Belgian cost-effectiveness analysis of hydroxocobalamin (Cyanokit) in known or suspected cyanide poisoning
- 1Panacea Officinalis BVBA, Antwerp, Belgium
- 2Emergency Department, ZNA Stuivenberg Hospital, Antwerp, Belgium
- 3I-CHER Interuniversity Centre for Health Economics Research (UGent-VUB), Gent and Brussels, Belgium
- 4Merck Serono, Lyon, France
- Correspondence to Severien Drieskens, Panacea Officinalis BVBA, Meir 80, Antwerp 2000, Belgium;
- Received 24 August 2012
- Revised 10 December 2012
- Accepted 3 January 2013
- Published Online First 29 January 2013
Background No published burden of illness or cost-effectiveness studies on cyanide (CN) poisoning exist. A health economic model has therefore been developed to determine the cost-effectiveness of hydroxocobalamin (Cyanokit) in suspected and known CN poisoning.
Objectives To assess the cost-effectiveness of hydroxocobalamin versus standard treatment in known or suspected CN poisoning in Belgium.
Methods Probabilities for neurological sequelae and mortality were applied based on phase III trials, literature data and the hydroxocobalamin preclinical trial. Since no cost and utility data for CN poisoning exist, costs and utilities of very comparable or related diseases derived from publicly available Belgian sources and literature were applied. Direct medical costs from the public healthcare payer's perspective were used. The time horizon was 1 year. Sensitivity analyses were performed to assess the robustness of the results.
Results Base-case analyses versus standard treatment revealed cost-effective results (incremental cost-effectiveness ratio=9921€/QALY) in suspected CN poisoning and dominance (more effective and cost-saving) in known CN poisoning. It was determined that 17 lives could be saved and one sequel prevented per year on a national level using hydroxocobalamin treatment. One-way sensitivity analyses varying efficacy, costs, utilities and time horizon demonstrated the robustness of the results. The results were most sensitive to the probability of death and neurological sequelae, but remained within acceptable limits of cost-effectiveness. Furthermore, it was shown that a longer time frame (5 or 10 years) leads to even more favourable cost-effective results.
Conclusions Hydroxocobalamin appears cost-effective to dominant compared with standard treatment in CN poisoning from a healthcare payer's perspective.
Cyanide (CN) pills are not only found in James Bond movies or Agatha Christie books. In the real world there are various sources of CN poisoning including inhalation of fire smoke, accidental or intentional ingestion, occupational exposure and terrorism. CN is a potent and rapidly acting poison that inhibits the cytochrome oxidase complex.1–3
There is no method for the rapid detection of CN poisoning in the prehospital setting or in the emergency department, so the diagnosis of acute CN toxicity is primarily clinical.2 ,3 Prompt and empirical treatment with an antidote is critical due to the rapid action of CN toxicity. Recent European guidelines advocate empirical treatment of CN toxicity from fire smoke inhalation by administration of hydroxocobalamin (Cyanokit).2 ,3
Cyanokit is the only CN antidote to contain hydroxocobalamin and is approved by the European Medicines Agency and the Food and Drug Administration for suspected or known CN poisoning. It acts by binding with CN to form cyanocobalamin, which is excreted in the urine, and is used to treat multiple sources of CN poisoning including smoke inhalation and ingestion. Hydroxocobalamin has demonstrated effectiveness in the prehospital setting with a reported survival rate ranging from 42% to 72% in victims of suspected CN poisoning.4 ,5 It appears to have the ideal antidote characteristics; it is easily administered, it rapidly neutralises CN with no effect on cellular oxygen, and is safe for prehospital use in smoke inhalation victims.6 Other antidotes include cobalt edetate, methaemoglobin-forming agents and thiosulfate; however, none of these agents is suitable for empirical use because of serious toxicity issues or a slow onset of action.
Patients who have serious CN poisoning may be at risk of death or CNS sequelae7 ,8 resulting in an important reduction in health-related quality of life (HRQoL) and financial cost from a healthcare perspective. Cardiac complications are also common in CN poisoning; the majority of cardiac disorders observed in a study by Fortin et al9 were cardiac arrests.
However, in the current environment where healthcare budgets are strictly limited, medicines are no longer measured on therapeutic added value (efficacy and safety) alone. Cost-effectiveness analyses are becoming increasingly important and, in Belgium, they are a mandatory part of the reimbursement request for drugs with an added therapeutic value compared with existing therapeutic alternatives.10 Furthermore, economic evaluation will help to raise awareness of the need to treat with hydroxocobalamin. This study aims to assess the cost-effectiveness of hydroxocobalamin versus standard treatment (ST).
