Acta Oncologica
ISSN: 0284-186X (Print) 1651-226X (Online) Journal homepage: http://www.tandfonline.com/loi/ionc20
The Cost of Treating Pancreatic Cancer Jonas Hjelmgren, Jeanette Ceberg, Ulf Persson & Thor A. Alvegård To cite this article: Jonas Hjelmgren, Jeanette Ceberg, Ulf Persson & Thor A. Alvegård (2003) The Cost of Treating Pancreatic Cancer, Acta Oncologica, 42:3, 218-226, DOI: 10.1080/02841860310000386 To link to this article: http://dx.doi.org/10.1080/02841860310000386
Published online: 08 Jul 2009.
Submit your article to this journal
Article views: 927
View related articles
Citing articles: 2 View citing articles
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ionc20 Download by: [37.44.207.158]
Date: 26 January 2017, At: 12:38
ORIGINAL ARTICLE
æ
æ
The Cost of Treating Pancreatic Cancer A Cohort Study Based on Patients’ Records from Four Hospitals in Sweden Jonas Hjelmgren, Jeanette Ceberg, Ulf Persson and Thor A. Alvega˚rd From the Swedish Institute for Health Economics (IHE) (J. Hjelmgren, U. Persson) and the Department of Cancer Epidemiology, Lund University Hospital (J. Ceberg, T.A. Alvega˚rd), Lund, Sweden Correspondence to: Jonas Hjelmgren, The Swedish Institute for Health Economics, SE-220 02 Lund, Sweden. Tel: /46 46 329 106. Fax: /46 46 121 604. E-mail:
[email protected]
Acta Oncologica Vol. 42, No. 3, pp. 218 /226, 2003 An estimate of the average cost of treatment (COT) was assessed for 53 patients with pancreatic cancer treated between 1997 and 1999 in four hospitals in southern Sweden. Average COT was estimated to t18 947, 55% of which was attributable to hospitalization (including surgical procedures), 20% to long-term care and 11% to chemotherapy. Diagnostics and radiotherapy accounted for 9% and 4%, respectively. Median survival was 5.6 months (mean 6.3 months). Treatment costs per patient were negatively correlated with age but were higher for patients receiving chemo/radiotherapy and surgical treatment than for patients receiving only standard supportive care. Disease stage and type of hospital (university versus regional/local hospitals) were not significant predictors of COT per se. Assuming that our estimate of the average cost is representative for Sweden, the total healthcare cost for pancreatic cancer was t16 million ($14 million), i.e. about 2 /3% of the COT for all cancer diseases in Sweden. In the USA the cost of pancreatic cancer accounted for the same proportion. However, our estimated cost per patient was about half the amount of the US estimate. The distribution of costs between the different types of treatment services did not differ greatly between Sweden and the USA. Received 10 April 2002 Accepted 31 January 2003
Pancreatic cancer is one of the most serious forms of cancer and in general the prognosis for a newly diagnosed patient is poor. In Sweden, 12 men and 14 women per 100 000 will be diagnosed with pancreatic cancer each year (1). Incidence and mortality rates are virtually identical because median survival is somewhere between 3 and 6 months and 5-year survival after diagnosis less than 5% (2 /4). One reason for the poor prognosis is the late appearance of symptoms, which include jaundice, abdominal pain, unintentional weight loss and weakness. The treatment of pancreatic cancer varies depending on factors such as tumour size and location, lymph node involvement, vascular infiltration and distant metastases. Pancreatic cancer is classified into three stages: resectable, locally advanced and metastatic disease (5). For the 10 / 15% of patients with resectable disease, first-line treatment is radical surgery (6). The primary treatment modalities for patients with locally advanced and metastatic disease include palliative radio- and chemotherapy. Although tumours in patients with locally advanced and metastatic disease are not resectable, palliative surgery is offered occasionally. Healthcare resource utilization in connection with pancreatic cancer is poorly documented. To our knowledge, no # Taylor & Francis 2003. ISSN 0284-186X DOI: 10.1080/02841860310000386
Swedish data on healthcare consumption and its costs in connection with pancreatic cancer during the whole treatment process have been published. Besides increased knowledge of the overall cost level for patients at different stages of disease and in different demographic groups (sex and age), the distribution of costs for different healthcare services, and the evolvement of costs over time are of interest. The purpose of this study was (i) to estimate the average cost of treatment (COT) for patients with pancreatic cancer, generally and by initial disease stage, (ii) to determine which factors are correlated with COT and (iii) to analyse how costs are distributed between different healthcare services over time.
