and bone marrow - Nature

Bone Marrow Transplantation, (1997) 20, 975–982  1997 Stockton Press All rights reserved 0268–3369/97 $12.00

Cost comparative study of autologous peripheral blood progenitor cells (PBPC) and bone marrow (ABM) transplantations for nonHodgkin’s lymphoma patients MC Woronoff-Lemsi1, P Arveux2, S Limat1, E Deconinck3, P Morel4 and JY Cahn3 1

Department of Pharmacy, 2Doubs Cancer Registry, 3 Department of Hematology, Hopital Jean Minjoz, CHU Besançon; and 4Blood Bank, Besançon, France

Summary: Intensive high-dose chemotherapy with autologous stem-cell support has become a common treatment strategy for non-Hodgkin’s lymphomas. A cost-identification analysis was conducted comparing 10 patients autografted with PBSC to 10 others autografted with BM. The analysis included harvest and graft until graft day +100 and was carried out from the point of view of the hospital setting. Resources used, logistic and direct medical costs per patient were identified, and sensitivity analyses performed. The cost distribution was different. Stem cell harvest was more expensive for PBPC ($9030) and BM ($4745); on the other hand, hospitalization from graft to discharge from hospital cost savings with PBSC were about $10 666. After discharge from hospital, costs were similar and cheaper in both groups. For the overall study the PBPC procedure was less expensive than ABMT, $35 381 and $41 759 respectively, with cost savings of $6378. The number of days spent in hospital and blood bank costs were the major cost factors. This study was based on a single pathology, non-Hodgkin’s lymphoma, and the actual hospital records for each patient situation as opposed to a clinical trial, and our results were consistent with different previous studies carried out in different health care systems. Keywords: lymphomas; cost analysis; ABMT; PBSC; hematopoietic growth factor

Intensive high-dose chemotherapy (HDC) with autologous peripheral blood progenitor cells (PBPC) and bone marrow (BM) support is a common treatment strategy for intermediate- or high-grade non-Hodgkin’s lymphoma (NHL) in first remission or after relapse.1,2 Hematopietic growth factors (HGF) such as G-CSF or GM-CSF facilitate PBPC mobilization and accelerate engraftment after transplantation. To date, PBPC transplant has been shown to be as safe and effective as ABMT for Hodgkin disease (HD) and

Correspondence: M-C Woronoff-Lemsi, Department of Pharmacy, Hopital Jean Minjoz, CHU Besançon, 25 030 Besançon, France The results of this study were presented in part at the American Society of Hematology Meeting (Seattle, December 1995) Received 17 March 1997; accepted 14 July 1997

NHL, representing significant cost savings in several studies.3 PBPC involves no anesthesia risk compared with ABMT which requires hospitalization and anesthesia. In recent years, PBPC has become the standard care in France over ABMT. The French National Health Insurance is concerned, as in many countries, by the increasing high costs of health care. Hematological departments are commonly the most expensive wards in French university medical school hospitals. Studies comparing cost treatment in patients with NHL have been published from different health care systems,3–5 including one in France by a private Cancer Center.6 The aim of this study was to carry out a cost-identification analysis comparing PBPC and BM autologous transplantations in NHL for patients treated with successive protocols. A retrospective economic analysis was carried out in the Besançon university medical school hospital. Ten patients autografted with BM were compared with 10 others autografted with HGF primed PBPC. The economic assessment was based on a cost analysis using the hospital perspective.

Materials and methods Patients Consecutive patients treated in the university of FrancheComte´ medical school hematology unit with HDC for NHL between 1992 and 1994 were identified. Every patient who underwent high-dose chemotherapy and autologous bone marrow or peripheral stem cell transplantation was included. Analysis was performed for patients for whom data were available until graft day +100 post-graft. All patients included were treated in first line, except one patient with a delayed relapse. Ten patients were autografted with unpurged marrow between 1992 and the end of 1993. Subsequently, 10 patients received PBSC transplantation between the end of 1993 and July 1994. Patient characteristics are summarized in Table 1. Clinical data All patients had received previous chemotherapy regimens before high-dose chemotherapy; eight patients were treated chronologically according to the LNH 87 protocol2 and 12

Cost study of PBPC and ABMT for NHL patients MC Woronoff-Lemsi et al

976

Table 1

Patient characteristics and stem cell harvest outcomes BMT group

PBPC group

No. of patients

10

10

Mean age years (range)

