Mortality related to pharmacotherapies for opioid dependence: a comparative analysis of coronial records

Drug and Alcohol Review (July 2007), 26, 405 – 410 Mortality related to pharmacotherapies for opioid dependence: acomparative analysis of coronial records National Drug and Alcohol Research Centre, University of New South Wales, Australia AbstractIntroduction and Aims. The aim of this study was to compare the mortality associated with oral naltrexone, methadoneand buprenorphine in opioid dependence treatment, employing a retrospective data analysis using coronial and prescriptiondata. Design and Methods. The number of deaths were identified through national coronial data and number of treatmentrecipients were estimated from 2000 to 2003 prescriptions and restricted medications data. Mortality rates were expressed asdeaths per number of treatment episodes and per person-years at high and low risk of fatal opioid overdose. Results. Thirty-twooral naltrexone, one buprenorphine and 282 methadone-related deaths were identified. Mortality rates in the highest risk periodin deaths per 100 person-years were 22.1 (14.6 – 32.2) for oral naltrexone following treatment cessation and 3.0 (2.3 – 3.9) formethadone during treatment induction. Rates in the lowest risk period in deaths per 100 person-years were 1.0 (0.3 – 2.2)during oral naltrexone treatment and 0.34 (0.3 – 0.4) during post-induction methadone treatment. The relative risk of death fororal naltrexone subjects was 7.4 times (high-risk period, p 5 0.0001) or 2.8 times (low-risk period, p ¼ 0.055) that ofmethadone subjects. Discussion and Conclusions. This is the first comparison of mortality associated with these threepharmacotherapies for opioid dependence. The risk of death related to oral naltrexone appears higher than that related tomethadone treatment. [Gibson AE, Degenhardt LJ. Mortality related to pharmacotherapies for opioid dependence: acomparative analysis of coronial records. Drug Alcohol Rev 2007;26:405 – 410] Key words: buprenorphine, methadone, mortality, naltrexone, opioid dependence.
the primary drug for opioid maintenance pharmaco- therapy [5] and plays an important role in retaining Opioid dependence is a chronic, relapsing disorder patients in treatment, improving health, reducing cri- [1,2] associated with elevated mortality risk, with death minal activity and decreasing heroin and other drug use most commonly from opioid overdose [2 – 5]. The [7]. Buprenorphine is a more recently developed medi- mortality rate among untreated opioid-dependent in- cation that is being prescribed increasingly in many dividuals has been estimated at 0.9 per 100 person- countries [8,9] and is similarly effective to methadone in years [6]. Reducing this mortality risk is an important terms of retention and suppression of heroin use [10].
aim of opioid dependence treatment. However, users Naltrexone, in contrast, has been available for many often cease treatment; and that mortality risk will differ years but has remained little-used due to low interest during and following the treatment episode depending among opioid-dependent individuals and poor compli- upon the nature of the treatment modality. The current ance with treatment [5,11,12]. No significant benefit of study compares mortality risk related to three pharma- naltrexone over placebo has been found in terms of cotherapies for opioid dependence: naltrexone, bupre- retention, side effects or relapse to heroin use [13].
Each of these treatments bears some mortality risk, The range of maintenance pharmacotherapy treat- which differs over the course of treatment due to the ments available for opioid dependence include opioid different mechanisms of drug action. Methadone antagonists such as naltrexone, partial opioid agonists provides some cross-tolerance to opioids, so once such as buprenorphine and full opioid agonists such as inducted into treatment recipients have a lower over- methadone. For several decades, methadone has been dose mortality rate than untreated opioid-dependent Amy E. Gibson MPH BSc (Hons) BA, Senior Research Officer, National Drug and Alcohol Research Centre, University of New South Wales,Australia, Louisa J. Degenhardt PhD, MPsychol (Clinical), BA (Hons), Senior Lecturer, National Drug and Alcohol Research Centre, Universityof New South Wales, Australia. Correspondence to Amy Gibson, NDARC, UNSW, Sydney, NSW, Australia 2052.
E-mail: [email protected] Received 17 July 2006; accepted for publication 18 December 2006.
