Cost-effectiveness analysis of oxycodone with naloxone versus oxycodone alone for the management of moderate-to-severe pain in patients with opioid-induced constipation in Canada

Abstract

Background:

Approximately 20–30% of Canadians suffer from chronic pain. Guidelines for the management of chronic pain support the use of controlled-release (CR) opioids to treat chronic pain. Although effective in managing chronic pain, oxycodone is associated with high rates of opioid-induced constipation (OIC). The cost-effectiveness of a combination of oxycodone for the management of pain and naloxone for the relief of OIC has not previously been evaluated for Canada.

Methods:

A decision analytic model was developed to estimate the cost-utility of combination oxycodone/naloxone compared to oxycodone alone in four populations. Drug costs for managing pain and healthcare costs related to managing OIC were included in the analysis and the primary measure of effectiveness was quality adjusted life years (QALYs) derived from OIC rates observed in clinical trials. The analysis was conducted from a healthcare system perspective, used a 1-year time horizon, and results were expressed in 2015 Canadian dollars.

Results:

In all four patient populations, there was a trade-off between slightly higher total expected costs for Targin treated patients compared to oxycodone treated patients, but also improved clinical benefits in terms of reduced OIC, which resulted in higher QALYs for patients. Although analgesic costs were found to be slightly higher for Targin treated patients, Targin also resulted in cost offsets to the healthcare system in terms of less rescue laxative drug use and other resources required for the management of OIC. The resulting 1-year cost-utility of Targin compared to oxycodone ranged from $2178–$7732 per QALY gained in the base case analysis, and it was found that these cost-utility results remained robust and at low values throughout a series of one-way deterministic analyses of uncertainty.

Conclusion:

The clinical effectiveness of oxycodone/naloxone in managing pain and OIC compared to CR oxycodone alone resulted in low cost-utility estimates.

Keywords::

• Chronic pain
• Opioid-induced constipation
• Targin
• Oxycodone
• Economic evaluation
• Cost-utility analysis
• Canada

Introduction

Background on chronic pain

The International Association for the Study of Pain1 (IASP) defines pain as ‘an unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage’. Acute pain, which does not dissipate after 6 months, is considered to be chronic2. Approximately 20–30% of Canadian adults (or 7–10.5 million adults) suffer from chronic pain3–7. Analysis of the Canadian Community Health Survey (CCHS)8, and the Canadian Chronic Pain Study II (CCPSII)3 have both confirmed that the prevalence of chronic pain in Canada increases with age, and is greater in females compared to males.

Burden of chronic pain in Canada

Chronic pain has significant emotional and psychological effects and can be so debilitating that it impairs every aspect of one’s life, from the ability to go to work or school, to performing day-to-day tasks, to quality-of-life (QoL) impairment in general9. In a survey of Canadian adults with chronic pain, 49% said that they found it difficult to attend social or family functions, 61% reported that they could not participate in their usual recreational activities, and 58% said that they could not carry out typical daily activities at home5. Pain has been reported to interfere with normal work in 93% of patients10, and an estimated 40–50% of people with chronic pain also have mood disorders9.

Chronic pain also inflicts a substantial cost burden. A Canadian study found that the average cost for patients on waitlists for multidisciplinary pain treatment facilities was over $3000 per month ($37,000 per year), and that only 9% of this amount was covered by public or private insurance plans11. Combined with treatment costs and lost productivity, chronic pain is estimated to cost more than cancer, heart disease and HIV combined12. The total direct and indirect (i.e. lost productivity) cost of pain in Canada has been estimated to be $56–63 billion per year4^,13.

Treatment of chronic pain

The National Opioid Use Guideline Group2 (NOUGG) recommends that a stepped approach be used when selecting the most appropriate opioid therapy for pain management. These guidelines for the management of chronic non-cancer pain, and the World Health Organization’s14 (WHO) guidelines for cancer pain management, both support the use of controlled-release (CR) opioids in appropriately selected patients. CR opioids generally provide better analgesia control over the course of the day, and are used when immediate-release (IR) formulations do not provide sufficient pain relief for a given period of time15 Clinical practice guidelines from both the American Geriatrics Society16 and the American Pain Society17 also recommend the use of opioid analgesics for the treatment of moderate-to-severe pain in appropriately selected and monitored patients.

Opioid-induced constipation

Although opioids are effective in the management of chronic pain, they are frequently associated with gastrointestinal (GI)-related AEs. This phenomenon is termed opioid-induced bowel dysfunction (OIBD), and can include the following symptoms: opioid-induced constipation (OIC); incomplete evacuation; abdominal distension; bloating; cramping; gastric reflux; nausea; and vomiting18^,19. While OIC is the most common, persistent and clinically important aspect of OIBD, it is under-recognized by healthcare professionals. A systematic review of placebo-controlled randomized controlled trials (RCTs) found that 41% of subjects with non-cancer pain who received opioids reported constipation as an AE compared to 11% in the placebo group20. OIC was reported as a symptom by 50% of all patients admitted to hospices, and 95% of advanced cancer patients taking strong oral opioids require laxatives21. In a recent prospective study of 489 pain patients from Canada, the US, Germany, and the UK, 73% of patients reported OIC requiring the use of laxatives22.

One of the main reasons that prevalence rates for OIC vary is due to the fact that a universal definition of OIC is not used in clinical studies or in routine clinical practice23. Often, studies define constipation as less than three bowel movements per week; however, this does not address many of the other issues associated with OIC. In contrast, the ROME II and III diagnostic criteria for constipation assess the frequency of bowel movements and discomfort associated with constipation24. More recently, the bowel function index (BFI) has been developed to better detect OIC in chronic pain patients because it takes into account three factors: ease of defecation; feeling of incomplete bowel evacuation; and personal judgment of constipation25. The BFI is a validated patient-assessment scale, which has been shown to be effective for managing patients and for assessing OIC in clinical trials25. A change in the BFI score of 12 or more points is considered to be clinically meaningful26, and a BFI score of greater than 28.8 has been shown to be an important cut-off point to distinguish between OIC and non-OIC patients27.

