Critical Reviews in Oncology/Hematology 33 (2000) 99 103 www.elsevier.com/locate/critrevonc Assessing the relative costs of standard open surgery and laparoscopic surgery in colorectal cancer in a randomised controlled trial in the United Kingdom M.L. Stead a *, J.M. Brown a, N. Bosanquet b, P.J. Franks c, P.J. Guillou d, P. Quirke e, D. Jayne d, J.R.T. Monson f, A.V. Webb a a Northern and Yorkshire Clinical Trials and Research Unit, Uni ersity of Leeds, Arthington House, Hospital Lane, Leeds LS16 6QB, UK b Imperial College School of Medicine, London, UK c Thames Valley Uni ersity, London, UK d St James s Uni ersity Hospital, Leeds, UK e Uni ersity of Leeds, Leeds, UK f Castle Hill Hospital, Hull, UK Accepted 19 October 1999 Contents 1. Introduction... 100 2. Methods... 100 2.1. Trial design... 100 2.2. Economic evaluations... 100 2.2.1. Cost analysis... 101 2.2.2. Q-TWIST analysis.... 101 2.2.3. Cost-effectiveness analysis... 102 3. Discussion.... 102 4. Reviewer... 103 Acknowledgements... 103 References... 103 Biography.... 103 Abstract The role of laparoscopic surgery for the treatment of colorectal cancer is being explored in a multi-centre, randomised clinical trial in the UK, the MRC CLASICC Trial (Conventional versus Laparoscopic-assisted Surgery in Colorectal Cancer). An important end-point of the trial is the cost-effectiveness of laparoscopic surgery compared with that of conventional open surgery. The economic evaluation of this trial has been modelled on that in a similar trial being conducted in the USA in colon cancer. The aim of this paper is to discuss the rationale for modelling the UK trial on the US trial, and to describe the adaptations necessary for the UK trial. The parallel design of the economic evaluation in both trials will provide a unique opportunity to compare the cost implications of incorporating laparoscopic surgery in the UK and the USA, and to determine any cross-cultural * Corresponding author. Tel.: +44-113-3924411; fax: +44-113-2619006. 1040-8428/00/$ - see front matter 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S1040-8428(99)00057-8
100 M.L. Stead et al. / Critical Re iews in Oncology/Hematology 33 (2000) 99 103 differences. The UK trial will also provide information about the cost-effectiveness of laparoscopic surgery in rectal cancer. 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Laparoscopic surgery; Colorectal cancer; Controlled trial; United Kingdom 1. Introduction A combination of the development of expensive new treatments for patients with diseases such as cancer, the limited budgets available for health care, and the increasing demands on the health system has led to an increase in the number of clinical trials which evaluate the cost of implementing new treatments into the health service. Whilst it is a relatively straightforward matter to determine the actual costs of a new treatment and to compare those with the costs of the standard treatment, there are other components of cost-benefit and cost-effectiveness which require detailed evaluation. This requires that measurement of quality of life be incorporated into the evaluation. The innovative application of laparoscopic procedures has reinvigorated general abdominal surgery in recent years. Surgeons with an interest in colorectal disease have begun to explore the boundaries of this technology within their speciality. Procedures such as laparoscopic cholecystectomy, appendectomy and inguinal hernia repair are being rapidly incorporated into routine clinical practice. The role of laparoscopic surgery in colorectal cancer, however, remains unclear. Colorectal cancer is the second most common malignancy in the western world with over 34 000 new cases per year in the UK. Adequate surgical resection is the only curative treatment, with overall survival rates of just under 50% at 5 years. Surgical technique is critical both in respect of cure and local recurrence. In 1995 the USA National Institute of Health [1] launched a randomised controlled trial to evaluate the role of laparoscopic surgery in the management of patients with colon cancer in the USA. The need to obtain answers to such a technology-led modification of surgical practice which may have a major impact on survival also prompted the Medical Research Council (MRC) to fund a similar trial in the UK. This is the MRC CLASICC Trial (Conventional versus Laparoscopic-assisted Surgery in Colorectal Cancer), which opened to recruitment in 1996. The UK trial protocol incorporates many features of the US protocol in order to permit direct comparisons and to contribute to future meta-analyses. In particular, the health care economics aspect of the UK protocol was specifically modelled on that of the US protocol to enable direct comparisons of the cost implications of incorporating laparoscopic surgery into the management of colorectal cancer in the UK compared with the USA. The design and methodology of the economic evaluation in the US trial has been described in detail previously [2]. The purpose of this paper is to describe the rationale for modelling the UK economic evaluation on the US trial and to describe the adaptations necessary to meet the requirements of the study in the UK. 2. Methods 2.1. Trial design The MRC CLASICC Trial is a randomised, controlled, multi-centre trial of laparoscopic versus conventional open resection for colorectal cancer. It involves 15 hospitals throughout the UK. The Northern and Yorkshire Clinical Trials and Research Unit (an independent central trials office) is responsible for the co-ordination, data management and analysis of the trial. The primary end-points of the trial are pathological resection margins, 30-day post-operative mortality, local recurrence rates, disease-free survival and overall survival at 3 years. The secondary end-points include quality of life and cost-effectiveness which will provide additional information and will allow the investigation of the potential implications of incorporating laparoscopic surgery into routine clinical practice. The trial aims to recruit 1000 patients and currently there are 420 randomised patients, 220 of whom have colon cancer and 200 of whom have rectal cancer. All randomised patients will contribute to the economic evaluation, except for the detailed collection of theatre costs, which will be carried out in a smaller sample of ten consecutive patients per centre per trial arm, i.e. a total of approximately 300 patients. The US trial aims to recruit 1200 patients with colon cancer and there are currently 500 randomised patients. Data on resource utilisation will be collected for all patients. Patient bills and quality of life data will be obtained from sub-samples of approximately 20 and 35% of patients respectively. 