MATERIAL AND METHODS Cost of treatment analysis COT analyses incorporate all the costs of treating a cohort of patients with a particular disease borne by the healthcare sector. In this study, a longitudinal COT approach was applied, which means that individual patients are followed during the whole treatment process. By using individual Acta Oncologica
Acta Oncologica 42 (2003)
data such as patient records we were able to break down the average COT for the whole treatment period to average cost per patient-day for a cohort of patients. This is done by dividing the total cost per day by the number of patients surviving at any given time. The longitudinal approach enables us not only to identify cost-drivers but also the timing of costs in the treatment process. Many services in the treatment of pancreatic cancer are expensive and offer only modest benefits for a handful of patients. For example, the effectiveness of chemotherapy depends on when and to whom it is used. Consequently, the distribution of services and thus costs during the treatment process can serve as a guide to focus on clinical as well as economic research in the future. Patients Patients diagnosed with pancreatic cancer during 1997 and treated between 1997 and 2000 were identified from the Swedish Cancer Registry, which records all cases of cancer according to the international classification of diseases (ICD). There were 161 registered cases of cancer of the pancreas (ICD-10 code C25) in 1997 in the Southern Sweden Hospital Region (SSHR). Patient records from 4 hospitals were chosen: the university hospitals (UH) in Lund (n /19) and Malmo¨ (n/30) and the regional/local hospitals (RLH) in Helsingborg (n /11) and Kristianstad (n /8). We excluded the charts of patients who were alive at the time of the study (n /1), patients with multiple diagnoses (n /4), patients with incomplete chart information (n /7) and patients who were diagnosed post mortem (n /3), leaving a total of 53 individuals. Healthcare resource consumption Patient characteristics such as age, home district, date of diagnosis and disease stage (resectable, locally advanced or metastatic) and date of death were extracted from the patient charts. Detailed information about when a specific treatment service was carried out, as well as the intensity and frequency of the same service, was also extracted. The following information was of particular interest for our purpose: diagnostic procedures (ultrasound, ultrasound including fine-needle aspiration biopsy (FNA), computerized tomography (CT), percutaneous transhepatic cholangiography (PTC), abdominal x-ray, endoscopic retrograde cholangiography (ERCP), oesophagogastroduodenoscopy (OEGD), cytology, and pulmonary x-ray, ventricle passage), surgery, radio- and chemotherapy, pain relievers and utilization of anaesthetic care, admissions and discharges from hospital, hospice and patient hotel, and transportation (bus, own car or taxi) in connection with diagnostics, radioand chemotherapy, and hospitalization. Information on palliative nursing in the patient’s home performed by personnel employed by the municipal authorities or the hospital was obtained from the patient’s nursing
Cost of treating pancreatic cancer
219
chart. The nursing chart contained information about when (the date) the patient received basic palliative nursing and the mix of personnel at each visit. A nurse at the Southern Swedish Regional Tumour Registry (Lund University Hospital) reviewed the patients’ charts and nursing charts Information from patients’ charts was matched against a protocol specially designed to include both the timing and quantity of resource consumption.
Calculation of costs Healthcare services utilized by patient with pancreatic cancer were allocated to one of the following 7 categories: diagnostics, surgery, radio- and chemotherapy, analgesics, hospitalization and long-term care. Diagnostics. All procedures or services with the purpose of detecting or diagnosing the disease were included under diagnostics. The total costs were calculated as the sum of the quantities of each procedure multiplied by its respective unit cost. Surgery. The primary cost-driver for surgery was the number of minutes in surgery. Surgery was therefore estimated by multiplying the number of minutes by the cost per minute. Radiotherapy. Radiotherapy was performed intraoperatively or postoperatively. Intraoperative radiotherapy is a one-time event (rendering a per patient cost), whereas postoperative radiotherapy is carried out as a series of treatment episodes on each treatment occasion (rendering costs per treatment field). Patients receiving radiotherapy and whose homes were at some distance away could stay at the hotel for patients. The costs for accommodation at the patient hotel are included in the cost for radiotherapy. Chemotherapy. In addition to cytotoxic agents, we included antiemetics, i.e. pharmaceuticals to alleviate side effects (vomiting and nausea) from cytotoxic agents. Analgesics. Analgesics included prescribed analgesics such as paracetamol, dextropropoxyphene and morphine, and intensive anaesthetic care given at the hospital. Hospitalization. The cost of intensive hospital care includes a flexible per day cost and a fixed enrolment cost per admission. Both the flexible and the fixed enrolment costs depend on the ward in which the patient is treated. Long-term care. Long-term care could take place at a hospice, nursing home or in the patient’s home. The cost was calculated according to the cost per day and the length of stay when the patient was treated in the hospice or in a nursing home. When the patient was treated at home, the cost was estimated according to the mix of personnel at each home visit. There were four possible combinations of personnel: hospice team (consisting of physician and nurse), physician, nurse, and other personnel (e.g. welfare officer, occupational therapist, dietician). Irrespective of whether the patient was treated by a physician or a nurse employed