Treatment of opioid-induced constipation

The use of over-the-counter (OTC) or prescription laxatives is the most common treatment recommended for constipation. However, laxatives are not nearly as effective in patients with OIC, and a combination of laxatives is commonly attempted to treat persistent OIC. Therapeutics choices offer recommendations on different types of laxatives to use in the management of constipation28. Canadian recommendations state that the use of oral naloxone (i.e., a μ-opioid receptor antagonist) may improve OIC without creating systemic opioid withdrawal, and injections with methylnaltrexone (i.e., another μ-opioid receptor antagonist) is a treatment option for patients with OIC who have failed on laxative therapy29^,30. It has been found that the use of a combination of different laxatives with different mechanisms of actions may be more effective in the treatment of resistant constipation than laxative monotherapy alone31.

While laxatives are frequently used for the management of constipation, their ability to treat OIC is limited, particularly among patients with poor health status and advanced illness32. Laxatives may offer some symptom relief, but they do not specifically address the underlying mechanisms that lead to OIC (i.e., their actions do not address the μ-opioid receptor), and many patients continue to suffer from OIC despite laxative therapy. A survey determined that 81% of patients who take opioids regularly experience constipation, despite taking laxatives prophylactically33. Furthermore, an estimated 54% of patients treated for OIBD do not achieve a desired result with laxatives even 50% of the time23.

Oxycodone/naloxone

^NTargin (oxycodone hydrochloride (HCl)/naloxone HCl – (OXN)) is a CR fixed dose combination tablet that has a dual therapeutic effect. The oxycodone (OXY) component in OXN is indicated for the relief of moderate-to-severe pain in adults who require continuous around-the-clock opioid analgesia for several days or more. The naloxone component in OXN is indicated for the relief of OIC. Four strengths of OXN are currently available in Canada (i.e., 5/2.5 mg, 10/5 mg, 20/10 mg, and 40/20 mg), and OXN is the only oral preparation currently approved and marketed in Canada for the concurrent management of pain and OIC34.

Clinical trial data has demonstrated that OXN is both efficacious and well tolerated in the management of moderate-to-severe chronic pain. The efficacy of OXN has been demonstrated in randomized double-blind trials involving over 1300 patients with extensive open-label data. Five Phase III trials are included in OXN’s product monograph34. Four of the randomized, double blind Phase III studies were parallel group, multi-center trials with CR OXY as an active comparator (OXN340135, OXN300136, OXN300637, and OXN200138). OXN9001 was a prospectively designed, pooled analysis of OXN3001 and OXN300639. In all of these pivotal trials, OXN demonstrated equal analgesic efficacy compared to CR OXY alone, based on using the Pain Intensity Scale.

The main added benefit of OXN compared to CR OXY alone is that OXN has been shown to result in clinically significant improvements in the BFI as early as 1 week after treatment initiation, and statistically significant improvements in bowel function over the course of the 4- or 12-week double-blind phases of the trials. For example, a mixed-model repeated measures (MMRM) analysis demonstrated that, during the first 4 weeks of the double-blind phases of OXN3001 and OXN3006, the difference between the mean BFI scores of the two treatment arms was statistically significant in favor of OXN (OXN3001: −15.2 [95% CI = −18.2 to −12.2]36, p < 0.0001; OXN3006: −14.9 [95% CI = −17.9 to −11.9], p < 0.0001)37, and remained significant to the end of the studies. The improvement in BFI was also found to be significant in the pooled analysis of these studies (−15.1 [95% CI = −17.3 to −13.0], p < 0.0001)39, and in the study on cancer patients (−14.8 [95% CI = −23.0 to −6.5], p < 0.01)38. Analgesic efficacy aside, the improvement in BFI for OXN compared to CR OXY treated patients was consistent and statistically significant across all the studies.

Other peripherally-acting mu-opioid receptor antagonists (PAMORAs) (e.g., Relistor and Movantik) are not covered by provincial drug plans in Canada and are currently not under review by Canada’s national Common Drug Review program. Therefore, in keeping with guidelines for economic evaluations in Canada40, these PAMORAs were not included in the analysis.

Previous published economic evaluations including OXN

There have been two previously published economic evaluations including OXN. Dunlop et al.41 conducted a cost-utility analysis comparing OXN to CR OXY in patients with moderate-to-severe non-malignant pain using trial OXN3001 from a UK perspective. Based on the trial results, the analysis assumed equal analgesic efficacy, but different rates of OIC between OXN and CR OXY. It was found that over 1 year, OXN cost more and produced additional quality adjusted life years (QALYs), such that the incremental cost-utility ratio (ICUR) of OXN compared to CR OXY was £5842 per QALY gained41. In a more recent economic evaluation, Coluzzi and Ruggeri42 compared extended-release (ER) OXN, ER tapentadol, and CR OXY in patients with musculoskeletal pain from an Italian healthcare system perspective. The 1-year ICUR of OXN compared to CR OXY was estimated to be €475 per QALY gained42.

Objectives of the study

Although economic evaluations of OXN have been published for the UK and Italy, no economic evaluations have previously been published on OXN in Canada. In addition, the previously conducted economic evaluations have either focused on the patient population from a specific trial (e.g., OXN3006)41, have pooled results across trials with different patient populations (e.g., OXN3401, OXN3001, OXN3006)42, or have not included cancer patients in their assessment (e.g., OXN2001).

