104 Journal of Pain and Symptom Management Vol. 41 No. 1 January 2011

Transcription

1 104 Journal of Pain and Symptom Management Vol. 41 No. 1 January 2011 Original Article Methylnaltrexone in the Treatment of Opioid-Induced Constipation in Cancer Patients Receiving Palliative Care: Willingness-to-Pay and Cost-Benefit Analysis Michael Iskedjian, BPharm, MSc, Shrividya Iyer, PhD, S. Lawrence Librach, MD, CCFP, FCFP, Mike Wang, BSc, Bechara Farah, MSc, and Jade Berbari, BSc PharmIdeas Research and Consulting Inc. (M.I., B.F., J.B.), Oakville, Ontario, Canada; Pfizer, Inc. (S.I.), Collegeville, Pennsylvania, USA; Division of Palliative Care (S.L.L.), University of Toronto, and Temmy Latner Centre for Palliative Care (S.L.L.), Toronto, Ontario; and Pfizer Canada Inc. (M.W.), Markham, Ontario, Canada Abstract Context. When laxative regimens have failed, methylnaltrexone may be indicated for the relief of opioid-induced constipation (OIC) in patients with advanced illness receiving palliative care. Objectives. A cost-benefit analysis (CBA), based on a willingness-to-pay (WTP) approach, was performed to determine if methylnaltrexone should be added to the formulary list of drugs being reimbursed by third-party payers in Canada for the treatment of cancer patients in palliative care suffering from OIC. Methods. The WTP study had two components: a decision board explaining treatment options (Component A) and a questionnaire to measure individual WTP using a bidding game approach (Component B). Component A had two options: Option 1 (laxatives only) and Option 2 (laxatives þ methylnaltrexone injection). Only participants choosing Option 2 were invited to complete Component B. The results of the WTP survey were then incorporated into a CBA. Within a hypothetical cohort, additional monthly premiums that individuals were willing to pay for methylnaltrexone were compared with the monthly costs to the insurer for providing methylnaltrexone to all patients who would potentially be using it. Results. Four hundred one Canadians, of age 18 years and older, were surveyed and yielded a WTP in additional monthly insurance premiums of Canadian dollar (CAD) $8.65 (95% confidence interval: CAD$6.17eCAD$11.13). The CBA resulted in additional CAD$89,307 with a cost of CAD$139,840 and benefits of CAD$229,147. A set of 10,000 Monte Carlo simulations resulted in average CBA savings of CAD$145,011 with a 99.86% probability of dominance. Conclusion. The present CBA provides pharmacoeconomic evidence for the adoption of methylnaltrexone for treating OIC in terminally ill cancer Address correspondence to: Michael Iskedjian, BPharm, MSc, PharmIdeas Research and Consulting Inc., 1175 North Service Road West, Suite 211, Oakville, Ó 2011 U.S. Cancer Pain Relief Committee Published by Elsevier Inc. All rights reserved. ON, L6M 2W1, Canada. miskedjian@ pharmideas.com Accepted for publication: April 8, /$ - see front matter doi: /j.jpainsymman

2 Vol. 41 No. 1 January 2011 WTP-Based CBA of a Treatment for OIC 105 patients. J Pain Symptom Manage 2011;41:104e115. Ó 2011 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved. Key Words Cost-benefit analysis, economic, willingness to pay, opioid-induced constipation Introduction As cancer progresses in terminally ill patients, distressful and troublesome symptoms become more frequent and more important, often resulting in hospitalization and increased prescribing. 1e3 Pain, fatigue, anorexia, constipation, dyspnea, and delirium are the most frequently reported symptoms in palliative care patients. 1,3e6 Pain is unequivocally the most feared symptom of cancer 7 and contributes to the suffering of the patients. 5 According to the World Health Organization s (WHO s) ladder of pain management, a combination of shortand long-acting opioids (often classified as strong opioids) appear as the third option for the treatment of moderate-to-severe pain, after nonsteroidal anti-inflammatory drugs in mild pain, and short-acting opioids (also referred to as weak opioids) in moderate pain. 8 In addition, opioids are the most frequently prescribed pain medication for patients with moderate-to-severe cancer pain. 8 Although pain is alleviated by opioids, other symptoms still remain. Constipation, one of those drug-induced systemic adverse effects, results in increased discomfort for the terminally ill cancer patient. 9 In 2006, Sykes et al. 9,10 reviewed the pathogenesis of constipation specifically in cancer patients and discussed three major factors causing persistent constipation in this population: factors associated directly and indirectly with the cancer and factors arising from the treatment. Although the factors associated with cancer itself (age, diet, exercise, nerve damage; biochemical and neuroendocrine disturbance) are important in the pathogenesis of constipation, researchers have focused intensively on drug-induced opioid-induced constipation (OIC). 10e12 Constipation affects more than 50% of cancer patients in palliative care. 7 In the presence of opioids, the prevalence of OIC can rise to 90%. 9,13,14 Opioids affect the intestine by reducing motility, reducing the amount of secretions, and increasing fluids absorption and blood flow. 10,15,16 If constipation is not treated, it can lead to very serious complications ranging from bloating to bowel perforation. 