220
J. Hjelmgren et al.
by the county council (hospital-based home care) or by the municipality, the same unit price was applied. Since there were no standardized prices for house calls performed by other personnel, it was assumed that the cost was equivalent to two conventional outpatient visits to the caregiver in question (even if the duration of a house call is no longer than a conventional outpatient visit, some of the caregiver’s productive time is lost in travel). Physician and nurse visits. The cost of a physician or nurse visit depended on the type of clinic the patient visited. During treatment, outpatient visits to the physician or nurse were made at three types of clinics: the primary care clinic, the oncology clinic and the surgery clinic at three stages in the treatment process: diagnostics, treatment and follow-up. The costs of physician and nurse visits were allocated to diagnostics if the visits were undertaken before the date of diagnosis irrespective of the physician’s/nurse’s speciality, to radiotherapy if the visits were undertaken during treatment or follow-up by an oncologist, or to chemotherapy if the visit was undertaken during treatment or follow-up by a surgeon or oncologist. Unit costs for inpatient care and treatment services performed at the hospital were obtained from an official price list of regional hospital fees for southern Sweden (Table 1) (7). Capital costs, administrative services, nursing care, etc., are included in hospital fees and are assumed to represent a reasonable approximation of average cost per unit. Prices of pharmaceuticals were obtained from the Swedish FASS 2001 (8) and were costed according to the price per dose (mg). Since the intention here is not to show the distribution of costs for different parties (i.e. providers of healthcare, third-party payers and patients), patient fees are included in both hospital fees and pharmaceutical prices. Transportation costs. Travel costs in connection with diagnostics, radio- and chemotherapy and hospitalization were estimated according to mode of transportation and the distance from the patient’s home district to the hospital in Lund, Malmo¨, Helsingborg or Kristianstad. According to the county councils’ statistics (9), the three most common modes of transport were taxi (67% of all transportation in the southern hospital district; t15 per 10 km), car (15%; t1.75 per 10 km), and bus (18%; t1.2 per 10 km, according to the regional bus company). The cost of transportation was estimated by the weighted average of the share of all transports of each mode of transport and its price. If the patient resided in Lund, Malmo¨, Helsingborg or Kristianstad, it was assumed that the distance from the patient’s home to the hospital would be 10 km. The calculations were made according to the 2001price level in Euro (t) (the average exchange rate in 2001: t1 / $0.89, t1 /SEK 9.13 (Swedish krona), t1 /CAD 1.52 (Canadian dollar)).
Acta Oncologica 42 (2003)
Table 1 Selection of unit prices used in the calculation of costs Resource item Diagnostics Ultrasound including FNA Abdominal CT PTC OEGD ERCP (diagnostic/therapeutic stenting) Ventricle passage Surgery Anaesthesia minute Physician visit Radiotherapy Intraoperative Postoperative (per treatment field) Physician visit Chemotherapy Gemcitabine (per g) 5-FU (per g) Hospitalization Surgery ward (per day) Surgery ward (per admission) Oncology ward (per day) Oncology ward (per admission) Long-term care Hospice (per day) Nursing home Home care (per visit by hospice team) Home care (visit by nurse)
Price Euro (t)
191 168 966 71 341 268 12.6 99 1 550 49 144 177 5.5 275 198 383 200 630 146 278 160
Abbreviations: FNA/Fine needle aspiration biopsy; PTC/ percutaneous transhepatic cholangiography; OEGD /oesophagogastroduodenscophy; ERCP/endoscopic retrograde cholangiography.
Statistical analyses For each patient, treatment costs were correlated with patient demographics, disease stage, and treatment pattern. Of particular interest was the distribution and timing of costs with respect to different kinds of services. Survival analysis in the form of Kaplan /Meier survival curves and log-rank tests, two sample t -tests, partial correlation tests, and ordinary least square (OLS) multiple regression analysis were used for the statistical analyses. RESULTS Patient characteristics and healthcare utilization The patient characteristics and healthcare utilization are presented in Table 2. In total, the cohort of pancreatic patients consisted of 53 patients. Five patients (10%) had resectable disease, 24 (45%) had locally advanced and 24 (45%) had metastatic disease. The distribution of disease groups did not vary by age or sex. A majority of patients (38 out of 53: 72%) were treated at specialized university hospitals in either Lund or Malmo¨. There were no statistically significant differences with
Acta Oncologica 42 (2003)
Cost of treating pancreatic cancer
221
Table 2 Characteristics of the patient population and utilization of healthcare services for all patients and by disease stages Treatment
