clinical and methodological issues in a cost-utility analysis alongside a
randomized clinical trial in patients with rectal cancer undergoing
total mesorectal excision
Brink, Mandy van den
Citation
Brink, M. van den. (2005, June 28). Economic evaluation of preoperative radiotherapy in
rectal cancer : clinical and methodological issues in a cost-utility analysis alongside a randomized clinical trial in patients with rectal cancer undergoing total mesorectal excision. Retrieved from https://hdl.handle.net/1887/4273
Version: Corrected Publisher’s Version
License: Licence agreement concerning inclusion of doctoral thesis in theInstitutional Repository of the University of Leiden Downloaded from: https://hdl.handle.net/1887/4273
COST-
UTI
LI
TY ANALYSI
S OF
PREOPERATI
VE RADI
OTHERAPY I
N
PATI
ENTS W I
TH RECTAL CANCER
UNDERGOI
NG TOTAL MESORECTAL
EXCI
SI
ON:
A STUDY OF THE DUTCH
COLORECTAL CANCER GROUP
Mandy van den Brink, Wilbert B. van den Hout, Anne M. Stiggelbout, Elma Klein Kranenbarg, Corrie A.M. Marijnen, Cornelis J.H. van de Velde, Job Kievit
Abstract
Purpose To compare the societal costs and the (quality adjusted) life expectancy of patients with rectal cancer undergoing total mesorectal excision (TME) with or without short-term preoperative radiotherapy (5×5 Gy).
Methods We used a Markov model to project the clinical and economic outcomes of preoperative radiotherapy. Data on local recurrence rates, quality of life, and costs were obtained from the patients of a multi-center randomized clinical trial. In this trial, 1861 patients with resectable rectal cancer from 108 hospitals were randomized for TME-surgery with or without preoperative radiotherapy. Outcome measures of the model were life expectancy, quality adjusted life expectancy, lifetime costs per patient, and the incremental cost-effectiveness ratio.
Results The base case model estimates that the loss of quality of life due to preoperative radiotherapy is outweighed by the gain in life expectancy. Life expectancy increases by 0.67 years, quality adjusted life expectancy by 0.39 years, and costs increase by $9,800 per patient. The corresponding cost-effectiveness ratio is $25,100 per QALY. Sensitivity analyses indicate that the cost-effectiveness ratio remains acceptable under a wide range of assumptions.
Introduction
In the treatment of rectal cancer, local recurrence is of major concern, as it causes severe disabling symptoms, is difficult to treat, and is often fatal (1). The best chance for cure in patients presenting with rectal cancer is standardized total mesorectal excision (TME-surgery). Since the introduction of TME-surgery, local recurrence rates have decreased to rates as low as 5%-8% (2;3), as compared to 15%-45% after conventional surgery (4-6). Two recent meta-analyses have shown that adjuvant radiotherapy also reduces local recurrences rates by almost 50% and reduces overall mortality by 2%-10% (7;8). Preoperative radiotherapy (PRT) with higher biologically effective doses (of at least 30 Gy) showed larger reductions than preoperative radiotherapy with lower doses and than postoperative radiotherapy. However, the trials included in these meta-analyses all started before the broad introduction of surgery. The only randomized comparison of TME-surgery with or without preoperative radiotherapy is the trial conducted by the Dutch Colorectal Cancer Group (9;10). At two years follow-up the estimated local recurrence rates in this trial were 2.4% with PRT, compared to 8.2% without PRT. This relative reduction of local recurrence is similar to that in the meta-analyses, but the absolute reduction is smaller and the current follow-up does not show a difference in survival.
From a health policy perspective, clinical benefits of treatments should be estimated in terms of survival benefits, and balanced against the effects on quality of life and costs. This is especially true in oncology, where costs and other disadvantages of treatment can be considerable (11;12). PRT has been associated with increased postoperative morbidity such as perineal dehiscence, wound infection, cardiovascular complications and prolonged hospital stay (13-19). Information on long-term complications and quality of life is scarce (20).
The objective of this study was to compare the societal costs and the (quality adjusted) life expectancy of patients with rectal cancer undergoing TME with or without PRT.
Methods
We analyzed lifetime costs, life years (LYs) and quality adjusted life years (QALYs) for patients undergoing TME-surgery with or without PRT, from a societal perspective following the recommendations from the Panel on Cost-Effectiveness in Health and Medicine (21).
months (range 13-68 months). A state-transition Markov model was developed to integrate all available data from the TME-study, and to facilitate long-term extrapolation and sensitivity analyses, using the software package DATA 3.5 (TreeAge Software, Williamstown, Mass). Future costs and QALYs were both discounted at 3% (21).
Model overview
In a Markov model, a cohort of patients moves through a set of health states, defined to capture important clinical characteristics (22). Starting from the date of TME-surgery, the lifelong time horizon was divided into monthly cycles. After each cycle, transitions can occur from one health state to another, and costs, LYs and QALYs are accrued.
