57.139€
31.7€
Abbreviations: APPA ⫽ antero posterior opposed fields; RT ⫽ radiotherapy; CRS ⫽ care-related support; EBRT ⫽ external beam RT.
brachytherapy simulations, localizations, and making masks) and other rooms are only used a few hours throughout the day (e.g., the meeting room) and to integrate the productivity factor into the calculation, the following approach was used. After having defined the activities—treatment-related, as well as supporting and non-EBRT activities—taking place within a specific room, these activities were linked to the “critical personnel” (i.e., the personnel category most closely related to the use of the room). The calculation of the proportional space cost was subsequently performed in analogy with the cost calculation of the wage costs. The time consumption of all activities performed in a certain room, on the basis of the critical personnel, was summed. The proportional time (and thus cost) for each separate activity could thus be easily calculated. Third, the equipment costs were allocated using a comparable approach, that is by defining the activities performed with a type of equipment and then linking these to the critical personnel using this equipment. Second-stage cost drivers: allocation of activity group costs to activities. This allocation was simply based on the time consumption of each specific activity per specific product and personnel category. As an example, the technologists’ cost in the simulation of a tangential breast irradiation was calculated as follows (Table 4). The number of minutes spent by the technologists on a simulation for tangential
breast irradiation was divided by the total yearly time consumption of technologists for the activity group of simulation. Then, the resulting proportion was multiplied by the annual simulation cost of the technologists (proportional to the time technologists yearly spent on simulation). This calculation was then performed for every type of personnel, as well as for the building and equipment cost, and repeated for all treatment-related activities and all products. Third-stage cost drivers: allocation of activity costs to products. The final step between the activities and treatment-related product costs was basically a summation of the different calculated activity cost components. Some factors correcting for complexity (defined as third-stage cost drivers) were nevertheless introduced in this last step. The fractionation-related activities were multiplied by the number of fractions (or weeks) necessary to produce the product and the field-related activities were multiplied by the number of fields used in the setup of the treatment. The addition results in the treatment-related product cost based on wage, equipment, and space consumption. Overhead costs. The support costs (both departmental and hospital support) were allocated to the products in a single step using a cost driver that was thought to adequately capture the product complexity (i.e., number of fractions). In other words, each product was assigned a proportion of the support costs on the basis of the number of fractions necessary to deliver that product.
ABC in radiotherapy
● Y. LIEVENS, W. VAN DEN BOGAERT, and K. KESTELOOT
Table 5. Comparison of costs at activity group level: costs of treatment-related, care- and non– care-related departmental support and non-external beam radiotherapy activities Cost type
Cost (€)
%
Treatment-related activities Treatment preparation Simulation Planning Quality assurance Administration Treatment delivery Care-related support activities Non–care-related support activities Non–EBRT activities Total
2,956,848 923,031 356,872 340,099 213,586 199,859 1,620,372 318,254 566,790 275,824 4,117,716
71.81 22.42 8.67 8.26 5.19 4.85 39.35 7.73 13.76 6.70 100
Abbreviation: EBRT ⫽ external beam radiotherapy.
Direct costs. The specific material costs (immobilization masks, materials used for making shielding blocks, and X-ray films) were directly assigned to the products using the defined consumption of these materials per product. RESULTS Cost data of Leuven radiotherapy department Resource costs. The distribution of the cost inputs of the Leuven radiotherapy department were wage costs (52.06%), equipment costs (28.48%), hospital overhead costs (12.51%, of which 9.34% and 3.17%, respectively, were allocated by the number of square meters and FTEs), space costs (4.71%), and material costs (2.24%). Activity costs. Table 5 shows the costs at the activity group level, consisting of wage, equipment, and space costs. The treatment-related activities were organized into the larger categories of treatment preparation, quality assurance, administration, and treatment delivery. As could be expected, treatment-related activities consumed the greatest proportion of the costs (72%), with treatment delivery the most important component. Within treatment preparation, simulation and planning roughly consumed the same amount of costs.
