Can emergency departments not afford to carry essential antidotes?

Toxicology

CJEM 2002;4(1):23-33

Abstract

Résumé

Presented at the Canadian Association of Emergency Physicians Annual Scientific Meeting, Saint John, NB, June 2000, and at the North American Congress of Clinical Toxicology, Montreal, Que., October 2001.

Introduction

Although the vast majority of acutely poisoned patients need only supportive care, the successful management of a selected subgroup of poisonings may require a specific antidote. Indeed, the timely administration of an antidote can occasionally be life-saving, even in a patient already receiving optimal supportive care. Such cases are uncommon,^1-5 but are often complicated by deficiencies in the quantity and accessibility of local antidote stocks. Recent surveys in a variety of jurisdictions have demonstrated glaring inadequacies in the availability of common antidotes in the majority of emergency departments (EDs).^4-14

The 2 greatest barriers to adequate antidote stocking are a lack of awareness of the deficiencies, due in part to the low likelihood of antidote utilization, and a belief that maintaining such a stock would be excessively costly.^4,8,9,15 Both of these barriers are accentuated in smaller centres. The high purchase price of certain antidotes such as anti-digoxin F_abfragments and fomepizole contribute to the perception of excessive cost. Moreover, publications addressing the issue of deficient antidote stocking have focused exclusively on purchase price. For example, recently issued US consensus guidelines report a cost of US$19 808.60 to acquire essential antidotes.¹⁶ However, purchase price is only one component of the cost. Once purchased, it is shelf life and frequency of utilization that determine the cost to maintain such medications on formulary, as these factors determine the probability of actually using the antidote prior to its expiry. For rarely used antidotes with long shelf life, the annual maintenance cost is only a small fraction of the purchase price. In an effort to more accurately estimate the true budgetary impact of maintaining a stock of antidotes from the perspective of the ED pharmacy, we set out to model these costs for a Canadian community hospital.

Methods

The following input parameters were included in the cost model: antidote formulation, purchase price, frequency of administration, shelf life and supplier replacement policy for expired antidote returned unused. The setting was assumed to be a Canadian civilian setting with no exceptional local industrial hazards or venomous animals. Cases were assumed to present in a random fashion. Costs were divided into initial acquisition costs to purchase the antidote stock, and maintenance or ongoing costs needed to replace used or expired antidote. All figures are in Canadian dollars.

Community size and location

Antidote costs for 4 different ED configurations, categorized by size (small, medium) and location (suburban, remote), were modelled. Size was based on the hospital's catchment population, with small being 20 000 persons and medium 100 000. A suburban ED was expected to stock sufficient antidote for the initial 1 hour from presentation, whereas a remote centre was expected to manage the patient for 4 hours -- until arrival of additional supplies from a central source or until transfer of the patient to a referral centre.

Frequency of antidotal administration

In the absence of a Canadian equivalent, estimates of antidote utilization frequency were based primarily on the American Association of Poison Control Centers Toxic Exposure Surveillance System (TESS) data.^1-3 This US national database collects information reported to approximately 65 poison centres, and is estimated to reflect 95% of the US population. Cases that involved the use of an antidote and a call to a poison centre were collected during calendar 1996 to 1998, and divided by the population base serviced during that interval. The inherent limitations of the database precluded correcting for antidote use outside the initial 1- or 4-hour window, for the occasional inappropriate or unnecessary use of certain antidotes, and for antidote not being administered due to lack of availability. For commonly administered (more than once/year/100 000 population) antidotes like N-acetylcysteine, activated charcoal, naloxone, flumazenil and NaHCO₃, TESS data was adjusted upwards by 30% to correct for the underreporting that occurs when emergency physicians administer these agents without consulting a poison centre.

Estimates of the use of D₅₀W (dextrose 50% in water) and thiamine, part of the so-called "coma cocktail" for altered mental status of unclear etiology, could only be approximated to within an order of magnitude as use of these agents is not reported by TESS. These agents were included, however, to provide a comparison for commonly used but inexpensive antidotes. The antidotal use of octreotide and cyproheptadine, 2 novel antidotes for sulfonylurea-induced hypoglycemia and serotonin syndrome respectively, are not reported by TESS. Intravenous calcium and glucagon are also not separately reported by TESS. Estimated frequency of use for these 4 agents was derived from the number of moderate to severe poisonings in the appropriate agent exposure data provided by TESS. Fomepizole was assigned a frequency estimate equal to ethanol, and hydroxocobalamin equal to sodium thiosulfate. No adjustment was made for the use of agents like glucagon and atropine for non-poisoned patients.

