Pii: s0959-8049(98)00260-3

European Journal of Cancer, Vol. 34, No. 12, pp. 1894±1901, 1998 # 1998 Elsevier Science Ltd. All rights reserved Cost-eVectiveness Analysis of Paclitaxel and Cisplatin Versus Cyclophosphamide and Cisplatin as First-line Therapy in Advanced Ovarian Cancer. A European Perspective 1Medical Economics Research Group, Prinzregentenst. 72, D-81675 Munich, Germany; and 2Centre for Pharmacoeconomics, School of Pharmacy, University of Milan, Milan, Italy Paclitaxel is a new cytotoxic agent that has demonstrated signi®cant activity in advanced ovariancancer. The aim of this study was to determine the cost structure of advanced ovarian cancer and thecost-eVectiveness of paclitaxel±cisplatin (PC) combination therapy compared with a standard cyclo-phosphamide±cisplatin (CC) regimen as ®rst-line therapy. The analysis was performed separately forsix European countries: Germany, Spain, France, Italy, The Netherlands and the U.K. The study wasconducted from the national health service payer's perspective. The total cost of treatment per patient(six cycles of chemotherapy) in the six European countries varied between a minimum of US$4,926 inthe U.K. and US$12 578 in Germany for the CC regimen and between US$13 038 and US$24 487 for thePC regimen (April 1996). Since the new regimen improved life expectancy by 1.283 years comparedwith CC, the incremental cost-eVectiveness of PC was calculated to be between US$6,403 per 5-yearsaved in the U.K. and US$11 420 per life-year saved in Italy. Overall, the cost-eVectiveness of PCcompares favourably with other oncological interventions. The ®ndings of this study suggest thathealthcare decision makers should consider paclitaxel, in combination with cisplatin, as a cost-eVec-tive ®rst-line therapy for patients with advanced ovarian cancer. # 1998 Elsevier Science Ltd. Allrights reserved.
Key words: cost-effectiveness analysis, paclitaxel, cyclophosphamide, cisplatin, advanced ovariancancer, European perspectiveEur J Cancer, Vol. 34, No. 12, pp. 1894±1901, 1998 cessful management. Whilst initial response rates are high, Who data show an incidence of 645 000 cases of cancer in however, the long-term prognosis of patients with advanced the European Community in 1990. Included in this are ovarian cancer remains poor: only 10±25% of women survive approximately 28 000 cases of ovarian cancer [1]. Ovarian cancer is the sixth most common form of cancer worldwide Paclitaxel (Taxol1), which is produced semisynthetically and has the highest mortality rate of all gynaecological can- from needles and twigs of taxus plants, is a new cytotoxic cers [2]. Early stages of ovarian cancer are usually asympto- agent with a novel mechanism of action that has demon- matic, and most symptoms of the disease are manifestations strated signi®cant activity in advanced ovarian cancer. Initial of advanced disease [3]. Thus, the majority of patients with trials in patients with advanced disease, who were either ovarian cancer present with advanced disease (stage III±IV heavily pretreated with platinum or had refractory disease, according to International Federation of Gynaecology and showed a dramatic and prolonged response to paclitaxel Obstetrics (FIGO) classi®cation). In these patients, typically monotherapy [6]. In a randomised clinical study, the Gynae- after careful cytoreductive surgery, systemic platinum-based cologic Oncology Group compared paclitaxel and cisplatin combination chemotherapy forms the cornerstone of suc- (PC) as ®rst-line chemotherapy with the standard therapy ofcyclophosphamide and cisplatin (CC) (GOG 111) [7]. Theresults showed a signi®cant improvement in overall response Correspondence to K. Berger.
Received 15 Oct. 1997; revised 18 May 1998; accepted 12 Jun. 1998.
rate, complete response, progression-free survival and overall Paclitaxel plus Cisplatin in Advanced Ovarian Cancer survival for the paclitaxel group [7]. These ®ndings have led (135 mg/m2 over 24 h) for advanced ovarian cancer in the to the suggestion that the PC combination should be given in GOG 111 study (Table 1). To be included in the GOG 111 preference over CC as ®rst-line chemotherapy for patients study, patients had to have histologically con®rmed stage III epithelial ovarian cancer with residual masses > 1 cm after As the focus on cost containment among decision makers surgery or stage IV disease. No patient had received prior becomes more intense, the medical bene®ts could easily be chemotherapy or radiation therapy for ovarian cancer and all diminished by the fact that therapy with paclitaxel might patients had a GOG performance status score of 0, 1 or 2.
appear to be more expensive than other chemotherapies.