529
Twenty-one percent of the activity group costs were related to departmental support activities that could not be directly allocated to the specific RT treatment (e.g., research and teaching). About 7% of the wage, equipment, and space costs of the RT department were used to perform non-EBRT activities. Product costs. Table 6 shows the total yearly cost per organ system. It also presents the percentage of the costs and products per organ system as a function of the total costs and total products. The relative index divided the percentage of the costs by the percentage of the products, giving an indication of treatment complexity. Figures of ⬎100% were found in the relatively more expensive treatments (i.e., that consumed more resources than expected on the basis of the number of delivered treatments). This was most clearly seen in organ systems in which treatments are predominantly curative, with long irradiation schedules (between 6 and 7 weeks) and a boost, including secondary simulation and planning, such as headand-neck and breast RT. Conversely, if the relative index was ⬍100%, this means that, on the basis of the production volume, fewer costs (resources) were consumed than average. Bone and softtissue RT, largely composed of palliative treatments that combine very short schedules and easy treatment techniques, were found in this category. Although predominantly curative, the cumulative effect of short fractionation regimens (between 3 and 4 weeks) and of shorter time slots on the treatment machines (because of predominantly APPA irradiations) explain why treatments for patients with hematologic malignancies are also found to be less costly than the “average” product. In Table 7, some specific product costs are shown, as well the global costs, as the different components: treatmentrelated costs (wage, equipment, and space costs and differentiating between various subactivities), material costs, and overhead costs (both departmental and hospital support costs). As a matter of comparison, the average departmental product cost is also shown. Breast treatments were chosen as an example of curative treatments. A distinction was made between breast and
Table 6. Cost per organ system Organ system
Total cost (€)
Cost (%)
Products (%)
Relative index
CNS Head and neck Breast Lung Hematology Gastrointestinal Urology Gynecology Bone and soft tissue Total
385,774 619,059 1,709,988 358,550 126,356 444,878 298,933 139,370 471,581 4,554,489
8.47 13.59 37.55 7.87 2.77 9.77 6.56 3.06 10.35
9.21 8.93 25.49 10.63 4.18 9.16 5.14 3.05 24.19
91.92 152.18 147.27 74.08 66.32 106.66 127.59 100.25 42.80
Abbreviation: CNS ⫽ central nervous system.
530
I. J. Radiation Oncology
● Biology ● Physics
Volume 57, Number 2, 2003
Table 7. Examples of specific costs per product Treatment-related costs* (€) Product Breast Breast Thoracic wall Separate IM-MS Additional IM-MS Electron boost Photon boost Lung Palliative Involved field Mediastinum Boost Head and neck 3D-3-field setup Parotid-sparing Boost 3 fields Boost 4 fields Palliative (Gy) 1⫻8 4⫻5 10 ⫻ 3 Average cost
Quality Radiotherapy Material Overhead Total Preparation Simulation Planning assurance Administration delivery Total costs (€) costs (€) costs (€) 71 71 114 89 16 16
146 130 130 100 66 99
157 171 124 38 77 110
107 69 88 68 33 71
122 122 122 4 35 35
966 966 966 445 296 296
1567 1528 1544 744 523 626
18 2 12 12 2 18
966 966 966 966 309 309
2551 2496 2522 1721 834 953
110 134 110 87
178 398 178 157
110 439 439 68
71 159 107 99
81 165 129 35
396 1264 898 366
946 2559 1866 807
24 36 24 36
386 1275 888 386
1356 3869 2778 1229
74 230 87 110
315 401 105 105
351 442 63 63
125 143 89 107
128 131 35 35
1449 1894 551 729
2442 3241 929 1149
67 79 36 47
966 966 386 386
3475 4286 1351 1583
48 95 95
127 127 127
77 77 77
47 59 71
33 52 72
59 207 390
392 618 833
24 24 24
39 193 386
454 835 1243 2575
Abbreviation: IM-MS ⫽ internal mammary and medial supraclavicular lymph node chain. * Includes wage, equipment, and space costs.