A schematic representation of the model is shown in figure 1. For the time from randomization until TME-surgery, pre-treatment and treatment costs, LYs and QALYs were assigned by randomization group. After TME-surgery, patients start in one of the initial states R0 (microscopically negative resection margins of more than 1 mm), R1 (microscopically positive resection margins at 1 mm or less), R2 (macroscopically incomplete local resection or distant metastases at surgery), or death (for all patients dying before leaving the hospital). In case of recurrence, the R0 and R1 patients move to the states local recurrence, distant recurrence or local and distant recurrence. Recurrences were not modelled explicitly for R2 patients, because any reduction in local recurrence rates due to PRT is not expected to improve survival. From rectal cancer or non-rectal cancer related causes, all patients eventually die.
Initial probabilities and transition rates
The initial probabilities and transition rates for the Markov model were estimated from all TME patients. For the initial R0, R1, and R2 states, transition rates were estimated using Gompertz distributions to allow for rates that increase or decrease with time. For the non-initial states, constant transition rates were used. All analyses were performed for patients with specific initial age, and weighted for the age distribution in the TME-study (mean age 64 years, SD 11 years).
For the recurrence rates, proportional hazards were considered for PRT, for the inferior margin (the distance from the anal verge to the tumor), and for their interaction. These variables have been shown to be significant predictors of local recurrence risk (10), and may be determined by preoperative staging procedures. For the mortality rates, proportional hazards for PRT and age were considered. For the base case analysis, all non-significant proportional hazards were excluded (stepwise backward-forward regression, p>0.05 for removal and p<0.05 for entry). Table 1 summarizes the parameter estimates. Sensitivity analyses were performed on the inclusion of all estimated proportional hazards (whether significant or not), and on the proportional hazard of PRT for the local recurrence rate after R0 and R1 resection (95% confidence intervals 0.15-0.65 and 0.10-0.72 respectively).
Table 1. Estimated initial probabilities and transition rates for the base case analysis
State* Initial probability Transition rate† to LR‡ to DR to LDR to RCM to NRM§ R0 0.750 PRT+TME: 0.010 TME: 0.033 [-0.571] 0.069 [-0.286] PRT+TME: 0.001 TME: 0.012 [-0.627] 0.006 [-0.090] 0.0076 [0.006] R1 0.139 PRT+TME: 0.038 TME: 0.142 [-0.490] 0.351 [-0.680] 0.026 [-0.447] 0.046 [-5.648] 0.013 [-0.351] R2 0.078 - - - 0.425 [-0.133] 0.014 [0.022] LR - - - PRT+TME: 0.967 TME: 0.349 0.219 0.000 DR - - - 0.049 0.423 0.020 LDR - - - - 0.656 0.034 Death 0.033 - - - -
-* LR=Local recurrence, DR=Distant recurrence, LDR=Local and distant recurrence, RCM=Rectal Cancer mortality and NRM=Non-rectal cancer mortality.
† Numbers between square brackets denote the Gompertz decay parameter. A mortality rate r with decay parameter d denotes the time-dependent transition rate r×exp(d×time). For recurrences, the transition rates are r×exp(d×(time -0.25)), with r=0 for time<0.25 to account for the absence of recurrences during the initial months in the TME-study. ‡ For R1-patients local recurrence rates were significantly dependent of the inferior margin. Compared to the presented
average rates, the respective proportional hazards for patients with inferior margin in the categories 0-5 cm, 5-10 cm, and 10-15 cm, were 1.41, 0.89, and 0.23. The interaction with PRT was not significant.
The observed non-rectal cancer mortality in the TME study was low compared to the average Dutch mortality, possibly due to the exclusion of patients with co-morbidity. Therefore, long term non-rectal cancer mortality rates were obtained from 1995-1999 Dutch life tables (provided by Statistics Netherlands and weighted for the 67% male TME-study population). We used the short term non-rectal cancer mortality from the TME-study (until a follow-up time of 4 years), the long term mortality from Dutch life tables (after a follow-up time of 8 years), with linear interpolation in between.
In our base case model a reduction in local recurrence rates by PRT will automatically lead to a gain in survival. To investigate the possibility that reduced local recurrence rates might not lead to a difference in survival, we conducted an alternative analysis in which all mortality rates after PRT were uniformly increased, to such an extent that the life expectancy for both randomization groups was identical.
To investigate the impact of local recurrence risk on the estimated QALYs and cost-effectiveness (CE) ratio, the local recurrence rates were uniformly varied in both randomization groups. A plot of the QALYs and CE-ratios by local recurrence risk allows the reader to assess the effect of PRT for subgroups of patients, based on their 5-year local recurrence risk without PRT. This figure can be used to assess the (cost-)effectiveness of PRT in particular groups of patients that were not explicitly included in our Markov model (e.g. by tumor stage or inferior margin). Note however that the estimates provided by the plot are subject to larger error than the other results presented in this paper, because subgroups may differ by more than only their local recurrence rate.
