Radiation costing methods: a systematic review

Objective Costs for radiation therapy ( rt ) and the methods used to cost rt are highly diverse across the literature. To date, no study has compared various costing methods in detail. Our objective was to perform a thorough review of the radiation costing literature to identify sources of costs and methods used. Methods A systematic review of Ovid medline , Ovid oldmedline , embase , Ovid HealthStar, and EconLit from 2005 to 23 March 2015 used search terms such as “radiation,” “radiotherapy,” “neoplasm,” “cost,” “ cost analysis,” and “cost benefit analysis” to locate relevant articles. Original papers were reviewed for detailed costing methods. Cost sources and methods were extracted for papers investigating rt modalities, including three-dimensional conformal rt (3D- crt ), intensity-modulated rt ( imrt ), stereotactic body rt ( sbrt ), and brachytherapy ( bt ). All costs were translated into 2014 U.S. dollars. Most of the studies (91%) reported in the 33 articles retrieved provided rt costs from the health system perspective. The cost of rt ranged from US$2,687.87 to US$111,900.60 per treatment for imrt , followed by US$5,583.28 to US$90,055 for 3D- crt , US$10,544.22 to US$78,667.40 for bt , and US$6,520.58 to US$19,602.68 for sbrt . Cost drivers were professional or personnel costs and the cost of rt treatment. Most studies did not address the cost of rt equipment (85%) and institutional or facility costs (66%). reporting of rt costs. More work to promote comparability and consistency across studies is needed.


INTRODUCTION
Based on the World Health Organization's World Cancer Report 2014, the burden of cancer rose to approximately 14 million incident cases per year in 2012 and is expected to rise to 22 million annually by the mid-2030s 1 . Given this striking increase in incident cancer cases, it becomes imperative to properly manage finances and resources for timely and appropriate patient care.
An integral part of cancer treatment is radiation therapy (rt). Approximately 50% of all cancer patients will receive rt at some point during the course of their treatment 2 . Using ionizing and non-ionizing radiation, rt kills cells or damages dna to prevent cancerous cell growth 3 . Delivery of rt can be achieved using various clinical procedures: three-dimensional conformal rt (3D-crt), intensity-modulated rt (imrt), stereotactic body rt (sbrt), brachytherapy (bt), and so on. The end goal of rt is to cure or shrink early-stage cancer, to prevent cancer recurrence, and to treat symptoms caused by advanced cancer 3 .
Radiation therapy requires high capital expenditure and is staff-and resource-intensive 2 . Because the costs associated with health systems have to be economically sustainable, cost becomes an important factor to take into consideration 4,5 . Given that health care costs are consuming a rising share of government budgets, understanding the true costs of rt, a procedure so common and so necessary in cancer treatment, is important for managing drivers related to cancer costs. Internationally, the operationalization, definition, and costs of rt show large variation, which emphasizes the importance of using rigorous evidence-based methods to develop an accurate representation of the cost of rt 2 . Methods for costing rt are inconsistent, making it difficult to compare and contrast rt costs and to determine their accuracy.
The objective of the present study was to conduct a systematic review of the literature to critically assess various rt costing methods for various cancer types. Specific costs and sources of costs were identified for each study, as were the costing methods used.

Database Search
A systematic review of the published literature identified studies assessing the costs of rt in any type of cancer ( Figure 1). A number of electronic databases were used: medline (resources from 1946 onward) and Ovid oldmedline (resources from 1946 to 1965) were searched for the combined terms "radiation," "neoplasm," "cost," and "cost analysis"; embase (resources from 1974 onward) was searched for the combined terms "radiotherapy," "neoplasm," and "cost effectiveness analysis" or "cost benefit analysis"; Ovid HealthStar (resources from 1966 onward) was searched for the combined terms "radiation," "neoplasm," "cost," and "cost analysis"; and a basic search of the EconLit database for "cost of radiotherapy" was also performed. All searches (excluding EconLit) were limited to studies with human subjects and were searched for the years from 2005 to 23 March 2015. Searches of Ovid medline, embase, EconLit, and Ovid HealthStar were limited to the English language.