Price and shelf life

Drug costs reported reflect hospital pharmacy costs at our hospital based on current supplier contracts. For medications over $20/dose, the manufacturer's recommended shelf life and the current supplier's policy for replacement of expired stock were obtained.

Drug dose and formulation

A dose sufficient to treat one 80-kg adult in the majority of cases was estimated based on published recommendations and the authors' experience.^{6-9,11-13,15-18} Although the occasional severely poisoned patient might require massive doses of a certain antidote,¹⁶ the level of stocking was selected to provide the capacity to initiate treatment in almost all cases encountered.

The available drug formulation most appropriate to stocking was selected, based on convenience of administration and on minimizing cost. For most antidotes the entire dose stocked was assumed to be used with each case encountered, and any excess either wasted or transferred with the patient to a referral centre. With several antidotes, however, the minimum recommended stocking level is far greater than the dose typically administered. Moreover, some of these antidotes (activated charcoal, anti-digoxin F_abfragments, D₅₀W, glucagon, NaHCO₃ and naloxone) are supplied in a format that permits partial use without waste, so that any remaining medication would be used prior to expiry or replaced for free. These antidotes were arbitrarily modeled to a mean consumption of 50% per case encountered. This figure was selected to compensate for administration of partial doses titrated to effect, as well as for use in patients weighing less than 80 kg. The time required to restock administered or expired antidote was assumed to be negligible.

Cost model

The full derivation of the cost model used to estimate the ongoing maintenance costs for each antidote is shown in Appendix A. For those antidotes where full credit is provided by the supplier on expiry, shelf life can be considered to be unlimited, and maintenance costs reflect actual use. For the remaining antidotes, the probability that a given antidote will be used 0, 1, 2 or more times during their finite shelf life is given by the Poisson probability distribution, which describes rare independent events such as self-poisonings.

Cost reduction strategies

For expensive (over $100/dose) antidotes not replaced free upon expiry (i.e., fomepizole and the cyanide antidotes), the impact of a change in supplier policy to free replacement was modeled. For expensive antidotes in which the smallest formulation available was more than double the minimum recommended stocking amount (i.e., fomepizole), the impact of distributing partial lots between centres to defray costs was also modeled.

Results

The list of antidotes considered, the recommended stocking level, purchase price and shelf life are shown in Table 1, in roughly decreasing frequency of expected use. This list was derived from the World Health Organization International Programme on Chemical Safety,¹⁹ recent US expert consensus guidelines,¹⁶ and other sources.^8,9,15,17,18Each agent listed could reasonably be expected to be administered in a Canadian ED, generally in consultation with a poison information centre or toxicologist, shortly after patient presentation and without the benefit of sophisticated toxicology testing. The initial cost to purchase the entire set of listed antidotes is $9250 for suburban centres, based on a decision to stock fomepizole rather than ethanol. Remote centres face an additional cost of $940 (10.2%) to enable them to treat a patient from 1 to 4 hours after presentation (Table 2).

Table 2 shows that annual maintenance costs increase with increasing antidote price and frequency of use. For antidotes that commonly expire prior to use, these costs decrease substantially as the shelf life of the antidote increases. Obviously there is cost without benefit when the antidotes expire unused. Centres electing to stock the entire list of antidotes shown, based on the parameters given, would experience average annual maintenance costs ranging from $2130 to $5410. Distance to a referral centre has a more substantial impact on annual maintenance costs than on initial cost. Figure 1 illustrates the interplay of input variables on cost per unit time.

