Acts Anaesthesiol Scand 199 1: 35: 584-590
A placebo-controlled trial of flumazenil given by continuous infusion in severe benzodiazepine overdosage *J. HOJER,S. BAEHRENDTZ, A. MAGNUSSON' and L. L. GUSTAFSSON~ I>rpartments of Internal hledicinr and 'Clinical Chemistry, Southrrn Hospital. and 'Department of Clinical Pliarinac~ology,Hudclingr Hospital, Karolinska Institute, Stockholm, Sweden
A controlled trial was perforrnrd to study thr rfect and any adverse reactions of continuously inliwcl flumazenil in severe twnzodiaxpine overdosage. The study comprised 51 adults admitted to a n interisivr carr unit with poisoning. Enrolmrnt criteria were unconsciousness on admission and ii clear rrsponsr within 5 min aftrr an i.v. bolus injection of I mg flumazenil. T h r levrl of consciousness was assessed according to a modifird Glasgow coma scale immediatrly before and at defined intrrvals up 10 12 h aftrr injection. A doublr-blind infusion was staried 15 min after the bolus injertiun and administerrd for 5 h. T h e patients werr rantlomly allocatrd to onc. of three groups: a group given Ilumazenil 0.5 mg/h, a group given Iluniazrnil 0.1 mg/h and a placebo group. The groups were roniparabk i n age, sex and toxicological laboratory results. I here wrrr no significant differrncrs between the groups in thr average coma scow on admission or 15 min after the holus injection. In thc flumazenil 0.5 mg/h group the level of consciousness remained unrhangrd between the 15-min recording xnd the subsequent assessmrnts. I n the two other groups the levrl of consciousness drcrrased significantly during the course of the infusion. The infusions werr wrll tolrratrd. I t is cnncludrd that a continuous infiision of 0.5 mg of flumazenil per hour can prevent rrlapsr into coma in patients with scvrre hrnzodiazepinr poisoniiig aroused with a singlr in.jection of the antagonist. I
.
Rireirwd 19 Orlobrr 1990, accepted for publicalion 9 janua,:v 1991
Kcr u1ord.r: Benzodiazcpine; flumazenil; infusion; poisoning
Benzodiazepines (BZDs) are the most commonly used drugs in self-poisoning ( I , 2 ) . The hospital mortdity following acute BZD overdosage is low, but the rate of complications is not negligible (3, 4). Respiratory dcpression and prolonged coma are sometimes seen, particularly in elderly persons (3, 5) and in patients with chronic pulmonary disease (6, 7), and these sy mptoms always involve a risk of aspiration and other respiratory complications. Close obscrvation of these patients, preferably in an intensive care unit, is therefore mandatory. Flumazenil, a newly available BZD antagonist (8), is a valuable tool in the management of these cases of poisoning (7, 9). The antagonist is effective in reversing CNS depression and BZD-induced acute respiratory insufficiency (7, 10, 1 I ) . However, flumazenil has a short duration of action compared with the tranquillizing RZIIs, the plasma half-life being less than 1 h i 12). As a consequcncc, in some patients flumazenil hits to be given in repeated i.v. injections or by continuous infusion, to avoid resedation and return of respiratory insufficiency ( 1 1, 13-1 5). For continuous infusion of flumazenil in adults, the doses recommended b) the manufacturer are 0.1-0.4 mg per hour. However, clin-
ical experience and data concerning flumazenil in infusion arc sparse and no controlled clinical study of the drug administered in this form has hitherto been published. The purpose of the prescnt study was to evaluate the effects and any adverse reactions of continuously infused flumazenil in severe BZD poisoning. An attempt was also made to establish the optimal dose level of the infusion.