Therefore, the aim of this study was to develop a general model that could evaluate the cost-effectiveness of OXN (i.e., Targin) compared to CR OXY (i.e., ^NOxyNEO) in different patient populations suffering from moderate-to-severe chronic pain and OIC, and then use the findings from the pivotal clinical trials and other data sources to populate the model. In particular, the specific objectives were to assess:

1. The cost-utility of OXN compared to CR OXY in non-malignant chronic pain patients who are experiencing OIC (i.e., study OXN9001);

2. The cost-utility of OXN compared to CR OXY in non-malignant chronic pain patients requiring lower doses of analgesia for pain relief who are experiencing OIC (i.e., study OXN3001);

3. The cost-utility of OXN compared to CR OXY in non-malignant chronic pain patients requiring higher doses of analgesia for pain relief who are experiencing OIC (i.e., study OXN3006); and

4. The cost-utility of OXN compared to CR OXY in malignant chronic pain patients who are experiencing OIC (i.e., study OXN2001).

Methods

Type of economic evaluation

The economic model was designed to assess the following:

• Drug and other healthcare costs of treatment;

• Health outcomes, including rates of constipation and QALYs; and

• Incremental costs and outcomes.

The primary analysis was a cost-utility analysis comparing cost and QALYs and, if a trade-off existed between costs and outcomes, ICURs expressed as a cost per QALY gained were calculated.

Target populations

The target population is adults (i.e., aged 18 years or greater) with moderate-to-severe chronic pain who are experiencing OIC that is difficult to manage with preventive measures (e.g., increased fluid and fiber intake) and laxatives. Although laxatives are used for OIC, laxatives are not indicated for long-term treatment of chronic pain and OIC in Canada. Four primary patient populations were considered in the analysis, corresponding to four pivotal clinical trial populations:

• Non-malignant chronic pain patients who are experiencing OIC (i.e., study OXN9001);

• Non-malignant chronic pain patients requiring lower doses of analgesia for pain relief who are experiencing OIC (i.e., study OXN3001);

• Non-malignant chronic pain patients requiring higher doses of analgesia for pain relief who are experiencing OIC (i.e., study OXN3006); and

• Malignant chronic pain patients who are experiencing OIC (i.e., study OXN2001).

Comparators

OXY is a well-established opioid for the management of moderate-to-severe pain. CR ^NOxyContin has been demonstrated to be effective for the management of nociceptive, cancer-related, and neuropathic pain43–47. OxyContin was selected as the active CR opioid comparator for use in the OXN pivotal clinical trials. However, patients in Canada were transitioned to OxyNEO in 2012, and OxyContin is no longer on the market. Therefore, OxyNEO was chosen as the primary analgesic current standard of care (SOC) CR OXY for the analysis. OxyNEO is bioequivalent to OxyContin, but has unique physical and chemical properties34. Since the overall SOC for the management of moderate-to-severe chronic pain and OIC includes both a CR opioid and a rescue laxative, rescue laxatives were permitted in both treatment arms in the studies, and their use as observed in the clinical trials were included in this analysis.

Cost perspective (viewpoint)

The model is conducted from the perspective of a provincial healthcare payer in Canada. As such, only direct medical care costs were included. We specifically modeled unit costs from Ontario in the analyses and this report.

Analytic perspective (time horizon)

The time horizon of the model is 1 year. The comparative component of the pivotal trials was either 4 or 12 weeks and, as such, it was believed that extrapolating the results beyond 1 year would involve a fair amount of speculation and, therefore, would be inappropriate.

Discounting of future costs and effects

Since the time horizon is 1 year, discounting of costs and effects were not included in the analysis.

Decision analytic model structure and inputs

A decision analytic model was used to estimate the expected cost and QoL (i.e., utility) per person for each treatment and in each of the four studied patient populations. If appropriate (i.e., a trade-off existed between cost and outcomes), the model was designed to also calculate ICURs. Since it has been established that OXN and CR OXY (i.e., Targin and OxyNEO) have equal analgesic effect, pain control was not included as a health state in the model and was assumed to be equal across treatments. However, as OXN and OXY have a different effect on the rates of constipation, it was assumed that the primary health state influencing cost and outcomes in the model was whether patients were ‘constipated’ or ‘not constipated’.

As illustrated in Figure 1, patients were assumed to start off in the ‘constipated’ health state and over time either transitioned to ‘not constipated’ or remained in the ‘constipated’ health state based on the efficacy observed in the pivotal clinical trials. The model is structured using weekly cycle lengths over the entire analytic time horizon (i.e., 1 year in the base case). Mortality, or difference in mortality, was not included in the model. As illustrated in Figure 1 (and discussed in more detail below), the main components of the model were: rates of constipation; cost of analgesia treatment, cost of rescue laxative treatment; other healthcare resource use and cost for treatment of OIC; and QoL associated with constipation or no constipation.

Figure 1. Model structure for patients with moderate-to-severe pain and opioid-induced constipation (OIC).

Rates of constipation

The primary outcome in studies OXN2001, OXN3001, and OXN3006, and the OXN9001 pooled analysis, was the BFI, which includes three parameters (i.e., ease of defecation, feeling of incomplete defecation, and personal judgment of constipation). The BFI is a validated numerical rating scale (NRS) for use in pain patients to detect important changes in OIC25. At regularly scheduled follow-ups, patients were asked to complete the BFI for the preceding 7-day period, and the mean of the measures provides a final NRS score between 0–100. As such the BFI is a continuous measure of bowel function, where a higher score indicates a worse degree of constipation. However, the score itself does not explicitly identify a clinical cut-off point as to when a patient would be considered constipated or not constipated. In a separate study, Ueberall et al.27 used the BFI to estimate whether patients were constipated or not. A sample of 474 patients with an average age of 58 years completed the BFI following an interview with a health professional regarding their constipation. It was found that 95% of patients who reported they were not constipated had a BFI score less than or equal to 28.8 mm.