9,15 A retrospective cross-sectional study performed in a palliative care unit in Canada, consisting of a chart review of 255 consecutive cancer patients admitted to this service, revealed that opioids were prescribed to 67% of patients and laxatives/stool softeners to 54% of patients. 3 These rates differ depending on hospital protocols, trials, the type of opioid prescribed, and guidelines followed. 17e20 Laxatives or/and stool softeners (i.e., lactulose, sorbitol, sennosides, and bisacodyl) are the recommended initial treatment for constipation in advanced cancer patients. 9,15 Laxatives and stool softeners are efficacious in more than 50% of patients, and for the remaining patients, additional treatment is required. 18,21 Methylnaltrexone (RELISTOR Ò ) is a novel quaternary derivative of naltrexone, which acts as a selective peripheral opioid receptor antagonist and is restricted in its ability to cross the blood-brain barrier. Methylnaltrexone, a first in class treatment, targets the underlying cause of OIC. 12,21,22 It is available as a methylnaltrexone bromide subcutaneous injection and is indicated for the treatment of OIC refractory to laxatives in patients with advanced illness receiving palliative care. In these cases, methylnaltrexone can be given as an adjunct therapy to rapidly induce bowel movement. 23 Two recent double-blind placebo-controlled studies have assessed the effectiveness of methylnaltrexone against placebo in patients with advanced illness and OIC. 24,25 Methylnaltrexone and placebo subcutaneous injections were given along with oral laxatives. The primary outcome was laxation within four hours. Both studies, respectively, reported increased laxation within four hours of 48% 24 and 62% 25 for a methylnaltrexone dose of 0.15 mg/kg compared with 15% 24 and 14% 25 for placebo (P < in both studies).

3 106 Iskedjian et al. Vol. 41 No. 1 January 2011 Although methylnaltrexone has been approved for the treatment of OIC in patients with advanced illness receiving palliative care, very little is known on whether patients, or society, would be willing to pay an extra amount of money to benefit from this therapy, when and as needed. Accordingly, various approaches can be taken to assess the willingness to pay (WTP) for such a treatment, in other words, the maximum amount an individual or a group of individuals are willing to pay, often in additional monthly premiums, to have the treatment reimbursed when needed. Moreover, the results of a WTP analysis can be applied in a formal economic evaluation using financial terms for both costs and outcomes, also known as cost-benefit analysis (CBA). A CBA would, in turn, provide decision makers from third-party payers the essential pharmacoeconomic evidence with a view to adopting methylnaltrexone into their formularies. A WTP study was conducted on healthy subjects in Canada to evaluate their propensity to pay for methylnaltrexone as a novel treatment for opioid-induced constipation. In turn, a CBA, based on the WTP results, was performed on the use of methylnaltrexone in terminally ill cancer patients receiving palliative care. Methods Stepwise Approach A two-step approach was applied in this study. The first step was used to measure both the preference and WTP, from a societal perspective, for methylnaltrexone, presented as a novel treatment for OIC in terminally ill cancer patients. A WTP survey measures the value of a given intervention in monetary terms within a given population. It is expressed as the dollar amount of money the subject would hypothetically pay to have the chosen treatment available. In making decisions on allocation of competing health care resources or services, the WTP approach is an effective tool evaluating benefits and costs while estimating the economic value of the implementation of a specific health policy in a health care setting or system. 26e28 In a second step, the potential cost of using methylnaltrexone was projected in a hypothetical cohort, and then compared with the WTP amounts applied to that cohort. That approach amounts to a CBA, where a positive difference when subtracting the total costs from the WTP amount for the cohort would result in savings, and thus the treatment be considered cost saving or dominant. In 2007, the Treasury Board of Canada updated a Canadian CBA Guide recommending the use of WTP as a guiding principle in measuring the benefits of a new policy. 29 Development of WTP Survey/Tool The WTP study tool had two components: a decision board explaining treatment options to patients and a questionnaire to measure individual WTP using a bidding game approach. The development of the tool followed the guidance proposed by Levine et al., 30 O Brien and Gafni, 26 and Gafni. 27 The decision board, herein presented in Fig. 1, was developed with the guidance of key clinical opinion leaders and based on data collected in the clinical trials of methylnaltrexone. 31e33 The decision board described a palliative state and the OIC associated with that palliative state. Two treatment options for OIC, as well as the likely response and adverse events associated with these two treatment options, were presented. The first treatment option consisted of liquids and capsules (i.e., laxatives), whereas the comparator treatment consisted of liquids, capsules, and an injection of methylnaltrexone (i.e., laxatives þ methylnaltrexone injection). The medications listed on the decision board were not identified. The second part of the tool was a bidding game, that is, an algorithm to frame the WTP. It consisted of asking how much in additional insurance premiums a participant in the survey was willing to pay for the combination of laxatives in liquid and capsule form and methylnaltrexone. Only participants choosing the second treatment option (laxatives þ methylnaltrexone injection) were invited to do the bidding game. The additional monthly insurance premiums started at Canadian dollar (CAD) $2.50, with an overall range of CAD$0.50eCAD$15. A positive response to the maximum value on this range was followed by an open-ended question asking the participant s maximum WTP. Participants were finally asked how confident they were of their answers on a scale of