Health related quality of life
The assessment of quality of life was based on the 1530 Dutch patients (the Dutch sample). Patients were asked to fill out a quality of life questionnaire (23) before treatment and at 3, 6, 12, 18 and 24 months after surgery. The questionnaire included several descriptive and disease specific questions, a 100 mm visual analogue scale (VAS) valuing health from 0 (death) to 100 (perfect health), and the EuroQoL questionnaire (24). The EuroQoL questionnaire consists of 5 descriptive items (mobility, self-care, daily activities, pain/discomfort, and anxiety/depression), rated on 3-point scales (no, some, or extreme problems). From the EuroQoL questionnaire TTO-values as assigned by the general public were inferred (25). To obtain more detailed data for the cost-utility analysis (CUA), from February 1999, Dutch patients (the CUA sample) were asked to participate in interviews, before surgery, and at 3 and 12 months after surgery. During 308 interviews in 112 patients' valuations of quality of life were assessed using the VAS and the time trade-off method (TTO). During hospitalization the CUA sample was asked to fill out the VAS and a EuroQoL questionnaire weekly.
values measured during hospitalization in the CUA sample were used. Utility values for the R0 and R1 states until 9 months after TME-surgery differed significantly by randomization group, time and stoma type (i.e. patients with no stoma, a diverting stoma, a removed diverting stoma and a permanent stoma, weighed by their distribution over time). After 9 months, utility values remained relatively constant over time and no significant differences were observed between randomization groups. Therefore, for the time following 9 months after surgery, we calculated a weighted average of the 12, 18, and 24 month scores and assumed that utility values within each health state were constant thereafter. Utility estimates for the R2 state and all recurrence states were calculated independent of time, because of the smaller number of observations. Utility values for the R2, distant recurrence, and combined local and distant recurrence states only differed by randomization group. For the local recurrence state, utilities did not differ by randomization group.
Table 2. Utility estimates for the base case analysis*
PRT+TME TME
Stoma type† All NS RS DS PS All NS RS DS PS
State and time
Randomization to PRT or surgery 0.78 0.78 PRT to surgery 0.70 - R0 Surgery to discharge 0.11 0.21 Discharge to 4.5 months 0.77 0.71 0.83 0.74 0.89 0.85 0.79 0.80 4.5 to 9 months 0.85 0.85 0.78 0.79 0.90 0.80 0.76 0.85 >9 months 0.86 0.86 0.80 0.85 0.86 0.86 0.80 0.85 R1 Surgery to discharge 0.09 0.17 Discharge to 4.5 months 0.88 0.73 0.77 0.80 0.63 0.69 0.76 0.88 4.5 to 9 months 0.86 0.82 0.81 0.88 0.85 1.00 0.84 0.86 >9 months. 0.89 0.89 0.75 0.88 0.89 0.89 0.75 0.88 R2 0.73 0.80 Local recurrence‡ 0.67 0.67 Distant recurrence 0.70 0.64
Local and distant recurrence 0.48 0.45
* Shown are the average utility estimates from the general public based on the EuroQoL questionnaire
† NS = no stoma, RS = removed diverting stoma, DS = diverting stoma, PS = permanent stoma
We performed sensitivity analysis on the utility estimate for the local recurrence state, since local recurrence is the main outcome measure of the TME-study and since this utility estimate was based on a relatively small number of 15 observations available from 12 local recurrence patients without a distant recurrence (median follow-up time after local recurrence diagnosis ranges from of 0.7 to 10.6 months). We also analyzed the impact of using the patients' preferences, based on their VAS scores. These VAS scores were transformed to TTO scores using the power transformation TTO = 1-(1-VAS)1.75, estimated from the CUA sample (26;27).
Costs
From the health care perspective, costs include primary treatment, continuing care, and recurrence treatment. From the societal perspective, costs in addition include productivity losses (for both paid and unpaid labor), informal care costs, travel costs, time costs, and out-of-pocket costs. For the base case analysis, all costs were included. Sensitivity analysis was performed on the exclusion of non-health care costs. If not mentioned otherwise, volumes of health care utilization were multiplied by standard cost prices if available (28), or by standard tariffs from the Dutch National Health Authority, as a proxy for true resource costs. All costs were updated to the year 2002 using the price index rate for the Dutch health care sector (obtained from Statistics Netherlands), and converted from Euros to U.S. dollars using the exchange rate on November 1, 2002 (1 Euro = 1 U.S. dollar).
Costs of primary treatment were estimated conditionally on randomization group and R-status. For all Dutch patients, volume data on preoperative screening, type of TME-surgery (e.g. low anterior resection, abdomino-perineal resection), duration of first hospital stay including re-interventions, and postoperative adjuvant radiotherapy and chemotherapy were available. For the costs of PRT we performed a true resource cost calculation using the cost pricing analysis of the Dutch bone metastasis study (29;30). In order to adapt the calculation for specific features of the irradiation used in the TME-study, all Dutch radiotherapy institutes participating in the TME-study (n=18, 100% response) filled out a mailed questionnaire on the use of CT-planning, shielding, and portal imaging. Sensitivity analysis was performed on the costs of PRT by increasing and decreasing the estimated costs by 50%. The costs of continuing care were estimated as time-dependent monthly costs, conditionally on randomization and model state. Data on hospital readmissions were available for the total Dutch sample. Volumes of outpatient visits and follow-up investigations were modeled according to the protocol of the TME-study. Visits to medical specialists, GP’s, paramedics, hours of home help and district nursing were obtained from weekly (0 to 3 months after surgery) and monthly (3-12 months after surgery) diaries filled out by the CUA sample. Medication use and the use of stoma care products were estimated from the pharmacists' and stoma suppliers' registrations of patients in the CUA sample.