2.2. Economic e aluations Two prospective economic evaluations are being carried out in both trials: a cost analysis and a cost-effectiveness analysis. In the cost analysis the differences in costs and resources used in treating patients receiving laparoscopic resection versus conventional open resection will be calculated. In the second analysis a cost-
M.L. Stead et al. / Critical Re iews in Oncology/Hematology 33 (2000) 99 103 101 effectiveness ratio will be calculated by comparing the difference in costs between the two procedures with the difference in benefits, as derived from a Q-TWIST analysis (quality-adjusted time without symptoms of disease and toxicity of treatment) [3,4]. 2.2.1. Cost analysis In a cost analysis the cost of the resources used to treat patients in both trial arms is estimated. In both the US and UK studies a number of key resources expected to differ most between the two surgical procedures is being assessed. These are the number of days in hospital (both for the primary surgery and for post-operative complications), the number of days in an intensive care unit (ICU) following the primary surgery, the length of time in theatre, and details of the laparoscopic supplies used in theatre. In the UK study details of the surgical team and ancillary staff are also being recorded. Within the National Health Service (NHS) in the UK, support networks for patients discharged from hospital are provided. These may be resource intensive and thus have an impact on the cost of treating the patient. Thus, in the UK study, additional resource use data are also being collected. These are the number of GP visits, the number of out-patient visits, use of physiotherapy departments, use of occupational therapy departments/services, the number of times the patient was seen by a district nurse and/or a stoma care nurse, use of hospital transport and use of social services. The use of these associated resources provided by the NHS infrastructure is being assessed for the 3 months prior to the surgery, and then at 2 weeks, 3 months, 6 months and 18 months after the operation. Details of GP visits and hospital stays are also recorded 3 years after the operation. In both studies, data regarding the number of days in hospital for the primary surgery and the number of days in an ICU are collected in the main data collection forms which are completed by the surgeon, the research fellow or the data manager after the patient is discharged from the hospital. These data are being collected for each patient in the trial. Data regarding theatre times, theatre staff and the use of equipment are being collected for the sub-set of patients. Data collection for the UK economic evaluation started in June 1998 and to date 62 patients have been recruited into this sub-group. In the UK, a study-specific health resources questionnaire was designed on which to record the number of days in hospital for post-operative complications and the additional details on resource use. This questionnaire is completed by each patient in hospital before the operation and is sent to the patient s home address at each follow-up time. In contrast, only the number of days in hospital for post-operative complications is recorded by patients in the US study. This is performed at 2 months and 18 months after the operation by a telephone call to the patient from the data manager. The current overall compliance with the health resources questionnaires in the UK is 73%. It is planned that unit costs specific to each centre involved in the UK trial will be assigned at the end of the trial. In the USA mean unit costs for each resource used will be derived from the patient-specific charges in several representative US centres. 2.2.2. Q-TWIST analysis The average quality-adjusted survival in each arm of the trial will be calculated using the Q-TWIST method to provide an estimate of the benefit of each surgical procedure [2 4]. The first stage of the Q-TWIST analysis is to define health states in which patients receiving surgery for colorectal cancer are likely to experience different levels of quality of life. In the US trial five health states were defined, and adopted for the UK trial [2]: 1. the perioperative period: the first 30 days after the operation; 2. adjuvant therapy: the time between the start date of the first therapy to 30 days after the last dose of therapy; 3. time without symptoms and toxicity (TWIST): the time from the end of the perioperative period to recurrence, death or end of follow-up, whichever occurs first, minus the duration of adjuvant therapy; 4. late complications: any complications resulting in hospitalisation will be assigned arbitrarily a duration of 1 month; 5. relapse: the time from the diagnosis of recurrence to death or end of follow-up, whichever occurs first. Utility scores are also required for the Q-TWIST analysis. These are being obtained from patients in both trials, but using slightly different methods. In the UK, utilities are obtained from the Euroqol questionnaire [5] which is a simple standardised instrument designed for use as a measure of health outcome for a wide range of health conditions and treatments, and validated for use in European studies. The Euroqol is a two-part instrument which describes and values healthrelated quality of life and provides a single index value for the patient s current health state. Domain-specific quality of life is assessed using five questions, the results of which may be converted to a single score to estimate patient utilities. In addition, the Euroqol has a visual analogue scale which is a 0 100 rating scale of overall quality of life. This can also be used to estimate patient utilities. In the US trial, patient utilities are assessed in a similar way using the Q-tility Index [6] and a 0 100 rating scale of overall quality of life. In addition, a time trade-off utility question is also included. The Euroqol is completed by the patient at 2 weeks, 3 months, 6 months, 18 months and 3 years post-operatively. The utility to be used for the perioperative