PATIENTS AND METHODS Fifty-one patients with a mean age of44 years (rangr 17-83) adniiitrd consecutively to thr blrdical Intensivr Cwr Unit of thr Southrrn Hospital in Stockholm with drug poisuning, werr enrollrd i n a double-blind randomisrd study. l'hr rnrolmrnt criteria were unconsciousness on admission, defined as less than I I points on 3 modilied Glasgow coma scale (MCCS) graded from 4 to 20 (16, 17) ['l'ablr I ) , and a clear rrsponsr ( a n improvenicnt by niorr than J points on the coma scale) within 5 niin after the end of a fractionated i.v. bolus injrctioii of I.0 mg of flumazenil givrn i n 3 min. Exclusion critcria werr known pregnancy and epilepsy. The patients werc randomly allocatrd to one of thrrr groups: n group givrn Ilumxzrnil 0.5 mg/h ( 7 men. I 1 womrn; inran age .l:j years), a group given Ilumazenil 0. I mg/h ( 7 mrn. I I w t ~ m r n ;niran age 43). and a placrbo group ( 4 mcn, I I women; niran agr 45). 'l'hr
FLUMAZENIL IN CONTINUOUS INFUSION patiriitr were enrolled and evaluated by the physician on duty at the itirriisive care unit. Blood and urine samples for toxicological examinations were taken on admission. An electrocardiogram was moniioi-rd continuously throughout the study in all patients. Slit Ir1l.l after admission and immediately before the flumazenil bolus injection, the MGCS score was assessed. After the bolus injertion. qastric lavage was perfornird in two patients in the placebo group m d in three in each of the other two groups. Adverse reactions (specilird and graded as severe, moderate or mild) and the MGCS score. ,IS well as the heart rate, blood pressure and rate of respiration werr rworded 5, 15 and 30 min and I , 2, 3, 4, 5, 7 and 12 h after ~ l i rflumazenil bolus injection. Fifteen minutes after the bolus injerrion. a double-blind continuous i.v. infusion was started, and this M .IS continued for 5 h. T h r infusion solution consisted of flumazenil 5 0 pg/ml (0.5 mg of flumazenil/h), flumazenil 1.0 pg/ml (0.1 mg (11 Ilumazenil/h) or placebo (saline alone). Tht. tandomisation was created by a computer at the flumazenil manulnc.rurer’s Department for Clinical Research in Basel, Switzerland. I’hc infusion bottles were prepared, coded and stored at the Phariiwy Production Unit of thc Southern Hospital ( 18). The stability tll’che flumazenil in the infusion solutions used was checked by detertiiining roncentrations of the drug by a gas chromatographic methctd 19).The codes of the infusion bottles were not broken until the full investigation, including analyses of the results, was completed. WiL oxon’s test for paired or unpaired values was used for statistical calculatiuns, and differences were considered significant at P < 0.05. Thc, study was approved by the local Ethics Committee and conductrtl i n acrordance with the Helsinki Declaration. Analvtir al methods In all patients the serum concentrations of the BZD derivatives were assayid in duplicate by gas chromatography by electron-capturedetetllJt I 20, 21). This method is capable of detecting all tranquilliz-
Tablr I Level o1 ( onsciousness, assessed acrording to a modified Glasgow coma r(.ile (MGCS). Minimum MGCS score=4 (deep coma), maxinium MGCS srore=20 (fully awake). Best uerbal response: I = none 2 =incomprehensible 3 =inappropriate 4 = confused 5 =adequate Orientation in time nnd space: 1 = n o response 2 =total disorientatic!n 3 =only orientation in space 4 =only orientation in time 5 =excellent orientation Best motor reJpunse: I = none 2 =extending 3 =flexing 4 = withdrawing 5 = localising 6 =obeying commands Eye opening: I = none 2 = to pain 3 = to speech 4 = spontaneous
585
ing BZDs on the market in Sweden, except the low-dose derivativc triazolam. (Triazolam is one of the least commonly prescribed BZD compounds in Sweden.) The serum conrentrations of ethanol, methanol and paracetamol were also determined in all patients, using standard procedures. A qualitative screening method for urinary amine was used for detection of phenothiazines (22). This method gives false positive results for antihistamines and for high levels of antidepressants. In patients with suspected intake of other agents, thr serum and/or urine levels of these were also assayed.
RESULTS The study was performed over a period of 18 months, during which a total of 107 unconscious patients received a bolus injection of flumazenil. Fifty-one of these were included in accordance with the enrolment criteria. The toxicological examination confirmed that all patients in the study, except one in each study group, had taken an overdose of BZD (in two patients the serum levels of BZD were mistakenly never measured, and one patient had taken a n overdose of the low-dose derivative triazolam). In nine of the 18 patients in the flumazenil 0.5 mg/h group ethanol was present in the serum (mean 47 mmol/l). Eleven of the 18 patients in the flumazenil 0.1 mg/h group had ethanol in the serum (mean 46 mmol/l), compared with eight of the 15 patients in the placebo group (mean 42 mmol/l). Six, seven and four patients in the three study groups, respectively, had ingested other drugs with or without alcohol in addition to BZD (Table 2a-c).