retrospective examination of the medical records of all 69 patients diagnosed with a local recurrence (with and without distant recurrence, median follow-up time after recurrence diagnosis 6 months, range 0-29 months). These costs included the costs of diagnostic procedures, outpatient visits, surgery, radiotherapy, chemotherapy, hospitalization, and admissions to nursing homes. For distant recurrences, total costs were estimated by multiplying the available re-admission costs by a factor 2.2 (in accordance with the ratio of total and re-admission costs for the 69 local recurrence patients).
The quality of life questionnaire, administered in the Dutch sample, included questions on paid productivity losses. Paid labor was significantly lower in the PRT+TME group, even two years after surgery. Unpaid productivity losses were assessed in the CUA sample using the Health and Labour questionnaire (31) (before treatment, and at 3 and 12 months after TME-surgery). There was a trend towards less unpaid labor per year in the PRT+TME group. After discharge, volumes of paid and unpaid labor were modeled time-dependently, conditionally on randomization group and model state. Paid labor was valued using the friction cost method (32). In this method each employee is considered replaceable and productivity costs are calculated only for the friction period of 4 months, that is the estimated time needed to find a replacement (32). Sensitivity analysis was performed on the exclusion of productivity costs.
Hours of informal care and out-of-pocket costs (e.g. clothing) were retrieved from the diaries in the CUA sample. Time and travel costs were based on average durations and travel distances for different types of health care in the Netherlands. During hospitalization, time costs were calculated based on 8 hours per day. The valuation of time was assessed in the CUA sample using the Willingness-to-Pay method, eliciting the amount of money that patients would be willing to pay to save the time needed for an outpatient visit to a medical specialist.
Results
Validation of the model
Figure 2. Results from the base case model as compared to the observed results from the TME-study. Observed and modeled survival, cumulative distant recurrence rate and cumulative local recurrence rate (including combined recurrence) are shown for a 65-year old person. The observed results are the Kaplan-Meier curves up to 6 years. The results from the model are the smooth curves that extend beyond 6 years.
Base case analysis
The model estimates that the average life expectancy of patients undergoing TME-surgery with or without PRT is about 13 years (table 3), compared to 18 years if they had not had rectal cancer. About 35% of the patients eventually die from causes related to rectal cancer, with or without diagnosed recurrence. The addition of PRT leads to an estimated decrease of the long term probability of local recurrence by 5.2%. The model estimates that the reduction of local recurrences will translate into a survival benefit of 2.8% and 3.2% after 5 and 10 years respectively. The mean gain in life expectancy and quality adjusted life expectancy respectively are estimated at about 8 months and 5 months. The quality adjusted gain is smaller, because of the quality adjustment during gained life years, but also because of the short term quality loss due to the PRT.
Table 3. Effectiveness and cost-effectiveness for the base case analysis
PRT+TME TME Difference
Local recurrence† (%)
Distant recurrence‡ (%)
Any recurrence (%)
Mortality related to rectal cancer (%)
4.4 17.9 22.3 32.7 9.6 17.1 26.7 36.9 - 5.2 0.8 - 4.4 -4.2
Life expectancy (years)§
QALYs
Costs (U.S. dollar) Costs/QALY 13.59 8.34 115,000 12.92 7.95 105,200 0.67 0.39 9,800 25,100 † Lifelong, including combined local and distant recurrences
‡ Lifelong, excluding combined local and distant recurrences § Undiscounted
The addition of PRT leads to an estimated increase of the societal costs by $9,800 (table 4). This increase is not only caused by the costs of the PRT itself, but also by a slightly longer hospitalization period after TME-surgery, by considerably higher continuing care costs (hospital admissions, outpatient visits, home help, district nursing, and medication use) and by higher time costs. These higher costs are partly compensated by lower costs of postoperative adjuvant therapy, local recurrence treatment, and unpaid labor. Patients in the TME alone group more often received post-operative radiotherapy. The lower local recurrence costs in the PRT+TME group are not only explained by the reduction of local recurrences but also by less intensive treatment after recurrence for these patients. Costs of unpaid labor are lower in the PRT+TME group, because the annual unpaid productivity losses due to PRT are more than compensated by the unpaid labor performed during gained life years.
Table 4. Estimates of volumes and costs for the base case analysis*
Volumes Costs($)
PRT+TME TME PRT+TME TME Diff.