Taken together, the most expensive antidotes not presently replaced free on expiry, fomepizole and the nitrite/thiosulfate cyanide kit, could account for over 80% of annual antidote costs for a small centre, so cost reduction strategies would logically begin with these agents. Table 3 shows the estimated impact of a change in supplier policy to free replacement for these antidotes. Another possible cost reduction strategy is to share antidote stocks between hospitals. For example, hospitals could reduce the purchase price of fomepizole -- which is supplied as 4 vials, each containing 1500 mg -- by sharing the vials among 3 other hospitals. Because each vial contains sufficient antidote to fully load a 100-kg patient and inhibit alcohol dehydrogenase for over 12 hours in the absence of hemodialysis,²⁰ this strategy would allow each hospital an adequate stock to initiate treatment. Table 3 shows different possible scenarios regarding replacement of the opened or partially used fomepizole 4-packs. In combination, these strategies could reduce the total annual antidote maintenance cost to as low as $737 for a small suburban centre.

The alternative cyanide antidote, high-dose hydroxocobalamin, is sold in France as a 2.5-g vial (Cyanokit, Lipha Santé, Lyon, France) for 2000 French francs, has a 30-month shelf life and is not replaced for free on expiry. Two vials, in addition to 8 g of sodium thiosulfate (available without purchasing the nitrite kit), have been recommended as the typical adult dose.²¹ This antidote may be obtained via the Special Access Programme of the Therapeutic Products Directorate of Health Canada, and the manufacturer has applied to the US Food and Drug Administration for approval as an orphan drug (M. Mosnier, Lipha Santé: personal communication, Aug. 30, 2001).

Discussion

In managing poisoned patients, one of the most frustrating experiences for emergency physicians and consulting toxicologists alike is to encounter the lack of timely availability of a potentially life-saving antidote. Unfortunately, this problem of inadequate antidote stocking is common and widespread. Indeed, only half of Massachusetts hospitals reportedly stock sufficient pyridoxine to treat a single patient with isoniazid-induced seizures,⁹ and nearly one-third to one-half of hospitals in several US states do not stock antidote with which to treat a methanol or ethylene glycol ingestion.^8,9,11,14 Even fewer hospitals have reported adequate stocking of glucagon and anti-digoxin F_ab fragments.^8,13,14 Recently published surveys suggest that the situation is no better in Canada. Only 1 of 179 acute care hospitals in Ontario reported adequate stocks of 10 essential antidotes.⁶ From a list of 13 essential antidotes, Quebec hospital pharmacies reported adequate stocks of a median of 3, and not more than 9 antidotes.⁷

Barriers to antidote stocking

Important barriers exist to antidote stocking.^9,15,18Foremost is the rarity with which these antidotes are required. Indeed, the vast majority of poisoned patients can be successfully managed with supportive care alone. Because the need for antidotal therapy is rare, most physicians and pharmacists are unaware of antidote deficiencies in their practice setting. Moreover, even after a sentinel event demonstrates the lack of a certain antidote for a given patient, corrective measures may mistakenly target only that particular antidote due to lack of recognition of the plural nature of the problem. When the deficiency is acknowledged, a distorted estimate of the related costs can prevent remedial action. Physicians' perceptions of medication and investigation costs are often skewed, and most Canadian emergency physicians significantly overestimate the price of common antidotes.²² The high purchase price of medications such as anti-digoxin F_ab fragments contribute to this distortion. However, the rarity of antidote administration makes it important to look beyond the purchase price in calculating the actual costs of maintaining an adequate stock of these medications. Our model demonstrates that shelf life, supplier replacement of expired stock, and frequency of use are far more important determinants of the true cost to the departmental pharmacy budget.

Expensive antidotes

Table 2 illustrates that the antidotes with the highest acquisition costs are not necessarily the most expensive antidotes over time. Rather, 2 inexpensive antidotes used often (D₅₀W and activated charcoal) and 2 expensive antidotes that are not replaced for free upon expiry (fomepizole and the nitrite/thiosulfate cyanide kit) account for the large majority of annual maintenance costs, regardless of hospital size or location. The most expensive antidote, anti-digoxin F_ab fragments, is not very costly on an annual basis (despite its very short shelf life) because the supplier accepts returned expired product for full credit, and the small formulation reduces waste.

Alternative agents exist for the 2 most costly antidotes. High-dose hydroxocobalamin may be used for victims of cyanide poisoning, and ethanol may be given instead of fomepizole. Hydroxocobalamin is thought to be safer than the nitrite/thiosulfate cyanide kit,²¹ but hydroxocobalamin may be even more expensive.