ISSN 0959-5236 print/ISSN 1465-3362 online/07/040405–06 ª Australasian Professional Society on Alcohol and other DrugsDOI: 10.1080/09595230701373834 subjects [14,15]. During induction, however, the risk of as used in routine clinical practice. This report uses fatal opioid overdose is high, with one study estimating coronial data to quantify and directly compare the 7.4 deaths per 100 patient-years in the first 2 weeks of mortality associated of oral naltrexone with that of methadone treatment, compared to 0.07 deaths per 100 buprenorphine and methadone treatment for opioid patient years beyond 2 weeks [16], and another study dependence in Australia, including deaths both during finding that 21% of deaths in methadone treatment and after an episode of treatment. This will provide the occurred during the first week [17]. Death is caused first estimate of the mortality rate related to oral primarily by respiratory failure or complications arising from opioid overdose, and methadone is usually detected in post-mortem toxicology [18,19].
Buprenorphine has a longer mechanism of action estimate the number of people receiving phar- than methadone, with a flatter dose – response curve, macotherapy for the treatment of opioid depen- so that high doses prolong but do not increase the peak effects [20], making subjects less likely to experience a estimate the number of deaths related to these fatal overdose [21]. The great majority of all fatal over- doses have involved the concomitant administration estimate mortality rates according the number of other respiratory-depressant drugs, usually ben- of treatment episodes, and among periods of buprenorphine-only deaths have been reported [23].
Opioid overdose deaths in France have decreased sub- stantially since the introduction of buprenorphine [24]and the mortality rate attributed to buprenorphine treat- The number of deaths related to naltrexone, buprenor- ment has been estimated at 0.24 per 1000 patients [25].
phine and methadone were determined by keyword Naltrexone is an opioid antagonist, used primarily as searches of the National Coronial Information System a maintenance drug to aid opioid abstinence. Overdose (NCIS). The NCIS is a regularly updated electronic rates associated with naltrexone have been less well database allowing access to all coronial cases in documented than with methadone and buprenorphine, Australia [31]. Between 2000 and 2003, inclusive, an perhaps because the greatest risk occurs after the estimated 88% of a total of 66 659 coronial cases cessation of treatment. While compliant with naltrex- nationally were closed coronial cases (cases no longer one, the effects of any opiates administered are blocked under coronial investigation), and so were able to be or reduced substantially [26]. The primary cause of used in this analysis (NCIS, unpublished data).
naltrexone-related death is by opioid overdose after National registration data were used for the number cessation of naltrexone treatment [27]. Such deaths are of buprenorphine and methadone treatment episodes.
unlikely to have naltrexone detected in post-mortem As naltrexone is only available privately for the toxicology, and only the mention of recently ceased (or treatment of opioid dependence and registration data non-compliant) naltrexone treatment identifies the are not kept, the number of private prescriptions of oral death as naltrexone-related. Less commonly recognised naltrexone and expert clinical and research experience causes of naltrexone-related death include fatal opioid [21] of the typical naltrexone treatment retention were overdose during naltrexone treatment [27], and death used to calculate the number of naltrexone treatment from a severe adverse reaction to naltrexone [28].
episodes. Mortality relating to unregistered forms of There has been limited comparative work examining naltrexone, such as depot formulations, was not mortality rates across these three pharmacotherapies for opioid dependence. Higher rates of opioid overdose Mortality rates were calculated using: (a) a crude rate (mainly non-fatal) in naltrexone subjects compared to of deaths per 1000 treatment episodes and (b) a methadone and buprenorphine subjects was noticed in a stratified rate of deaths per 100 person-years at high group of Australian studies [29]. Subjects leaving naltr- or low risk of death. All deaths were classified as exone treatment had an overdose rate eight times that of occurring in either the high- or low-risk period of death; subjects leaving agonist treatment, and naltrexone high-risk period deaths occurred if the date of death was recipients were six times more likely to experience an recorded as being within 2 weeks after the cessation of overdose out of treatment than in treatment. Three of naltrexone treatment episode or in the first week of the 27 overdoses proved fatal [29]. Another Australian methadone or buprenorphine treatment.
study reported higher rates of non-fatal overdose in Deaths during naltrexone treatment or after the first naltrexone recipients compared to opioid agonist treat- week of methadone or buprenorphine treatment were classified as occurring during the low-risk period.