Patient demographics Age at diagnosis mean (min /max) years Sex (male/female) Treatment site (no. patients) Lund Malmo¨ Helsingborg Kristianstad Diagnostics Procedures (no. patients; events per patient) Ultrasound Ultrasound including FNA Abdominal CT PTC PTC-control Plain abdominal x-ray ERCP OEGD Cytology Pulmonary x-ray Ventricle passage Physician visits (no. patients; visits per patient) Primary care Oncology Surgery Surgery Surgical operations (no. patients; minutes per patient) PTC stent insertion (no. patients; procedures per patient) Physician visits (no. patients; visits per patient) Radiotherapy Intraoperative (no. patients) Postoperative (no. patients; treatment fields) Nurse visits (no. patients; visits per patient) Oncology Physician visits (no. patients; visits per patient) Oncology Patient hotel mean (min /max) days Chemotherapy Drug regimen (no. patients; mg per patient) No chemotherapy Only 5-FU Only Gemzitabine 5-FU/Gemzitabine Hospitalization Surgery ward Mean (min /max) days Mean (min /max) admissions Oncology ward Mean (min /max) days Mean (min /max) admissions Long-term care Hospice (no. patients) Mean (min /max) days Nursing home (no. patients) Mean (min /max) days Palliative home care (no. patients; no visits) Hospice team Only physician Only nurse Other personnel
All patients n /53
(A) Resectable (B) Locally advanced n/5 n/24
(C) Metastatic n /24
66.4 (41 /88) 26/27
64.2 (57 /72) 3/2
64.5 (41 /88) 10/14
67.7 (46 /84) 13/11
14 24 8 7
1 1 2 1
7 14 1 2
6 9 5 4
49; 1.0 33; 0.70 39; 1.0 7; 0.19 9; 0.41 12; 0.32 28; 0.71 19; 0.40 51; 1.5 44; 2.2 16; 0.51
3; 2; 5; 2; / 3; 4; 1; 5; 4; 2;
0.6 0.80 0.20 2.0 4.2 0.60
23; 1.9 14; 0.67 17; 1.1 3; 0.17 5; 0.46 2; 0.13 14; 0.92 12; 0.58 23; 1.7 20; 1.8 5; 0.42
23; 1.6 17; 0.75 16; 0.88 2; 0.17 4; 0.46 7; 10; 0.50 6; 0.25 23; 1.2 20; 2.1 9; 0.58
23; 0.72 1; 0.057 26; 0.89
2; 0.40 / 2; 0.20
10; 0.62 / 14; 1.08
11; 0.79 1; 0.13 10; 0.75
27; 164 3; 0.075 28; 1.4
5; 538 0; / 4; 2.2
12; 133 1; 0.042 12; 1.3
12; 123 2; 0.13 8; 1.3
4 9;
0 0
3 7; 17.2
1 2; 2.7
3; 0.19
0; /
3; 0.86
0; /
8; 0.49 0.27; 0 /14
0; / 0; /
7; 0.92 0.58; 0 /14
1; 0.17 0; /
32 4; 4292 12; 19 853 5; 4 602/14 627
5 0; / 0; / 0; /
10 4; 4 292 6; 19 309 4; 4 250/15 917
17 0; / 6; 20 403 1; 6 009/9 465
23.9 (0 /79) 2.5 (0 /7)
40.2 (23 /72) 3.6 (2 /7)
23.1 (0 /73) 2.8 (0 /6)
20.8 (0 /56) 2.4 (0 /7)
1.9 (0 /40) 0.22 (0 /5)
0; / 0; /
4.0 (0 /40) 0.38 (0 /5)
0.11 (0 /2) 0.083 (0 /1)
7 3.0 (0 /68) 6 2.5 (0 /90)
0 0; / 1 0.80 (0 /4)
4 5.4 (0 /68) 1 3.8 (0 /90)
3 1.3 (0 /14) 4 1.7 (0 /32)
2; 8; 7; 2;
1; 1; 1; 0;
0; 4; 3; 0;
1; 3; 3; 2;
0.11 0.45 8.4 0.42
1.0 0.40 1.2 0.4
0.40 0.20 34.4 /
/ 0.67 6.1 /
0.17 0.45 5.2 0.92
Abbreviations: FNA/Fine needle aspiration biopsy; PTC/percutaneous transhepatic cholangiography; OEGD/oesophagogastroduodenscophy; ERCP /endoscopic retrograde cholangiography.
222
J. Hjelmgren et al.
respect to disease stage between treatment sites, though a larger proportion of patients (3 out of 15: 20%) in Helsingborg and Kristianstad were resectable compared with 2 out of 38 (5%) in Lund and Malmo¨. Resectable patients received more surgery (anaesthesia minutes) than patients with locally advanced and metastatic disease. All but one patient receiving intraoperative and postoperative radiotherapy lived in municipalities near to Lund. Intraoperative and postoperative radiotherapy as well as healthcare services related to radiotherapy, i.e. physician and nurse visits, was almost exclusively directed towards locally advanced patients (differences between disease groups were statistically significant). All patients receiving chemotherapy were treated in either Lund or Malmo¨ (data not shown). Thus, nearly half (47%) of the patients not receiving chemotherapy were treated in Helsingborg and Kristianstad. Patients were mainly hospitalized in surgery wards rather than in oncology wards. The average length of stay in the surgery ward and oncology ward was 9.4 and 0.97 days, respectively (data not shown). Survival rates Resectable patients lived longer on average than patients with locally advanced and metastatic disease, by 2.4 and 5.2 months, respectively, and patients with locally advanced disease lived an average of 2.8 months longer than patients with metastatic disease. The survival curves in Fig. 1 illustrate the rate at which patients die, by initial disease stage. The curve for patients with metastatic disease is steeper than the curves for patients with resectable and locally advanced disease, indicating a faster pace of death. Significant differences in survival rates between resectable patients and the other disease groups were not detectable by the statistical log-rank test (p/0.38 between patients with resectable and locally advanced disease; p/0.095 between those with resectable and metastatic disease). There was, however, a significant difference in survival between patients with locally advanced and metastatic disease (p / 0.022). Univariate analyses of treatment costs Total treatment costs by treatment services and disease stage. The total average treatment cost for pancreatic cancer was t18 947 (Table 3), of which t10 369 (55% of total treatment cost) was attributable to hospitalization, including surgical procedures and t3 958 (20%) was attributable to long-term care, either in a hospice/nursing homes or as palliative care in the patient’s own home. Chemotherapy treatment, including the cytotoxic agents, outpatient visits, transportation and antiemetics, amounted to t2 106 (11%) and the remaining costs were associated with
Acta Oncologica 42 (2003)
Fig. 1. Survival curves by disease stage for patients with pancreatic cancer.