102 M.L. Stead et al. / Critical Re iews in Oncology/Hematology 33 (2000) 99 103 period will be the mean utility score for all patients at the 2-week assessment. This will also be used as the utility for the time spent with late complications. The mean utility score for patients receiving adjuvant therapy at the time of the 3-month assessment will be used as the utility for the adjuvant therapy period. The utility for the TWIST period will be the mean utility score for all patients remaining disease-free at the time of the 18-month assessment. The utility for the relapse health state has been arbitrarily set at 0.5, as in the US trial. This is because, by definition, patients quality of life at the beginning of this period (i.e. immediately before the relapse occurs) will be approximately 1.0 and their quality of life will decline to zero at death. However, in the trial utility scores are obtained from patients until 3 years after surgery or to death, whichever occurs first, and so in some patients utilities will be obtained after relapse. Thus we will be able to calculate utility scores from patients who have relapsed and, if the mean utility score at relapse is different to the arbitrary value of 0.5, then the real utility score will be used in the final analysis. In order to calculate the quality-adjusted survival for each trial arm, the time spent in each of the defined health states is first multiplied by the mean reported utility for that state, and then the weighted survival times are added together. 2.2.3. Cost-effecti eness analysis The cost-effectiveness ratio will be calculated by comparing the incremental cost of the more expensive procedure (as estimated in the cost analysis) with the incremental benefit (as estimated in the Q-TWIST analysis). This ratio can then be compared with generally accepted norms for other medical interventions. The analysis plan for the economics evaluations in the UK trial will be based on that of the US trial [2]. 3. Discussion It is not usually feasible for direct comparisons of health economics data derived from two similar but separate trials to be made. Thus, the design of the economic evaluation in the UK trial was deliberately modelled on that of the US protocol to enable crosscultural comparisons of the cost-effectiveness and cost implications of incorporating laparoscopic surgery for colon cancer into routine clinical practice. The UK trial also includes patients with rectal cancer, and so information on resource use for this group of patients will be available. The key resources identified in previous analyses as differing most between laparoscopic surgery and conventional open surgery [7] are being collected both trials. However, in the UK the majority of health care is provided by the NHS and so we were also interested in the implication of the two surgical techniques on a wider range of resources, such as visits by district nurses and use of social services. The financial implications of involving specialised stoma care nurses for patients receiving laparoscopic surgery is also relevant and must be assessed. The availability of this additional resource use data will enable us to assess the resources used in the two surgical techniques in colorectal cancer, and identify any other UK-specific differences. The timings of the assessment of resource use differs between the two trials. In the UK, use of resources is assessed at 2 weeks, 3 months, 6 months, 18 months and 3 years after the surgery, to mirror the timing of the clinical follow-up. In the US trial resource use is assessed at 2 weeks, 2 months and 18 months post-operatively. However, all patients are required to record resource use for the period of time in between each follow-up and so the data collected up to 18 months in both trials should be comparable. The three-year assessment of GP visits and hospital stays in the UK trial will provide information on any long-term cost implications of the two surgical procedures. Information regarding baseline use of resources is also being collected in the UK for the 3 months prior to the operation to verify that resource use by patients before randomisation is similar in the two trial arms. The overall compliance with the patientcompleted questionnaires in the UK trial is good, with over 70% of patients returning data. If this compliance rate is maintained to the end of the trial, economic data will be available for approximately 700 patients, which is estimated to be more than adequate for the purposes of the study. It is anticipated that differences in theatre and technical costs between the two types of surgery can be estimated from the subsample of 300 patients. Patient utilities to be used in the Q-TWIST analysis are assessed using different but similar tools. In both trials domain-specific quality of life is being assessed, from which a single score as an estimate of utilities will be generated. The domains assessed by the Euroqol and the Q-tility Index are different. The Euroqol assesses mobility, self-care, activity, pain/discomfort and anxiety/depression. In the Q-tility Index activity, daily living, health, support and outlook are assessed. A 0 100 rating scale to provide a single index of overall quality of life is being used in both trials. The US trial also includes a time trade-off utility question. The use of these different methods and instruments to determine patient utilities in the two trials provides an opportunity for a sub-study to examine any differences between the scales.