4 Vol. 41 No. 1 January 2011 WTP-Based CBA of a Treatment for OIC 107 Fig. 1. Decision board presenting a palliative state, the opioid-induced constipation associated to it, and both treatment options for it.

5 108 Iskedjian et al. Vol. 41 No. 1 January to 10, with 0 being not confident at all to 10, being very confident. Data Collection Institutional review board approval was sought and obtained before approaching members of the general public for participation in this study. Members of the general public were surveyed to determine the overall WTP for methylnaltrexone. Participants were required to be Canadian citizens, 18 years or older, with a basic understanding of English. They were recruited through opportunistic sampling in the following urban centers in Ontario, Canada: the Greater Toronto Area, the Greater Ottawa Area, Hamilton, and London. Demographic data were collected before the administration of the tool. Statistical Analyses Descriptive statistics (mean, median, standard deviation [SD], 95% confidence intervals [CI], minimum, and maximum) were calculated to describe the demographic characteristics and WTP values of the study population. A Chi-square test was performed to assess if any significant differences in preference between the two treatment options existed based on demographic characteristics of the cohort. A Kruskal-Wallis or Wilcoxon rank-sum test was used to identify significant differences in WTP. Finally, a regression analysis was performed to identify significant predictors of WTP. All statistical analyses were performed at an alpha error level of 5%. Cost-Benefit Analysis The CBA was undertaken with the data collected from the WTP component. The CBA considered the establishment of a new policy (i.e., injections of methylnaltrexone) from a third-party payer perspective by comparing the additional monthly insurance premiums that individuals were willing to pay for the product, methylnaltrexone subcutaneous formulation, with the monthly costs to the insurer in a hypothetical cohort of a population of 100,000. To determine the potential cost of adding methylnaltrexone to the formulary list of drugs being reimbursed by third-party payers in Canada, the number of potential patients using methylnaltrexone to treat OIC had to be evaluated. The potential patient population was defined as cancer patients above the age of 18, who were in palliative care, using opioids, and nonresponsive to laxative treatment. Demographic data were obtained through Statistics Canada, and Canadian cancer prevalence rates were provided by the Canadian Cancer Society. 34,35 A literature search was conducted to obtain the proportion of patients on palliative care, those among them who were using opioids, those among the opioid users who were on laxatives, and those among this latter category who did not respond to laxatives. The final figure represented the potential users for methylnaltrexone. To evaluate the benefits to third-party payers of adding methylnaltrexone to the formulary list of reimbursed drugs, a contributing population had to be determined. The contributing population was defined as members of the general public, 18 to 64 years of age, participating in the workforce and willing to pay additional monthly premiums for methylnaltrexone to be included on the formulary list of drugs reimbursed by third-party payers. This figure was obtained by imputing the proportion of subjects who opted for the methylnaltrexone option in the WTP survey. The total WTP amount was then calculated for the hypothetical cohort, using the mean WTP value determined in the survey. On estimating the costs and benefits for third-party payers of adding methylnaltrexone to their formulary, a CBA was then undertaken to weigh costs against benefits. One-way sensitivity analyses (SAs) were performed on key parameters to test the robustness of this model. In addition, in a probabilistic SA, a set of 10,000 Monte Carlo simulations were carried out by simultaneously varying parameter values. Results Willingness to Pay Four hundred one (n ¼ 401) participants were enrolled in the survey, all of whom completed the survey successfully with usable responses (response rate, 100%). When compared with Canadian population characteristics, this study sample was younger (average age ¼ years), predominantly male