Study Selection
Studies selected for inclusion had to have provided the cost of rt for any type of cancer. No geographic restrictions were used. The method used for determining costs had to be documented in enough detail to outline the resources for or the sources of the costs (or both) used to cost rt and could come from burden-of-illness studies or comparative analyses. Abstracts were further reviewed, and the studies were included if they investigated at least one of 3D-crt, imrt, sbrt, bt, or if they mentioned rt in general; studies were excluded if they assessed chemoradiation, radiosurgery, or Calypso 4D (Varian Medical Systems, Palo Alto, CA, U.S.A.). To remain current, studies published before 2005 were excluded. Review articles were also excluded, but were checked for relevant articles within the reference sections. Duplicates and studies that lacked relevant content were also excluded.

Data Extraction
The selected articles were thoroughly reviewed (FR, SJS, SYC), and any study-relevant methods-such as patient population, modality, costing method, cost of the various modalities of rt, costing source or sources, costing outcome measures, and year of costs-were extracted. The extracted information was amalgamated into a comprehensive table and critically analyzed for the purposes of the present study. Using the Consumer Price Index 6 , the cost per treatment for each rt modality was inflated to 2014 U.S. dollars for comparison purposes.

Study Characteristics
Ovid medline and Ovid oldmedline generated 268 results, embase produced 206 results, Ovid HealthStar produced 256 results, and EconLit generated 11 results. Of the 741 studies located, 304 were duplicates and were removed. Of the 437 remaining studies, 386 were removed because of lack of relevant content, and another 18 were removed because only abstracts were available ( Figure 1). Of the thirty-three original articles included in the analysis, twenty-seven had been conducted in the United States 7-33 (one of which used international data 26 ), four were from Canada 34-37 , and two were from the Netherlands 38,39 .

Methods in the Included Studies
The costing methods varied widely in the articles reviewed, largely because of the perspective from which the costs were reported (that is, reimbursed costs, charged costs, billed costs, and so on) and also because of the data sources and components included in determining the overall cost of rt.
Of the thirty-three studies considered, thirty (91%) used costing methods that took a health care perspective (that is, Medicare for U.S. studies, and provincial ministry of health for Canadian studies); the remaining three 9,25,26 took a societal perspective.
In the determination of rt costs, almost all studies (n = 32, 97%) included professional or physician fees in their cost analyses, which consisted of personnel such as physicians, radiation oncologists, physicists, and nurses. All studies included treatment and planning costs, per the objective of the studies. Only 33% of the studies incorporated institution or facility costs into their costing model, which most often included inpatient, outpatient, and technical costs of the hospital 7,16,18,22,24,26,[33][34][35][36][37] . Only five studies (15%) included equipment costs in their cost analyses (that is, computed tomography scanner and planning system, capital cost, specialized construction cost, and maintenance and operating costs of the radiation equipment) 22,24,34,35,37 . Most studies (70%) accounted for other costs (overhead, administration, and so on), which included minor equipment such as port films, immobilization devices, multileaf collimator, and other complex treatment devices; however, ten studies did not account for such items in their costing method 13,16,18,21,23,26,28,30,34,37 .
Based on modality type, cancer type, and costing components used, costs showed large variability between the studies. The cost for imrt ranged from US$2,687.87 to US$111,900.60 per treatment, followed by 3D-crt at US$5,583. 28 20 and Shah et al. 24 included institutional costs in addition to the hospital-specific reimbursement for rt treatment.