thoracic wall RT, internal mammary and medial supraclavicular lymph node chain (IM-MS) RT (treated separately or in addition to breast or thoracic wall RT) and boosts with electrons or photons. Breast, thoracic wall, and lymph nodes were irradiated to 50 Gy in 2-Gy fractions, boosts delivered a dose of 16 Gy in eight fractions. All curative intent breast RT sessions cost grossly the same (i.e., around 2500), irrespective of the setup (breast, thoracic wall, or separate IM-MS RT). The cost of a breast boost ranged between 834 and 953. The cost of additional IM-MS RT was one-third lower than the cost of a separately irradiated IM-MS. Moreover, because of the specific structure of the ABC program, an equal amount of overhead costs was allocated to additional and separate treatments. Hence, this calculated cost of additional IM-MS might even be too high; if no overhead cost was allocated to additional RT treatments, the cost would drop to 756€. For lung treatment, a palliative fractionation schedule (30 Gy in 3-Gy fractions) was compared with the curative-intent fractionation (involved field of 66 Gy in 2-Gy fractions and mediastinal RT to 46 Gy followed by a boost of 20 Gy in 2-Gy fractions). It demonstrated that a palliative schedule costs about one-third of a curative schedule, in other words, that the costs correlated well with the number of fractions. This could obviously be expected, because the model explicitly defined the number of fractions as a factor of complexity. A similar impact of the number of fractions on the cost
was found in the palliative bone schedules (i.e., one fraction of 8 Gy; 20 Gy in 4-Gy fractions, respectively, without and with use of an immobilization mask; and 30 Gy in 3-Gy fractions). It was assumed that each of these RT sessions was performed with two parallel-opposed fields. Moreover, a comparable number of fractions translated into a similar cost, irrespective of the indication. A treatment delivering 10 fractions costs 1242€ in the case of bone metastases and 1356€ for lung treatment. The curative treatments showed that the complexity of the treatment preparation (e.g., CT scan in treatment position and 3D planning) was of less importance in the total cost picture. The cost of curative-intent, 3D pulmonary treatment hovered between 3869 and 4007 and that of 3D head and neck RT amounted to 4826; the cost of breast RT (without treatment CT scan) followed by an electron boost was 3385. If additional IM-MS RT was performed, the total cost amounted to 5106 or 4141, respectively, with and without the inclusion of the overhead costs in the latter. The more complex parotid-sparing approach for head-and-neck cancer was even more expensive at 5869€. For most RT, about one-third of the total costs were consumed by overhead costs. As defined, this proportion correlated with the number of fractions. It should be acknowledged that within the model, mean equipment cost estimates were used, not accounting for the type of linear accelerator (simple, dual-energy, or MLC) on which the product was delivered. This decision was taken
ABC in radiotherapy
● Y. LIEVENS, W. VAN DEN BOGAERT, and K. KESTELOOT
531
Table 8. Impact on costs of replacing all actual treatment machines with multileaf accelerators Variable Cost inputs Wage Space Equipment Material Support Treatment-related costs Treatment preparation Quality assurance Administration Treatment delivery Total Product costs Tangential breast Involved field cerebral Palliative lung Bone metastasis Average cost Total EBRT cost
Baseline cost (€)
After replacement cost (€)
After correction cost (€)
2,514,702 227,454 1,375,560 108,329 605,268
2,514,702 (0) 227,454 (0) 1,719,349 (25) 80,441 (⫺25.7) 605,268 (0)
2,483,254 (⫺1.3) 227,454 (0) 1,719,349 (25) 80,441 (⫺25.7) 601,668 (⫺0.6)
923,031 213,586 199,859 1,620,372 2,956,848 2551 3785 1356 454 2575 4,554,489
1,000,180 251,723 201,557 1,852,306 3,305,767
(8.4) (17.9) (0.8) (14.3) (11.8)
2797 (9.6) 3554 (⫺6.1) 1481 (9.2) 494 (8.8) 2759 (7.1) 4,880,882 (7.4)
996,828 248,913 200,297 1,833,726 3,279,763
(8) (16.6) (0.2) (13.2) (10.9)
2777 (8.9) 3529 (⫺6.7) 1471 (8.1) 491 (7.9) 2740 (6.4) 4,846,896 (6.4)
Abbreviation: EBRT ⫽ external beam radiotheraphy. Data in parentheses are percentages of differences in cash.