M.L. Stead et al. / Critical Re iews in Oncology/Hematology 33 (2000) 99 103 103 Unit costs to be used in the cost analyses of multicentre trials can be obtained from national tariffs, from a limited selection of centres involved in the trial or from every centre participating in the trial. It would be preferable to collect unit costs from each centre involved. However, this process itself has cost implications for a trial. In the UK trial an attempt will be made to collect unit costs from every centre participating in the trial to calculate the mean charge for each resource used by patients in each arm of the trial. However, costs derived from two different types of hospital, for example a small district general hospital and a large teaching hospital, will be compared to those derived from all participating centres to determine whether it is always necessary to use all centres. In conclusion, the MRC CLASICC Trial economic evaluation has been designed to obtain information about the cost-effectiveness of laparoscopic surgery and the potential implications of incorporating laparoscopic surgery into routine clinical practice in a way that will enable direct comparisons with a similar trial being carried out in the USA. Additional data relevant to the UK s health service system and to UK patients are also being collected. Utility scores are being obtained from all patients as part of a more extensive quality of life assessment which is an integral part of the trial. Reliable and valid economic information about the relative value for money of laparoscopic surgery will be available when the results of the clinical evaluation are known. The collaboration between the USA and UK groups will enable direct cross-cultural comparisons of both the clinical and economic data and will provide an opportunity to pool data in a meta-analysis. 4. Reviewer This paper was reviewed by Richard Stephens, Medical Statistician, MRC Cancer Trials Office, 5 Shaftesbury Road, Cambridge CB2 2BW, UK. Acknowledgements We would like to thank Jane Weeks for her help in the preparation of this manuscript and for the original design of the economic evaluation; Richard Gelber for designing the Q-TWIST analysis; Heidi Nelson for allowing us to model our protocol on the US protocol; Trish Shevlin for commenting on a first draft of this manuscript; and Alice Hamar for her secretarial support. References [1] Nelson H, Weeks J, Wieand HS. Proposed phase III trial comparing laparoscopic-assisted colectomy versus open colectomy for colon cancer. JNCI Monogr 1995;19:51 6. [2] Weeks J. Taking quality of life into account in health economic analyses. Monogr Natl Cancer Inst 1996;20:23 7. [3] Gelber RD, Goldhirsch A. A new endpoint for the assessment of adjuvant therapy in post-menopausal women with operable breast cancer. J Clin Oncol 1986;44:1772 9. [4] Goldhirsch A, Gelber RD, Simes RJ, et al. Costs and benefits of adjuvant therapy in breast cancer; a quality adjusted survival analysis. J Clin Oncol 1989;7:36 44. [5] Euroqol Group. Euroqol-a new facility for the measurement of health-related quality of life., Health Policy 1990;16:199 208. [6] Weeks J, O Leary J, Fairclough D, et al. The Q-tility Index: a new tool for assessing health-related quality of life and utilities in clinical trials and clinical practice. Proc ASCO 1994;13:436. [7] Senagore AJ, Luchtefeld MA, Mackeigan JM, et al. Open colectomy versus laparoscopic colectomy: are there differences? Am Surg 1993;59:549 54. Biography Maxine Stead graduated in 1993 with a IIi in Pharmacology and Physiology from the University of Leeds. She has worked at the Northern and Yorkshire Clinical Trials and Research Unit since leaving university, where she has been the trial co-ordinator for the MRC CLASICC trial. She has also been carrying out a part time Ph.D. investigating quality of life issues for patients with multiple myeloma and ovarian cancer..