37.5 (26–54)

43.2 (25–60)

Disease status at graft CR1 PR Conditioning regimens

8 2 5 BEAM/4 CBV/1 HDM + TBI

7 3 9 BEAM/1 TAM 12

Interval in months between diagnosis and graft in CR1 patients, mean (range)

6.1 (4–10)

6 (4–10)

Mean length of hospitalization, days (range)

3.3 (3–4)

3.0 (2–5)

Previous first line Goelam 07 LNH 87

4 (CEEP + MINE) 6 ACBV

8 (CEEP + MINE) 1 ACVB, 1 VCAP

Second lines (salvage) Mean No. of HGF days (range) HGF (5 mg/kg/day)

3 (MIV) NA

Mean leukaphereses (range)

NA

4 (MIV or DHAP) 12.9 (7–20) 8 FIL, 2 MOL 3 (2–5)

CR1 = first complete response; PR = partial response; BEAM = BCNU: 300 mg/m2 day 1, cytosine-arabinoside: 400 mg/m2 continuous infusion days 2–5, etoposide: 400 mg/m2 days 2–5, melphalan: 140 mg/m2 day 6; CBV = cyclophosphamide: 1500 mg/m 2 days 2–5, BCNU: 300 mg/m2 day 1, etoposide: 200 mg/m2 days 2–5; HDM = high-dose melphalan: 140 mg/m2; TBI = total body irradiation; 12 Gy, 6 fractions, 3 days; TAM 12 = total body irradiation: 12 GY, 6 fractions, 3 days, cytosine-arabinoside: 12 g/m2 , melphalan: 140 mg/m2 ; ACVBP = Adriamycin: 75 mg/m 2 day 1, cyclosphosphamide: 1200 mg/m2 day 1, vindesine: 2 mg/m2 days 1 and 5, bleomycin: 10 mg/m2 days 1 and 5, methylprednisolone: 60 mg/m 2 days 1–5; CEEP = cyclophosphamide: 1200 mg/m2 day 1, epirubicin: 100 mg/m 2 day 2, eldisine: 3 mg/m2 days 1 and 3, methylprednisolone: 80 mg/m2 days 1–5; MINE = methotrexate: 100 mg/m2 day 1, ifosfamide: 1.5 g/m2 days 1–3, novantrone: 10 mg/m2 days 1–2, etoposide VP16:100 mg/m2 days 1–3; MIV = mitoxantrone: 10 mg/m2 day 1, ifosfamide: 1.5 g/m2 days 1–3, VP16: 150 mg/m2 days 1–3; DHAP = dexamethasone: 40 mg/day days 1–4, cytosine arabinoside: 2 g/m2 days 1 and 2, cis-platinum: 100 mg/m2 day 1; HGF = hematopoietic growth factor; FIL = filgrastim; MOL = molgramostim; NA = non-applicable.

according to the GOELAM 07 protocol. 7 Fifteen were intensified in complete response (BMT, eight; PBPCT, seven) and five in partial remission (PR). The mean period between diagnosis and graft was around 6 months, except for one patient who relapsed after 5 years and had a subsequent second complete response. PBPC collections were consecutive after mobilization by chemotherapy and HGF (8 filgrastim, 2 molgrastim). In the PBPC group, HGF was initiated at the end of mobilization chemotherapy and maintained daily until the end of cytapheresis procedures. Four patients had two leukapheresis procedures performed, three patients needed three and one patient four procedures. Two patients failed at the first procedure and needed another different cycle of HGF stimulation for priming PBPC. Bone marrow harvests were consecutive after cycles of chemotherapy, according to the LNH 87 (n = 6) and the GOELAM 07 protocol (n = 4) (Table 1). Supportive care During transplant and post-transplant periods, all patients were hospitalized in an intensive care hematological unit. Platelet transfusion units were given when the platelet count was less than 20 × 109/l or in cases of bleeding with a low platelet count. Packed red blood cells were transfused to maintain the hemoglobin level over 8 g/dl. All blood components were leukocyte-reduced and irradiated and red blood cells were phenotyped. Broad-spectrum i.v. antibiotics were given in cases of temperature >38°C and