This constipation cut-off point was applied in this analysis to calculate the number of subjects who were in either the ‘constipated’ or ‘not constipated’ health state for each treatment arm in each of the patient populations studied in the pivotal trials. Any missing BFI values between weeks 1–4 for study OXN2001 and between weeks 1–12 for studies OXN3001 and OXN3006 were estimated by last observation carried forward (LOCF). The extension phases of the pivotal trials demonstrated that subjects receiving OXN continued to experience further reductions in BFI during the extension phases48–50. However, since subjects who received OXY during the double-blind phase of the studies were switched to OXN during the extension phases, there is no longer-term follow-up data available for OXY subjects. Therefore, it was conservatively assumed that the BFI values at the end of the study period would remain constant over time in both the OXY and OXN study groups (i.e., to 1 year in the base case) and the method of LOCF was used for these projections. The resulting percentage of subjects constipated in each of the trials over time is summarized in Table 1.

Table 1. Percentage of subjects constipated over time, by treatment group and study.

Week	OXN300136^,69		OXY300637^,70		OXN900139^,49		OXN200138^,70
	OXY	OXN	OXY	OXN	OXY	OXN	OXY	OXN
0	88%	91%	94%	97%	91%	93%	94%	96%
1	82%	63%	90%	67%	86%	65%	73%	73%
2	79%	56%	87%	65%	83%	60%	78%	63%
4	77%	58%	82%	65%	80%	61%	83%	69%
8	75%	51%	78%	58%	76%	54%	–	–
12	67%	49%	73%	53%	70%	51%	–	–

OXY, oxycodone; OXN, oxycodone/naloxone.

Cost of analgesia treatment

The cost of pain medication was estimated by multiplying the average daily doses of OXY and OXN observed in the pivotal trials by a weighted average cost per milligram (mg) for OXY and OXN. The average daily doses of OXY and OXN from the pivotal clinical trials are presented in Table 2. Not surprisingly, the mean daily doses were lower in study OXN3001 than in study OXN3006, reflecting the admission criteria into the trials (i.e., patients requiring lower or higher doses of analgesia for pain relief). The weighted average cost per milligram of OXY was estimated from the number of claims in the 2014 national public payer IMS Brogan Inc. PharmaStat database51. As presented in Supplemental Table 1, the total number of public claims in 2014 was 268,067. The doses of OxyNEO which are equivalent to the doses used in the OXN pivotal clinical trials are 10, 20, and 40 mg. Using the proportion of claims for these doses only and the costs per tablet from the Ontario Drug Benefit Formulary (ODBF)52, the weighted average cost of OxyNEO is estimated to be $0.0687 per mg. Generic CR oxycodone is not covered by drug plans in Canada.

Table 2. Mean daily doses (in mg), by treatment group and study.

OXN300136^,69		OXY300637^,70		OXN900139^,49		OXN200138^,71
OXY	OXN	OXY	OXN	OXY	OXN	OXY	OXN
33.0 mg	32.3 mg	71.9 mg	74.4 mg	50.3 mg	50.6 mg	43.1 mg	46.6 mg

OXY, oxycodone; OXN, oxycodone/naloxone; mg, milligram.

The weighted average cost per mg of OXN was estimated from the number of claims in the 2014 national private payer IMS Brogan Inc. PharmaStat database51. As presented in Supplemental Table 2, the total number of private claims in 2014 was 15,865. The four strengths of OXN currently marketed in Canada are 5/2.5 mg, 10/5 mg, 20/10 mg, and 40/20 mg. Using the proportion of private claims for OXN in Canada in 2014 and the costs per tablet obtained from Purdue Canada53, the weighted average cost of OXN is estimated to be $0.0862 per mg.

Using the average daily doses reported in the pivotal trials (Table 2) and the weighted average costs per mg for OXY and OXN (Supplemental Tables 1 and 2), the average weekly cost of analgesic therapy used in the model is presented in Table 3 for each treatment group and study.

Table 3. Mean weekly cost of therapy (2015 CDN dollars), by treatment group and study.

OXN3001		OXY3006		OXN9001		OXN2001
OXY	OXN	OXY	OXN	OXY	OXN	OXY	OXN
$15.86	$19.43	$34.56	$44.90	$24.17	$30.53	$20.71	$28.11

OXY, oxycodone; OXN, oxycodone/naloxone; CDN, Canadian denomination.

Cost of rescue laxative treatment

Although there are no detailed and widely accepted guidelines that are specific for the management of OIC23, in clinical practice patients would be managed daily with a stimulant laxative and stool softener combination. In each of the pivotal trials, rescue laxative therapy was permitted as needed by the patient. For standardization across countries and across trials, laxative bisacodyl was used in the form of increments of 5 mg tablets. The mean number of tablets for each treatment group and study were calculated (i.e., for each patient, the total number of tablets was calculated, and then the mean across all patients was calculated). Using the average number of bisacodyl (5 mg) tablets observed in the pivotal trials and the 2015 ODBF cost of bisacodyl (i.e., $0.045 per tablet)52, the average weekly cost of laxative treatment is presented in Supplemental Table 3 for each treatment group and study.

