|Economic evaluation of gemcitabine in the treatment of pancreatic cancer in the UK
|Aristides M, Lees M, Botwood N, McKendrick J, Stephenson D A, Maniadakis N
This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn.
The use of gemcitabine as chemotherapy in patients with pancreatic cancer. Gemcitabine was administered at a dose of 1,000 mg/m2 once weekly for 7 weeks, followed by once weekly for every 3 out of 4 weeks. It was compared with 5-fluorouracil (5-FU) at a dose of 600 mg/m2 once weekly.
Economic study type
The population comprised patients with advanced symptomatic pancreatic cancer who had not received chemotherapy.
The setting was tertiary care. The economic study was carried out in the UK.
Dates to which data relate
The effectiveness evidence was from a trial published in 1997 (Burris et al., see Other Publications of Related Interest). Resource use was taken from the 1997 trial and prices were expressed in year 2000 values.
Source of effectiveness data
The effectiveness data were derived from a single randomised controlled trial.
Link between effectiveness and cost data
The costing appears to have been undertaken retrospectively on the same patient sample as that used in the effectiveness study.
Very little information on the study sample was provided in this paper. Most of the following information was obtained from the original effectiveness paper (Burris et al., see Other Publications of Related Interest).
Sample size calculations were not reported in the published study. Patients with a pathologic diagnosis of pancreas cancer that was locally advanced or metastatic, and not suitable for curative surgical resection, were selected for the study. The patients underwent a pain stabilisation period of up to 7 days before the trial started. The study sample appears to have been appropriate for the clinical study question. Overall, 160 patients were included in the study. There were 63 in the gemcitabine group, 63 in the 5-FU group, and 34 who did not proceed beyond the pain stabilisation period. These included 17 patients who were no longer eligible, 10 in whom pain control could not be achieved, 4 who developed other medical problems and 3 who decided not to proceed.
Very little information on the study design was provided in this paper. Most of the following information was obtained from the original effectiveness paper (Burris et al., see Other Publications of Related Interest).
The trial was a single-blind, randomised controlled trial (RCT). The patients did not know which drug they had been assigned to receive. Randomisation took place after the pain stabilisation period and immediately before starting the drug treatment under study, and was performed at a central location. The unit of randomisation was not reported. The patients were followed up for at least 12 months. No patients were reported to have been lost to follow-up.
Analysis of effectiveness
The authors did not state whether an intention to treat analysis was used. The primary outcome was a combined measure of clinical benefit. This was derived from the measurement of pain (assessed by pain intensity and analgesic consumption), functional impairment (assessed by Karnofsky performance status) and weight loss. For patients to achieve an overall rating of positive clinical benefit response, they had to be positive for at least one parameter, without being negative for the others, and the improvement had to last for at least 4 weeks. Mean survival and mean time to progressive disease and toxicity were also calculated using Kaplan-Meier survival curves. The groups were stated to be well balanced for prognostic factors at baseline, although the baseline characteristics were not mentioned in the published economic evaluation.
Clinical benefit response was reported by 23.8% of gemcitabine patients and 4.8% of 5-FU patients.
The median survival duration was 5.65 months for gemcitabine patients and 3.72 months for 5-FU patients.
Progression-free survival was 2.44 months for gemcitabine patients and 1.27 months for 5-FU patients.
The authors concluded that gemcitabine was more effective than 5-FU in alleviating some disease-related symptoms in patients with advanced symptomatic pancreas cancer. Gemcitabine also conferred a modest survival benefit over 5-FU.
Measure of benefits used in the economic analysis
The outcome measures used in the economic model were the life-years gained (LYG), progression-free LYG and clinical benefit response. Clinical benefit response was a composite measure of pain intensity, analgesic use, Karnofsky performance score and body weight change. To estimate the difference between treatments and the variance, life-years and progression-free life-years were calculated from the area under the Kaplan-Meier survival curves that were presented in the trial. Incremental survival benefit was the difference between the mean survival times, with a variance estimated as the sum of the two group variances.
The economic evaluation was carried out from the perspective of the UK NHS. Therefore, only the direct health care costs borne by the NHS were included. The total treatment costs included all direct costs associated with pancreatic cancer treatment, such as chemotherapy, infusion, hospitalisation, visits to health professionals and use of concomitant medications.
Individual patient data from the trial were used to calculate the number of vials of chemotherapy drug used. The doses were rounded up to account for wastage, and it was assumed that the drugs were given on an outpatient basis. The use of concomitant medications (costed from the British National Formulary) and the frequency of toxicity-related hospitalisations were also calculated using individual patient data from the trial.
The cause and duration of each hospitalisation was recorded and the costs were applied using a case-mix approach. Visits to oncologists were assumed to occur at the same time as chemotherapy administration. The patients were assumed to visit their general practitioner once monthly, before and after disease progression.