modified according to standard practice for neutropenia. Patients were treated in single rooms and received oral decontamination with non-absorbable antibiotics (gentamicin) and amphothericin B. No growth factor was systematically prescribed post-transplantation. Patients were discharged when the absolute neutrophil count reached .0.5 × 109/l, if they were without fever or infections. After discharge from hospital, patients were either rehospitalized (one patient) or followed in outpatients, on the day ward or both. Economic assessment Data collection concerned the harvest period and the period from HDC until discharge from hospital, and finally the post-graft period until graft day +100. Since January 1983, French public hospitals have been managed on the basis of a global budget allocated by the National Health Insurance every year. However, an analytic accounting system still has to be developed. Therefore, costing units of health care utilization in the hospital setting in France relies on a mix of the per diem method and standard cost for specific services and procedures.8 Resources used and direct medical costs were identified for each patient. All treatment consequences during study periods were accounted for. The components of care that were not affected by the autograft within 1 day of hospitalization (eg hotel and basic care) and those that were likely to be affected by interventions such as drugs and transfusion were identified. The complete hospital structure


cost (eg overheads, financial expenses), fixed medical or logistic costs and variable logistic cost (eg food, laundry) were part of the analytic account. Medical costs were identified from clinical records observed and computer systems and included laboratory tests, transfusions and drugs. This analysis was performed from the point of view of the Hospital Institution. Monetary values for 1994 French prices were used for all components. The exchange rate used was 1 US$ = 5 French francs (FRF). Medical costs (medical resource utilization and costs) (Table 2) The following broad categories of service were considered when assessing resources used: • The room and overhead analytic accounting system of the ‘Assistance Publique-Hoˆpitaux de Paris’ (AP-HP), which was the only one available in France, obtained from 50 hospitals, was used.9 These changes included non-medical and medical costs such as medical and nonmedical personnel, overheads, hotel, small equipment (the costs of devices is included in small equipment in the analytic accounting system), laundry, food, devices except the following medical costs which were obtained from clinical record data (drugs, blood products, laboratory tests). This avoids double accountancy. • The cost of leukapheresis for PBPC collection and BM harvest was obtained from the Blood Bank tariff and included harvest, processing and cell conservation, then thawing the cells to graft. • Drugs–chemotherapy (used for conditioning regimens), anti-infectious agents (to prevent infections and treat febrile patients with empiric broad-spectrum antibiotics), parenteral nutrition, HGF (used to mobilize PBPC) and Table 2

Unit prices (US $)

Nature

Unit price US $

Hospitalization Intensive Hematological Care Unit Normal Hematological Care Unit Medical Intensive Care Unit

565 493 916

Platelet transfusion Irradiated and leukocyte depleted 2 to 4 × 1011 Platelets per unit 4 to 6 × 1011 Platelets per unit

411 869

Red cell transfusion Irradiated, leukocyte depleted and phenotyped

195

Harvest 2 leukaphereses 3 leukaphereses Bone marrow

2160 3020 2640

Conditioning regimen BEAM CBV TAM TBI + M

1120 701 1702 1310

Medical visits

30

supportive drugs were recorded. All costs were obtained using the unit price multiplied by the total number of units. Unit prices were obtained from wholesale price lists applied to the Besançon Hospital Pharmacy. • The official tariff for Blood products as published in the ‘Journal Officiel’ on 22 December 1993 and applied during 1994 was used (cost per unit multiplied by total number of units). • Bacteriological, virology, biochemistry, hematology and immunology tests were evaluated according to the relevant charge index ‘Nomenclature des Actes de Biologie Me´dicale’ (NABM). Analysis Differences in mean values of clinical results (number of days of hospitalization, number of days to reach 0.5 × 109/l neutrophils, 1 × 109/l white blood count (WBC), 20 × 109/l and 50 × 109/l platelets, number of days of i.v. antibiotics, and number of units transfused) were tested using the nonparametric Mann–Whitney test. Data were expressed in each group as a mean ± standard deviation (s.d.). The mean rather than the median was used as the mean represents the most adequate measurement to represent the cost of patients whose consumption is high (care) and where there is a level of heterogeneity. Costs are reported as average per patient ± s.d. No statistical testing was performed on costs, but extensive sensitivity analyses were conducted by varying the clinical and cost parameters in order to test their effect on overall results of the study. Four types of sensitivity analyses were conducted: (1) varying HGF stimulation mean days and unit cost of ABM harvest; (2) varying mean PBSC leukapheresis; (3) varying post-graft hospitalization; and (4) varying post-graft platelet units transfused. This sensitivity analysis was performed to test the robustness of the economic findings.