for practical purposes within the development of the program, and also because, except for intensity-modulated radiotherapy (IMRT) of the prostate, there are no absolute clinical reasons why specific products delivered in our department should be irradiated on the MLC. That this approach may, however, yield some distortions in the calculated product costs is illustrated for the GI and urologic patient populations. In our department, these patients are specifically treated on the more-expensive MLC. The treatment time slots used on this machine are, however, shorter. As a result, because the model does not account for the type (and related cost) of the treatment machines, the cost of GI and urologic RT was relatively underestimated. In the baseline calculation, both groups were shown to have resource consumption grossly in line with the expectations based on the number of delivered treatments as expressed by a relative index of 107 and 128, respectively. However, after rerunning the program with time slots as for a regular linear accelerator, higher relative costs were calculated (i.e., corresponding index of 114 and 145, respectively). The same finding was observed for the absolute costs. When using longer time slots, the cost of coplanar rectal RT, for example, increased from 2747€ to 3119€ or 14%. Some practical applications of the program The ABC model can also be used to analyze how costs are affected by changes in cost inputs, in practice. Some examples are presented. Different resource costs. As an example of analyzing how the introduction of different resource inputs affects the costs, it was assumed that MLCs, including on-line portal imaging, would replace standard collimators on the three
non-MLC treatment machines of the department. With MLC, shorter time slots can be used for some indications, implying that less technologist time would be needed for treating the same number of patients. Because the actual workload is very high, a calculation was also made assuming that the level of staffing for technologists was not reduced. Table 8 shows the costs and the difference in costs compared with baseline, without and with correction for a lower number of needed technologists. The calculations show that replacing all standard collimators with a MLC, including on-line portal imaging, would lead to an increase in the annual EBRT costs of 7.4%. The total equipment costs would increase by 25% (343,789€) and the material costs would decrease by 25% (27,888€), because portal films would no longer be used. After the introduction of reduced treatment time slots in those RT sessions in which this would be possible, it was calculated that the technologist staff could be reduced by (only) 0.74 FTE, while maintaining the same workload as in the baseline calculation. This represented a decrease in annual wage cost of 31,448€ and of FTE-related support of 3600€. Hence, these savings would not compensate for the increased equipment costs. As expected, the costs of treatment delivery and quality control would increase by introducing more costly treatment machines and on-line portal imaging. For most products, the increase in equipment costs translated into an increase of product cost. Only for the products for which the treatment time slots shortened after introducing MLC, costs conversely decreased. Because, in the Leuven department, most of the more complex treatments are done on the MLC, the products for which the actual treatment time slots (and thus costs) would decrease, would be limited.
532
I. J. Radiation Oncology
● Biology ● Physics
Volume 57, Number 2, 2003
Fig. 4. Variation in cost per patient as a function of department size.