A health economic analysis was performed to compare costs and health outcomes of CN poisoning treated with and without hydroxocobalamin. Two decision trees were developed based on the two registered indications of the drug; one for suspected CN poisoning resulting from smoke inhalation and one for known CN poisoning following intentional/accidental exposure. It was deemed necessary to use two separate decision trees as the probabilities of health outcomes (death, survival with or without neurological sequelae) differ due to different circumstances and dose of CN received. Two treatment arms in each decision tree were compared: hydroxocobalamin added to ST (hydroxocobalamin arm) and ST alone (ST arm).
ST consisted of 100% oxygen administration and, if necessary, artificial ventilation. If needed, advanced life support was started including aggressive cardiovascular support, control of epileptic activity and correction of metabolic acidosis. Carbon monoxide poisoning was treated with normobaric or hyperbaric oxygen therapy.
The health economic model estimates the cost and utility, expressed in quality-adjusted life years (QALYs), for an average patient over a period of 1 year in the base-case and over a period of 5 and 10 years in the sensitivity analyses. The basic model structure is outlined in figure 1. Owing to a lack of data in the literature regarding cardiac arrest, the ST arm was simplified.
Cost-effectiveness is typically expressed in terms of the incremental cost-effectiveness ratio (ICER), where the denominator is the incremental gain in health and the numerator is the incremental cost or saving obtained with a new treatment versus the current standard of care. A cost-utility analysis is a form of cost-effectiveness analysis in which costs are reported in monetary units (€) and health outcomes are converted to QALYs gained, thereby incorporating a measure of HRQoL (utility) into the health outcomes. The ICER is thus reported as the incremental cost per QALY gained (€/QALY), referring to the amount of money needed to produce one additional QALY.
Suspected CN poisoning (smoke inhalation)
In the hydroxocobalamin arm, efficacy data were based upon a prospective single-centre non-comparative phase III study.5 This study sampled 69 patients admitted to the Fernand Widal Hospital in Paris between June 1987 and February 1994, who were treated with hydroxocobalamin at the fire scene or in the ICU at hospital admission. No changes to the formulation of hydroxocobalamin have been made since this time. Supportive therapy (eg, 100% oxygen, catecholamines, artificial ventilation, hyperbaric oxygen therapy) was administered according to need and at the discretion of the treating physician. Forty-one patients survived without neurological sequelae, nine patients survived with neurological sequelae and 19 patients died. A placebo control group was not included as it was not considered ethical to withhold potentially lifesaving treatment from patients assigned to placebo. To obtain data for the ST arm, a MEDLINE literature review was performed for publications on death due to enclosed space fires in countries where hydroxocobalamin was unavailable (table 1).11–16
Neurological sequelae were only sporadically reported and therefore the conservative assumption was made that the probability of neurological sequelae in the ST arm equals the probability observed in the hydroxocobalamin arm. Although hydroxocobalamin demonstrated efficacy on the prevention of sequelae in a preclinical study,17 it is difficult to quantify the exact effect of hydroxocobalamin on neurological sequelae as these can occur due to a combined effect of various toxic components following smoke inhalation.
Known CN poisoning (intentional/accidental exposure)
In the hydroxocobalamin arm, efficacy data were based on a retrospective two-centre study in patients with confirmed exposure to CN via sources other than smoke inhalation.2 After a systematic review of the ICU admissions at the Fernand Widal and Lariboisière hospitals, 14 patients were identified as being treated for pure CN poisoning with hydroxocobalamin between 1988 and 2003. Nine patients survived without neurological sequelae, one patient survived with neurological sequelae and four patients died.
An extensive search of the literature was insufficient to generate significant amounts of data to quantify the health outcomes in the ST arm, possibly due to the lower incidence of known versus suspected CN poisoning. Consequently, the health outcomes from the placebo group of the controlled preclinical study, in which dogs received pure CN, were used. The dog model can be considered as a good predictor of the antidote effect in humans, based both on the pharmacokinetics and pharmacodynamics of hydroxocobalamin in dogs.17 The protocol for this study determined that dogs with severe neurological problems should be euthanised; however, in this model these dogs were considered to be survivors with sequelae as euthanasia would not occur in humans.
A systematic review of the literature found no cost of illness studies for CN poisoning. Based on expert opinion it was therefore assumed that, of the observed neurological sequelae in the phase III trials (psychomotor retardation, memory loss, confusion, intellectual degeneration, dementia, loss of speech, cerebral syndrome and anoxic encephalopathy),2 ,5 50% could be classified as being similar to Alzheimer's disease and 50% as being similar to neurological sequelae following a stroke. Consequently, it was assumed that 50% of the yearly average costs of Alzheimer's disease plus 50% of the yearly average costs of stroke would be a good estimate of the average cost of neurological sequelae due to CN poisoning.