Level of consciousness In the flumazenil 0.5 mg/h group the MGCS score on admission averaged 6.6. Fifteen minutes after the flumazenil bolus injection and immediately before the blind infusion was started, it had increased to an average of 17.9 (P
A placebo-controlled trial of flumazenil given by continuous infusion in severe benzodiazepine overdosage *J. HOJER,S. BAEHRENDTZ, A. MAGNUSSON' and L. L. GUSTAFSSON~ I>rpartments of Internal hledicinr and 'Clinical Chemistry, Southrrn Hospital. and 'Department of Clinical Pliarinac~ology,Hudclingr Hospital, Karolinska Institute, Stockholm, Sweden
A controlled trial was perforrnrd to study thr rfect and any adverse reactions of continuously inliwcl flumazenil in severe twnzodiaxpine overdosage. The study comprised 51 adults admitted to a n interisivr carr unit with poisoning. Enrolmrnt criteria were unconsciousness on admission and ii clear rrsponsr within 5 min aftrr an i.v. bolus injection of I mg flumazenil. T h r levrl of consciousness was assessed according to a modifird Glasgow coma scale immediatrly before and at defined intrrvals up 10 12 h aftrr injection. A doublr-blind infusion was staried 15 min after the bolus injertiun and administerrd for 5 h. T h e patients werr rantlomly allocatrd to onc. of three groups: a group given Ilumazenil 0.5 mg/h, a group given Iluniazrnil 0.1 mg/h and a placebo group. The groups were roniparabk i n age, sex and toxicological laboratory results. I here wrrr no significant differrncrs between the groups in thr average coma scow on admission or 15 min after the holus injection. In thc flumazenil 0.5 mg/h group the level of consciousness remained unrhangrd between the 15-min recording xnd the subsequent assessmrnts. I n the two other groups the levrl of consciousness drcrrased significantly during the course of the infusion. The infusions werr wrll tolrratrd. I t is cnncludrd that a continuous infiision of 0.5 mg of flumazenil per hour can prevent rrlapsr into coma in patients with scvrre hrnzodiazepinr poisoniiig aroused with a singlr in.jection of the antagonist. I
.
Rireirwd 19 Orlobrr 1990, accepted for publicalion 9 janua,:v 1991
Kcr u1ord.r: Benzodiazcpine; flumazenil; infusion; poisoning
Benzodiazepines (BZDs) are the most commonly used drugs in self-poisoning ( I , 2 ) . The hospital mortdity following acute BZD overdosage is low, but the rate of complications is not negligible (3, 4). Respiratory dcpression and prolonged coma are sometimes seen, particularly in elderly persons (3, 5) and in patients with chronic pulmonary disease (6, 7), and these sy mptoms always involve a risk of aspiration and other respiratory complications. Close obscrvation of these patients, preferably in an intensive care unit, is therefore mandatory. Flumazenil, a newly available BZD antagonist (8), is a valuable tool in the management of these cases of poisoning (7, 9). The antagonist is effective in reversing CNS depression and BZD-induced acute respiratory insufficiency (7, 10, 1 I ) . However, flumazenil has a short duration of action compared with the tranquillizing RZIIs, the plasma half-life being less than 1 h i 12). As a consequcncc, in some patients flumazenil hits to be given in repeated i.v. injections or by continuous infusion, to avoid resedation and return of respiratory insufficiency ( 1 1, 13-1 5). For continuous infusion of flumazenil in adults, the doses recommended b) the manufacturer are 0.1-0.4 mg per hour. However, clin-
ical experience and data concerning flumazenil in infusion arc sparse and no controlled clinical study of the drug administered in this form has hitherto been published. The purpose of the prescnt study was to evaluate the effects and any adverse reactions of continuously infused flumazenil in severe BZD poisoning. An attempt was also made to establish the optimal dose level of the infusion.