Other inclusion criteria included normal baseline blood Therefore, economic evaluations are necessary. Economic counts and normal renal and hepatic function. Patients with a evaluations compare the costs and consequences of alter- history of cardiac arrhythmias and those receiving anti- native treatments for advanced ovarian cancer with respect to arrhythmic drugs were not eligible for inclusion. All che- medical improvements. The cost implications of treating motherapeutic agents were administered intravenously and advanced ovarian cancer with PC have already been exam- patients received a total of six cycles of chemotherapy, with ined from a Canadian perspective: the ®rst study concerning each cycle administered at 3 weekly intervals. Because second- and third-line chemotherapy with diVerent regimens administration of paclitaxel can cause hypersensitivity reac- including PC showed total average costs of Can$53 000 tions, patients assigned to the PC group were premedicated (US$37 900) per patient [9]. Another cost analysis showed according to local treatment patterns.
that the incremental cost-eVectiveness of PC over conven- Among 216 patients with measurable disease, 73% of PC tional ®rst-line therapy was Can$20 355 (US$14 500) per recipients responded to therapy compared with 60% of life-year saved [10]. In a diVerent economic analysis of the patients in the CC group (P = 0.01), and the median pro- GOG 111 study conducted from the perspective of the US gression-free survival was signi®cantly longer (P < 0.001) healthcare provider, combination therapy with PC cost an among PC recipients (18 months versus 13 months among additional US$19 820 per life-year saved compared with CC CC treated patients). The median overall survival was also when patients were treated in the hospital, and US$21 222 signi®cantly longer (P < 0.001) in the PC group (38 months per life-year saved when patients were treated in an out- versus 24 months among CC treated patients) [7]. For the patient setting [11]. Finally, in an Italian study, the cost purposes of the present economic analysis, the outcome of structure of PC versus CC as ®rst-line therapy was analysed.
interest was median overall survival.
They calculated the incremental cost-eVectiveness for pacli-taxel with US$19 603 per life-year saved [12].
However, the cost of care structure of usual care of The objectives of this economic analysis, conducted from advanced ovarian cancer in Europe, and the economic out- the national health service payer's perspective, were to deter- comes of treatment with PC chemotherapy, have not been mine the cost structure of treating advanced ovarian cancer examined previously. Thus, the aim of this retrospective and to determine the cost-eVectiveness of PC as ®rst-line study, conducted from the national health service payer's chemotherapy for women with advanced ovarian cancer perspective of six European countries, was to determine the compared with the standard platinum-based combination cost structure of advanced ovarian cancer and the cost-eVec- tiveness of PC combination therapy as ®rst-line chemother-apy for advanced ovarian cancer compared with a standard CC regimen, using the results of the GOG 111 study [11].
In general, costs are de®ned as the result of resource con- sumption multiplied by the price for the resource. Therefore, the quantitative resource consumption as well as the prices were collected separately. The resource consumption was In this study, a retrospective analysis was performed on determined quantitatively on the basis of structured face-to- 386 women treated with cisplatin (75 mg/m2 at a rate of 1 mg/ face interviews with experts (Table 2) treating advanced min) plus either cyclophosphamide (750 mg/m2) or paclitaxel ovarian cancer, delivering the information required to evalu-ate the direct costs induced by chemotherapy of advancedovarian cancer and the GOG 111. The interviews yielded Table 1. Patient clinical characteristics at baseline [7] country-speci®c data about duration and extent of drug treatment (e.g. co-medication) as well as frequency of con-sultations and laboratory tests. GOG 111 was the data source CC, cyclophosphamide±cisplatin; GOC, Gynaecologic Oncology for information on chemotherapy dosage, number of treat- life expectancy based on actuarial methods, calculates the ment cycles and their toxicity pro®le (adverse events ). Addi- speci®c life expectancy of a patient from the disease-speci®c tional interviews and literature searches, depending on the mortality rate and the mortality rate of the standard popula- speci®c country, yielded prices for drugs and other resources, tion of a given age and gender. The GOG 111 study provided e.g. hospital stay, consultations, visits to validate the inter- the median age of both treatment groups and disease-speci®c view results. To achieve comparable results, all costs were mortality. Further tables of vital statistics provided the coun- try-speci®c life expectancy. Thus, the speci®c life expectancy The prices for chemotherapy were obtained from tele- could be determined in years for each treatment group phone interviews with hospital pharmacists and oYcial drug (Table 4). In the ®nal step of the analysis, the incremental price lists (Table 3). The calculation was performed for a life-years saved in the PC group was determined by subtract- body surface area of 1.76 m2 at doses of 135, 75 and 750 mg/ ing the speci®c life expectancy in the CC group from the m2 for paclitaxel, cisplatin and cyclophosphamide, respec- speci®c life expectancy in the PC group. The incremental tively. The cost of co-medication administered with each cost-eVectiveness of PC was subsequently calculated as cycle of chemotherapy (e.g. antibiotics, rehydration, anti- emetics), including premedication for PC recipients, was alsodetermined through the same sources. The costs of hospital or day clinic stay were obtained from interviews and hospital price lists (Table 3). The average costs of laboratory tests/ investigations were determined in the same way (Table 3).