The yearly direct medical costs (€) from the Belgian healthcare payer perspective per patient with Alzheimer's disease, stroke and cardiac arrest, derived from Belgian publicly available sources and literature, are shown in table 2 (2007 figures). An acute cost for neurological sequelae due to CN poisoning of €3507 and follow-up costs of €10 166 (year 1) and €8665 (years 2–10) were calculated. Additionally, an acute cost of €4286 and a follow-up cost of €4208 (year 1) and €2162 (years 2–10) was applied for a cardiac arrest. A cost of €667 was used per hydroxocobalamin administration.
Quality of life was taken into account by attributing each health state in the model a utility value between 0 and 1, where 0 represents death and 1 represents perfect health. The model then multiplies the time horizon with the weighted average utility value, based on the proportional distribution of patients over the different health states, to obtain QALYs.
A similar assumption was made for the utility of neurological sequelae as for the cost; 50% of the utility for Alzheimer's disease21 plus 50% of the utility for stroke22–24 would be a good estimate of the utility of neurological sequelae due to CN poisoning. A utility of 0.51 for neurological sequelae due to CN poisoning was calculated. The utility following a cardiac arrest (0.88) was based on a study of 67 patients who recently had a myocardial infarction.25
For patients who did not have an initial cardiac arrest and survived without neurological sequelae, utilities from the general population were used.26 The utilities for the age groups corresponding to the mean age of the patients in each of the two phase III trials were applied (0.93 for suspected CN poisoning (age 40–49 years) and 0.94 for known CN poisoning (age 30–39 years)). As the chance of survival following cardiac arrest was very low in the ST arm, we assumed that the utility of survival without sequelae in the ST arm was the same as the utility for survival without sequelae and without cardiac arrest in the hydroxocobalamin arm. This assumption will have led to an underestimation of the cost-effectiveness of hydroxocobalamin.
Because of the various assumptions used in the health economic models, one-way sensitivity analyses were carried out to assess the sensitivity of the ICERs obtained in the base-case to variations of several key input variables, leading to a validation of the robustness of the model. These key inputs included probability of survival with neurological sequelae and death (ST arm), costs and utilities. As the health outcomes and costs of neurological sequelae and cardiac arrest may last longer than 1 year, the time horizon of the model (5 or 10 years) was also varied to investigate its effect on the cost-effectiveness results. Table 3 summarises the parameter variability used in the one-way sensitivity analyses.
The results of the cost-effectiveness analysis demonstrated that hydroxocobalamin use in the treatment of suspected CN poisoning resulting from smoke inhalation is cost-effective in comparison with ST (table 4). At the 1-year time horizon there is an incremental cost of €772/patient and an incremental average gain in health of 0.08 QALYs compared with ST. The calculated ICER of 772/0.08=€9921/QALY remains considerably below (and thus is clearly favourable to) the threshold of €30 000–50 000/QALY, which is a commonly applied threshold for this region of Europe.27 ,28
Hydroxocobalamin use in the treatment of known CN poisoning following intentional/accidental exposure has been shown to be dominant (meaning more effective and less expensive than ST) and therefore money-saving. At the 1-year time horizon there is an incremental saving of €542/patient and an incremental average gain in health of 0.35 QALYs compared with ST.
Based upon the probabilities used in the decision trees, it could also be calculated that, for suspected CN poisoning, the number of patients needed to treat in order to save one extra life was 13. For known CN poisoning, hydroxocobalamin treatment would save one extra life for every three patients treated, and would also prevent neurological sequelae developing in one out of six patients treated.
The one-way sensitivity analyses varying efficacy, costs, utilities and time horizon demonstrated the robustness of the results (table 5). Results were most sensitive to the probability of death (due to its profound effect on HRQoL) and neurological sequelae (due to its important effect on HRQoL and costs) in the ST arm but remained within acceptable limits of cost-effectiveness, even with the very conservative assumption of a probability of 0.03 (suspected CN poisoning) and 0.00 (known CN poisoning) for neurological sequelae in the ST arm. Furthermore, it was demonstrated that, with a longer time frame, more favourable cost-effectiveness results were obtained.
CN poisoning is frequently associated with mortality and neurological sequelae and consequently has a profoundly negative effect on HRQoL. Moreover, CN poisoning has economic consequences due to the additional costs of the management of these sequelae. The outcome measure of this cost-effectiveness analysis is cost per QALY gained. Although CN toxicity is acute, the use of cost per QALY is considered appropriate as HRQoL is the important outcome for this condition. The target population for this analysis was the Belgian healthcare payer, as the model was constructed as part of the Belgian reimbursement request for hydroxocobalamin and, as such, only direct medical costs from the healthcare payer's perspective were included.