Other healthcare resource use and cost of treatment for OIC

Other healthcare resources associated with OIC were collected using a questionnaire developed through qualitative interviews conducted with Canadian clinicians54. The survey of 101 physicians was conducted to confirm and quantify any additional healthcare resource use which would be expected to be incurred by a patient with OIC. The healthcare resources which were reported, the proportion of patients expected to require the resource and frequency of expected use of the resource for every 6 months as found in the survey is presented in Supplemental Table 4. Also presented is the estimated cost of the resource, weighted by the proportion of patients using the resource, the frequency of use, and the unit cost of the healthcare resource.

Unit cost estimates for healthcare resources were estimated from various sources. The cost of a pharmacist visit was estimated using the Ontario Drug Benefit Formulary dispensing fee55. Hydration status and manual evacuation were estimated assuming a 30-min procedure and an average nurse hourly wage of $36.1856. An enema administered by a medical professional at home was estimated using a previously published estimate for a home nurse visit57, inflated to 2015 prices using the healthcare component of the consumer price index (CPI)58, and adding in the cost of one fleet enema ($6.79)59. For the remaining healthcare physician services presented in Supplemental Table 4 that are covered under the Ontario Health Insurance Act, the Ontario Schedule of Benefits (OSB) was used using the following codes and assumptions60:

• Physician telephone call—repeat consultation (A006);

• Emergency room visit—physician emergency department assessment (A100);

• Additional GP consultation—repeat consultation (A006);

• X-ray—hospital and professional components (average of X108 and X113);

• Ultrasound—hospital and professional (P1) components (J435);

• Barium enema—hospital and professional components (X112);

• Barium swallow—hospital and professional components (average of X103 and X104);

• Referral to surgeon—family physician and general surgeon consultation (A005 + A035);

• Colectomy—average of: S166 = 83 min procedure; S167 = 83 min average procedure time for anesthesiologist and/or assistant (A&A); E796 = 105 min average procedure time for A&A; S169 = 120 min average procedure time for A&A. Plus $20 per minute operating room cost61; and

• Colonoscopy—no signs or symptoms or risk factors (Z555 = 45 min average procedure time for A&A).

The resulting cost of additional healthcare resources required for OIC patients was estimated at $113.34 per 6 months, or $4.39 per week.

Quality-of-life

The measure of treatment effectiveness used for the economic evaluation was the number of QALYs. To estimate QALYs, the expected time, with and without OIC, was weighted by utility values for each state on a 0–1 scale, where 1 represents perfect health and 0 represents death. Since the pivotal trials did not include a robust QoL assessment for use in a CUA, utility weights for the constipated and not constipated health states were derived from the literature. Guest et al.62 rated patients experiencing symptoms of constipation at 0.74 compared to 0.90 for patients suffering from constipation but being managed well on medication. Van der Linden et al.63 found that the median utility value for cancer patients with constipation was 0.39, compared to 0.63 for patients without constipation. Finally, Penning-van Beest et al.64 found that the median utility value for patients with constipation and non-advanced illness was 0.31, compared to 0.65 for patients who did not experience constipation. Since the Penning-van Beest et al. study is the largest utility study to date conducted on opioid-treated patients, it was decided to use the utility values from this study for the base case analysis.

Analysis of the decision analytic model

The model was constructed so that it could be run in either deterministic or probabilistic mode. In the deterministic mode, the ‘best’ or ‘most likely’ point estimate of each model parameter and analytic modeling assumption are drawn upon in the model. In the probabilistic mode, the same analytic modeling assumptions are used, but stochastic parameters in the model (i.e., random variables with a defined distribution) are specified according to probability distributions that reflect the underlying uncertainty of these model parameters (i.e., model parameter uncertainty).

In the following sections, the base case analysis (i.e., probabilistic model), the deterministic model and the other analyses of uncertainty are discussed.

Base case analysis—probabilistic model

There are three main types of inputs into the decision analytic model: analytic assumptions; fixed model parameters; and stochastic model parameters. Model values or assumptions for each of these types of inputs for the probabilistic base case analysis are discussed below.

Only two modeling assumptions were relevant for the current model (i.e., discounting and time horizon). Since a 1-year time horizon was chosen for the base case analysis, no discounting of costs or effects in future years was necessary. The time horizon chosen for the base case analysis was 1 year, based on extrapolations from the 4 week (i.e., OXN2001) or 12 week (i.e., OXN3001, OXN3006, OXN9001) pivotal trial data.

The main difference between fixed and stochastic model parameters relates to the certainty of the model input value. For example, unit costs for drugs listed on the ODBF52, or physician fees listed on the OSB60 are fixed prices and reimbursement fees. The drug costs presented in Supplemental Table 1 for OXY and Supplemental Table 2 for OXN were treated as fixed costs in the model. Similarly, the cost for a 5 mg tablet of bisacodyl (i.e., $0.045) as a laxative and the professional fees of other healthcare costs for managing OIC listed in Supplemental Table 4 were treated as fixed costs in the analysis.

The remaining variables in the model were treated as stochastic and distributions were calculated or specified for each of these model parameters. The type of distribution defined depended on the variable. For example, probabilities, which are constrained between 0–1, were defined using Beta distributions. Cost variables, which are constrained to be greater than or equal to zero, were defined according to Gamma distributions. Finally, analgesic drug use for the management of pain and OIC was defined according to normal distributions. The stochastic model variables, the assumed distributions, the values used to define the distributions, and the data sources used for these specifications are presented in Supplemental Table 5.