6 Vol. 41 No. 1 January 2011 WTP-Based CBA of a Treatment for OIC 109 Parameter Table 1 Inputs and Sensitivity Values Used Base Case Value Sensitivity Value(s) Reason for Variation Source Number of patients 100,000 Unit cost of MNTX ($) Manufacturer a Frequency of use of MNTX (per week) Median from clinical trials Manufacturer a Percentage of prevalence of cancer (all cancer) Percentage of patients in palliative care Estimation Estimation 36 Percentage using opioids a 3 Percentage using laxatives Percentage of nonresponder to laxatives Different treatment protocol 17 Demographic characteristics Age group, years 18 and older (%) e64 (%) Proportion participating in workforce (%) Proportion in workforce choosing MNTX (%) Proportion of employed WTP analysis WTP in additional monthly insurance premiums ($) All costs are in Canadian dollars. LB ¼ lower bound; MNTX ¼ methylnaltrexone; UB ¼ upper bound. a Wyeth Pharmaceuticals. participants choosing MNTX LB 95% CI WTP analysis UB 95% CI (61.8%), more had completed a postsecondary education, and more were in the labor force. Most study participants (n ¼ 241, 60.1%) preferred the treatment with laxatives and a methylnaltrexone injection (Option 2 on thedecisionboard).therewere146participants (36.4%) who stated that they would prefer to treat their OIC condition with laxatives only (Option 1), whereas 14 participants (3.5%) were indifferent between options. These results were significantly different (X 2 ¼ , P < 0.001). The rate of selection of Option 2 within various demographic parameters was analyzed to determine whether a difference between the survey sample and the population demographics for Canada might have introduced a bias. With regard to gender, the rate of selection of Option 2 for males was 60.5% vs. 59.5% for females. Regardless of differences between the survey sample and the 2006 Canadian Census with regard to male/female distribution of population, both groups reported similar preference (P ¼ 0.841) in the survey sample. When the sample was divided into working vs. not working, the preference for Option 2 in the survey sample was 58.5% and 62.3%, respectively, with a nonsignificant difference (P ¼ 0.491). With regard to the level of education when comparing any university level education vs. no postsecondary degree þ nonuniversity certificate holders, the rates of selection of Option 2 were 62.2% and 57.7%, respectively, again with the difference being nonsignificant (P ¼ 0.435). When the answers from the survey sample were adjusted according to three demographic parameters (age, income, and ethnicity) obtained from the 2006 Canadian Census figures, new rates of 60.3%, 59.2%, and 57.8% were respectively obtained for subjects choosing Option 2. The differences between these rates and the obtained survey rate of 60.1% were nonsignificant (P ¼ 0.964, 0.794, and 0.518, respectively). Hence, it is not unreasonable to assume that a bias was probably not introduced in the selection of Option 2 because of differences between the survey sample and the demographics obtained from the 2006 Canadian Census. The average WTP in monthly insurance premiums for the injection was CAD$8.65 (95% CI, CAD$6.17eCAD$11.13), with a median of CAD$4 and a range from zero to CAD$280. The certainty expressed by the study participants regarding the premiums they were willing to pay was of 10 (median ¼ 8). Among those who selected Option 2 and were accordingly surveyed for the WTP, individuals reporting prior use of laxatives were more often willing to pay for additional

7 110 Iskedjian et al. Vol. 41 No. 1 January 2011 insurance premiums, when compared with those who had not used laxatives before, with an average WTP of CAD$10.10 for prior laxative users and CAD$8.29 for participants who did not report prior use of laxatives (Z ¼ 2.639, P ¼ 0.008). No other significant differences were observed in any other demographic category. Age (P ¼ 0.044) and prior laxative use (P ¼ 0.04) were the only significant predictors of WTP in additional monthly insurance premiums. Cost-Benefit Analysis A diagram illustrating the framework of the CBA is presented in Fig. 2. According to a retrospective population-based study by Johnston et al. 36 published in 1998, 14.2% of patients who died of cancer in Nova Scotia, Canada were enrolled in a palliative care program. However, as not all patients undergoing palliative care are necessarily enrolled in a formal program, the proportion of cancer patients in palliative care should exceed the above figure. An assumption was made that roughly one-third of all cancer patients in palliative care would be registered in a formal program. Consequently, 42.6% of cancer patients were estimated to be in palliative care. This approach was conservative as such an estimation would not favor methylnaltrexone in this CBA; it increases the number of potential methylnaltrexone users, adding to the total costs to third-party payers. Data pertaining to the proportion of cancer patients taking opioids were obtained from a study by Riechelmann et al. 3 In 1998, Sykes 13 published a study in which 94% of cancer patients on opioid treatment were shown to be taking laxatives. The proportion of cancer patients who do not respond to laxatives was published in a study by Hawley and Byeon 17 in That study stated that 40% of patients using sennosides and 56.7% of patients using a combination of sennosides and docusate required additional interventions (lactulose, suppositories, and enemas). 17 In this analysis, the data pertaining to the Fig. 2. Diagram illustrating the framework of the cost-benefit analysis. MNTX ¼ methylnaltrexone; N ¼ number.