No extrapolations beyond the timeframe of the trial were included (maximum 18 months). The outcomes and costs were discounted at a rate of 6% per annum. Any costs not initially expressed in year 2000 values were adjusted for inflation using the consumer price index of the Office for National Statistics. The resource quantities were not reported in the evaluation, but the authors stated that resource use data from the published study were combined with unit cost data from UK national sources (see Other Publications of Related Interest).
Statistical analysis of costs
The costs were presented as point estimates in the economic evaluation.
The indirect costs were not included in the evaluation.
Sensitivity analyses were conducted on key assumptions and values. The upper and lower values from the 95% confidence intervals (CIs) of the key values were used.
Estimated benefits used in the economic analysis
There was a survival benefit with gemcitabine of 0.189 life-years (95% CI: 0.059, 0.318) and of 0.116 progression-free life-years (95% CI: 0.038, 0.194). Clinical benefit response showed an increase of 19% (95% CI: 7.3, 30.8) with gemcitabine. Side effects were considered to some extent in the clinical benefit response measure. The benefits were measured only for the duration of the trial (18 months).
The average total cost was 3,568.56 per patient treated with gemcitabine and 1,261.57 per patient treated with 5-FU. The incremental total cost was 2,306.99 per patient.
The costs of treating adverse events were included in the analysis.
Synthesis of costs and benefits
The average and incremental cost-effectiveness ratios were derived from the total costs and the outcome measures.
The incremental cost for gemcitabine over 5-FU was 12,206 per LYG, 19,888 per progression-free LYG and 12,172 per clinical benefit responder.
Sensitivity analyses demonstrated that the results were relatively robust and insensitive to variations in the key parameters. The most sensitive parameter was the survival benefit. The incremental cost per LYG was 10,062 at the upper 95% CI and 25,316 at the lower 95% CI.
The treatment of advanced pancreatic cancer with gemcitabine is cost-effective. However, the gain in survival with gemcitabine results in greater use of health care resources. Despite the significant gain in life expectancy, the survival at 1 year was relatively low (around 20%) in this study. The potential overall financial impact on the National Health Service of listing gemcitabine for first-line therapy for pancreatic cancer is likely to be modest.
CRD COMMENTARY - Selection of comparators
The comparator was justified on the grounds that it was the most commonly used chemotherapy. However, there are several alternative drugs and they may be relevant. You should decide whether 5-FU is a widely used health technology in your own setting.
Validity of estimate of measure of effectiveness
The analysis used an RCT, which was appropriate for the study question. Few details of the study design and study sample were reported in the published economic evaluation; reference was only made to the original clinical paper (Burris et al., see Other Publications of Related Interest). It was therefore impossible to evaluate the validity of the clinical evidence from this paper
Validity of estimate of measure of benefit
The estimates of measure of benefit were the LYG, progression-free LYG and clinical benefit responders. These were obtained directly from the effectiveness analysis, with some transformation of the data. This selection of estimate was justified, although quality of life is an important outcome consideration in this case. It was unclear in the reporting of the benefits whether or not the results, as reported, accounted for the intended discounting.
Validity of estimate of costs
The NHS perspective was adopted and all the categories of costs relevant to the perspective adopted were included in the analysis. The authors acknowledged that the time horizon may have been insufficient to capture any extra costs to the NHS due to increased survival. Resource use quantities were taken retrospectively from the effectiveness trial, which is second best to prospective costing. No statistical analysis of the quantities was performed. Discounting was appropriately conducted. The unit costs were taken from published sources and a sensitivity analysis of the unit costs was conducted. The unit costs were reported, as was the date to which the prices related. This enhances the generalisability of the results. However, it would have been useful had resource use also been reported.
The authors made reference to only one other costing analysis on this subject. The results were comparable to those in this study. The authors noted that the findings of the effectiveness trial may not be generalisable to other settings. They stated that they conducted extensive sensitivity analyses, although only the main results were reported. The study included patients with advanced pancreatic cancer and the authors' conclusions reflected this. Four of the six authors of the economic evaluation worked at Eli Lilly and Company, the manufacturers of gemcitabine, and the study was supported by funding from Eli Lilly. The results were adequately reported, but the information on the quality of the evidence was scant.
Implications of the study
The authors make no explicit recommendations for changes in policy or practice, or for further research.
Source of funding
Supported by Eli Lilly and Company Ltd.
Aristides M, Lees M, Botwood N, McKendrick J, Stephenson D A, Maniadakis N. Economic evaluation of gemcitabine in the treatment of pancreatic cancer in the UK. European Journal of Health Economics 2003; 4: 216-221
Other publications of related interest
Burris HA, Moore MJ, Anderson J, et al. Improvement in survival and clinical benefit with gemcitabine as first line therapy for patients with advanced pancreatic cancer: a randomised trial. Journal of Clinical Oncology 1997;15:2403-13.
British Medical Association, Royal Pharmaceutical Society of Great Britain. British National Formulary 2000;40.
Netten A, Dennett J, Knight J. Unit costs of health and social care. Canterbury: University of Kent, Personal Social Services Research Unit; 1999.
Subject indexing assigned by NLM
Date bibliographic record published
Date abstract record published