Second-order Monte Carlo simulations using 1000 iterations of the probabilistic distributions defined in Supplemental Table 5 were conducted. Mean simulation results for costs and effects (i.e., QALYs) were calculated for OXY and OXN for each of the four populations. The cost and QALY outputs from the model were presented in tabular format and expressed as mean per patient values. Incremental costs and QALYs (OXN minus OXY) were also calculated and an ICUR calculated if a trade-off existed between incremental costs and effects. The results from the probabilistic analysis of uncertainty was also expressed as cost-effectiveness acceptability curves (CEACs) showing the probability of OXY or OXN being cost-effective for different decision-maker willingness-to-pay (WTP) thresholds (i.e., cost per QALY gained).

Deterministic model

In the deterministic mode of the model, the same modeling assumptions and same values for the fixed model parameters were utilized for the deterministic analysis as for the probabilistic analysis. The main difference with the deterministic analysis is that fixed point estimates were used for the model parameters listed in Supplemental Table 5 instead of probability distributions. For the probability of constipation, the percentage constipated at each time point from the pivotal trials was used (see Table 1). For analgesic drug use, the mean (µ) daily dose reported in the pivotal trials, as shown in Supplemental Table 5, was used. For laxative drug cost, the mean laxative cost per week calculated using data from the pivotal trials was used (see beta values in Supplemental Table 5). For the cost of other healthcare resource use for managing OIC, the mean cost estimate from the physician survey was used (i.e., $4.39 or the beta value in Supplemental Table 5). Finally, for utility values, the mean utility weights calculated in the Penning-van Beest et al.64 study were used (i.e., 0.65 for not constipated and 0.35 for constipated).

One-way deterministic analyses of uncertainty

To investigate the robustness of the model results against reasonable ranges of model input parameter values and to assess which input parameters have the greatest effect on the model results, one-way sensitivity analyses were conducted. The model parameters tested and the ranges used in the one-way sensitivity analyses are presented in Supplemental Table 6. In conducting the one-way sensitivity analyses, the deterministic (as opposed to probabilistic) model was employed utilizing base case point estimates for each model parameter. A total of 12 one-way sensitivity analyses were conducted for each of the four patient populations (see list in Supplemental Table 6). For brevity of presentation, the values listed for the probabilities of constipation are a sub-set of the values used in the sensitivity analyses (i.e., end of study values at 12 weeks are presented only). It is also noteworthy to mention that two of the one-way sensitivity analyses conducted (e.g., mean daily dose of analgesic used and percentage constipated) represent extreme scenarios where values from the lower 95% CI of OXY and upper 95% CI of OXY are used. These two sensitivity analyses should be interpreted as conservative scenarios for the assessment of OXN relative to OXY based on the trial findings, but they have been included as ‘worse case’ scenarios.

The results from the one-way sensitivity analyses have been presented as Tornado diagrams for each of the four patient populations. A Tornado diagram shows sequentially the variables with the largest impact on the cost-utility results (i.e., largest impact on the top).

Results

Probabilistic base-case and deterministic results are presented in the next section, while the results for the one-way sensitivity analyses are presented in the following section. All costs are presented in 2015 Canadian denomination dollars (CDN), and the probabilistic results were calculated as the means from 1000 iterations of the second-order Monte Carlo simulations.

Base case, probabilistic, and deterministic analyses

The expected cost and number of QALYs per patient is presented in Table 4 for the entire non-malignant population and for journal space considerations, Supplemental Tables 7–9 for each of other three patient populations. The expected cost is also broken down into three categories: pain/OIC treatment costs; laxative treatment costs; and other healthcare costs related to the management of OIC. For each of the four patient populations, it was found that OXN treated patients resulted in higher expected costs, but also additional QALYs over the 1-year time horizon, compared to OXY treated patients. Pain/OIC treatment cost represented the largest component of total cost, followed by other costs for managing OIC and then laxative treatment. The incremental QALYs gained were ∼0.07 for non-malignant patients (Table 4, Supplemental Tables 7 and 8) and approximately 0.05 for malignant patients (Supplemental Table 9). The smaller gains in QALYs for cancer patients reflect the findings from study OXN2001, where the difference in the rates of constipation was not as pronounced as for non-malignant patients (i.e., Table 1). It was also found that the pain/OIC treatment costs represented between 85–90% of total expected costs and other healthcare costs related to OIC management represented between 5% and 15% of total expected costs. Laxative treatment cost represented less than 0.5% of total expected 1-year cost. It was found that laxative treatment and other healthcare costs related to OIC management were higher for OXY treated patients. Overall, the incremental cost of OXN compared to OXY-treated patients ranged from $141 (Supplemental Table 7) to $494 per patient (Supplemental Table 8).

Table 4. Cost-utility results for non-malignant chronic pain patients.

Model output	OXN	OXY	Incremental (OXN − OXY)
Probabilistic base-case analysis
QALYs	0.4705	0.4048	0.0657
Cost
Pain/OIC treatment	$1591.79	$1258.95	$332.84
Laxative treatment	$1.16	$2.39	($1.23)
Other costs for OIC management	$120.86	$165.18	($44.31)
Total costs	$1713.82	$1426.52	$287.30
Incremental cost per QALY gained	–	–	$4376
Deterministic analysis
QALYs	0.4706	0.4045	0.0660
Total costs	$1709.50	$1424.37	$285.13
Incremental cost per QALY gained	–	–	$4319

QALY, quality-adjusted life-year; OXY, oxycodone; OXN, oxycodone/naloxone. Additions/subtractions may not match due to rounding.

For all four patient populations, there was a trade-off between higher expected costs for OXN treated patients compared to OXY treated patients, but higher expected QALYs as well. The 1-year cost-utility of OXN ranged from $2178–$7732 per QALY gained, for non-malignant patients treated with low dose analgesics and malignant patients, respectively.