Brie¯y, the interviews were performed to determine the The robustness of the results of this economic analysis medication: chemotherapy (paclitaxel, cisplatin and were tested using a series of sensitivity analyses. These tests cyclophosphamide), treatment of adverse events, take into consideration, due to uncertainties, lack of precise including allergic reactions, neurological symptoms, cost of illness data and, therefore, use data from expert nephrotoxicity, alopecia, fever, gastrointestinal symp- interviews and literature research. First, the increase in spe- toms, anaemia, thrombocytopenia, leucopenia and ci®c life expectancy associated with PC therapy was varied by neutropenia (drugs, dosage and duration of treat- ‹ 50%. Second, the costs of medication (chemotherapy and co-medication) and hospitalisation were varied by ‹ 20%, as hospitalisation (during administration of chemother- described for a previous economic analysis of the GOG 111 apy): average duration of hospital stay per cycle; data and the prices in most of the included countries varied consultations/laboratory tests/investigations: average number of consultations/laboratory tests/investiga-tions per patient for an overall course of treatment (six In patients receiving standard CC, the overall net expen- Determination of eVectiveness and cost-eVectiveness analysis diture over all cycles varied between a minimum of approxi- The eVectiveness endpoint in this study was life-years mately US$4,926 in the U.K. and a maximum of US$12 578 saved under PC therapy. The ®rst step in the analysis was to in Germany. (Approximately US$9,290 are obtained in calculate the speci®c life expectancy of PC and CC treated Spain, US$8,502 in France, US$6,578 in Italy and US$6,537 patients using the DEALE (declining exponential approx- in The Netherlands). The use of PC yielded minimum total imation of life expectancy) approach [13]. This algorithm, costs of approximately US$13 038 in the U.K. and maximal which is simple to use and has been shown to closely estimate total costs of US$24 487 in Germany (approximately Table 3. Source of costs for medication, hospital stay and consultation and laboratory tests departments of ®ve hospitals (of whichtwo were private) Table 4. Life expectancy for Germany, Spain, France, Italy, The Netherlands and the U.K.
*PC À CC. PC, paclitaxel±cisplatin; CC, cyclophosphamide±cisplatin.
Table 5. Incremental costs for paclitaxel±cisplatin (PC) versus cyclophosphamide±cisplatin (CC) in US$ Table 6. Cost per life-year saved for paclitaxel±cisplatin (PC) versus cyclophosphamide±cisplatin (CC) in US$ Incremental life expectancy of PC group versus CC group US$21 230 in Italy, US$17 520 in Spain, US$17 150 in US$10 477, respectively (Table 8). Increasing the costs of PC France and US$16 547 in The Netherlands).
chemotherapy and hospitalisation by up to 20% revealed a Compared with the CC regimen, drug costs, hospitalisa- predictable increase in the respective cost-eVectiveness ratios.
tion costs, costs of consultations/laboratory tests/investiga- The sensitivity analyses of Germany, Spain, France, The tions and the costs of treating adverse events were higher Netherlands and the U.K. brought similar results (Table 8).
among PC recipients. Overall, the incremental cost of a Nevertheless, the incremental cost-eVectiveness of PC under complete course of PC chemotherapy was US$11 909 in these conditions remained comparable with other oncological Germany, US$8,230 in Spain, US$8,648 in France, US$14 652 in Italy, US$10 010 in The Netherlands and The incremental increase in speci®c life expectancy asso- ciated with PC was also subjected to sensitivity analyses.