Fomepizole is superior to ethanol with regard to safety and ease of administration, and may obviate the need for hemodialysis when administered early after ethylene glycol ingestion.^20,23However, a high purchase price, large formulation and the absence of free replacement on expiry make fomepizole very costly. This is especially true for small centres, where it is more likely to expire than to be used during its 36-month shelf life. Ironically, it is in these small centres where fomepizole has the greatest potential benefit relative to ethanol, as these centres typically lack hemodialysis, intensive care facilities, familiarity with ethanol therapy, and the ability to measure methanol or ethylene glycol levels. Moreover, in selected cases of exposure rather than toxicity, interfacility transfer could be facilitated or perhaps averted if fomepizole were used.

The large drug cost difference between fomepizole and ethanol creates a particular dilemma for Canadian emergency physicians. The US manufacturer of fomepizole provides free replacement to US hospitals for unused expired product.²⁴ At the same time, the cost of pharmaceutical grade intravenous ethanol can be over 7 times greater in the US than in Ontario,^6,16reducing the cost disparity between these 2 antidotes in the US but not in Canada.

Interestingly, hospital location had only a minor impact on overall costs. Because almost all of these antidotes should be administered within the first hour for optimal efficacy, proximity to a referral centre does not preclude the need for an ED to stock most of these agents. This time dependence also has implications for rapid access to antidote stocks on a 24-hour basis, raising questions about current strategies that rely upon antidote stockpiled outside the hospital.

Limitations

The safety, efficacy and cost of a given antidotal strategy should be compared to those of alternate therapies. Because this analysis is not a cost-benefit or cost-effectiveness analysis, we did not perform such comparisons. Our analysis does not take into consideration the fact that increased availability of any medication increases the likelihood of its overuse, misuse or abuse, which would increase related costs without benefit and, at times, with harm. Flumazenil and anti-digoxin F_abfragments are examples of antidotes prone to unnecessary use, which could be minimized by restricting access to them. Other antidotes (e.g., nitrites in cyanide antidote kit) have a narrow therapeutic window, and have the potential to harm even patients for whom the antidote is indicated, especially when used by physicians unfamiliar with the agent.

The decision to stock certain antidotes is, by its nature, somewhat arbitrary. Although based on published recommendations,^15-17,19 this report is not intended to provide the definitive list of antidotes and doses to be carried in every Canadian ED. Rather, this analysis is intended to provide emergency physicians with some of the tools needed to help make these decisions. Antidotes vary with regard to efficacy and redundancy.²⁵ Several antidotes, including antivenoms, chelating agents and exotic agents, have been excluded as either unlikely to be administered annually in all of Canada, or unlikely to be given empirically in smaller centres within 4 hours of presentation. Some EDs may elect not to carry some antidotes included herein. Even experts disagree on the need to stock flumazenil and physostigmine, and whether to stock ethanol or fomepizole.^16,26

The frequency estimates used in our models are approximate. While the Toxic Exposure Surveillance System (TESS) database represents the highest quality information on antidotal administration, it only reports cases in which the poison centre was aware of the use of an antidote. Underreporting occurs in cases where a poison centre was not consulted, and cases in which antidote should have been administered but was not, perhaps due to limited availability or due to failure of the treating physician to follow poison centre recommendations. Conversely, over-reporting occurs in cases of inappropriate antidote use. There is evidence demonstrating the TESS dataset to be incomplete when compared to poisoning fatalities as ascertained from death certificates filed with the US National Center for Health Statistics.²⁷Specifically, poison centre data vastly underreport these poisoning fatalities, especially in unintentional drug and non-drug deaths. The role of antidotes in these fatalities cannot be directly calculated from the aggregate data reported, but it is likely that a disproportionately large number of these severely poisoned patients would qualify for antidotal administration. Finally, even after removing venomous animals from consideration, it is likely that differences exist in the epidemiology of poisoning between Canada and the US. Unfortunately, there is no comparable Canadian mechanism to obtain information on antidote administration on a national level. We believe the frequency estimates provided correspond to the situation in most Canadian EDs to within an order of magnitude, and are derived from the best available data.