8 Vol. 41 No. 1 January 2011 WTP-Based CBA of a Treatment for OIC 111 sennosides-only cohort was used in the base case, as a sennosides-only treatment protocol was deemed by Hawley and Byeon 17 to be most appropriate for standard practice. Base case values and proportions are presented in Fig. 2 and Table 2. From a hypothetical cohort of 100,000 Canadians, 230 were shown to be potential methylnaltrexone users. Monthly potential costs to third-party payers were then calculated by factoring in a CAD$38 unit cost of methylnaltrexone and 3.5/week rate of methylnaltrexone use. 37,38 As presented in Fig. 2 and Table 2, from a hypothetical cohort of 100,000 Canadians, 26,491 satisfy the criteria (i.e., Canadians between the ages of 18 and 64, participating in the workforce, who were willing to pay for methylnaltrexone) of our contributing population. The monetary value of the benefits of adding methylnaltrexone to formulary lists was then evaluated by taking into account the average WTP in monthly insurance premiums for the injection obtained through the WTP analysis. Using the results of the WTP analysis, a monetary value for both costs and benefits to thirdparty payers was then generated. The CBA values for the base case and all the one-way SAs are presented in Table 3. The base case CBA was CAD$89,307 with a cost of CAD $139,840 and benefits of CAD$229,147. The lowest CBA value (CAD$23,609) was associated to the SA where the lowest bound of the WTP monetary value was used, whereas the highest CBA value (CAD$181,723) occurred when 14.2% of cancer patients were deemed to be in palliative care. The highest costs (CAD $198,208) were associated to a 56.7% rate of nonresponders to laxatives, whereas the lowest costs were found when the rate of cancer patients in palliative care was 14.2%. The upper and lower bounds of the 95% CI of the WTP monetary value yielded the highest (CAD $294,845) and lowest (CAD$163,449) benefit values, respectively. Sensitivity Analyses One-Way Sensitivity Analyses. One-way SAs were performed on various parameters used for the estimation of costs. These SAs were applied to the frequency of methylnaltrexone use, the estimation of Canadian cancer patients who are in palliative care, the percentage of cancer patients using opioids, and the rate of nonresponse to laxative. The first SA was undertaken by varying the frequency of methylnaltrexone use described in the product monograph (3.5 times a week) to two times a week as per the mean frequency of use in Wyeth clinical trials. 39 The second and thirdsaspertainedtotheestimatedproportion of Canadian cancer patients who are in palliative care; 14.2% and 28.4% sensitivity values were applied. According to product information obtained from Wyeth Pharmaceuticals, 39 the percentage of cancer patients using opioids was 90% rather than the 67% published by Riechelmann et al.; an SA was undertaken to test the consequences of this assertion on our model. Two rates of nonresponders to laxatives were published by Hawley and Byeon in 2008, one pertaining to a cohort following a sennosides-only protocol Parameter Table 2 Table of Cost-Benefit Outcomes Cost of MNTX Used ($) WTP ($) CBA (Additional WTP Amount) ($) Base case 139, ,147 89,307 One-way SA MNTX used twice a week 78, , , % of cancer patients are in palliative care 47, , , % of cancer patients are in palliative care 93, , ,123 (double the previous number) 90% are using opioids 187, ,147 41, % are nonresponders to laxative treatment 198, ,147 30,939 70% in workforce choose MNTX 139, , ,056 LL WTP $ , ,449 23,609 UL WTP $ , , ,005 All costs are in Canadian dollars. LL ¼ lower limit; MNTX ¼ methylnaltrexone; UL ¼ upper limit.