Drug costs (chemotherapy plus co-medication) accounted Under base-case conditions, increasing the incremental for the largest proportion of treatment costs among PC reci- increase in speci®c life expectancy by 50% to 1.925 years in pients (average 59.2%; maximum 77.4% in the U.K.), whilst Italy decreased the cost-eVectiveness of PC to US$7,611 per hospitalisation costs accounted for an average of 53.9% with life-year saved. Whilst decreasing the speci®c life expectancy a maximum of 75.7% (France) for overall treatment costs by 50% to 0.642 years increased the cost-eVectiveness of PC to US$22 822 per life-year saved in Italy, the cost-eVective-ness ratio of PC versus CC under such conditions remained within acceptable limits. There was only a small diVerence The median age of PC and CC treated patients in the between the other ®ve countries concerning the calculation GOG 111 study was 59 and 60 years, respectively [7]. The with 1.925 and 0.642 years of life expectancy. For the age, gender and race adjusted life expectancy of these patients assumed higher and lower life expectancies, sensitivity ana- was 22.1±24.4 years [14], whilst the median survival among lyses were also carried out for decreased and increased drug PC and CC treated patients in the GOG 111 study was 3.2 and 2 years, respectively. Using the DEALE approach, thecalculated speci®c average life expectancy in the PC and CC treatment groups was 3.848 and 2.565 years, respectively, On the basis of the ®ndings of the GOG 111 study, we giving an average incremental life expectancy among PC decided to perform an incremental cost-eVectiveness analysis recipients of 1.283 years (Table 4). The incremental cost- (i.e. cost per life-year saved). In contrast to cost-bene®t and eVectiveness (cost per life-year saved) of PC compared with cost-minimisation analyses, which express costs and con- CC was, therefore, calculated at US$9,362 in Germany, sequences in monetary units, a cost-eVectiveness analysis US$6,395 in Spain, US$6,642 in France, US$11 420 in expresses costs in monetary units and eVectiveness (out- Italy, US$7,796 in The Netherlands and US$6,403 in the comes) in non-monetary units. If two or more interventions U.K. (Table 6). The cost-eVectiveness ratios of other onco- have the same treatment objective but diVerent degrees of logical therapies for comparison are shown in Table 7.
preferred method. One of the advantages of incremental cost- eVectiveness analysis is that it provides healthcare decision- As the greatest diVerence between the two regimens related makers with a common denominator by which various to the costs of chemotherapy/co-medication and hospitalisa- healthcare interventions can be tentatively compared, which tion, these parameters were further investigated in sensitivity may aid decision making with regard to optimal allocation of analyses (Table 8). Presenting, for example, the cost sensi- scarce resources. In a recent analysis of 500 life-saving inter- tivity analysis of Italy with an assumption of an incremental ventions in the U.S.A., for example, Tengs and colleagues increase in speci®c life expectancy of 1.283 years and that the reported that the cost of postsurgical chemotherapy for pre- costs of drugs and hospitalisation for the PC regimen are each menopausal women with breast cancer was US$18 000 per reduced by an arbitrary ®gure of 20%, then the cost per life- life-year saved [14]. Among non-oncological interventions, the year saved among PC recipients decreases to US$9,551 and incremental cost-eVectiveness of b-adrenoreceptor antagonists Table 7. Cost-eVectiveness ratios of diVerent oncological therapies ABMT for relapse metastatic Hodgkin's disease Bone marrow transplant and high (versus standard) chemotherapy for breast cancer Postsurgical chemotherapy for woman with breast cancer age 60 years Postsurgical chemotherapy for premenopausal woman with breast cancer Interferon alpha-2b in hairy cell leukaemia Paclitaxel as ®rst-line chemotherapy (six European countries) CEA, carcinoembryonic antigen; ABMT, autologous bone marrow transplantation; CMF, cyclophosphamide, methotrexate ¯uorouracil;ANLL, acute nonlymphocytic leukaemia.