Results Clinical outcome (Table 3) Lymphoma status at the time of graft was homogenous in both groups. Patients were grafted approximately at the same period after diagnosis, except for one patient grafted after a late relapse (5 years). Stem cell collection in both groups required 3 days of hospitalization; no patient in the BMT group received HGF. Patients in the PBSC group received a mean of 12.9 days of HGF (range 7–20), three needed a second cycle to collect stem cells and a total of 20 days of HGF (Table 1). Aplasia was shorter in PBSC patients. Hospitalization time between the two groups was not statistically significant but patients receiving peripheral stem cell reinfusion were hospitalized for a shorter period of time compared with the ABMT group (27.9 days vs 34.3 days, NS). This shorter time is linked to the significantly different duration needed by the two groups to recover a white blood cell count >1 × 109/l (21.7 days vs 13.2 days, P = 0.011) and to reach an absolute neutrophil count ,0.5 × 109/l (mean

977


978

Table 3

Clinical results

Mean (s.d.) or median (range)

BMT group

PBPC group

P

Days to reach WBC .1 × 109/l mean (± s.d.) median (range)

21.7 (±9.38) 18.5 (11–45)

13.2 (±4.83) 12 (9–26)

0.011

Days to reach ANC .0.5 × 109 /l mean (± s.d.) median (range)

23.5 (±10.26) 20 (11–48)

15.8 (±6.89) 14 (9–33)

0.031

Days to reach platelets .20 × 109/l mean (± s.d.) median (range)

20.9 (±17.81) 14 (12–70)

14.1 (±16.48) 9 (5–60)

0.01

Days to reach platelets .50 × 109/l mean (± s.d.) median (range)

29.2 (±22.14) 21 (16–90)

21.4 (±14.05) 17 (12–60)

0.054 (NS)

Duration of i.v. antibiotics use (days) mean (± s.d.) median (range)

18.7 (±5.25) 18.5 (10–26)

14.1 (±4.86) 13.5 (8–21)

0.089 (NS)

WBC = white blood cells; ANC = absolute neutrophil count.

15.8 days, range 9–33 vs mean 23.5 days, range 11–48, P = 0.031). In the same way, the length of time that intravenous antibiotics were required differed between the two groups (14.1 days vs 18.7 days, NS) but was not statistically significant. The mean number of days to recover and maintain a platelet count >20 × 109/l was significantly shorter in the PBPC group (mean 14.1 days, range 5–60) in comparison with the BMT group (mean 20.9 days, range 12–70, P = 0.01), which reflected significantly fewer platelet transfusions (6.1 vs 18.7 units, P = 0.012). The mean number of red blood cell transfusions (6.6 vs 8.8 units, NS) was similar in the two groups (Table 4). For delayed hematological recovery one patient needed HGF post-graft for 12 days in the PBPC group, in comparison with four patients in the BMT group using HGF for a mean of 19 days (Table 4). Table 4

Medical consultations and number of laboratory tests were similar in the two groups. After discharge from hospital two patients needed transfusion support; one in each group was transfused as an outpatient. One patient in the BMT group was readmitted within 100 days post-graft and stayed in hospital for 4 days. Economic assessment Total costs are presented in Table 5. For the overall study, the PBPC procedure was less expensive than autologous bone marrow transplantation, with a cost savings per patient of $6378, but the cost distribution was different. Stem cell harvest was more expensive for PBPC ($9030) than BM ($4745). The cost increased in PBSC collection ($3020 = 33% of total harvest costs) as compared with BM

Cost items

Mean (s.d.) or median

BMT group

PBPC group

P

Hospitalization days mean (± s.d.) median (range)

34.3 (10.67) 33 (21–60)

27.9 (5.24) 28 (21–39)

0.104 (NS)

within post-graft hospitalization mean (± s.d.) median (range)

24.9 (10.2) 22.5 (13–50)

19.1 (5.2) 18.5 (13–30)

0.11 (NS)

Transfused units Red blood cells mean (range) Platelets mean (range)

8.8 (4–22) 18.7 (4–51)

6.6 (2–20) 6.1 (3–21)

0.059 (NS) 0.012

HGF post-graft No. of patients days mean (± s.d.) median (range)