9 112 Iskedjian et al. Vol. 41 No. 1 January 2011 Table 3 Details and Results of the Monte Carlo Simulation Approach to Variations Parameter Distribution Type Selected Range Frequency of use Uniform 2e3.5 Percentage of cancer patients in palliative care Uniform 14.2e42.6 Percentage using opioids Uniform 67e90 Percentage of nonresponders to laxative treatment Uniform 40e56.7 Proportion in workforce who chose MNTX (%) Uniform 60.1e70 WTP additional monthly insurance premiums ($) Triangular 6.17e11.13 (Likeliest: 8.65) Results Average WTP ($) Average Cost ($) Average CBA ($) 248, , ,011 Outcome Number Percentage Cost benefit No cost benefit All costs are in Canadian dollars. MNTX ¼ methylnaltrexone. (40%) that was used in the base case and another (56.7%) pertaining to a cohort following a sennosides and docusate protocol that was tested in another one-way SA. Other one-way SAs were undertaken to test the model s evaluation of benefits. These SAs were undertaken on the percentage of the general public willing to pay for methylnaltrexone treatment and on the monetary value that they were willing to pay for the treatment. In the first SA, the percentage of members of the general public willing to pay for methylnaltrexone obtained through the WTP analysis was varied from the base case value of 60.1% to a sensitivity value of 70% that represents the proportion of employed participants who were willing to pay for methylnaltrexone. Two additional one-way SAs applied the lower bound and the higher bound of the 95% CI, respectively, of the monetary value that participants in the WTP analysis Fig. 3. Results of the Monte Carlo simulation presenting the willingness to pay vs. the cost of the resources. NB: The diagonal line represents the break-even line, with results to the top left displaying that methylnaltrexone is the option of choice (i.e., providing a cost benefit). An equation defining the line is provided.

10 Vol. 41 No. 1 January 2011 WTP-Based CBA of a Treatment for OIC 113 were willing to pay for methylnaltrexone. The results of all the above-mentioned one-way SAs are presented in Table 3. Additionally, SAs were undertaken after making adjustments in the WTP amounts within subgroups for the following demographic parameters: age, gender, employment status, income, education level, and ethnicity. These adjustments were made in a two-step process. First, for each subgroup, the preference for Option 2 was adjusted according to new weights based on the 2006 Canadian Census and, in a second step, multiplying these adjusted rates by the average WTP that had been determined for each of these subgroups in the survey sample. Finally, an average WTP was calculated for the entire sample. After adjusting for age, the WTP was determined to be CAD$8.98 in additional monthly insurance premiums, as opposed to the initial average of CAD$8.65. As for adjustments based on gender, employment status, income, education level, and ethnicity, the new WTP amounts were determined to be CAD$8.97, CAD$8.49, CAD$8.54, CAD$8.10, and CAD$9.35, respectively. None of these amounts displayed a statistically significant difference from the base case WTP amount. Probabilistic Sensitivity Analysis. In addition to the one-way SAs presented above, a set of 10,000 Monte Carlo simulations were carried out, by simultaneously varying all the abovementioned parameter values, according to ranges and types of distributions described in Table 1, which also presents results. The average CBA was CAD$145,011 with a 99.86% probability of showing a cost benefit. Fig. 3 graphically presents the probability of methylnaltrexone displaying cost benefit, based on the Monte Carlo simulation results. Discussion The results of the CBA, based on a contingent valuation design and a WTP approach, provide pharmacoeconomic evidence for the adoption of a novel treatment (subcutaneous methylnaltrexone) for OIC among palliative care cancer patients. In our study, the WTP projected over a hypothetical population more than offset the cost of the potential use of the medication by the same population. Although methylnaltrexone can be used in the treatment of OIC for patients suffering from any advanced illness, this study focused on a cancer population because such a population is reasonably definable and represents a large proportion of patients in palliative care. If all patients in palliative care suffering from OIC were included, the costs to third-party payers would rise, as there would be more patients seeking treatment; however, broadening the population of potential methylnaltrexone users also would increase the WTP of members of the general public, resulting in a similar CBA profile. One-way SAs confirmed the robustness of the model. Furthermore, Monte Carlo simulations indicated that more than 99% of the expected results would generate cost benefit, that is, a WTP amount for the population that would exceed the expected costs associated with the use of the medication. Instead of determining cost-effectiveness in the traditional methods with cost/clinical outcomes or cost/quality adjusted life years, the approach taken in this study first determines the WTP by subjects of the general population in terms of additional monthly insurance premiums for the drug of interest and then assesses the potential use of the medication in a given population. Finally, calculations are performed to determine whether the amount generated by applying the expected WTP value over that population would be sufficient to cover the expected costs of using the medication by the same population. This approach has been increasingly used and requested in Canada to inform decision makers on the WTP of their covered population for a specific drug in comparison to the expected costs arising from such a decision. 