DISCUSSION
This literature review was able to retrieve thirty-three papers representing studies conducted over 10 years with the objective of costing rt in a number of cancer types. The results show that costing methods are vastly different across studies and countries, resulting in wide variations in cost estimates for similar treatments. Our findings demonstrate the need for consistent agreed-on costing methods for future economic studies of rt.
A study by Paravati et al. 40 identified sources of variation in rt costing for Medicare beneficiaries with cancer. Another study by Amin et al. 41 systematically reviewed the literature to identify articles that performed cost-effectiveness analysis of rt options for prostate cancer to identify the main cost drivers. Both studies also found large variations in the cost of rt between studies because of factors unrelated to the patient and the       various treatment-related morbidities; however, neither study performed a thorough investigation of differences in costing components and sources in rt.
To our knowledge, the present study is the first to review rt costing components and sources across all cancer types. It shows that most costing was based on inputs into hypothetical models from pre-existing original costing studies. Original data would allow for a more accurate representation of cost outcomes based on the cohort of interest and the cancer type, which might otherwise be subject to unreliable statistics when model inputs are used. Such inputs might differ based on institution, geography, and adopted care or clinical pathways, and thus original patient-level data would provide the most unbiased costing results.
In addition, although some societal costing was found, most studies were conducted from the health system perspective. The most consistent variables used in the costing analyses were the costs associated with treatment and planning, followed by professional or personnel and other fees. Costing studies rarely considered the costs of equipment and facility or institutional fees. Such omissions caused the final cost of rt treatment to appear inconsistent across studies, with large variability in costs being observed within and between rt modalities. The cost drivers therefore included the costs of the various personnel required during the course of rt and the actual costs of the delivery and planning of rt.
Notably, rt often requires the delivery of services by a variety of personnel (the physician, radiation therapist, medical oncologist, nurse, etc.) that were more often reflected in the Canadian than in the Dutch and U.S. studies. In addition, although all studies included treatment and planning costs, many did not identify the components that fell within the treatment and planning phase of rt; the reader is therefore unable to identify what the costs truly encompass. Both of the foregoing costing components are cost drivers in the overall cost of rt and thus should be considered for inclusion in future rt costing studies.
All in all, the inconsistencies identified here can lead to the drawing of incorrect and inappropriate conclusions about the cost of rt when the largest variability in costs can be attributed to the differences in rt components between studies. Our study's Figure 2 provides evidence of the wide variability in costs between studies, which might become more comparable if rt costing components were to be more inclusive, complete, and consistent from study to study. We recommend that rt costing studies aim to be as inclusive as possible in their costing methods. At a minimum, components should include detailed treatment costs, capital costs, operational costs (that is, equipment and overhead), detailed personnel costs, institutional or facility costs, and other costs (administration, etc.). To promote comparability between studies and an understanding of the cost drivers of rt, costing studies should be as transparent and comprehensive as possible.
Our study uncovered vast differences in rt costing components across studies, which draws attention to the fact that rt costing studies have room to improve and to be more inclusive in their costing components and methods. The limitations discussed and the variation in costing components between studies creates difficulty in comparing, contrasting, and understanding the true costs associated with rt. Even within countries, the heterogeneity between studies using the same health care perspective does not allow for ease of interpretation and application, oftentimes involving underestimations and overestimations in costs. Future research requires a more comprehensive costing analysis that encompasses as many elements of rt costing as possible for thorough inclusion and standardization. Such inclusivity will allow for efficient comparisons and informed evidence-based public health changes. More comprehensive costing is important for producing the good inputs required for policy decision-making and economic analyses.

CONCLUSIONS
The literature review presented here demonstrates that rt costing is diverse and complex between studies and especially between countries, which results in differing costing units and wide ranges in rt costs. The summarized findings provide insight into the costing frameworks and methods used by such studies and the accuracy and usefulness of those methods of rt costing. Based on the perspective used, the data available, the components used, and the aims of the study, rt can be costed in a variety of ways. Such variation makes understanding the true cost of rt at a per-patient or per-visit level quite difficult. Future research has to focus on using patient-level data and including as many of the cost drivers of rt as possible to arrive at a true cost. Given the increasing cost of health care delivery, it is necessary to understand the current financial burden and to pinpoint areas that require improvement to prevent negative effects on health care delivery and to support good management of the health care system.