Table 8. Sensitivity analysis for Germany, Spain, France, Italy, The Netherlands and the U.K. All costs are in US$ (April 1996) Decrease cost of hospitalisation for PC regimen by Increase cost of hospitalisation for PC regimen by *Cost of chemotherapy and co-medication. PC, paclitaxel±cisplatin.
for low-risk patients following myocardial infarction was esti- sion-free survival compared with standard CC chemotherapy mated at US$16 897 per life-year saved [14]. Whilst the latter in the GOG 111 study [7]. Thus, the ability of PC to delay study was performed in the U.S.A. and, therefore, its ®ndings disease progression, which is the most important factor in are not directly comparable with other countries, these ®gures determining the quality of life of patients with advanced can- indicate the acceptable cost to society of many life-saving cer, is likely to have improved the quality of life of surviving medical interventions. The incremental cost-eVectiveness of patients. Therefore, the cost utility of PC (i.e. cost per qual- US$6,403 to US$11 420 per life-year saved for patients trea- ity-adjusted life-year saved) as ®rst-line therapy in advanced ted with PC for advanced ovarian cancer in six European ovarian cancer warrants further investigation. Indeed, a cost countries, therefore, compares favourably with other life-sav- utility analysis would enable the eVects of this innovative ing interventions. Furthermore, even if the improvement in chemotherapy regimen on patients' quality of life and survival survival is reduced by 50% to 0.642 years, the incremental to be considered together, by converting both in a common cost-eVectiveness of PC continues to remain acceptable when unit of measure. Moreover, recent clinical trials have shown compared with other oncological therapies. The analysis that carboplatin, a platinum compound with a better ther- shows some diVerences between the European countries apeutic index than the one of cisplatin, is safe and eVective concerning the proportion between chemotherapy drug costs when used in combination with paclitaxel in patients with and hospitalisation costs. In France, the hospitalisation costs advanced ovarian cancer and the cost-eVectiveness of this are very high (75.7% of all costs) in the CC treatment group, combination regimen warrants further investigation.
whereas in the U.K. the hospitalisation costs are nearly as In conclusion, paclitaxel, in combination with cisplatin, is high as the other costs (32.6% for hospitalisation versus clinically more eVective than a standard CC regimen (in 21.1% for treatment and 37.1% for laboratory and consulta- terms of duration of progression-free and overall survival) as ®rst-line chemotherapy following incomplete surgical resec- Whilst the present study is limited by the fact that it was a tion of advanced ovarian cancer. From the national health retrospective analysis and that the cost structure was princi- service payer's perspective of six selected European countries pally determined by expert opinion from interviews with a (Germany, Spain, France, Italy, The Netherlands and the limited number of physicians, it does provide an interesting U.K.), the incremental cost-eVectiveness of PC was calcu- insight into the cost structure of treating advanced ovarian lated at US$6,403 to US$11 420 per life-year saved, which cancer. We found that the main cost drivers in the treatment compares favourably with other oncological and non-oncolo- of advanced ovarian cancer in the six European countries are gical interventions. These ®ndings suggest that healthcare drug and hospitalisation costs, which con®rms the ®ndings of decision makers should consider PC as a cost-eVective ther- previous economic analyses of the GOG 111 study from apeutic option for ®rst-line management of advanced ovarian Canadian and US healthcare perspectives [9±11]. When the contribution of the diVerent costs included in the presentanalysis are compared between the two regimens, it can be 1. International Agency for Research on Cancer. Facts and ®gures seen that drug-related costs accounted for the greatest diVer- of cancer in the European community. Lyon, 1993.
ence between the PC and CC regimens.
2. Parkin DM, Pisani P, Ferlay J. Estimates of the worldwide inci- The magnitude of the diVerence in hospitalisation costs dence of eighteen major cancers in 1985. Int J Cancer 1993, between the two chemotherapy regimens may be explained by the fact that paclitaxel was given as a continuous 24 h 3. Lorigan PC, Crosby T, Coleman RE. Current drug treatment guidelines for epithelial ovarian cancer. Drugs 1996, 51(4), 571± intravenous infusion, which requires more intensive (and, therefore, more costly) nursing time. Reducing the duration 4. The new FIGO stage grouping for primary carcinoma of the of infusion of paclitaxel may be one way in which the costs of ovary (1985). Gynecol Oncol 1986, 25, 383.
hospitalisation could be reduced. Indeed, Eisenhauer and 5. Allen DG, Baak J, Belpomme D, et al. Advanced epithelial ovarian cancer: 1993 consensus statements. Ann Oncol 1993, associates reported similar clinical eYcacy by 3 h and 24 h infusions of paclitaxel [23]. The superiority of a 3 h infusion 6. Spencer CM, Faulds D. Paclitaxel: a review of its pharmaco- of paclitaxel (175 mg/m2) in combination with cisplatin over a dynamic and pharmacokinetic properties and therapeutic poten- standard CC regimen was recently con®rmed in a large mul- tial in the treatment of cancer. Drugs 1994, 48(5), 794±847.