Initial treatment
Preoperative screening (nr. of tests) Preoperative radiotherapy (%)† TME-surgery (hospital days) Postoperative radiotherapy (%) Postoperative chemotherapy (%)‡ 6.5 96.5 22.6 0.6 4.3 6.3 1.6 21.1 9.9 4.9 900 2,600 9,700 0 200 900 100 9,200 400 200 0 2,500 500 -400 0 Continuing care
Hospital admissions (nr. & hospital days) Follow-up (outpatient visits)
GP and paramedics (contacts) Home help (hours)
District nursing (hours)
Stoma care (% stoma at 0, 1 & 10 years) Medication§ 3.7 & 48 155 160 355 100 77, 45 & 33 -2.9 & 36 139 165 260 35 77, 45 & 34 -12,700 6,000 2,300 4,800 2,800 12,900 6,800 9,700 5,400 2,400 3,600 1,000 12,500 4,300 3,000 600 -100 1,200 1,800 400 2,500 Recurrence treatment Local recurrence (%) Distant recurrence (%) Local and distant recurrence (%)
1.8 19.8 4.1 6.0 19.0 7.3 300 3,600 700 2,300 3,400 1,400 -2,000 200 -700
Total health care costs 66,300 56,800 9,500
Paid productivity (%) 4 4 3,200 3,200 0
Unpaid productivity (hours) 5,200 5,700 30,600 33,200 -2,600
Informal care (hours) 300 170 1,900 1,200 700
Travel costs (miles) 3,800 3,400 700 600 100
Time costs (hours) 1,100 900 9,900 8,400 1,500
Out-of-pocket expenses§ - - 2,400 1,800 600
Total non-health care costs 48,700 48,400 300
Total societal costs 115,000 105,200 9,800
* Shown are the mean lifelong estimates by randomization group. The units for the estimates of volumes are shown between brackets. Estimates of costs are in U.S. dollars, discounted with 3%.
† The costs of preoperative radiotherapy were calculated for the patients that actually received radiotherapy, taking into account schedules other than 5×5 Gy. A small number of patients in the TME-surgery alone group were irradiated, although not randomized for PRT.
‡ Including combined radio- and chemotherapy.
Subgroup and sensitivity analyses
To investigate the potential benefit of improved diagnostics, subgroup analyses by R-status were performed (table 5 and figure 3). PRT was estimated to be most cost-effective for R1 patients, and not effective for R2 patients. Irradiation of only R0, or only R1 patients would lead to CE-ratios of $29,700 and $3,600 per QALY respectively.
Figure 3. Estimated effects of PRT on QALYs and the CE-ratio by the 5-year local recurrence risk without PRT. Difference in QALYs by PRT and the CE-ratio, for 5-year local recurrence risks without PRT ranging from 0% to 35%. The dotted lines indicate the results for the base case analysis and for several subgroups (TNM stage I, R0 and R1 patients).
Because the objective of PRT is to prevent local recurrences, the local recurrence rate is also the main determinant of the QALY gain and of the cost-effectiveness ratio. Figure 3 shows the effect of variations in the baseline 5-year local recurrence risk without PRT on these outcomes. For our base case analysis, the 5-year local recurrence risk without PRT was 8.6%. The figure shows the estimated QALY gain of 0.39 and the CE-ratio of $25,100 per QALY. For subgroups with different local recurrence risks the figure predicts different outcome estimates. As an illustration, consider patients with tumor node metastases (TNM) stage I tumors with a 5-year local recurrence risk of 1.7% (10). Figure 3 predicts that for these patients PRT still provides a QALY gain of 0.08, but with a cost-effectiveness ratio of more than $100,000 per QALY. This way the figure can be used as a method to assess the (cost-) effectiveness of PRT for other subgroups of patients, based on their 5-year local recurrence risk without PRT.
-0.5 0.0 0.5 1.0 1.5 2.0 0 5 10 15 20 25 30 35
5-year local recurrence risk without PRT
For older ages, the cost-effectiveness ratio is less favorable, but still in favor of PRT (table 5). Although the incremental costs are higher for younger ages, due to lifelong higher costs of continuing care, the CE-ratio is more favorable for younger patients, because they benefit more from PRT than older patients as a result of a longer life expectancy.