The cost-utility results for the deterministic model are also presented in Table 4 and Supplemental Tables 7–9. Similar results for the expected costs, expected QALYs, incremental costs, incremental QALYs, and cost-utility were found in the deterministic analysis compared to the results from the means of the simulations in the probabilistic analysis (i.e., base case results). For example, the 1-year cost-utility results of OXN based on the deterministic analysis ranged from $2194–$7897 per QALY gained, for non-malignant patients treated with low dose analgesics and malignant patients, respectively.

Based on the results from the Monte Carlo simulations, the cost-effectiveness scatterplots for each of the four patient populations were calculated and presented as CEACs. Although the results from the simulations differed slightly across the four patient populations, in general the results were similar and, therefore, only the scatterplot and CEAC for the total non-malignant population is presented in Figures 2 and 3. The scatterplots and CEACs for the other three patient populations are presented as Supplemental Figures 1–6. The results in Figure 3 demonstrate that, in nearly all of the simulations, OXN resulted in a trade-off between higher expected cost and higher expected QALYs. The CEAC results in Figure 3 confirm that, in the majority of the simulations, the cost-utility of OXN compared to OXY was low, with OXN approaching 100% of the simulations being cost-effective at fairly low decision-maker threshold values (i.e., cost per QALY gained). For example, at the commonly quoted decision-maker threshold of $50,000 per QALY gained (i.e., shown as the red vertical line threshold in Figure 3), OXN was found to be cost-effective in 100% of the simulations in all three non-malignant populations, and 95% cost-effective for cancer patients.

Figure 2. Cost-utility scatterplot for non-malignant chronic pain patients.

Figure 3. CEACs for non-malignant chronic pain patients. CEAC, cost-effectiveness acceptability curve; ICUR, incremental cost-utility ratio; QALY quality-adjusted life year.

One-way deterministic analysis of uncertainty

The results of the one-way sensitivity analyses are presented as Tornado diagrams. For journal space considerations, only the results for total non-malignant population are presented in Figure 4. The Tornado diagrams for the other three patient populations are presented in Supplemental Figures 7–9. For each of the four patient populations, the variable with the largest impact on the cost-utility results was using the lower 95% CI constipation rates for OXY over time and the upper 95% CI for OXN. In this scenario, the ICURs increased to $12,825 for the non-malignant low dose analgesic population, to $36,465 in the high dose population, to $10,602 for the combined low and high dose population, and for cancer patients, OXN became dominated (i.e., higher cost and lower QALYs) compared to OXY. It should be noted that this is a worst case scenario for OXN, and was tested only to examine the unlikely scenario where these constipation rates might be observed.

Figure 4. Tornado diagram for non-malignant chronic pain patients. ICUR, incremental cost-utility ratio; QALY, quality-adjusted life year; CI, confidence interval.

Finally, and as shown in Figure 4 for the total non-malignant population, for the 11 other variables examined in the one-way sensitivity analyses conducted on the deterministic model, none of the ranges considered for these variables made an appreciable impact on the ICURs. The ICURs for OXN compared to OXY remained low across the variables and ranges examined. As shown in Supplemental Figures 7–9, similar results were found for the 11 other variables considered in the one-way sensitivity analyses for the other three patient populations.

Discussion

A CUA was undertaken to compare the costs and outcomes associated with treating moderate-to-severe chronic pain patients with either OXY and laxative treatment or OXN and laxative treatment in Canada. Both OXY and OXN are indicated for the management of moderate-to-severe pain; however, OXN is the only oral preparation currently approved and marketed in Canada for the concurrent management of pain and OIC. Since OXY and OXN have been shown to provide an equivalent analgesic effect, this economic evaluation focused primarily on the health and cost consequences of OIC. It was found that the cost-utility of OXN was low in all four patient populations examined and remained well below the commonly quoted threshold of $50,000 per QALY gained throughout the sensitivity analyses conducted65. As such, OXN would be considered to represent good value for money with a fair degree of certainty.

The cost-effectiveness of OXN has been compared in two other publications. In a CUA conducted in the UK, the cost-utility of OXN compared to OXY in patients on low dose analgesia (i.e., OXN3001) was found to be £5842 per QALY gained41. Considering differences in unit prices and differences in the management of OIC between Canada and the UK, this finding is very similar to the finding of the current evaluation for the same patient population (i.e., $2178 CDN per QALY gained). In another CUA conducted in Italy on patients with musculoskeletal pain and based on a meta-analysis on patients on different doses of analgesia (i.e., OXN3401, OXN3001, and OXN3006 trials), the authors reported a 1-year ICUR of €475 per QALY gained for OXN compared to OXY42. Again, considering the differences in unit costs and differences in practice patterns between the healthcare systems, this finding is very similarly to the cost-utility findings of $4376 per QALY gained in the current evaluation of the overall pooled non-malignant patient population (OXN9001).

There are a number of strengths associated with this assessment from a Canadian healthcare decision-making perspective. First, it is based on Canadian unit costs and practice pattern estimates for managing patients with OIC. Second, the primary clinical data used to populate the analysis is based on multi-centered head-to-head RCTs in which the comparator is recommended by clinical practice guidelines. Third, conservative assumptions were used in modeling the patient progression beyond the duration of the trial (i.e., LOCF; a conservative imputation method). Fourth, extensive deterministic, structural and probabilistic sensitivity analyses were performed to gauge the stability and robustness of the cost-utility results. Finally, this assessment included evaluations in a broad range of four patient populations (i.e., non-malignant chronic pain patients requiring high doses of analgesia, non-malignant chronic pain patients requiring low doses of analgesia, non-malignant chronic pain patients requiring a mixture of low and high doses of analgesia, and chronic pain cancer patients), which have not previously been considered in economic evaluations of OXN.