29 When used in a CBA, the WTP estimates allow the valuation of benefits in the same unit as costs. The results of the CBAs provide greater clarity of the evidence that can help facilitate formulary decision making. A Canadian CBA Guide published in 1995 and updated recently supports researchers in producing consistent highquality CBAs to inform regulatory decisions. 29 In a recent Canadian study, a treatment for neuropathic pain in multiple sclerosis was found to be cost beneficial after evaluating the amount a covered population was willing to pay in insurance premiums to have access to that treatment. 40

11 114 Iskedjian et al. Vol. 41 No. 1 January 2011 As with any modeling-based analysis, there are inherent limitations to the study. First, the sample surveyed in the WTP component of the study did not perfectly match Canadian demographics. This sample was younger, predominantly male, more educated, and more participants were in the labor force. However, no significant difference in the selection of treatment option was found based on gender, age, employment, education, or even ethnicity. In addition, after adjustments made to the sample to match Canadian demographics, no statistically significant differences were observed in the new WTP amounts as compared with the average WTP determined in the base case analysis. Further, the modeling component for determining the cost of medication use applied data from the literature from various studies and different parts of Canada, whereas the WTP survey was carried out in Ontario. We are also confident that the large number of participants would, to a certain degree, compensate for the lack of perfect demographic representativeness. In addition, some data points were not very specific, as was the case for the proportion of cancer patients using palliative care. Nonetheless, for the sake of performing conservative analyses, the proportion of enrollees in palliative care programs was multiplied by three, as it was assumed that more cancer patients would be using opioids out of formal palliative care programs than within those programs. A similar conservative approach was taken in determining the weekly rate of methylnaltrexone use; the base case rate was 3.5/week (which is the maximum rate of use advocated by the product monograph), whereas the more beneficial rate of two/week (observed in clinical studies) was confined to SAs. Carrying out a bidding game for determining WTP values from participants also may have potentially caused a starting point bias, by influencing with the first numbers presented. However, there is no conclusive evidence in the published literature of starting point bias. 41 Furthermore, the cost of medication used in the calculations did not take into account any markup or dispensing fees. Finally, transferability of the results to other populations and other settings should be done with caution. In conclusion, under a range of assumptions and limitations, the results of this CBA provide evidence for considering methylnaltrexone an attractive drug to add on a formulary, as determined by the difference between the WTP in a given population cohort for providing additional insurance payments for the drug, based on the preferences of a sample from the general public, and the expected cost for using the medication when needed by all eligible members of the same population. Disclosures and Acknowledgments This study was undertaken with financial support from Wyeth Pharmaceuticals (currently Pfizer). The authors acknowledge Brigitte Desjardins for her input in the CBA component. They would also like to acknowledge the following people for their help in the WTP survey: Rima Aziziyeh, Tamar Boghossian, Adriana Costei, Sloan Karver, Frances Pairaudeau, and Ruben Tavares. References 1. Walsh D, Donnelly S, Rybicki L. The symptoms of advanced cancer: relationship to age, gender, and performance status in 1,000 patients. Support Care Cancer 2000;8:175e Mancini I, Lassignol D, Obiols M, et al. Supportive and palliative care: experience at the Institut Jules Bordet. Support Care Cancer 2002;10:3e7. 3. Riechelmann R, Krzyzanowska M, O Carroll A, Zimmermann C. Symptom and medication profiles among cancer patients attending a palliative care clinic. Support Care Cancer 2007;15:1407e Donnelly S, Walsh D, Rybicki L. The symptoms of advanced cancer: identification of clinical and research priorities by assessment of prevalence and severity. J Palliat Care 1995;11:27e Ross D, Alexander C. Management of common symptoms in terminally ill patients: part II. Constipation, delirium and dyspnea. Am Fam Physician 2001;64:1019e Strömgren A, Sjogren P, Goldschmidt D, et al. Symptom priority and course of symptomatology in specialized palliative care. J Pain Symptom Manage 2006;31:199e Vanegas G, Ripamonti C, Sbanotto A, Conno F. Side effects of morphine administration in cancer patients. Cancer Nurs 1998;21:289e Mercadante S, Fulfaro F. World Health Organization guidelines for cancer pain: a reappraisal. Ann Oncol 2005;16:iv132eiv135.