Table 5. Results of the sensitivity analyses
Incremental costs* Incremental QALYs CE-ratio (costs/QALY)
Base case analysis 9,800 0.39 25,100
R-status (base case is observed distribution) R0 only R1 only R2 only 11,600 3,100 6,700 0.39 0.86 -0.25 29,700 3,600 No PRT Age (base case is observed distribution)
45 years 65 years 85 years 12,800 9,700 6,400 0.66 0.38 0.15 19,400 25,500 42,700 Proportional hazard of PRT for local recurrence
(base case 0.31 for R0 and 0.27 for R1)
0.15 for R0 and 0.10 for R1 (lower bounds 95% CI) 0.65 for R0 and 0.72 for R1 (upper bounds 95% CI)
9,100 11,400 0.47 0.22 19,400 51,800 Inclusion of all non-significant proportional hazards 9,900 0.23 43,000 No survival benefit of PRT (base case survival predicted by model
based on reduced recurrence rates) 9,500 0.07 135,700
Utility for local recurrence (base case 0.67) 0.25 - 0.5 9,800 9,800 0.43 0.49 22,800 20,000 Patients' preferences obtained from the VAS 9,800 0.43 22,800
No productivity costs 12,400 0.39 31,800
Health care perspective 9,500 0.39 24,400
Duration of randomization differences in transition rates, utilities and costs (base case infinite)
5 years 10 years 3,600 5,800 0.38 0.40 9,500 14,500 Cost of PRT (base case $2,400)
50% decrease = $1,200 50% increase = $3,600 8,500 11,000 0.39 0.39 21,700 28,200 Discount rate (base case 3%)
The base case proportional hazards of PRT for local recurrences are 0.31 after an R0-resection and 0.27 after an R1-R0-resection. As shown in table 5, the CE-ratio ranges from $19,400 to $51,800 per QALY if we vary these proportional hazards over the 95% confidence intervals. Inclusion of all estimated significant as well as non-significant proportional hazards decreases the effectiveness of PRT, resulting in a CE-ratio of $43,000 per QALY. Increasing the mortality rates in the PRT group to balance the life expectancy in both groups still leads to a QALY difference in favor of PRT, but the CE-ratio is far less favorable at $135,700 per QALY. In this analysis, the short term loss in quality of life is 0.002 QALYs, and the long term gain in quality of life because of prevented local recurrences by PRT is 0.074.
The valuation of quality of life after local recurrence was estimated at 0.67, which is more favorable than we had expected. Assuming worse utilities for local recurrence leads to more favorable CE-ratios, down to $20,000 per QALY. Using patients’ valuations of quality of life instead of valuations from the general public also resulted in a more favorable CE-ratio of $22,800 per QALY.
The difference between the base case analysis and the analysis excluding productivity costs is mostly explained by the difference in unpaid labor. The analysis excluding productivity costs leads to less favorable CE-ratios, because of the unpaid labor irradiated patients provide during their improved survival.
The base case analysis was performed from the societal perspective, including both health care and health care costs. The analysis from the health care perspective, excluding non-health care costs, results in a slightly more favorable CE-ratio of $24,400 per QALY. This can be explained by higher informal care costs, travel costs, time costs and out-of-pocket expenses in irradiated patients that outweigh the long term gain in unpaid labour by PRT. Because late effects of radiotherapy cannot be excluded (20), the base case analysis assumed that estimated differences in costs and transition rates continued to exist indefinitely. If these differences were assumed to end after 10 or 5 years, then the CE-ratio would markedly improve to $14,500 and $9,500 per QALY respectively. This improvement is mainly caused by the fact that all costs of continuing care were higher after radiotherapy.
Decreasing and increasing the medical costs of PRT by 50% respectively, resulted in CE-ratios of $21,700 and $28,200 per QALY. Because the differences in costs mostly precede the difference in survival, giving the future more weight (0% discount rate) or less weight (5% discount rate) respectively renders the more favorable CE-ratio of $21,300 per QALY and the less favorable CE-ratio of $27,800 per QALY.
$11,000 (9,800+1,200) and 0.49 (0.39+0.10) respectively, resulting in a CE-ratio of $22,400 per QALY.
Discussion
Our analysis estimates that preoperative radiotherapy improves life expectancy and quality adjusted life expectancy, but also increases costs. Since cost-effectiveness is never the only decision criterion, no strict thresholds exist for acceptability of costs. Nevertheless, there is some consensus on the rule-of-thumb that less than $20,000 per QALY is definitely acceptable, less than $50,000 per QALY is acceptable, and less than $100,000 per QALY is possibly acceptable (12;33). According to this classification, the estimated base case costs of $25,100 per QALY for adding preoperative radiotherapy to TME-surgery are considered acceptable.
In our analysis an extensive amount of data was collected prospectively alongside a large randomized clinical trial. This largely guarantees the internal validity of the results. Although the large number of participating hospitals improves the external validity, generalizability may still be hampered by international differences and also by the fact that effectiveness was measured within a clinical trial. The benefits may be smaller if PRT is implemented under routine care. However, sensitivity analyses on the reduction of local recurrences by PRT maintained the acceptable cost-effectiveness.
The cost-effectiveness could improve if it were possible to identify R2 patients preoperatively. For these patients the model estimates that PRT increases costs and decreases effectiveness. A recent publication (34) suggests that accurate preoperative staging with specialized MRI techniques is possible. However, preliminary results based on the analysis of such subgroups should be confirmed prospectively before definite conclusions can be drawn.
Though QALYs are the preferred outcome measure for cost-effectiveness analyses, they may not accurately reflect the preferences of individuals. Clinicians emphasize the importance of preventing local recurrence for its severe disabling impact (1;8;35). This was not reflected in the estimated utility, possibly due to the relatively small number of patients and selective non-response. In a sensitivity analysis we examined the effect of a considerably worse valuation for local recurrence, which as expected improved the CE-ratio in favor of radiotherapy.
erroneously exclude group differences from the model). Moreover, the cost-effectiveness ratio including all observed proportional hazards remains within acceptable limits.