ticentre trial whose results were presented at the 1997 annual 7. McGuire WP, Hoskins WJ, Brady MF, et al. Cyclophosphamide and cisplatin compared with paclitaxel and cisplatin in patients meeting of the American Society of Clinical Oncology. The with stage III and stage IV ovarian cancer. N Engl J Med 1996, PC combination led to less febrile neutropenia than the CC 8. Neijt JP. New therapy for ovarian cancer (editorial). N Engl J Since the goal of modern medicine is to improve patients' 9. Doyle C, Stockler M, Pintilie M, Panesar P, Warde P, Sturgeon quality, as well as the quantity, of life, a potential limitation of J, Oza AM. Resource implications of palliative chemotherapy for the GOG 111 study (on which the present economic analysis ovarian cancer. J Clin Oncol 1997, 15, 1000±1007.
was based) is that patients' quality of life during the extended 10. Covens A, Boucher S, Roche K, et al. Is paclitaxel and cisplatin a survival period associated with PC chemotherapy was not cost-eVective ®rst-line therapy for advanced ovarian carcinoma? assessed. Indeed, an improvement in life expectancy does not 11. McGuire W, Neugut AI, Arikian S, et al. Analysis of the cost- necessarily confer an improvement in quality of life among eVectiveness of paclitaxel as alternative combination therapy cancer patients, since long-term adverse events of cytotoxic for advanced ovarian cancer. J Clin Oncol 1997, 15(2), 640±645.
drug therapy (e.g. peripheral neuropathy) can diminish a 12. Messori A, Trippoli S, Becagli P, Tendi E. Pharmacoeconomic patients' quality of life; furthermore, the management of such pro®le of paclitaxel as a ®rst-line treatment for patients withadvanced ovarian carcinoma. Cancer 1996, 78, 2366±2373.
adverse eVects may incur additional resource costs. In addi- 13. Beck JR, Kassirer JP, Pauker SG. A convenient approximation of tion to improving overall life expectancy, the PC regimen was life expectance (the `DEALE'). I. Validation of the method. Am associated with a signi®cantly increased duration of progres- Paclitaxel plus Cisplatin in Advanced Ovarian Cancer 14. Tengs TO, Adams ME, Pliskin JS, et al. Five hundred life-saving 19. Desch CE, Hillner BE, Smith TJ, et al. Should the elderly receive interventions and their cost-eVectiveness. Risk Anal 1995, 15, chemotherapy for node-negative breast cancer? J Clin Oncol 15. Kievit J, van de Velde CJH. Utility and cost of carcinoembryonic 20. Hillner BE, Smith TJ. EYcacy and cost-eVectiveness adjuvant antigen monitoring in colon cancer follow-up evaluation. A chemotherapy in woman with node-negative breast cancer. N Markov analysis. Cancer 1990, 65, 2580±2587.
16. Desch CE, Lasala MR, Smith TJ, et al. The optimal timing of 21. Ozer H, Golomb HM, Zimmerman H, et al. Cost-bene®t analy- autologous bone marrow transplantation in Hodgkin's disease sis of interferon alpha 2b in treatment of hairy cell leukemia. J patients following a chemotherapy relapse. J Clin Oncol 1992, 10, Natl Cancer Inst 1989, 81, 594±602.
22. Rees GJG. Cost-eVectiveness in oncology. Lancet 1985, 2, 1405± 17. Hillner BE, Smith TJ, Desche CE. EYcacy and cost-eVective- ness of autologous bone marrow transplantation in metastatic 23. Eisenhauer EA, Bokkel Huinink WW, Swenerton KD, et al.
breast cancer. JAMA 1992, 267, 2055±2061.
European-Canadian randomized trial of paclitaxel in relapsed 18. Welch HG, Larson EB. Cost-eVectiveness of bone marrow ovarian cancer: high-dose versus low-dose and long versus short transplantation in acute nonlyphocytic leukemia. N Engl J Med infusion. J Clin Oncol 1994, 12, 2654±2666.

Source: http://csrparticipantresources.joannabriggs.org/docs/2012-13_JBI_Courses/JBICSR_12-13/JBICSR_0004/Readings/JBICSR1213-0004-Reading01.pdf