There has been no epidemiological work completed Deaths where the timing was uncertain were classified to examine rates of mortality related to these treatments in the low-risk period of death and deaths occurring Mortality related to pharmacotherapies for opioid dependence more than 2 weeks post-treatment cessation were not Methadone treatment was associated with a mortality considered to be related to the treatment. Treatment rate of 3.0 per 100 person-years during the high-risk episodes were considered to have ended if the patient period (first week of treatment) and 0.34 per 100 had formally left treatment at the time of their death.
person-years during the period of low risk (the Using a mortality rate of deaths per person-years at risk remainder of the treatment episode). Buprenorphine reduces any bias caused by differing treatment reten- mortality rates were not calculated using this method tion between the three pharmacotherapies, as retention due to the low number of deaths detected. Naltrexone has been shown to be longer in methadone and was associated with a mortality rate of 22.1 per 100 buprenorphine than in naltrexone [21].
person-years during the period of high risk (2 weeks The significance of the difference in mortality rates post-treatment), and one per 100 person-years during was tested using Poisson regression models using SAS the period of low risk (during treatment). Mortality estimates and their 95% confidence intervals areincluded in Table 1.
Naltrexone subjects 2 weeks post-treatment (high- risk period) had 7.4 times the risk of dying than Searches of the NCIS revealed 282 methadone, one methadone subjects in their first week of treatment buprenorphine and 32 oral naltrexone-related deaths (high-risk period). This risk was highly significant during 2000 – 03 in Australia. Of these 258, one and 15 (p 5 0.0001, 95% CI: 4.6 – 11.5). Naltrexone subjects deaths, respectively, met the criteria for ‘known’ during treatment (low-risk period) have 2.8 times the methadone-, buprenorphine- and naltrexone-related risk of dying than methadone subjects in the post- deaths. This includes where the drug in question is induction treatment period (low-risk period). This risk mentioned as a cause of death in the coronial or autopsy approaches significance (p ¼ 0.055, 95% CI: 1.3 – 5.7).
document, an opioid overdose within 2 weeks ofcessation of treatment or an opioid overdose where the person is known to be in current treatment with the drugin question. A more detailed description can be found in Deaths related to methadone, buprenorphine and [32]. According to national records, an estimated 102 naltrexone have occurred in Australia, and clear 615 episodes of methadone and 49 948 episodes of differences in risk were observed. Whether estimated buprenorphine treatment occurred in Australia during as deaths per 1000 treatment episodes or per 100 that time. A total of 6337 private naltrexone prescrip- person-years of high risk, the mortality rates for tions were filled during 2000 – 03, each providing naltrexone treatment were significantly higher than medication for 1 month. Assuming mean treatment those for methadone treatment (p 5 0.0001). Our retention of 2 months, the number of oral naltrexone methadone mortality rate of 3.0 deaths per 100 treatment episodes during this time was 3169.
patient-years is comparable to the 7.4 deaths per 100 The crude estimated mortality rate was 2.7 deaths patient-years obtained by other researchers [16] in the per 1000 episodes for methadone, 0.02 per 1000 initial high-risk period of treatment. Our buprenor- treatment episodes for buprenorphine and 10.1 per phine mortality rate of 0.02 per 1000 episodes can be 1000 treatment episodes for naltrexone. Naltrexone compared to the French estimate of 0.24 per 1000 subjects had 3.7 times the mortality risk compared to patients [25] and 2.5 per 1000 patients in the NEPOD methadone subjects, a highly significant difference studies [29]. As only a single death in buprenorphine (p 5 0.0001). With 95% confidence, the true relative patients was noted in both this and the NEPOD risk falls between 2.5 and 5.2 times higher than studies, caution should be used when comparing rates and significant differences in buprenorphine mortality Table 1. Mortality rates per 1000 treatment episodes and per 100 person-years of exposure (stratified into periods of high and low risk of to methadone and naltrexone treatment were not that naltrexone implants carry a mortality risk. Nal- trexone implant deaths may be more difficult to identify The naltrexone-related mortality rate during treat- than oral naltrexone deaths due to poor reporting of the ment (one per 100 person-years) is very similar to the presence of an implant at autopsy [45]. Future work overdose mortality rate for non-treated opioid depen- needs to examine this issue carefully.