We believe that the most probable scenario is that a reduction in local recurrence rates does result in a survival benefit. This has recently been confirmed in meta-analyses of preoperative radiotherapy (7;8). Trials that found no or only small benefits in survival (8;17;36) studied different treatment regimens (e.g. postoperative radiotherapy, long term preoperative schedules, or wider irradiation fields) and used small sample sizes. The Swedish Rectal Cancer Trial (37) is the only other study that has investigated the same 5×5 Gy PRT. In combination with conventional surgery, PRT reduced local recurrence rates by 16% (from 27% to 11%) and increased survival at 5 years of follow-up by 10% (from 48% to 58%), with a mean survival benefit of 21 months. In our study, the reduction of local recurrences at 5 years of follow-up by about 5% (from 8.6% to 3.6%) corresponds to an estimated increase in 5-year survival of about 3% (from 67% to 70%), with a mean survival benefit of 8 months. Our results are generally in line with the results from the Swedish study, but the absolute benefits of PRT on local recurrence rates and survival are smaller due to the introduction of standardized TME-surgery. The TME-trial was designed and powered to detect a reduction of local recurrences from 10% to 5% by PRT. To detect the estimated 2.8% and 3.2% survival benefits at 5 and 10 years, sample sizes of 5,700 and 5,100 analyzable patients per arm would have been required (alpha 0.05, two-sided, power 0.90)(38). Therefore, longer follow-up of the TME-study may still not allow for more definite conclusions on whether there truly is a survival benefit. According to our model, even when we assumed no survival benefit by PRT, the short term loss in quality of life was still outweighed by the long term quality gain because of prevented local recurrences, but PRT might not be considered cost-effective according to the current acceptability thresholds. Recently, the cost-effectiveness for the Swedish trial was estimated at $3,700 per life year saved (39). Our cost-effectiveness ratio of $25,100 per QALY is less favorable. This can be explained not only by smaller benefits of PRT on local recurrence rates and survival, but from differences in the assessment of costs and quality of life as well. In the Swedish trial, the assessment of costs was limited to direct in- and outpatient hospital costs. Also, life years were not adjusted for loss in quality of life, hence overestimating the value of an increase in survival.
acceptable under a wide range of assumptions. We conclude that short-term preoperative radiotherapy in patients with rectal cancer undergoing total mesorectal excision is both effective and cost-effective.
References
(1) Wiggers T, de Vries MR, Veeze-Kuypers B: Surgery for local recurrence of rectal carcinoma. Dis Colon Rectum 39(3):323-328, 1996
(2) Enker WE: Potency, cure, and local control in the operative treatment of rectal cancer. Arch Surg 127(12):1396-1401, 1992
(3) Heald RJ, Karanjia ND: Results of radical surgery for rectal cancer. World J Surg 16(5):848-857, 1992 (4) Harnsberger JR, Vernava VM, Longo WE: Radical abdominopelvic lymphadenectomy: historic
perspective and current role in the surgical management of rectal cancer. Dis Colon Rectum 37(1):73-87, 1994
(5) Kapiteijn E, Marijnen CAM, Colenbrander AC, et al: Local recurrence in patients with rectal cancer diagnosed between 1988 and 1992: a population-based study in the west Netherlands. Eur J Surg Oncol 24(6):528-535, 1998
(6) Phillips RK, Hittinger R, Blesovsky L, et al: Local recurrence following 'curative' surgery for large bowel cancer: I. The overall picture. Br J Surg 71(1):12-16, 1984
(7) Camma C, Giunta M, Fiorica F, et al: Preoperative radiotherapy for resectable rectal cancer: A meta-analysis. JAMA 284(8):1008-1015, 2000
(8) Colorectal cancer collaborative group: Adjuvant radiotherapy for rectal cancer: a systematic overview of 8,507 patients from 22 randomised trials. Lancet 358(9290):1291-1304, 2001
(9) Kapiteijn E, Kranenbarg EK, Steup WH, et al: Total mesorectal excision (TME) with or without preoperative radiotherapy in the treatment of primary rectal cancer. Prospective randomised trial with standard operative and histopathological techniques. Dutch ColoRectal Cancer Group. Eur J Surg 165(5):410-420, 1999
(10) Kapiteijn E, Marijnen CA, Nagtegaal ID, et al: Preoperative radiotherapy combined with total mesorectal excision for resectable rectal cancer. N Engl J Med 345(9):638-646, 2001
(11) van den Hout WB, van den Brink M, Stiggelbout AM, et al: Cost-effectiveness analysis of colorectal cancer treatments. Eur J Cancer 38:953-963, 2002
(12) Earle CC, Chapman RH, Baker CS, et al: Systematic overview of cost-utility assessments in oncology. J Clin Oncol 18(18):3302-3317, 2000
