Real-world cost analysis of chemotherapy for colorectal cancer in Japan: detailed costs of various regimens during the entire course of chemotherapy

We undertook a real-world cost analysis to compare the detailed costs of different chemotherapy regimens used for advanced colorectal cancer from first-line treatment to the end of chemotherapy in a practical setting. The analysis was retrospectively performed from a payer’s perspective. The costs of healthcare-related services were calculated from the payer side using billing data obtained from Showa University Hospital. Showa University Hospital is a general hospital located in Tokyo that has approximately 800 beds as well as outpatient clinics. The fees are the same for all payers and virtually all providers in Japan and are set by the government in the national fee schedule [20]. The study was approved by the Medical Ethics Committee of Showa University Hospital (approved number: 981).

We retrospectively collected clinical and claim data on patients with advanced colorectal cancer in this observational study. Because most of the patients died before this study began, it was difficult to obtain written consent from the patients. We analyzed the data after removing all personal information to protect privacy. This approach was in accordance with the Japanese Ethical Guidelines for Epidemiological Studies [21]; moreover, before the study began the Internal Review Board determined that consent is not required.

The study group comprised patients who were given a diagnosis of advanced colorectal cancer and received chemotherapy until the end of the final line of chemotherapy in the Department of Surgery, Showa University Hospital during the period from July 2004 through October 2010. We extracted data on all lines of chemotherapy given to all patients and classified the regimens into the following three groups: oxaliplatin-based regimens, irinotecan-based regimens, and regimens including molecular targeted agents.

Resource utilization data on each chemotherapy regimen was extracted from Showa University Hospital billing data. The following costs were calculated on a monthly basis and were summed up on the basis of the Japanese National Health Insurance fee-for-service system (reimbursement price) in 2010: costs related to outpatient visits (physician consultations and outpatient visits); costs for hospitalization (basic bed charges, medical examinations, nursing care, and basic treatments during hospitalization); costs for operations and procedures (including transfusions); costs for laboratory tests and diagnostic imaging tests (radiography, computed tomography, magnetic resonance imaging, ultrasonography); costs for drugs and administration (anticancer drugs, antiemetics, and other drugs, including preparation and administration costs). Costs unrelated to cancer, such as costs required for the management of hypertension and hyperlipidemia, and operation fees, such as those required for the creation of an artificial anus, were not considered. Extra bed fees and meal fees were collected from billing data as a reference.

In this study, we collected data on regimens regardless of the treatment line, and then found that the duration of each regimen was influenced by the treatment line. The duration of the first line of treatment was longer than those of second and third lines. Therefore, the costs of each chemotherapy regimen were compared on a monthly basis to avoid the effects of the different durations of treatment.

The monthly average costs of all chemotherapy regimens from first-line to the end of the final line of treatment were analyzed over the course of time. Treatments costs during the period of terminal care were not considered because it was difficult to collect relevant data during that period, and the costs varied depending on the status of individual patients during terminal care, regardless of the regimens used previously. Trends in the breakdown of costs were also analyzed in detail. Costs for more than 1 year were discounted by 3 %.

The uncertainty of the results was explored by sensitivity analyses of uncertain factors. Several qualitative, one-way sensitivity analyses were conducted to explore the impact of alternative parametric assumptions on the results.

The characteristics of the 30 patients identified from the patient database of Showa University Hospital are shown in Table 1. The average age was 62.8 years. The major sites of colorectal cancer were the cecum in 4 patients, the ascending colon in 3, the transverse colon in 2, the descending colon in 1, the sigmoid colon in 7, and the rectum in 13. The rate of hospitalization was 23 %. The average duration of the entire course of chemotherapy was 13.5 months. Among the 30 patients, 20 received up to second-line chemotherapy, and 8 received up to third-line chemotherapy. The following different regimens were identified from among all courses of treatment in all patients: 13 oxaliplatin-based regimens (12 FOLFOX regimens and 1 XELOX regimen); 31 irinotecan-based regimens (27 IRIS regimens and 4 FOLFIRI regimens); 11 regimens including molecular targeted agents (6 FOLFOX + Bev regimens, 4 FOLFIRI + Bev or Cet regimens and 1 XELOX + Bev regimen); and 3 other regimens (2 Bev regimens and 1 Cet regimen). The schedules of each regimen are presented in Additional file 1.

The numbers of regimens and the durations of each treatment line are shown in Table 2. Two oxaliplatin-based regimens (2 FOLFOX regimens), 27 irinotecan-based regimens (27 IRIS regimens), and 1 regimen including molecular targeted agents (1 FOLFIRI + Bev regimen) were given as first-line treatment. Eleven oxaliplatin-based regimens (10 FOLFOX regimens, 1 XELOX regimen), 8 regimens including molecular targeted agents (6 FOLFOX + Bev regimens, 1 FOLFIRI + Cet regimen, 1 XELOX + Bev regimen) and 1 Bev regimen were given as second-line treatment. Four irinotecan-based regimens (4 FOLFIRI regimens), 2 regimens including molecular targeted agents (1 FOLFIRI + Bev regimen, 1 FOLFIRI + Cet regimen), 1 Bev regimen and 1 Cet regimen were given as third-line treatment. The average duration of each line was 9.0 months for first-line, 3.9 months for second-line, and 3.8 months for third-line. The average duration of chemotherapy according to the type of regimen was 3.9 months for oxaliplatin-based regimens, 8.8 months for irinotecan-based regimens, and 4.3 months for regimens including molecular targeted agents.

The results of sensitivity analysis are shown in Table 5. As mentioned above, the treatment line differed among the regimens in this study. There was some bias, which most likely depended on the treatment line. Regimens used as first-line during the first month of treatment tended to be administered in the hospital, and the hospital fee accounted for a large proportion of total costs and might have influenced the results. We therefore initially performed a simple sensitivity analysis to examine the hospital fee. Two scenarios were considered: 3 days’ hospitalization and an outpatient setting. As a result, the orders of the total costs among the regimen groups given on an inpatient basis and an outpatient basis were basically same as the order of the total costs in the regimen groups for the base-case. As for oxaliplatin-based regimens and regimens including molecular targeted agents, the base-case costs were slightly higher than those of a 3-day hospitalization setting because the period of hospitalization was longer than 3 days for the base-case. The costs for diagnostic imaging tests were higher for irinotecan-based regimens than for the other regimen groups because IRIS, which was the most commonly used irinotecan-based regimen, was mainly used for first-line treatment, during which diagnostic imaging tests were frequently performed to assess disease progression. Therefore, we performed a simple sensitivity analysis of the costs for diagnostic imaging tests. The results demonstrated that the cost of diagnostic imaging tests did not have an impact on the total costs. In addition, because the use of generic anticancer drugs may have influenced the results, we performed a simple sensitivity analysis comparing the use of generic leucovorin with that of brand leucovorin. The results did not change.