dent subjects of 0.9 per 100 person-years [6], suggest-ing that naltrexone offers little benefit in terms of mortality risk. In the high-risk period after treatmentcessation, naltrexone-related mortality increased to 22.1 per 100 person-years, a rate clearly elevated [28,29,34,36,46] have been accompanied by concerns compared both to other pharmacotherapies and also about the inability to monitor overdose deaths after to active, dependent heroin use [6]. The NEPOD naltrexone treatment cessation [37]. Both inadequa- studies reported a naltrexone mortality rate of 4.8 cies in the data and assumptions in calculating the deaths per 100 person-years after treatment cessation mortality rates have the potential to bias, in particular, [29]. The lower mortality rate in these studies could be our estimates of naltrexone mortality. First, identifying partially a reflection of standardised treatment proto- naltrexone-related deaths is difficult. Not only do cols and intensive monitoring present in the research naltrexone-related deaths rarely have naltrexone de- rather than the general clinical setting. Mortality rates tected in post-mortem toxicology, they rely on a past following naltrexone treatment can also be compared to episode of naltrexone treatment being recorded in mortality in other reportedly opioid-abstinent situa- coronial databases, something not conducted system- tions, such as shortly after release from prison. The rate atically in Australia. We were limited to keyword of five deaths per 1000 prisoner releases in the 2 weeks searching of the coronial database and so relied on following release in a Scottish study [33] is lower than correct spelling of the search terms in the files. For our crude estimate of 10.1 deaths per 1000 treatment this reason we believe we may have underestimated the number of naltrexone-related deaths using the The mortality rates we have found are plausible given both the pharmacology of these drugs and previous research [16,29,30,34]. Naltrexone is a treatment that prescriptions of naltrexone were for opioid dependence reduces tolerance to opioids, and reduces opioid effects treatment. If we assume that half of opioid-dependent during treatment. Buprenorphine and methadone, in patients were also alcohol-dependent [47] and so contrast, provide tolerance to all other opioids during eligible for a public prescription, the number of treatment. It is not surprising, then, that there is a treatment episodes would be increased by 50%, higher potential for more deaths to occur post- reducing the naltrexone mortality estimate by a factor of two. Thirdly, the mean length of a naltrexone Retention in naltrexone treatment is poor: approxi- treatment episode was estimated by a number of mately one-third of subjects remain in naltrexone clinical experts. Longer mean retention in treatment treatment after 3 months [12,35]. The mortality risk would elevate the naltrexone-related mortality.
associated with naltrexone treatment is of particular Our selection of the first week of methadone concern, considering the dangerous combination of treatment as the period of high risk was influenced by poor naltrexone compliance and sporadic heroin use the limitations of the data and previous research [17].
This does not necessarily imply that the high-risk It should be noted that naltrexone treatment may be period does not extend to the first 2 weeks of treatment, a useful option in some well-motivated patient sub- as discussed by other authors [16,48]. No additional groups, such as medical professionals [27,38], with deaths were noted in the first 2 weeks of methadone strong imperatives to remain abstinent. However, these treatment compared to the first week. This may be a subgroups represent the minority of dependent opioid reflection of a certain bias in the NCIS: the further users, and successful abstinence attempts are not the separated a death was from the commencement of norm, even in the well-motivated subjects [39].
treatment, the less likely the timing of treatment was Implant technologies have been proposed as alter- noted as salient in the coronial records.
native methods for delivering naltrexone [40 – 44].
These are not registered for use in Australia, and due pharmacotherapies include deaths occurring among to a lack of data on the number of naltrexone implant people using diverted medication. In the case of recipients this study was unable to make estimates of methadone, 53 deaths (19%) occurred among people mortality rates related to naltrexone implants. How- using diverted medication and 37% had an unknown ever, three naltrexone implant-related deaths were treatment status. As the level of diversion is likely to be identified in the NCIS over the same period, suggesting different between methadone and naltrexone, this Mortality related to pharmacotherapies for opioid dependence inflates the mortality rate associated with methadone [8] Davids E, Gastpar M. Buprenorphine in the treatment of opioid dependence. Eur Neuropsychopharmacol 2004;14:209 – 16.
[9] Fiellin DA, Kleber H, Trumple-Hejduk JG, et al. Con- sensus statement on office-based treatment of opioiddependence using buprenorphine. J Subst Abuse Treat Deaths related to both antagonist and agonist pharma- cotherapy for opioid dependence can and do occur.