Mean (min /max) All patients Resectable Locally advanced Metastatic
Median 6.3 (0.4 /22.3) 9.7 (4.6 /22.3) 7.3 (0.9 /17.8) 4.5 (0.4 /13.3)
5.6 6.1 6.0 3.8
diagnosis and radiotherapy, t1 545 (9%) and t743 (4%), respectively. Although none of the resectable patients received radiotherapy or chemotherapy, they still cost t4 363 (p /0.61) and t12 519 (p /0.16) more on average than patients with locally advanced and metastatic disease, respectively. Patients with locally advanced disease had t8 156 higher treatment costs than those with metastatic disease (p / 0.07). Multivariate analyses of treatment costs Treatment patterns and patient characteristics. The patients could be divided into five groups according to their pattern of treatment: surgery, chemotherapy and radiotherapy (SCR) (15% of all patients); surgery and chemotherapy (SC) (13%); surgery only (S) (28%); chemotherapy only (C) (13%); and standard supportive care only (SSC) (28%). Correlation tests were carried out to determine whether there was a systematic relationship between treatment patterns and patient characteristics such as age and gender, treatment site and disease stage. The correlation coefficients are not shown in Table 4 but mean values of the subgroups that are smaller than the mean value of all patients indicate a negative correlation and vice versa. The test showed that there was a negative relationship between age and SCR (p /0.02) and a positive relationship between age and SSC (p /0.001), indicating that younger patients were more likely to receive SCR and older patients SSC. Treatment
Acta Oncologica 42 (2003)
Cost of treating pancreatic cancer
223
Table 3 Total average treatment cost in Euro (j) by treatment services and disease stage Treatment service
Whole sample n /53
(A) Resectable n/5
(B) Locally advanced n /24
(C) Metastatic n /24
p-values1
Diagnostics Procedures Physician visit Transportation Surgery Anaesthesia minutes PTC stent insertion Physician visits Radiotherapy Intraoperative Postoperative Physician visits Nurse visits Patient hotel Transportation Chemotherapy Cytotoxic-agents Antiemetics Physician visits Nurse visits Transportation Analgesics Anaesthetic care unit Pharmaceuticals Hospitalization Surgery ward Oncology ward Transportation Long-term care Hospice Nursing home Palliative home care Average total treatment cost
1 545 1 330 147 68 2 388 2 096 154 138 743 117 446 71 11 21 77 2 106 1 112 252 575 35 132 226 31 195 7 981 7 116 757 108 3 958 1 903 494 1 561 18 947
1 776 1 624 116 36 6 996 6 778 0 218 0 0 0 0 0 0 0 0 0 0 0 0 0 177 72 105 11 643 11 523 0 120 5 800 0 117 5 683 26 392
1 551 1 323 152 76 1 884 1 671 85 128 1 408 194 850 133 24 45 162 2 959 1 478 355 921 39 166 285 31 254 8 837 7 097 1 607 133 5 105 3 388 549 1 168 22 029
1 490 1 270 148 72 1 933 1 545 256 132 232 65 136 24 0 0 7 1 632 976 184 331 40 101 160 22 138 6 362 6 217 65 80 2 064 814 244 1 006 13 873
ns
p/0.01***
p/0.006*, p/0.028***
p/0.005*, p/0.02**, p /0.08****
n.s.
p/0.10****
p/0.082**
p/0.072****
1
*A /B; **A /C; ***A /B and A /C; ****B /C; *****A /B and A /C and B /C. Abbreviation: PTC/percutaneous transhepatic cholangiography.