As with all decision analytic models and economic evaluations, the analysis also has its limitations. First, the constipation status for patient inclusion in the clinical trials was assessed using the ROME criteria which use the frequency of bowel movements and discomfort associated with constipation24. For this economic evaluation, the BFI was used because it was felt it was a more complete measure of constipation because it accounts for three factors: ease of defecation; feeling of incomplete bowel evacuation; and personal judgment of constipation25. Although both metrics are validated, and match quite closely, patients in a ‘not constipated’ health state according to BFI at the start of the model were assigned a ‘not constipated’ utility value. As shown in Table 1, there were more patients in the ‘constipated’ health state in the OXN treatment arms of the studies, which resulted in a bias against OXN using BFI as a measure of constipation. Since no adjustment was made to the base line differences in constipation using the BFI, the results from this analysis can be considered conservative for OXN. There are also some limitations with respect to the design of the clinical trials and data collected during the studies of OXN. For example, the clinical trials were short in duration—either 4 or 12 weeks.

Second, no resource use information was collected and no utility-based QoL assessment tool was used in the clinical trials. As such, resource use associated with the management of OIC was obtained from a separate survey of 101 clinicians in Canada, and the utility values used in the analysis were obtained from the literature and not directly from trial participants. Although the analysis of uncertainty suggested that alternative resource use estimates for managing OIC or alternative utility values from other studies made little difference for the ICUR results, direct measurement from patients in the trials is preferred.

Third, this model under-values cost savings that could arise through reductions in prescription opioid abuse and misuse. Abuse and misuse are significant concerns with OXY. Originally, Targin was developed to treat OIC in patients with moderate-to-severe chronic pain, and, thus, the clinical trials were designed to assess analgesia and bowel function. Consequently, this pharmacoeconomic model is based on the constipation state of the patient. Based on subsequent in vitro and human abuse liability studies, Targin has been demonstrated to have abuse-deterrent properties: it combines an opioid agonist and an opioid antagonist. Targin’s formulation deters abuse and misuse through various routes of administration (e.g., intranasal, intravenous, rectal)34. US and Canadian regulators have recognized the significant problem that prescription opioid abuse and misuse poses, and have called for the development of prescription opioids with abuse-deterrent (also known as tamper-resistant) properties to help address this multi-faceted problem. Recently, CADTH published a Rapid Review Response on the clinical and cost-effectiveness of tamper-resistant CR oxycodone formulations66. In its report, CADTH stated that, based on the publicly available clinical evidence for Targin and OxyNEO, there is reduced potential for misuse and abuse of tamper-resistant formulations of CR oxycodone. Furthermore, CADTH cited two cost studies which projected that tamper-resistant CR oxycodone would result in significant annual medical cost savings of $430 million (US)67, and reductions in healthcare service utilization68. Based on these results, CADTH concluded that ADF CR oxycodone may result in cost savings in Canada. Furthermore, CADTH stated that, ‘[t]amper-resistant oxycodone is likely to be an effective contributor to a broad opioid abuse and misuse strategy’66.

Fourth, long-term opioid administration may lead to analgesic tolerance, which may not have been observed during the trial durations. Longer-term follow-up data and more information on the pharmacokinetic and pharmacodynamic aspects of opioid tolerance development are needed in order to properly model tolerance in economic evaluations.

Finally, this analysis was conducted from a Canadian healthcare system perspective. As such, practices patterns and unit costs specific to Canada were used in the analysis. Although the clinical trial data are likely to be generalizable to other jurisdictions, resource use and cost used in the analysis may not be generalizable to other jurisdictions.

Conclusions

The cost-utility of Targin (OXN) plus laxative treatment compared to OXY plus laxative treatment was compared in four patient populations using the results from four pivotal clinical trials. Three non-malignant moderate-to-severe chronic pain populations were considered: patients requiring low dose analgesic (OXN3001); patients requiring high dose analgesic (OXN3006); and a combination of low and high dose analgesic patients (OXN9001). In addition, chronic pain cancer patients were also considered (OXN2001).

In all four patient populations, it was found there was a trade-off between slightly higher total expected costs for OXN treated patients compared to OXY treated patients, but also improved clinical benefits in terms of reduced constipation and resulting higher QALYs for patients. Although analgesic costs were found to be slightly higher for OXN treated patients, OXN also resulted in cost offsets to the healthcare system in terms of laxative drug use and other resources required for the management of OIC. The resulting 1-year cost-utility of OXN compared to OXY ranged from $2178–$7732 per QALY gained in the base case analysis, and it was found that these cost-utility results remained robust and at low values throughout a series of one-way deterministic analyses of uncertainty. It was only under an extreme assumption of using the most favorable assumptions for the constipation rates for OXY and least favorable assumptions for OXN that OXN no longer appeared to be cost-effective.

In conclusion, the superior clinical effectiveness of OXN in managing OIC compared to existing SOC (i.e., OXY and laxatives) indicates that OXN will provide patients with important clinical benefits over existing therapy for the management of moderate-to-severe pain and OIC at a small additional cost to the healthcare system.

Transparency

Declaration of funding

Purdue Canada provided funding support for the development of the model, analysis, and manuscript preparation. Purdue Canada provided copies of clinical study reports, results from a survey of physicians and relevant background references for the report. The authors had complete independent control over the methods, results, conclusions, and discussion.

Declaration of financial/other relationships

There are no other financial or other relationships to be declared for either of the authors.

Supplementary material available online

Acknowledgments

The authors wish to acknowledge Kristin Beard from Purdue Pharma (Canada) for her assistance in obtaining the clinical study reports, the results from a survey of physicians in Canada and for obtaining some of the relevant background references for the report.