4 19 (4.69) 19.5 (14–23)

1 12 12

Medical visits mean (range) 4.2 (3–7) Rehospitalization until graft day +100

1

3.8 (2–10) 0


Table 5

979

Average cost analysis per patient (in US$)

Resource average ± s.d. ABMT

PBPCT

Savings (ABMT-PBPCT) (US$)

Stem cell harvest Hospitalization (room and board) HGF BM collection Leukapheresis Harvest manipulation

4745 ± 238 1626 ± 238 0 479 ± 0 0 2640 ± 0

9030 ± 3171 1478 ± 519 2115 ± 1240 0 2417 ± 849 3020 ± 907

−4285 148 −2115 479 −2417 −380

Graft and post-graft period Hospitalization (room and board) Drugs Transfusions (RBC, platelets) Laboratory tests

36 177 ± 18 555 19 393 ± 6031 5552 ± 3824 8210 ± 8530 2828 ± 1434

25 511 ± 5030 15 844 ± 3130 3588 ± 1298 3769 ± 2450 2310 ± 857

10 666 3549 1964 4441 518

1030 ± 1298 126 ± 39

839 ± 2236 114 ± 71

191 12

Total costs Hospitalization Transfusions Drugs

21 364 ± 6031 8402 ± 9098 5552 ± 3824

16 830 ± 3130 4458 ± 3482 6145 ± 1312

4534 3944 −593

Total

41 759 ± 18 543

Out-patient follow-up Medical visits

Imputed cost

35 381 ± 6472

6378

HGF = hematopoietic growth factor; BM = bone marrow; RBC = red blood cells.

harvest ($2640 = 56% of total stem cell harvest costs). Hospitalization costs were similar (PBPC = $1478 vs BM = $1626), but represented only 16% of the total cost for PBSC whereas it was 34% for ABMT. PBPC costs were higher due to HGF priming (+$2115 corresponding to 23% of total cost of stem cell harvest) and three leukaphereses (+$2417). None the less, both harvest manipulation and leukaphereses represented 60% for the PBSC group ($5437) which can be compared with 56% for harvest bone marrow manipulation ($2640) (Table 5). Globally, harvest costs were approximately 12% of the total cost in ABMT and 26% in PBPC groups. The bulk of cost was for hospitalization from graft to discharge from hospital. Cost savings for this defined period with PBPC transplantation were about $10 666. The most significant savings were achieved with transfusion which decreased in PBPC transplant by $4441; transfusion components were twice as expensive with ABMT where transfusion represented 23% of the graft period cost, whereas it was only 15% in the PBPC group. Moreover, the decrease in the length of hospitalization with PBPC transplant explained the savings of $3549, either 62% in the PBPC ($15 844) or 53% in the ABMT group ($19 393). The reduced use of drugs saved $1964, essentially due to intravenous antibiotic use which was for a shorter time in the PBPC group than in the ABMT group. Severe sepsis occurring during neutropenia or slow engraftment led to the use of HGF for four patients in the ABMT group vs one in the PBPC group. Hospital costs represented 72% of overall costs in the PBPC group and 87% in the ABMT group. After discharge from hospital, costs were similar in both groups; either 2.5% of overall costs for ABMT or 2.4% for PBPC. Average total costs were $35 381 and $41 759 for patients receiving PBPC infusion or BMT, respectively.

The cost savings were $6378 representing 15% of total costs. The savings exceeded the costs of HGF priming and PBPC collection.

Sensitivity analysis Stem cell harvest cost: For the mean number of days of stimulation by HGF there were savings that were only offset in two cases, either when the cost of a BM graft was negative or if the mean number of days of stimulation was multiplied by four (13 × 4 = 52). For the mean number of leukaphereses there were savings that were only made on patients in the PBPC group which were offset if the cost of the BM harvest was negative or if the mean number of cytaphereses was multiplied by two (3 × 2 = 6). Graft: The length of hospitalization could vary in both directions. Varying the range of hospitalization from −50 to +50% did not alter the savings. For the mean duration of hospitalization there were savings which could only be offset if the PBPC group had to be hospitalized for 5 days longer than the BM group. To modify the direction of the savings, either the average number of platelet transfusion units had to be multiplied by 3.8 in the PBPC group (22.9 instead of 6.1 units) or decreased by 90% in the ABMT group (1.9 instead of 18.7 units). For the mean number of platelet transfusion units there were savings which could only be offset if the mean number of platelet transfusion units in the BM group was negative. This sensitivity analysis showed the robustness of our results.