praxis between the UH and the RLH also differed; patients treated in the RLH received surgery as the only mode of treatment, whereas patients in the UH received a combination of surgery and chemotherapy. There were also differences between genders; men received only surgery to a larger extent than women (p /0.002). Disease stage was related to treatment patterns; resectable patients received surgery only (p /0.001) and a greater fraction of patients with locally advanced disease received SRC (p /0.02). Multivariate analysis of total treatment cost In an initial OLS regression, age, treatment site (UH versus RLH), gender, disease stage (resectable versus locally advanced and metastatic), treatment pattern (SSB versus SCR, SC, S, and C) and survival were included as covariates. It was shown that treatment site, gender, and disease stage were not significant predictors of total
treatment costs (all p/0.28). We therefore revised the model to include only the significant predictors from the initial model. The result from the regression shows that COT decreases with age (p /0.016) and increases with survival (p /0.022) (Table 5). This implies that older patients are less costly (other things being equal) than younger patients and patients with longer survival are more costly than patients with shorter survival. The model also shows that patients who receive only SSC are less costly than patients in the one of the other treatment groups (p / 0.002). Longitudinal cost analyses Average cost per patient and day over the treatment time. The average daily costs per surviving patient distributed on categories of healthcare services for months 1 /18 are illustrated in Fig. 2. The treatment cost per day and
224
J. Hjelmgren et al.
Acta Oncologica 42 (2003)
Table 4 Relations between treatment patterns and age, treatment site and disease stage1
Age (mean) Sex (male/female) RLH (fraction of patients) Resectable (fraction of patients) Locally adv. (fraction of patients) Metastatic (fraction of patients)
SRC n /8
SC n/6
S n /15
C n/8
SSC n /15
All patients
57.8** 0.375 0.125 0 0.875** 0.125
59.5 0.33 0* 0 0.5 0.5
63.8 0.80*** 0.53*** 0.33*** 0.13** 0.53
67.0 0.25 0** 0 0.62 0.38
73.5*** 0.47 0.4 0* 0.47 0.53
66.4 0.49 0.283 0.094 0.453 0.453
1 Partial correlation test (subgroup values smaller (larger) than the total group values indicate negative (positive) correlations): ***Correlation significant at 0.01 level. **Correlation significant at 0.05 level. *Correlation significant at 0.10 level. Abbreviations: SRC /Surgery/radiotherapy/chemotherapy; SC/surgery/chemotherapy; S /surgery; C/chemotherapy; SSC/standard supportive care; RLH/regional/local hospital.
patient during the first month for patients with resectable disease, t405 per day and patient (not illustrated in the figure), was more than twice as high for patients with locally advanced and metastatic disease, t197 and t153, respectively. However, cost differences between the disease groups tend to level out after the first two months. A general pattern for all patients is that the costs for hospitalization (varying between t0 and t106 per day and patient during the treatment period) but also for long-term care (between t0 and t92) represent a large proportion of the treatment costs throughout the whole treatment time. The costs for radiotherapy (between t0 and t9) and chemotherapy (between t0 and t16) evolve from month 2 and onwards for patients with locally advanced and metastatic disease.
A large fraction of the treatment costs is concentrated in the first month; 36% of the COT for a patient with a survival of 6 months was incurred in the first month, of which 78% was attributable to hospitalizations and surgery. The average COT per patient and day during the first year and the whole treatment period (18 months) was t86 and t84, respectively. This means that the average costs of treating a patient with pancreatic cancer over one year and over 18 months are about t31 000 and t45 000, respectively. DISCUSSION We have estimated the average healthcare costs for treating 53 patients with pancreatic cancer in southern Sweden to be t19 000 (about $17 000). Approximately 55% of the total
Table 5 Regression analysis of total treatment costs in Euro (j) Variable
Intercept Treatment pattern2 SSC Demographics Age Survival3 Total survival
Regression coefficient (eb)1
95% CI
p-value
10.868
9.657 /12.079
0.000
/0.728
/1.185 //0.27
0.002
/0.02238
/0.04 //0.004
0.016
0.0001 /0.003
0.022
0.001581
1
The dependent variable Y/total treatment costs (Diagnostics/ Surgery/Radiotherapy/Chemotherapy/Hospitalization/Longterm care) in Euro (t). Owing to log transformation of Y, we estimated the following additative expression: ln Y/b0/b1X1/ b2X2/ /bkXk, where bi is the i:th estimated coefficient and Xi is the i:th variable. The original values of the coefficient are obtained by transformation; Y/exp(b0/b1X1/b2X2/ /bkXk). 2 The variable is expressed as a dummy: patient who received standard supportive care (SSC)/1; patients who received surgery (S), chemotherapy (C), surgery/chemotherapy (SC), and surgery/chemotherapy/radiotherapy (SCR)/0. 3 Survival expressed in total number of days from the first contact with healthcare to the date of death.
Fig. 2. Average treatment cost per day and patient in Euro (t) from first procedure and onwards. The treatment period in the figure is censored. Two patients had longer survival rates than 18 month (18.5 and 22.3 month).