Our model assumed equal frequency of administration of either ethanol or fomepizole, and of the cyanide antidotes. However, differences in safety, ease of use and cost may influence physicians to selectively use one antidote over another. For example, some physicians may have a lower threshold for initiating empirical treatment with fomepizole due to its favourable safety profile, while others may see its purchase price as a deterrent.

The model was not designed to handle multiple simultaneous casualties, as might occur due to a chemical spill, suicide pact or bioterrorist attack. Such events violate assumptions of independence used in the derivation of the model, as well as affecting the minimum recommended stocking level. Fortunately, these events are even less common than single patient poisonings, and they pose emergency preparedness challenges that go beyond antidote stockpiling. As with any model, the underlying assumptions are subject to criticism. For example, this model may overestimate costs to a small degree, as not all cases of antidote administration will require antidote to be administered during the initial 1 or 4 hours.

We did not perform a formal sensitivity analysis; however, we did provide the explicit mathematical formulae used in the model, allowing readers to determine the effect of any change in the underlying input parameters. For example, both initial and maintenance costs are linear in purchase price, and therefore a 20% increase in price similarly increases cost estimates by 20%. The effects of changes in shelf life and frequency of administration are more complex, but can be visually estimated from Figure 1.

Moreover, the model can accept any inputs, and be adapted to the local circumstances (population base, time to transfer, time from central repository of antidote), differences in supplier pricing and expiry policy, and selection of antidotes and formulation. An electronic, cost-calculating spreadsheet containing our model is available online for interested readers. This tool will accept user-defined inputs for all parameters including the ability to delist certain antidotes, or to select ethanol over fomepizole, and interactively recalculates Table 2. By modifying quantities of antidote stocked, the user can also increase the time interval of treatment beyond 4 hours for very isolated centres. This interactive tool is provided in lieu of a sensitivity analysis.

Future problems

Unfortunately, other barriers exist to comprehensive antidote stocking, which will limit the corrective impact of education and accurate cost models. The small market for antidotes relegates most to the status of orphan drugs. Manufacturers therefore have limited incentive to invest in the research necessary to develop new antidotes or to maintain current supplies. Fiscal restraint has pressured many institutions to downsize formularies, often removing antidotes that tend to expire unused, without due consideration to the availability and efficacy of alternative therapies. Even if a certain antidote is present on the formulary, it may be stocked in insufficient quantity or in an inaccessible location.

The issues involved in ED preparedness for rare events are complex and cannot be reduced to cost alone. The societal role and obligation of the ED to offer the ability to treat severely poisoned patients, and who should bear the cost of this obligation are matters for public debate and policy. In the absence of a regulatory requirement for mandatory antidote stocking, perhaps linked to hospital accreditation, it is unlikely that voluntary action alone will increase antidote stocking above current levels. The advantages of central regulatory oversight might include negotiating with suppliers for free replacement of unutilized antidote, monitoring and intervention during drug shortages, central drug stockpiling, rotating antidote to high-use centres prior to expiry, guaranteeing market for suppliers, and preventing monopolistic practices.

Conclusions

Our model provides a more accurate estimate of antidote stocking costs than other published guidelines.¹⁶ Where fiscal constraints preclude the stocking of every antidote listed, the cost model provides much of the information necessary to decide which antidotes to carry, or even to price alternative therapies. This information allows the motivated ED to analyze and optimize its current level of antidote availability. Clearly, when an antidote is truly essential (i.e., there is no antidote of comparable efficacy), cost alone cannot be used as the primary motive to delist the antidote.