12 Vol. 41 No. 1 January 2011 WTP-Based CBA of a Treatment for OIC Thomas J, Von Gunten C. Management of constipation in patients with cancer. Support Cancer Ther 2004;2:47e Sykes N. The pathogenesis of constipation. J Support Oncol 2006;4:213e Droney J, Ross J, Gretton S, et al. Constipation in cancer patients on morphine. Support Care Cancer 2008;16:453e Yuan CS. Clinical status of methylnaltrexone, a new agent to prevent and manage opioid-induced side effects. J Support Oncol 2004;2:111e Sykes N. The relationship between opioid use and laxative use in terminally ill cancer patients. Palliat Med 1998;12:375e Twycross R, Lack S. Symptom control in far advanced cancer: Pain relief. London, UK: Elsevier Health Sciences, 1983:1e Mancini I, Bruera E. Constipation in advanced cancer patients. Support Care Cancer 1998;6:356e Portenoy R. Constipation in the cancer patient: causes and management. Med Clin North Am 1987; 71:303e Hawley P, Byeon J. A comparison of sennosidesbased bowel protocols with and without docusate in hospitalized patients with cancer. J Palliat Med 2008; 11:575e Fallon M, Hanks G. Morphine, constipation and performance status in advanced cancer patients. Palliat Med 1999;13:159e Salvato C, Aretini G, Serraglia D, et al. Opioid prescription for terminally ill outpatients in a district of northern Italy: a retrospective survey. Pharmacol Res 2003;48:75e Nauck F, Ostgathe C, Klaschik E, et al. Drugs in palliative care: results from a representative survey in Germany. Palliat Med 2004;18:100e Foss J. A review of the potential role of methylnaltrexone in opioid bowel dysfunction. Am J Surg 2001;182:19Se26S. 22. Thomas J. Methylnaltrexone for opioid-induced constipation in advanced illness. N Engl J Med 2008; 359:1070e Anonymous. Methylnaltrexone: MNTX. Drugs R D 2006;7:374e Thomas J, Karver S, Austin Cooney G, et al. Methylnaltrexone for opioid-induced constipation in advanced illness. N Engl J Med 2008;358: 2332e Slatkin N, Thomas J, Lipman A, et al. Methylnaltrexone for treatment of opioid-induced constipation in advanced illness patients. J Support Oncol 2009;7:39e O Brien B, Gafni A. When do the dollars make sense? Toward a conceptual framework for contingent valuation studies in health care. Med Decis Making 1996;16:288e Gafni A. Willingness-to-pay as a measure of benefits. Relevant questions in the context of public decision making about health care programs. Med Care 1991;29:1246e Gafni A. The standard gamble technique. In: Armitage P, Colton T, eds. Encyclopedia of biostatistics. Baffins Lane, UK: John Wiley and Sons Ltd., 1998:4232e Treasury Board of Canada Secretariat. Canadian cost-benefit analysis guide: Regulatory proposals, Ottawa, Canada: Treasury Board of Canada Secretariat, Levine M, Gafni A, Markham B, MacFarlane D. A bedside decision instrument to elicit a patient s preference concerning adjuvant chemotherapy for breast cancer. Ann Intern Med 1992;117:53e Yuan CS, Foss J, O Connor M, et al. Methylnaltrexone prevents morphine-induced delay in oral-cecal transit time without affecting analgesia: a double-blind randomized placebo-controlled trial. Clin Pharmacol Ther 1996;59:469e Yuan CS. Methylnaltrexone for reversal of constipation due to chronic methadone use. A randomized controlled trial. JAMA 2000;283:367e Yuan CS, Wei G, Foss J, et al. Effects of subcutaneous methylnaltrexone on morphine-induced peripherally mediated side effects: a double-blind randomized placebo-controlled trial. J Pharmacol Exp Ther 2002;300:118e Stats Can. Estimates of population, by age group and sex for July 1, Canada, provinces and territories, annual (persons unless otherwise noted) (6210 series). CANSIMdTable , 1. Ottawa, Canada: Statistics Canada, Canadian Cancer Society/National Cancer Institute of Canada. Canadian cancer statistics Toronto: Canadian Cancer Society, Johnston G, Gibbons L, Burge F, et al. Identifying potential need for cancer palliation in Nova Scotia. Can Med Assoc J 1998;158:1691e Wyeth Pharmaceuticals. Methylnaltrexone cost, manufacturer data. Wyeth Canada, Markham, Ontario, Canada, Wyeth Pharmaceuticals, Relistor. Methylnaltrexone bromide. m-opioid receptor antagonist. Date of preparation: March 27, Wyeth Canada, Markham, Ontario, Canada, 1e Wyeth Pharmaceuticals. Internal data. Wyeth Pharmaceuticals, Collegeville, Pennsylvania, USA, Iskedjian M, Desjardins O, Piwko C, et al. Willingness to pay for a treatment for pain in multiple sclerosis. Pharmacoeconomics 2009;27:149e Drummond MF, Sculpher MJ, Torrance GW, O Brien BJ, Stoddart GL. Cost-benefit analysis. Methods for the economic evaluation of health care programmes. New York: Oxford University Press, 2005:203e231.