[10] Mattick RP, Kimber J, Breen C, et al. Buprenorphine maintenance versus placebo or methadone maintenance for Naltrexone treatment shows a higher mortality risk in opioid dependence. Cochrane Database Syst Rev 2003; comparison to both agonist pharmacotherapies and active-dependent heroin use. This is especially con- [11] Renault PF. Treatment of heroin-dependent persons with cerning, considering that the majority of opioid- antagonists: current status. Report No. 28: NIDA Research dependent individuals will return to opioid use soon Monograph no. 28. Maryland: National Institute on DrugAbuse, 1981.
after leaving naltrexone treatment. This high mortality [12] Tucker TK, Ritter AJ, Maher C, et al. Naltrexone rate should be emphasised to patients and considered maintenance for heroin dependence: uptake, attrition and by medical practitioners when determining a patient’s retention. Drug Alcohol Rev 2004;23:299 – 309.
suitability for naltrexone treatment. It is recommended [13] Minozzi S, Amato L, Vecchi S, et al. Oral naltrexone that future trials of opioid dependence treatments maintenance treatment for opioid dependence. CochraneDatabase Syst Rev 2006;Issue 1.
include monitoring of post-treatment mortality risk [14] Frischer M, ed. Estimating the prevalence of drug abuse for up to 12 months. Better systems to identify naltrexone-related death would be helpful, especially Luxembourg: Office for Official Publications of the Eur- those that capture deaths occurring soon after treat- [15] Frischer M, Hickman M, Kraus L, et al. A comparison of different methods for estimating the prevalence of proble-matic drug misuse in Great Britain. Addiction 2001;96: [16] Caplehorn JRM, Drummer OH. Mortality associated with Both authors are funded by the Australian Government New South Wales methadone programs in 1994: lives lost Department of Health and Ageing and received a small and saved. Med J Aust 1999;170:104 – 9.
[17] Zador D, Sunjic S. Deaths in methadone maintenance additional grant to fund this study. Ethics approval was treatment in New South Wales, Australia 1990 – 1995.
granted by both the University of New South Wales Human Research Ethics Committee and Monash [18] Corkery JM, Schifano F, Ghodse AH, et al. The effects of University National Centre for Coronial Information.
methadone and its role in fatalities. Hum Psychopharmacol Thanks go to Stuart Gilmour, who assisted with data [19] Milroy CM, Forrest ARW. Methadone deaths: a toxiciolo- gical analysis. J Clin Pathol 2000;53:277 – 81.
[20] Walsh SL, Preston KL, Stitzer ML, et al. Clinical pharmacology of buprenorphine: ceiling effects at highdoses. Clin Pharmacol Ther 1994;55:569 – 80.
[1] Oppenheimer E, Tobutt C, Taylor C, et al. Death and [21] Mattick RP, Digiusto E, Doran CM, et al. National survival in a cohort of heroin addicts from London clinics: a evaluation of pharmacotherapies for opioid dependence: 22-year follow-up study. Addiction 1994;89:1299 – 308.
report of results and recommendations. Sydney: National [2] Hser YI, Hoffman V, Grella CE, et al. A 33-year follow- Drug and Alcohol Research Centre, 2004.
up of narcotics addicts. Arch Gen Psychiatry 2001;58: [22] Lintzeris N, Clark N, Muhleisen P, et al. National clinical guidelines and procedures for the use of buprenorphine in [3] Degenhardt L, Hall W, Lynskey M, et al. Illicit drug use. In: the treatment of heroin dependence. Canberra: Department Comparative quantification of health risks: global and [23] Schifano F, Corkery J, Gilvarry E, et al. Buprenorphine regional burden of disease attributable to selected major mortality, seizures and prescription data in the UK, 1980 – risk factors. Chapter 13. Geneva: World Health Organiza- 2002. Hum Psychopharmacol 2005;20:343 – 8.
[24] Auriacombe M, Fatseas M, Dubernet J, et al. French field [4] Goldstein A, Herrera J. Heroin addicts and methadone treatment in Albuquerque: a 22-year follow-up. Drug [25] Auriacombe M, Franques P, Tignol J. Deaths attributable to [5] Ward J, Hall W, Mattick RP. Role of maintenance methadone vs. buprenorphine in France. JAMA 2001;285:3.
treatment in opioid dependence. Lancet 1999;353:221 – 6.
[26] Reisine T, Pasternak G. Opioid analgesics and antagonists.
[6] Caplehorn JRM, Dalton MSYN, Halder F, et al. Metha- In: Hardman JG, Limbird LE, eds. Goodman & Gilman’s done maintenance and addicts’ risk of fatal heroin overdose.