Acta Oncologica 42 (2003)
healthcare costs are attributable to hospitalization, including surgical procedures. Another 20% is taken up by the cost of long-term care, either in a hospice/nursing homes or as palliative care in the patient’s own home. Chemotherapy treatment, including the chemotherapeutic agent, outpatient visits, transportation and antiemetics account for 11% of the total treatment costs. The remaining costs are associated with the diagnosis and radiotherapy, 9% and 4%, respectively. The average survival (from diagnosis) was estimated to be 6.2 months (median 5.6 months). A large fraction of the treatment costs were concentrated to the first month. Our results indicate that the treatment of pancreatic cancer varies between individual patients, making it difficult to generalize about the driving forces behind COT. Differences in treatment praxis between treatment sites and the way hospitals prioritize between groups of patients can only to some extent predict the overall cost level. Patients treated at the UH were more likely to receive high-cost interventions such as chemo- and radiotherapy than patients treated at the RLH. Treatment site per se was not, however, a driving force for COT. Age, treatment pattern, and survival were the three main factors driving costs. Age was correlated with both treatment pattern and COT. Younger patients were more likely to receive radiotherapy, chemotherapy and surgery and older patients were more likely to receive standard supportive care. This implies that providers of healthcare have chosen to allocate resources towards patients who might benefit most from radio- and chemotherapy, often their younger patients. To some extent these differences could also be explained by differences in patient preferences with respect to treatment strategies. Since resectable patients on average survived longer after diagnosis, the relationship behind survival and disease stage is interesting and poses an interesting question: Does longer survival explain the higher COT or is it a question of being at a particular disease stage? Our results suggest that survival but not disease stage was a predictive factor of COT. However, our data included few resectable patients. A small sample size could lead us to accept the hypothesis that costs are equal between categories of patients even though differences really exist. This study shows that a large fraction of the average monthly treatment costs is incurred during the first month. The main driving factors behind treatment costs during this period are diagnostics, hospitalizations and surgery. After the first month there is a period where the average monthly costs diminish followed by a costly terminal period before death. This implies that the average monthly cost during the whole treatment period takes on a ‘U-shaped’ curvature. The initial costs are explained by surgery and subsequent hospitalisations, whereas the higher costs before death are incurred mainly as a result of palliative care.
Cost of treating pancreatic cancer
225
An interesting aspect of chemotherapy treatment, apart from its costs, is whether it improves quality of life (QoL). Improvements in QoL could justify more frequent use. Owing to the lack of convincing evidence that a large proportion of pancreatic cancer patients will benefit from palliative chemotherapy, The Swedish Council of Technology Assessment in Health Care (SBU) does not recommend the routine use of chemotherapy (10). Instead, the SBU argues for selective use of chemotherapy since there is evidence that between 20% and 35% of the patients will derive a short-lived QoL benefit. Thus from the perspective of the healthcare sector, the use of chemotherapy should not generally be recommended in patients with pancreatic cancer. Most of the earlier research on the economics of pancreatic cancer estimated only fractions of the treatment costs, e.g. for surgery (11, 12) or chemotherapy (13, 14), and these estimates are not directly comparable with our study. One study conducted in the US estimated the total healthcare cost for pancreatic cancer in the US to $881 million. With an annual incidence of 29 000 the estimated cost per patient in the USA was $30 000 in 1996 prices (about t34 000) (15), about twice as high as our estimate. Another US study estimated the COT to $48 803 (about t55 000) (16). This estimate is about three times greater than our estimate, although costs for long-term care were not included. However, this study only included patients from a large-volume urban referral centre, which focused on clinical research programmes and may not be representative for the US. A comparison between our estimate and the first US study (15) shows a number of similarities. For instance, the fraction of the costs attributable to hospitalization (including room, board and medication) was 44% in Sweden and 51% in the USA. The share of the costs for surgery (procedure costs) was 12% and 7% in Sweden and USA, respectively. The share of the costs for radiotherapy (procedure costs) was 3% in Sweden and 7% in the USA, respectively. The share of prescription medicines including cytotoxic agents, antiemetics and analgesics, and the costs related to long-term care including hospice and home care were 8% and 21% in Sweden, respectively. The corresponding figures for the USA were 6% and 14%. According to the US study and statistics in the USA, pancreatic cancer accounted for about 2 /3% of the COT for all cancer diseases in 1996 (17). Based on our cost estimates, an incidence of 800 /900 patients (2% of cancer patients) (1) and a Swedish cancer study (18), the corresponding figure for Sweden was of the same level. Another benchmark can be obtained by comparing our cost estimates with other cancer diseases. Non-small lung cancer is a cancer disease with a similar survival profile as pancreatic cancer. A Canadian study estimated the COT of non-small lung cancer to CAD 30 000 in 1993 (about