Number of patients. We ran the calculations varying the number of patients treated per year between 250 and 2500, while keeping a same product mix as in the original data. The resource requirements of personnel, equipment, and space were defined as a function of the number of patients according to the Belgian accreditation norms (17). The consumption of overhead was related to the defined numbers of FTEs and square meters; the material costs were varied proportional to the number of patients. Mean unit resource costs of the Leuven radiotherapy department were used. As an example, the cost of tangential breast irradiation is shown in Fig. 4. The variation in cost was similar in all other situations. Because the resource requirements of personnel, equipment, and space varied stepwise with the number of patients (semifixed costs, see “Discussion”), the cost per patient not only depends on the department size but also on the degree of use of the available resources. The resulting cost per patient is the lowest if all available staff and equipment is used fully. The variability in the costs, as well in absolute terms as related to the level of excess capacity, was most obvious in departments treating ⬍1000 new patients annually. For larger departments, an additional decrease in costs was observed, although less importantly. Product mix. In addition to the number of patients treated, the level of complexity also affects the costs. For example, the workload associated with treating 1000 patients with metastatic bone schedules will be lower than the workload for treating 1000 patients with conformal head-and-neck RT. It was analyzed how the total annual costs would vary as a function of a change in product mix, assuming that the cost per product would remain the same. The first example analyzed the impact of RT for the IM-MS. In the baseline data set, 128 patients received additional IM-MS irradiation. If the IM-MS would have
been irradiated as the standard in all patients treated to the breast or thoracic wall who did not receive RT to all locoregional lymph nodes (i.e., a total of 355 patients), the total department cost would have increased with 390,667€ (or 8.6%). Conversely, under the assumption that no additional IM-MS would have been delivered, the costs would have decreased by 4.8% (or 220,288€). The second example tested the impact of using different fractionation schedules for the treatment of bone metastases. Although in the baseline calculations, there was a mix of different metastatic bone RT schedules, it was subsequently assumed that all these patients would undergo RT with a single fraction of 8 Gy, 5 fractions of 4 Gy, or 10 fractions of 3 Gy. In the first two situations, the departmental costs would have decreased by 3.8% and 0.5% (or 172,764€ and 24,652€ for the standard use of a single fraction and 5 fractions, respectively). In the third situation, it would have increased by 3.1% (138,946€) if 10 fractions had been the standard. These simulations demonstrated that the total costs change proportionally with the complexity of the product mix. They, however, also assume that the resource consumption proportionally follows the product mix. This is correct for the material costs, but not for the wage, equipment, and space costs, which vary stepwise with the production and product mix. One could, moreover, assume that, because of budgetary restrictions, the total costs should remain constant. The second approach (Table 9) therefore analyzed the effect of the same changes in product mix under the assumption of fixed total resources. In this case, the costs per product decrease if a more complex product mix is delivered. In other words, more complex products can only be delivered at the same total costs by increasing productivity. There is, of course, a limit to increases in productivity,
ABC in radiotherapy
● Y. LIEVENS, W. VAN DEN BOGAERT, and K. KESTELOOT
533
Table 9. Product costs after change in product mix Variables
Breast (€)
1⫻8 Gy (€)
5⫻4 Gy (€)
10⫻3 Gy (€)
Baseline IM-MS Standard IM-MS No IM-MS Bone irradiation 8 Gy/8 Gy 20 Gy/4 Gy 30 Gy/3 Gy
2551
454
835
1243
2349 (⫺8.6) 2687 (5.1)
433 (⫺4.9) 467 (2.8)
781 (⫺6.9) 870 (4)
1152 (⫺7.8) 1303 (4.7)
2652 (3.8) 2568 (0.7) 2473 (⫺3.2)
462 (1.7) 455 (0.2) 448 (⫺1.3)
867 (3.7) 837 (0.2) 815 (⫺2.4)
1295 (4.1) 1248 (0.5) 1207 (⫺2.9)
Abbreviation: IM-MS ⫽ internal mammary and medial supraclavicular lymph node chain. Data in parentheses are percentages of change in product cost compared with baseline.
whether as a consequence of the complexity or the number of patients, that can be accommodated with a certain level of personnel, equipment, and space. If growth were substantial, additional resources would be needed regardless. It was calculated that by introducing standard IM-MS irradiation, the workload of most personnel would increase, with the largest increase for dosimetrists, technologists, and mold technicians (6.77%, 5.57%, and 2.99% increase, respectively). The standard use of single fractions for metastatic bone RT would lead to a 2.95% decrease in the workload for the technologists and 7.45% for the mold technicians. Considering the number of technologist working hours, this represents a decrease of 937 h (for the change in palliative practice) and an increase of 1771 h (for the change in IM-MS practice) yearly, which, assuming optimal productivity in the baseline situation, would free 0.58 FTEs or necessitate 1.09 FTEs extra, respectively. Furthermore, because most of these technologist hours (i.e., about 85%) would be devoted to treatment delivery, the former change would save about 1 h of machine time daily and the latter would require 2 h extra, using an assumption of three technologists per treatment machine.