980

Discussion Numerous published studies have already demonstrated the feasibility of PBPC autotransplantation.10,11 In the Sheridan study,12 all patients received both PBPC and bone marrow. Faucher et al6 and Uyl-de Groot et al13 found the mean day to reach 0.5 × 109/l ANC was 10 and 8 respectively, but these patients received autografts for solid tumours or malignant lymphomas and HGF post-graft. Other studies have analyzed the cost of HGF adjunction to autologous BMT or PBPC.6,13,14 Our study analyzed non-Hodgkin’s lymphoma in the first group of patients treated in our ward with high-dose chemotherapy and PBSC infusion and was directly based on data from clinical records and not from clinical trials. Although this analysis is based on a small number of patients, it has the advantage of using a non-selected population. It thus reflects daily medical practice and avoids the discrepancies between economic outcomes in a clinical trial and routine practice despite the fact that discrepancies are relatively small in hematology, a speciality where standard care is based on protocol. Sensitivity analysis allowed us to be relatively confident about the validity of our conclusions, even if cautious because of the small number of patients in each arm. Moreover, the population in this study had a homogeneous pathology requiring high-dose chemotherapy and autograft. This study did not compare the use of HGF post-graft, which were not given systematically, and differed from previous studies.4,6 Neutropenia, length of hospitalization and platelet units used decreased during the post-graft period confirming previous studies carried out in different health care systems.3,6,13 Our results are consistent with these studies on homogeneous groups of patients with a single pathology, high-grade non-Hodgkin lymphoma. Smith et al3 found that with a PBPC group, 11 days were needed to reach ANC .0.5 × 109/l, whereas 14 days were required for a BMT group in NHL and HD; days of platelet recovery (.20 × 109/l) were 16 and 23 respectively. Faucher et al6 found that 16 days were needed to recover ANC .0.5 × 109/l for BM followed by G-CSF and that only 10 days were required for G-CSF primed PBPC. Similarly, 24 days were needed to recover platelets over 20 × 109/l for BMT followed by G-CSF and only 16 days were required for G-CSF primed PBPC. Uyl-de Groot et al13 compared three different groups, PBPC or BMT followed by G-CSF or BMT alone and found that the number of days required to reach ANC .0.5 × 109/l were respectively 8, 12.5 and 15. Duration of hospitalization was 15 days in the PBPC plus G-CSF group and 20 days in the BMT plus G-CSF group in Faucher’s study.6 In the study by Smith et al3 the respective durations were 17 and 23 days. The Dutch study had durations of 22 and 36 days, respectively.13 No difference in terms of disease-free survival with either procedure was observed in the literature.15 Outpatient PBPC harvest may further improve the cost effectiveness of PBPC autologous graft and quality of life. Quality of life was better with the PBPC procedure as shown by Le Coroller et al.16 Anxiety and pain experienced