References

1. Litovitz TL, Klein-Schwartz W, Caravati EM, Youniss J, Crouch B, Lee S. 1998 Annual report of the American Association of Poison Control Centers Toxic Exposure Surveillance System. Am J Emerg Med 1999;17:435-87.
2. Litovitz TL, Klein-Schwartz W, Dyer KS, Shannon M, Lee S, Powers M. 1997 Annual report of the American Association of Poison Control Centers Toxic Exposure Surveillance System. Am J Emerg Med 1998;16:443-97.
3. Litovitz TL, Smilkstein M, Felberg L, Klein-Schwartz W, Berlin R, Morgan JL. 1996 Annual report of the American Association of Poison Control Centers Toxic Exposure Surveillance System. Am J Emerg Med 1997;15:447-500.
4. Ong HC, Yang CC, Deng JF. Inadequate stocking of antidotes in Taiwan: Is it a serious problem? J Toxicol Clin Toxicol 2000;38:21-8.
5. Nogue S, Munne P, Soy D, Milla J. [Availability, use and cost of antidotes in Catalonia.] Med Clin (Barc) 1998;110:609-13.
6. Juurlink DN, McGuigan MA, Paton TW, Redelmeier DA . Availability of antidotes at acute care hospitals in Ontario.CMAJ 2001;165:27-30.
7. Bailey B, Bussières JF. Antidote availability in Quebec hospital pharmacies: impact of N-acetylcysteine and naloxone consumption. Can J Clin Pharmacol 2000;7:198-204.
8. Dart RC, Stark J, Fulton B, Koziol-McLain J, Lowenstein SR. Insufficient stocking of poisoning antidotes in hospital pharmacies. JAMA 1996;276:1508-10.
9. Woolf AD, Chrisanthus K. On-site availability of selected antidotes: results of a survey of Massachusetts hospitals. Am J Emerg Med 1997;15:62-6.
10. Howland MA, Weisman R, Sauter D, Goldfrank L. Nonavailability of poison antidotes. N Engl J Med 1986;314:927-8.
11. Chyka PA, Conner HG. Availability of antidotes in rural and urban hospitals in Tennessee. Am J Hosp Pharm 1994;51:1346-8.
12. Santucci KA, Shah BR, Linakis JG. Acute isoniazid exposures and antidote availability. Ped Emerg Care 1999;15:99-101.
13. Lowe JN, Tandy TK. ß-Adrenoreceptor antagonist toxicity: a survey of glucagon availability. Ann Emerg Med 1993;22:267-8.
14. Teresi WM, King WD. Survey of the stocking of poison antidotes in Alabama hospitals. South Med J 1999;92:1151-6.
15. Bussières JF, Bailey B. Insufficient stocking of antidotes in hospital pharmacies: problem, causes, and solution. Can J Hosp Pharm 2000;53:325-37.
16. Dart RC, Goldfrank LR, Chyka PA, Lotzer D, Woolf AD, McNally J, et al. Combined evidence-based literature analysis and consensus guidelines for stocking of emergency antidotes in the United States. Ann Emerg Med 2000;36:126-32.
17. Olson KR, editor. Poisoning and drug overdose.Norwalk (CT): Appleton & Lange; 1994.
18. Leiken JB, Paloucek FP, editors. Poisoning and toxicology compendium. Hudson (OH): Lexi-Comp; 1998.
19. Pronczuk de Garbino J, Haines JA, Jacobsen D, Meredith T. Evaluation of antidotes: activities of the International Programme on Chemical Safety. J Toxicol Clin Toxicol 1997;35:333-43.
20. Sivilotti ML, Burns MJ, McMartin KE, Brent J. Toxicokinetics of ethylene glycol during fomepizole therapy: implications for management. Ann Emerg Med 2000;36:14-125.
21. Hall AH, Rumack BH. Hydroxocobalamin/sodium thiosulfate as a cyanide antidote. J Emerg Med 1987;5:115-21.
22. Innes G, Grafstein E, McGrogan J. Do emergency physicians know the costs of medical care? CJEM 2000;2:95-102.
23. Tenebein M. Recent advancements in pediatric toxicology.Pediatr Clin North Am 1999;46:1179-88.
24. Goldfarb DS. Fomepizole for ethylene-glycol poisoning [editorial]. Lancet 1999;354:1646.
25. Jacobsen D. The relative efficacy of antidotes. J Toxicol Clin Toxicol 1995;33:705-8.
26. Barceloux DG, Krenzelok EP, Olson K, Watson W. American Academy of Clinical Toxicology practice guidelines on the treatment of ethylene glycol poisoning. J Toxicol Clin Toxicol 1999;37:537-60. 27. Hoppe-Roberts JM, Lloyd LM, Chyka PA. Poisoning mortality in the United States: comparison of national mortality statistics and poison control center reports. Ann Emerg Med 2000;35:440-8.