DISCUSSION Benefits of ABC in RT Improved insight into departmental costs. The major advantage of using ABC for cost calculation in RT is that it yields more accurate costs and gives better insight into the departmental cost structure. In line with other literature reports, the Leuven analysis confirms that the major resource costs of a RT department are wage and equipment costs (3, 12, 18). The cost analysis of the different activity groups showed that almost threequarters of the total wage, equipment, and space costs were consumed by patient care, of which treatment delivery represented the largest part (almost 40%). This is not surprising, because the delivery of RT is a repetitive process that consumes important amounts of personnel time and expensive equipment and space. Treatment preparation, also com-
bining several highly complex and time-consuming activities of the RT process, consumed 22.42%. As a consequence of the former observation, the analysis of the costs per product group showed that long curative treatments had a higher relative cost compared with those with shorter curative schedules, which again cost more than those with predominantly palliative schedules. Also, the cost of the specific products was determined more by the length of treatment (i.e., number of fractions and, by extension, the use of the treatment machine) than by the treatment complexity as such (e.g., related to more complex planning using a treatment CT scan). As an example, in our calculation, three-field 3D prostate RT was not more expensive than standard tangential breast RT. However, if the treatment complexity also translates into more irradiation fields and thus longer treatment time slots, the product cost rose considerably, as for example seen in parotid-sparing headand-neck treatments, figuring in the top five of the most expensive treatments. These observations are important if we consider the cost of novel approaches that aim to obtain higher cure rates and/or diminish (long-term) side effects. From a radiobiologic point of view, it was found that, at least in certain cancer types, cure rates increased by altering the fractionation schedule, which resulted in the delivery of two (or more) sessions daily for a growing number of patients. Conformal (CRT) and intensity-modulated radiotherapy (IMRT) aim to improve outcome by optimizing physical dose delivery. In both described innovations, the treatment costs will increase because of longer treatment times. In hyperfractionation, the cost will increase as a consequence of the greater number of fractions and in CRT and IMRT because of the longer treatment time slots. Analysis of financial implications of resource and practice changes. The strength of such a detailed cost accounting program lies not only in the analysis of the actual situation, but also in the possibility to run simulations on variations in resources or practice patterns, as described in “Results.” The example for the changes in resource cost inputs demonstrated that the greater equipment costs incurred by replacing all standard machines with MLCs and on-line
534
I. J. Radiation Oncology
● Biology ● Physics
portal imaging were not compensated for by the savings in material, wage, and support costs. It can therefore be postulated that the introduction of more complex treatment techniques using high-tech equipment will inevitably induce higher costs. The other examples demonstrated the impact of the workload—in terms of the number of patients treated (department size) and treatment complexity (product mix)— on the product cost. Because of the indivisibility of some resources, largescale departments can treat their patients at a lower cost per case. The treatment cost also depends on the degree of resource use, with the cost lowest when all additional equipment and staff are fully occupied. That the product mix is equally important to the specific product costs was shown by simulating the effect of a more complex (standard use of IM-MS irradiation for all breast cancer patients) and a less complex (single fractions in the palliative treatment of bone metastases) product mix. As expected, we found that the former practice change would require extra treatment machine time, and the latter would liberate machine time. Because these two examples apply to the most frequent product groups in our department, these changes in resource needs must be at the outskirts of those expected by a single practice change. Moreover, they showed that different changes in practice should be weighed against each other to estimate their conjunctional effect on the product mix and resource requirements. Correction of costs for actual productivity. The Leuven ABC system used a mixed top-down and bottom-up approach in the allocation of the semifixed costs (i.e., costs that vary stepwise with the number of patients) of wage, equipment, space, and overhead. A top-down approach divides the total costs of the overall products. Because the total costs are known and remain constant, the cost of a product is inversely related to the production volume (i.e., the cost per product diminishes with increasing production volume). As such, a top-down calculation requires exact data on the delivered number of products, product mix, and relative resource consumption of the respective products. In a bottom-up approach, conversely, the cost of each final product is calculated on the basis of the activities used. Because the actually delivered production volume, however, is not necessarily equivalent to the expected one (on the basis of which the unit cost is calculated), this might result in higher (or lower) total (calculated) departmental costs than expected. In other words, in a bottom-up approach, the total calculated product costs, with unit costs based on the expected production volume, do not necessarily coincide with the actual total departmental input costs. The Leuven ABC program was mainly designed using the top-down allocation principle. All costs introduced in the model were totally allocated to the products. The major cost driver of the model (i.e., time consumption) was, however, defined using the bottom-up principle: “how much time is spent per type of personnel for each activity and per