by patients undergoing PBPC were significantly less severe than with bone marrow collection.16 Retrospective studies indicated that the main advantage of substituting BMT with PBPCs was a faster recovery of platelet counts, with a concomitant decrease in transfusion requirement.6,13–17 In our study, the number of days spent in hospital and blood bank costs were the major cost factors. As expected, the cost analysis showed savings for the PBPC group. Since PBPC patients experienced a trend toward less neutropenia and, hence, had a trend toward fewer hospitalization days, the related cost was $35 381 with savings of $6378 compared with the ABMT group. The cost savings would be expected to be higher if the PBPC patients had had only two leukapheresis procedures. The number of leukaphereses influenced stem cell harvest cost. In this study, the mean number of leukaphereses was three, but with greater expertise this number had decreased to two. The consequence was a reduction in the length of hospitalization for harvest and influenced the high cost identified with the PBPC procedure. For three leukaphereses the marginal cost was $4285 vs $1527 for two leukaphereses, reducing the total harvest cost from $9030 to $6272. Thus, the total cost savings increased from $6378 to $9137. Factors associated with decreasing cost were hospital staff, blood bank, and technology and processing improvements.18 Chemotherapy could be given as day care hospitalization for peripheral stem cell harvests and would decrease cost. Costs of conditioning regimens were lower in the ABMT group because of the use of the CBV protocol for four patients: the BEAM protocol was twice as expensive as the CBV protocol ($701 vs $1120) (Table 2). Sensitivity analysis showed that in the range observed in this study, the direction of results supporting PBSC grafting did not vary. Extreme or highly improbable values would be needed to change the direction of the results. In the same way, using the published data by Smith et al,3 Faucher et al6 and Uyl-de Groot et al13 for duration of hospitalization, the analysis was always in favor of PBPC with savings of $6159, $5000 and $10 664, respectively (Table 6). In a relapsed LNH trial, Bennett et al15 showed that using post-graft GM-CSF with BMT led to savings of approximately $6630. Forty percent cost savings were obtained from shorter hospitalization (3.5 days less). Two retrospective studies found that PBPC infusion as an alternative to BMT resulted in savings of 20% in total direct costs.6,11 Our study showed approximately 15%. However, in these studies, HGF was systematically used as an adjunct to BMT or PBPCT. An European study based on US estimates of costs compared PBPC and BM transplants for lymphomas (nonHodgkin’s lymphoma and Hodgkin disease) in 71 patients, with BEAM as a conditioning regimen.3 The reported estimated costs for PBPC collection or bone marrow harvest of $9500 and $4630 included G-CSF priming. This result was similar to $9030 and $4755 found in our study despite the differences in health care between different health systems. None the less, the significant cost savings were $13 600 due to shorter hospitalization with less supportive care.3 However, the estimated costs of ABMT vary greatly


Table 6

981

Comparison of costs in previously published studies (in US$) Faucher6

Cost item BMT Pretransplantation Cell collection G-CSF administration

Uyl-de Groot13 PBPCT

BMT

PBPCT

2770 2770 0

2780 1280 1500

5006 5006 0

5915 4537 1378

Transplantation Hospitalization Drugs HGF Transfusions Others

20 520 12 450 2200 0 3800 2070

16 990a 9720a 2000 1370 2170a 1730

25 586 17 578 3562 0 1584 2862

15 894a 9318a 1590a 1843 850a 2293a

Total costs

23 290

19 770a

30 592

21 809a

a

Significative statistical difference.

between French and US studies.14 For Westerman and Bennett, these variations can be attributed to differences in labor, pharmaceutical, overheads and the structure of socialized medicine. The average costs for treatment (high-dose therapy and autologous transplantation) of non-Hodgkin’s lymphoma in Nebraska was evaluated at $74 000 in 1991.5 Several limitations of our study must be discussed in evaluating the results. Firstly, the use of analytic accounting of the ‘Assistance Publique-Hoˆpitaux de Paris’ (AP-HP) which is explained by the fact that the Medical School Hospital did not have its own data for the study period. However, it must be noted that analytic accounting methods are currently used to evaluate the cost structure. Moreover, these costs are relatively homogeneous in hematology units in France. Secondly, the hospital perspective was used as it took into account all the modifications due to changes in techniques. Thirdly, the duration of patient follow-up did not enable us to carry out a cost-effectiveness study to obtain the ratio of the cost of an extra year of life. Finally, this economic evaluation reflects the actual hospital life situation as opposed to a clinical trial. It is important to consider that cost data from different countries cannot be directly correlated with other countries due to differences such as socialized medicine, worker’s wages and pharmaceutical costs.13 In spite of these differences, our study confirmed results from the private French system 6 and The Netherlands.4,13 In France the number of autografts with PBPC has changed considerably in less than 2 years, ie in 1992, 24% (287) of autografts used PBSC as opposed to 64% (662) in 1993 and 70% (1129) in 1994 for all pathologies (personal data). In the case of NHL autografts, 54% (228) used PBSC in 1993 and 75% (333) in 1994. This difference of 105 autografts using PBSC would give savings of nearly $670 000. The results of our unicenter study are confirmed by data from a large number of centers and demonstrate that PBPC transplantation is 15% cheaper than BMT. This confirms the importance of a pharmacoeconomic approach in the evaluation of all new protocols as their introduction does not systematically incur increasing costs.

Acknowledgements We thank Micheline Jacquet for her constant support, Dr Francis Fagnani for his methodological support and Dr Jean-Charles Brosse for his help.