(13) Goldberg PA, Nicholls RJ, Porter NH, et al: Long-term results of a randomised trial of short-course low-dose adjuvant pre-operative radiotherapy for rectal cancer: reduction in local treatment failure. Eur J Cancer 30A(11):1602-1606, 1994
(14) Stockholm Colorectal Cancer Study Group: Randomized study on preoperative radiotherapy in rectal carcinoma. Ann Surg Oncol 3(5):423-430, 1996
(15) Stockholm Rectal Cancer Study Group: Preoperative short-term radiation therapy in operable rectal carcinoma. A prospective randomized trial. Cancer 66(1):49-55, 1990
(16) Swedish Rectal Cancer Trial: Initial report from a Swedish multicentre study examining the role of preoperative irradiation in the treatment of patients with resectable rectal carcinoma. Br J Surg 80(10):1333-1336, 1993
(17) Cedermark B, Johansson H, Rutqvist LE, et al: The Stockholm I trial of preoperative short term radiotherapy in operable rectal carcinoma. A prospective randomized trial. Cancer 75(9):2269-2275, 1995
(18) Pahlman L, Glimelius B: Pre- or postoperative radiotherapy in rectal and rectosigmoid carcinoma. Report from a randomized multicenter trial. Ann Surg 211(2):187-195, 1990
(19) Marijnen CA, Kapiteijn E, van de Velde CJ, et al: Acute side effects and complications after short-term preoperative radiotherapy combined with total mesorectal excision in primary rectal cancer: report of a multicenter randomized trial. J Clin Oncol 20(3):817-825, 2002
(20) Holm T, Singnomklao T, Rutqvist LE, et al: Adjuvant preoperative radiotherapy in patients with rectal carcinoma. Adverse effects during long term follow-up of two randomized trials. Cancer 78(5):968-976, 1996
(22) Goldie SJ, Kuntz KM, Weinstein MC, et al: The clinical effectiveness and cost-effectiveness of screening for anal squamous intraepithelial lesions in homosexual and bisexual HIV- positive men. JAMA 281(19):1822-1829, 1999
(23) de Haes JC, van Knippenberg FC, Neijt JP: Measuring psychological and physical distress in cancer patients: structure and application of the Rotterdam Symptom Checklist. Br J Cancer 62(6):1034-1038, 1990
(24) The EuroQol Group: EuroQol--a new facility for the measurement of health-related quality of life. Health policy 16(3):199-208, 1990
(25) Dolan P: Modeling valuations for EuroQol health states. Med Care 35(11):1095-1108, 1997
(26) Stiggelbout AM, Eijkemans MJ, Kiebert GM, et al: The 'utility' of the visual analog scale in medical decision making and technology assessment. Is it an alternative to the time trade-off? Int J Technol Assess Health Care 12(2):291-298, 1996
(27) Torrance GW: Social preferences for health states: an empirical evaluation of three measurement techniques. Socio-Econ Plan Sci (10):129-136, 1976
(28) Oostenbrink JB, Koopmanschap MA, Rutten FFH: Manual for cost analyses, methods and standard prices for economic evaluations in health care (In Dutch). Amstelveen, The Netherlands, Dutch Health Insurance Executive Board, 2000
(29) van den Hout WB, van der Linden YM, Steenland E, et al: Single- versus multiple-fraction radiotherapy in patients with painful bone metastases: cost-utility analysis based on a randomized trial. J Natl Cancer Inst 95(3):222-229, 2003
(30) Steenland E, Leer JW, van Houwelingen H, et al: The effect of a single fraction compared to multiple fractions on painful bone metastases: a global analysis of the Dutch Bone Metastasis Study. Radiother Oncol 52(2):101-109, 1999
(31) van Roijen L, Essink Bot ML, Koopmanschap MA, et al: Labor and health status in economic evaluation of health care. The Health and Labor Questionnaire. Int J Technol Assess Health Care 12(3):405-415, 1996
(32) Koopmanschap MA, Rutten FF: A practical guide for calculating indirect costs of disease. Pharmacoeconomics 10(5):460-466, 1996
(33) Ubel PA, Hirth RA, Chernew ME, et al: What is the price of life and why doesn't it increase at the rate of inflation? Arch Intern Med 163(14):1637-1641, 2003
(34) Beets-Tan RG, Beets GL, Vliegen RF, et al: Accuracy of magnetic resonance imaging in prediction of tumour-free resection margin in rectal cancer surgery. Lancet 357(9255):497-504, 2001
(35) Minsky BD: Adjuvant radiation therapy for rectal cancer: is there finally an answer? Lancet 358(9290):1285-1286, 2001
(36) Wolmark N, Wieand HS, Hyams DM, et al: Randomized trial of postoperative adjuvant chemotherapy with or without radiotherapy for carcinoma of the rectum: National Surgical Adjuvant Breast and Bowel Project Protocol R-02. J Natl Cancer Inst 92(5):388-396, 2000
(37) Swedish Rectal Cancer Trial: Improved survival with preoperative radiotherapy in resectable rectal cancer. N Engl J Med 336(14):980-987, 1997
(38) Armitage P, Berry G: Statistical Methods in Medical Research (2nd ed). Blackwell Scientific Publications, 1987, pp 183