The pharmacological basis of therapeutics, 9th edn. Sydney: Subst Use Misuse 1996;31:177 – 96.
[7] Mattick RP, Breen C, Kimber J, et al. Methadone [27] Bell J, Kimber J, Lintzeris N, et al. Clinical guidelines and maintenance therapy versus no opioid replacement therapy procedures for the use of naltrexone in the management of for opioid dependence. Cochrane Database Syst Rev opioid dependence. Canberra: Australian Government Department of Health and Ageing, 2003.
[28] Hamilton R, Olmedo R, Shah S, et al. Complications of [39] Hulse GK, O’Neil G, Hatton M, et al. Use of oral and ultrarapid opioid detoxification with subcutaneous naltrex- implantable naltrexone in the management of the opioid one pellets. Acad Emerg Med 2002;9:63 – 8.
[29] Digiusto E, Shakeshaft A, Ritter A, et al. Serious adverse events in the Australian National Evaluation of Pharma- [40] Foster J, Brewer C, Steele T. Naltrexone implants can cotherapies for Opioid Dependence (NEPOD). Addiction completely prevent early (1- month) relapse after opiate detoxification: a pilot study of two cohorts totalling 101 [30] Ritter AJ. Naltrexone in the treatment of heroin depen- patients with a note on naltrexone blood levels. Addict Biol dence: relationship with depression and risk of overdose.
Aust NZ J Psychiatry 2002;36:224 – 8.
[41] Brewer C. Naltrexone implants for opiate addiction: new [31] Barker B, Conroy E, Degenhardt L, et al. Global indicators life for a middle-aged drug. Pharm J 2001;267:260.
for monitoring the illicit drug environment of NSW.
[42] Comer SD, Collins ED, Kleber HD, et al. Depot Sydney: National Drug and Alcohol Research Centre, naltrexone: long-lasting antagonism of the effects of heroin University of New South Wales, 2004.
in humans. Psychopharmacology 2002;159:351 – 60.
[32] Gibson A, Degenhardt L. Mortality related to naltrexone in [43] Carreno JE, Alvarez CE, San Narciso GI, et al. Maintenance the treatment of opioid dependence: a comparative analysis.
treatment with depot opioid antagonists in subcutaneous Sydney: National Drug and Alcohol Research Centre, implants: an alternative in the treatment of opioid depen- dence. Addict Biol 2003;8:429 – 38.
[33] Bird SM, Hutchinson SJ. Male drugs-related deaths in the [44] Hulse GK, Tait RJ. A pilot study to assess the impact of fortnight after release from prison: Scotland, 1996 – 99.
naltrexone implant on accidental opiate overdose in ‘high- risk’ adolescent heroin users. Addict Biol 2003;8:337 – 42.
[34] Miotto K, McCann MJ, Rawson RA, et al. Overdose, [45] Oliver P. Fatal opiate overdose following regimen changes suicide attempts and death among a cohort of naltrexone- in naltrexone treatment. Addiction 2005;100:560 – 3.
treated opioid addicts. Drug Alcohol Depend 1997;45: [46] Arnold-Reed DE, Hulse GK, Hansson RC, et al. Blood morphine levels in naltrexone-exposed compared to non- [35] Foy A, Sadler C, Taylor A. An open trial of naltrexone for naltrexone-exposed fatal heroin overdoses. Addict Biol opiate dependence. Drug Alcohol Rev 1998;17:167 – 74.
[36] Bell JR, Young MR, Masterman SC, et al. A pilot study of [47] Gossop M, Marsden J, Stewart D. Dual dependence: naltrexone-accelerated detoxification in opioid dependence.
assessment of dependence upon alcohol and illicit drugs, and the relationship of alcohol dependence among drug [37] Hall W, Wodak A. Is naltrexone a cure for heroin misusers to patterns of drinking, illicit drug use and health dependence? The evidence so far is not promising. Med J problems. Addiction 2002;97:169 – 78.
[48] Buster MC, van Brussel GH, van den Brink W. An increase [38] Ling W, Wesson DR. Naltrexone treatment for addicted in overdose mortality during the first 2 weeks after entering health-care professionals: a collaborative private practice or re-entering methadone treatment in Amsterdam. Addic- experience. J Clin Psychiatry 1984;45:46 – 8.

Source: http://www.iceers.org/docs/science/iboga/Gibson%20et%20al_2007_Mortality_related_pharmacotherapies.pdf