226
J. Hjelmgren et al.
t22 000) (19). This figure is somewhat higher than our estimate but is nevertheless of the same magnitude. We obtained data on treatment pattern from patients’ records. Because use of hospital resources, outpatient care and palliative home care are local, our cost estimates (obtained from 4 hospitals out of a total of about 80 hospitals in Sweden) may not be generalizable to other settings. However, our sample included patients treated at two specialized university hospitals, at one regional hospital and at one local hospital. Patients were also living in many different municipalities responsible for providing palliative home care and nursing home care. Another limitation is that our data consisted of relatively few patients. Consequently, the estimates of costs may be more sensitive to extreme values; patients with exceptionally high or low treatment costs can have a disproportionate impact on the results. A small sample size also reduces the power of statistical tests. The lack of unit prices for healthcare performed in the patients’ home is a problem when estimating the cost of palliative home care. We used the prices for hospital-based home care irrespective of whether the care was performed by personnel (physician or nurse) employed by the municipality or by the hospital. Since hospital-based health care is more specialized and more costly, we may have overestimated the cost for palliative home care. This study shows that although the survival duration for pancreatic cancer is short, it poses an economic burden for the healthcare system in Sweden. The COT for older patients is lower than that for younger patients, and patients with longer survival generate more costs. Patients receiving only standard supportive care are less costly than patients receiving curative treatment. A large fraction of the treatment costs are generated during the first month, most of which are attributable to hospitalization and surgery. Long-term care becomes an increasingly important item as treatment progresses. ACKNOWLEDGEMENTS The Janssen Research Foundation, Belgium, sponsored this study.
REFERENCES 1. The National Board of Health and Welfare. Cancer incidence in Sweden 1999. Stockholm: The National Board of Health and Welfare, 2001. 2. Rothenberg ML, Moore MJ, Cripps MC, et al. A phase II trial of gemcitabine in patients with 5-FU-refractory pancreatic cancer. Ann Oncol 1996; 7: 347 /53.
Acta Oncologica 42 (2003) 3. Williamson RCN. Pancreatic cancer: the greatest oncological challenge. Br Med J 1991; 296: 445. 4. Gold EB, Goldin SB. Epidemiology of and risk factors for pancreatic cancer. Surg Oncol Clin North Am 1998; 7: 67. 5. Evans D, Abbruzzese J, Willert C. Cancer of the pancreas. In: DeVita V, Hellman S, Rosenberg S, eds. Cancer: practice of oncology, 6th ed. Philadelphia: Lippincott-Raven, 2001: p. 1126 /61. 6. So¨dra sjukva˚rdsregionen. Treatment guidelines for pancreatic cancer 1999 (in Swedish) Va˚rdprogram fo¨r pankreascancer 1999. Lund: Onkologiskt Centrum, So¨dra sjukva˚rdsregionen; 1999. Available from http://www.ocsyd.lu.se/Vardprog/pancreas.pdf. 7. So¨dra Regionva˚rdsna¨mnden. Regional prices and reimbursements for the Southern Sweden Hospital Region 2001 (in Swedish) Regionala priser och ersa¨ttningar 2001 fo¨r so¨dra sjukva˚rdsregionen. Lund: So¨dra Regionva˚rdsna¨mnden, 2000. 8. FASS */pharmaceuticals in Sweden 2001 (in Swedish) FASS / la¨kemedel i Sverige 2001. Stockholm: LINFO; 2001. 9. Landstingsfo¨rbundet. Available from http://www.lf.se/nyheter/ download/sjukre98.xls. 10. Permert J, Hafstro¨m L, Nygren P, et al. A systematic overview of chemotherapy effects in pancreatic cancer. Acta Oncol 2001; 40: 361 /70. 11. Holbrook RF, Hargrave K, Traverso W. A prospective cost analysis of pancreatoduodenectomy. Am J Surg 1996; 171: 508 /11. 12. Raikar GV, Melin MM, Lettieri SZ, et al. Cost-effective analysis of surgical palliation versus endoscopic stenting in the management of unresectable pancreatic cancer. Ann Surg Oncology 1996; 3: 470 /5. 13. Ragnarson-Tennvall G, Wilking N. Treatment of locally advanced Pancreatic carcinoma in Sweden. A health economic comparison of palliative treatment with best supportive care versus palliative treatment with gemcitabine in combination with best supportive care. Pharmacoeconomics 1999; 15: 377 / 84. 14. Glimelius B, Hoffman K, Graf W, et al. Cost-effectiveness of palliative chemotherapy in advanced gastrointestinal cancer. Ann Oncol 1995; 6: 267 /74. 15. Wilson LS, Lightwood JM. Pancreatic cancer: total costs and utilisation of health services. J Surg Oncol 1999; 71: 171 /81. 16. Du W, Touchette D, Vaitkevicius VK, et al. Cost analysis of pancreatic carcinoma treatment. Cancer 2000; 89: 1917 /24. 17. Brown ML, Riley GF, Schussler N, et al. Estimating health care costs related to cancer treatment from SEER-Medicare data, Med Care 2002; 40: 8 Suppl: 104 /117. 18. Tennvall G, Karlsson G. Cancer treatment in Sweden */costs of drugs, inpatient and outpatient care from 1985 to 1996 and cost effectiveness of new drugs. Acta Oncol 1998; 37: 447 /53. 19. Houle C, Will BP, Bertelot JM, Evans WK. Use of POHEM to estimate direct medical costs of current practice and new treatments associated with lung cancer in Canada. Analytic Studies Branch Research Papers Series No. 99. Statistics Canada, 1997.