Volume 57, Number 2, 2003
specific product.” By equaling the total consumed time to 100% and by using percentages of this total time as the cost driver, the top-down approach was introduced at this level. Although the presented cost calculation model is by default post hoc and is unable to calculate the cost of isolated products, because knowledge of the global departmental productivity and product mix is required, the major advantage of such a combined top-down and bottom-up approach lies in the correction of the costs for productivity (i.e., account is made for the potential over or under use of resources). Because semifixed costs remain constant over a certain range, irrespective of the degree of resource use, a pure bottom-up calculation would result in an overestimation of the costs in the former situation (as more products are produced than expected) and an underestimation in the latter (because the unused resources would not have been allocated). The use of such a mixed top-down and bottom-up model is the major difference between the Leuven ABC calculation system and some other ABC-like exercises developed in the field of RT (19, 20), in which the calculation of the product costs, on the basis of activity consumption, was solely performed using the bottom-up principle. Potential drawbacks related to use of presented ABC model Model complexity. The precision of an ABC system depends critically on the level of detail of its constituting components: the more refined the definition of costs, activities, products, and cost drivers, the more accurate the resulting cost calculation. A high level of detail, however, unavoidably translates into a higher workload (and cost), not only during the development phase, but also in daily use. Although the decision to use detailed definitions of consumed resources, performed activities and delivered products—necessary to obtain the accurate cost estimates we aimed for—might have rendered the Leuven approach too complex to be practically manageable, this was avoided by limiting the number of cost drivers. This limitation in cost drivers was accomplished through the introduction of a third allocation step between the activity groups and activities. By doing so, time consumption, weighed by some factors of treatment complexity, became the main allocation principle, instead of using a large number of different cost drivers. Possible distortions. Although the detailed definitions and multistep allocation principle yielded a refined cost calculation, some assumptions (use of mean equipment cost estimates and allocation of overhead costs to the additional products in a similar way as for the other products) made during the development of the model, may result in a slight distortion of the calculated costs. The impact of these two potential shortcomings is described in “Results.” No attempt was made to correct for them, because this would have rendered the model much more complex, requiring programmatic adaptations when applying it to other RT departments. When adopting such a model for daily practice, however,
ABC in radiotherapy
● Y. LIEVENS, W. VAN DEN BOGAERT, and K. KESTELOOT
one should be aware of the potentially imbedded distortions related to the specificity of the model’s design. To estimate the importance of the distortions on the results, sensitivity analyses such as the ones described should be performed explicating how different options will have an impact on the costs.
Is the routine use of ABC practically achievable? The baseline results and described simulations illustrate that the use of an ABC model for RT allows one to solve questions on departmental activity and product costs. It provides a tool to evaluate the cost implications related to capital investments or changes in practice patterns. This knowledge can support the department in planning and controlling the costs of the services it provides (e.g., by reducing or eliminating nonessential activities of [nonprofitable] treatments or by adjusting the product mix). Although such a complex cost calculation program may be demanding in time and resource costs during the devel-
535
opment phase, once installed, its daily use should not require too much time and effort. A periodic (e.g., yearly) checkup should be performed to adapt the model to the changed input parameters. Most of these (the number of resources and their costs, the volume and types of products) are available in guidelines and/or routinely recorded within most institutions. The time estimates per activity may be more difficult to collect, but should not vary much from year to year. Moreover, standard time estimates, available within the department for appointment planning purposes, may be used instead of in-depth interviews. Finally, in the Leuven approach, a rather high level of detail was chosen, because the aim was to generate very detailed product costs (21). Within each specific department, however, different choices can be made about the desired accuracy. The less crucial the level of detail, the lower the workload associated with the data collection necessary for the routine use of such an ABC program.