References 1 Philip T, Guglielmi C, Hagenbeek A et al. Autologous bone marrow transplantation as compared with salvage chemotherapy in relapses of chemotherapy-sensitive non-Hodgkin’s lymphoma. New Engl J Med 1995; 333: 1540–1545. 2 Haioun C, Lepage E, Gisselbrecht C et al. Comparison of autologous bone marrow transplantation with sequential chemotherapy for intermediate-grade and high-grade nonHodgkin’s lymphoma in first complete remission: a study of 464 patients. J Clin Oncol 1994; 12: 2543–2551. 3 Smith TJ, Hillner BE, Yanovich S et al. Economic analysis of peripheral blood progenitor cell (PBPC) or autologous bone marrow (ABM) transplant in relapsed Hodgkin disease (HD) and non-Hodgkin lymphoma (NHL). Proc Am Soc Clin Oncol 1995; 14: 314 (Abstr. 923). 4 Uyl-de Groot CA, Hagenbeek A, Verdonck LF et al. Costeffectiveness of ABMT in comparison with CHOP chemotherapy in patients with intermediate- and high-grade malignant non-Hodgkin’s lymphoma (NHL). Bone Marrow Transplant 1995; 16: 463–470. 5 Bennett CL, Armitage JL, Armitage GO et al. Costs of care and outcomes for high-dose therapy and autologous transplantation for lymphoid malignancies: results from the University of Nebraska 1987 through 1991. J Clin Oncol 1995; 13: 969–973. 6 Faucher C, le Corroller AG, Blaise D et al. Comparison of G-CSF primed peripheral blood progenitor cells and bone marrow autotransplantation: clinical assessment and costeffectiveness. Bone Marrow Transplant 1994; 14: 895–901. 7 Milpied N, Lamy T, Gaillard F et al. Intensive short-term therapy for adults with disseminated intermediate or high-grade non-Hodgkin’s lymphomas (NHL): analysis of prognostic factors in a prospective multicenter trial. Blood 1995; 86 (Suppl. 1): 440a (Abstr. 1748). 8 Soueˆtre E, Quing W. Economic analysis of lenograstim in the correction of neutropenia following chemotherapy for non-


982

9

10 11 12

13

Hodgkin’s lymphoma. PharmacoEconomics 1994; 6 (Suppl. 1): 36–43. Direction des Finances, De´partement controˆle de gestion. Assistance publique-Hoˆpitaux de Paris. Comptabilite´ Analytique des Hoˆpitaux de l’AP-HP, Edition abre´geé. Re´sultats de l’anneé 1994. Paris, 1994. Bensinger W, Appelbaum F, Rowley S et al. Factors that influence collection and engraftment of autologous peripheralblood stem cells. J Clin Oncol 1995; 13: 2547–2555. Kumar L, Gulati SC. Peripheral stem-cells transplantation. Lancet 1995; 346 (Suppl.): 9. Sheridan WP, Begley GC, Juttner CA et al. Effect of peripheral blood progenitor cells mobilized by filgrastim (G-CSF) on platelet recovery after high-dose chemotherapy. Lancet 1992; 339: 640–644. Uyl-de Groot CA, Richel DJ, Rutten FFH. Peripheral blood progenitor cell transplantation mobilised by r-metHu G-CSF (filgrastim); a less costly alternative to autologous bone marrow transplantation. Eur J Cancer 1994; 30A: 1631–1635.

14 Westerman IL, Bennett CL. A review of the costs, cost-effectiveness and third-party charges of bone marrow transplantation. Stem Cells 1996; 14: 312–319. 15 Bennett CL, George SL, Vose JM et al. Granulocyte–macrophage colony-stimulating factor as adjunct therapy in relapsed lymphoid malignancy: implications for economic analyses of phase III clinical trials. Stem Cells 1995; 13: 414–420. 16 Le Coroller AG, Macquart-Moulin G, Auquier P et al. Comparative study of monetary costs and non-monetary costs of two procedures of hematopoietic stem cells collection: leucapheresis and bone marrow collection. Rev Epidem et Sante Publ 1996; 44: 133–143. 17 Frampton JE, Faulds D. Filgrastim. A reappraisal of pharmacoeconomic considerations in the prophylaxis and treatment of chemotherapy-induced neutropenia. PharmacoEconomics 1996; 9: 76–90. 18 Brice P, Godin S, Libert O et al. Impact of lenograstim on the cost of autologous bone marrow transplantation. PharmacoEconomics 1995; 7: 238–241.