The base-case analysis suggests that hydroxocobalamin versus ST is very cost-effective (9921€/QALY) in suspected CN poisoning and even dominant (thus resulting in life and cost savings) in known CN poisoning. The difference between both indications can be explained by the observed therapeutic effect on mortality and neurological sequelae in the hydroxocobalamin preclinical study in known CN poisoning whereas, in suspected CN poisoning, a conservative assumption that there would only be an effect on mortality was made due to the complications in quantifying the effect of hydroxocobalamin on neurological sequelae in smoke inhalation victims. Using the probability of death and neurological sequelae in both decision trees, it was also determined that 17 lives could be saved and one neurological sequel prevented per year using the Belgian incidence of CN poisoning (on average 200 suspected and seven known cases of CN poisoning per year).
Sensitivity analyses revealed that the base-case results are durable and resistant to a wide range of changes in input parameters, thus validating the robustness of the model. Only a minor effect on the ICER was observed when the costs and utilities were varied. However, the one-way analysis showed that the results had some sensitivity to variations in the probability of death and sequelae in the ST arm, but the ICER remained in most cases below the commonly applied threshold of €30 000–50 000/QALY.27 ,28 An ICER slightly >€50 000/QALY was only observed when assuming a higher number of sequelae if treated with hydroxocobalamin versus ST in suspected CN poisoning (0.13 vs 0.03). However, we would like to emphasise the strong possibility of overestimating the ICER in this case, as it was demonstrated in the placebo-controlled study in dogs that the occurrence of neurological sequelae was lower in the hydroxocobalamin arm than in the ST arm (75 mg/kg: 16%, 150 mg/kg: 0% vs placebo: 24%).17 Additionally, the phase III trial in smoke inhalation showed that the probability of sequelae in patients with CN poisoning was similar to the rate of sequelae in patients without CN poisoning (14% vs 14%), leading to the hypothesis that, when hydroxocobalamin is administered rapidly and with sufficient dosing following intoxication, the sequelae due to CN poisoning may be prevented, leaving only the sequelae due to other toxic components such as carbon monoxide.5 Furthermore, prolonging the time horizon to 5 or 10 years appeared to strongly improve the estimated ICER, as the major part of the costs are accumulated during year 1 whereas the benefits in terms of QALYs gained from lives saved and sequelae avoided are incurred over the duration of the time horizon.
Owing to the difficulty in obtaining data on efficacy, costs and utilities, it was necessary to make assumptions and this is therefore a limitation of the study. For example, based on the neurological sequelae observed in the phase III trials, it was assumed that 50% of the costs and utilities of Alzheimer's disease plus 50% of the cost and utilities of stroke would provide an accurate estimate for sequelae due to CN poisoning. Moreover, placebo-controlled studies in humans could not be executed for ethical reasons, so assumptions were made based on literature and on the placebo-controlled preclinical study to obtain probabilities of health outcomes in the ST arms of suspected and known CN poisoning, respectively. A further limitation of the study is that literature searches were performed using the MEDLINE database only. However, diverse sensitivity analyses were performed to evaluate the importance of uncertainty related to efficacy and other parameters. It is important to note that, based on the incidence of CN poisoning (approximately 207 Belgian cases/year), hydroxocobalamin is by definition an orphan drug (ie, a drug intended for diagnosis, prevention or treatment of a life-threatening or chronically debilitating condition affecting not more then 5/10 000 individuals; http://www.ema.europa.eu). Due to the low incidence of such conditions, orphan drugs typically do not produce extensive clinical data and consequently health economic data are not required for the reimbursement request of these drugs in Belgium. However, as hydroxocobalamin was not registered as an orphan drug (the first Marketing Authorisation was obtained in France in 1996 and thus, based upon the European orphan designation guidelines, it was not possible to request an orphan drug status), this study provides an exceptional Belgian cost-effectiveness analysis of a drug to treat a condition with an incidence rate of <5/10 000 inhabitants, leading to a positive mind-set of the healthcare payer towards accepting this large number of assumptions.
In conclusion, the findings of this study are relevant to both clinicians and healthcare policymakers. They indicate that administration of hydroxocobalamin is a very cost-effective to dominant strategy for the management of CN poisoning, and thus a potential method for improving resource allocation and reducing the health burden related to CN poisoning. However, health economic evaluations are only one element in the decision-making process of granting reimbursement, and must be considered together with other factors including clinical effectiveness, disease severity, budget impact and available treatment alternatives.29
Hydroxocobalamin appears cost-effective to dominant compared with standard treatment in cyanide poisoning from a healthcare payer's perspective.
The authors would like to thank IMC Healthcare Communication for their assistance in preparing this manuscript.
Contributors SD contributed to the literature search, model development and report writing. KA (clinical expert) contributed to the report review. LA contributed to the model development and report review.
Funding This work was funded by Merck Serono.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.