REFERENCES 1. Goddard M, Hutton J. Economic evaluation of trends in cancer therapy: Marginal or average costs? Int J Technol Asses Health Care 1991;7:594–603. 2. Goddard M, Maher EJ, Hutton J, et al. Palliative radiotherapy—Counting the costs of changing practice. Health Policy 1991;17:243–256. 3. Jonsson E. Cost of radiotherapy in Sweden. Acta Oncol 1986; 6(Suppl.):57–71. 4. Atherton L. Reply to the article by Greene on “The cost of radiotherapy treatments on a linear accelerator.” Br J Radiol 1984;57:106–107. 5. Greene D. The cost of radiotherapy treatments on a linear accelerator. Br J Radiol 1983;56:189–191. 6. Penn CR. Megavoltage irradiation in a district general hospital remote from a main oncology center: The Torbay experience. Clin Oncol R Coll Radiol 1992;4:108–113. 7. Van den Bogaert W. The material infrastructure of the radiotherapy departments in Belgium: Commission for material infrastructure of the Belgian Society of Radiotherapy and Oncology (BVRO-ABRO). Leuven, Belgium: BVRO-ABRO; 1983. 8. Wodinsky HB, Jenkin RDT. The cost of radiation treatment at an Ontario regional cancer center. Can Med Assoc J 1987; 137:906–909. 9. Drummond MF, O’Brien B, Stoddart GL, et al. Methods for the economic evaluation of health care programs. 2nd ed. Oxford: Oxford University Press; 1997. 10. Gold MR, Siegel JE, Russell LB, et al. Cost-effectiveness in health and medicine. 1st ed. Oxford: Oxford University Press; 1996. 11. Finkler SA. The distinction between costs and charges. Ann Intern Med 1982;96:102–109. 12. Perez CA, Kobeissi B, Smith BD, et al. Cost accounting in
13. 14. 15.
16. 17.
18. 19.
20. 21.
radiation oncology: A computer-based model for reimbursement. Int J Radiat Oncol Biol Phys 1993;25:895–906. Baker JJ. Activity-based costing and activity-based management for health care. 1st ed. Gaithersburg, MD: Aspen Publishers; 1998. Cooper R. The rise of activity-based costing—Part one: What is an activity-based cost system? J Cost Management 1988; Summer:45–54. Lefebvre C, van den Brande A. Activity based costing: een nieuw boekhoudkundig ABC. In: Theunisse H, editor. Activity based costing, beleidsgerichte kosteninformatie. 1st ed. Antwerpen: MAKLU Uitgevers; 1992. p. 17–50. Cooper R. The rise of activity-based costing—Part three: How many cost drivers do you need, and how do you select them? J Cost Management 1989;Winter:34 – 46. Koninklijk Besluit van 5 april 1991 tot vaststelling van de normen waaraan een dienst radiotherapie moet voldoen om te worden erkend als zware medisch-technische dienst. B. S. April 17, 1991. Kesteloot K, Lievens Y, van der Schueren E. Improved management of radiotherapy departments through accurate cost data. Radiother Oncol 2000;55:251–262. Azoury E. L’Activity based costing: Application au secteur medico-technique de la radiothe´ rapie des Cliniques Universitaires Saint-Luc. Bruxelles, Belgique: Universite´ Catholique de Louvain, Faculte´ de Me´ decine, Ecole de Sante´ Public; 1999 –2000. Valiquette L, Dunscombe P. The economic characterization of radiation treatment process within a cancer care program. Unpublished data. Lievens Y. Cost and economic evaluation in radiotherapy: Activity-based costing and modeling techniques. PhD thesis. Leuven: Katholieke Universiteit Leuven, Faculty of Medicine, Radiation Oncology Department; 2002.