PATIENTS AND METHODS Fifty-one patients with a mean age of44 years (rangr 17-83) adniiitrd consecutively to thr blrdical Intensivr Cwr Unit of thr Southrrn Hospital in Stockholm with drug poisuning, werr enrollrd i n a double-blind randomisrd study. l'hr rnrolmrnt criteria were unconsciousness on admission, defined as less than I I points on 3 modilied Glasgow coma scale (MCCS) graded from 4 to 20 (16, 17) ['l'ablr I ) , and a clear rrsponsr ( a n improvenicnt by niorr than J points on the coma scale) within 5 niin after the end of a fractionated i.v. bolus injrctioii of I.0 mg of flumazenil givrn i n 3 min. Exclusion critcria werr known pregnancy and epilepsy. The patients werc randomly allocatrd to one of thrrr groups: n group givrn Ilumxzrnil 0.5 mg/h ( 7 men. I 1 womrn; inran age .l:j years), a group given Ilumazenil 0. I mg/h ( 7 mrn. I I w t ~ m r n ;niran age 43). and a placrbo group ( 4 mcn, I I women; niran agr 45). 'l'hr
FLUMAZENIL IN CONTINUOUS INFUSION patiriitr were enrolled and evaluated by the physician on duty at the itirriisive care unit. Blood and urine samples for toxicological examinations were taken on admission. An electrocardiogram was moniioi-rd continuously throughout the study in all patients. Slit Ir1l.l after admission and immediately before the flumazenil bolus injection, the MGCS score was assessed. After the bolus injertion. qastric lavage was perfornird in two patients in the placebo group m d in three in each of the other two groups. Adverse reactions (specilird and graded as severe, moderate or mild) and the MGCS score. ,IS well as the heart rate, blood pressure and rate of respiration werr rworded 5, 15 and 30 min and I , 2, 3, 4, 5, 7 and 12 h after ~ l i rflumazenil bolus injection. Fifteen minutes after the bolus injerrion. a double-blind continuous i.v. infusion was started, and this M .IS continued for 5 h. T h r infusion solution consisted of flumazenil 5 0 pg/ml (0.5 mg of flumazenil/h), flumazenil 1.0 pg/ml (0.1 mg (11 Ilumazenil/h) or placebo (saline alone). Tht. tandomisation was created by a computer at the flumazenil manulnc.rurer’s Department for Clinical Research in Basel, Switzerland. I’hc infusion bottles were prepared, coded and stored at the Phariiwy Production Unit of thc Southern Hospital ( 18). The stability tll’che flumazenil in the infusion solutions used was checked by detertiiining roncentrations of the drug by a gas chromatographic methctd 19).The codes of the infusion bottles were not broken until the full investigation, including analyses of the results, was completed. WiL oxon’s test for paired or unpaired values was used for statistical calculatiuns, and differences were considered significant at P < 0.05. Thc, study was approved by the local Ethics Committee and conductrtl i n acrordance with the Helsinki Declaration. Analvtir al methods In all patients the serum concentrations of the BZD derivatives were assayid in duplicate by gas chromatography by electron-capturedetetllJt I 20, 21). This method is capable of detecting all tranquilliz-
Tablr I Level o1 ( onsciousness, assessed acrording to a modified Glasgow coma r(.ile (MGCS). Minimum MGCS score=4 (deep coma), maxinium MGCS srore=20 (fully awake). Best uerbal response: I = none 2 =incomprehensible 3 =inappropriate 4 = confused 5 =adequate Orientation in time nnd space: 1 = n o response 2 =total disorientatic!n 3 =only orientation in space 4 =only orientation in time 5 =excellent orientation Best motor reJpunse: I = none 2 =extending 3 =flexing 4 = withdrawing 5 = localising 6 =obeying commands Eye opening: I = none 2 = to pain 3 = to speech 4 = spontaneous
585
ing BZDs on the market in Sweden, except the low-dose derivativc triazolam. (Triazolam is one of the least commonly prescribed BZD compounds in Sweden.) The serum conrentrations of ethanol, methanol and paracetamol were also determined in all patients, using standard procedures. A qualitative screening method for urinary amine was used for detection of phenothiazines (22). This method gives false positive results for antihistamines and for high levels of antidepressants. In patients with suspected intake of other agents, thr serum and/or urine levels of these were also assayed.
RESULTS The study was performed over a period of 18 months, during which a total of 107 unconscious patients received a bolus injection of flumazenil. Fifty-one of these were included in accordance with the enrolment criteria. The toxicological examination confirmed that all patients in the study, except one in each study group, had taken an overdose of BZD (in two patients the serum levels of BZD were mistakenly never measured, and one patient had taken a n overdose of the low-dose derivative triazolam). In nine of the 18 patients in the flumazenil 0.5 mg/h group ethanol was present in the serum (mean 47 mmol/l). Eleven of the 18 patients in the flumazenil 0.1 mg/h group had ethanol in the serum (mean 46 mmol/l), compared with eight of the 15 patients in the placebo group (mean 42 mmol/l). Six, seven and four patients in the three study groups, respectively, had ingested other drugs with or without alcohol in addition to BZD (Table 2a-c).
Level of consciousness In the flumazenil 0.5 mg/h group the MGCS score on admission averaged 6.6. Fifteen minutes after the flumazenil bolus injection and immediately before the blind infusion was started, it had increased to an average of 17.9 (P
