`
`Original Contribution
`
`www.elsevier.com/locate/ajem
`
`Intranasal naloxone delivery is an alternative to
`intravenous naloxone for opioid overdoses☆
`Mark A. Merlin DO, EMT-P a,b,⁎, Matthew Saybolt BS c, Raffi Kapitanyan MD d,
`Scott M. Alter BS, EMT-P c, Janos Jeges MD d, Junfeng Liu PhD d,e, Susan Calabrese MICP b,
`Kevin O. Rynn PharmD b,f, Rachael Perritt PharmD b,f, Peter W. Pryor II MD, MPH d
`
`aDepartment of Emergency Medicine and Pediatrics, University of Medicine and Dentistry of New Jersey-Robert Wood
`Johnson Medical School, New Brunswick, NJ, USA
`bRobert Wood Johnson University Hospital, New Brunswick, NJ, USA
`cUniversity of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School Piscataway, NJ, USA
`dDepartment of Emergency Medicine, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical
`School, New Brunswick, NJ, USA
`eDepartment of Biostatistics, School of Public Health, University of Medicine and Dentistry of New Jersey, Piscataway, NJ, USA
`fDepartment of Pharmacy Practice, Rutgers University, School of Pharmacy, Piscataway, NJ, USA
`
`Received 17 August 2008; revised 25 October 2008; accepted 4 December 2008
`
`Abstract
`Introduction: This study proposes that
`intranasal (IN) naloxone administration is preferable to
`intravenous (IV) naloxone by emergency medical services for opioid overdoses. Our study attempts to
`establish that IN naloxone is as effective as IV naloxone but without the risk of needle exposure. We
`also attempt to validate the use of the Glasgow Coma Scale (GCS) in opioid intoxication.
`Methods: A retrospective chart review of prehospital advanced life support patients was performed on
`confirmed opioid overdose patients. Initial and final unassisted respiratory rates (RR) and GCS,
`recorded by paramedics, were used as indicators of naloxone effectiveness. The median changes in RR
`and GCS were determined.
`Results: Three hundred forty-four patients who received naloxone by paramedics from January 1, 2005,
`until December 31, 2007, were evaluated. Of confirmed opioid overdoses, change in RR was 6 for the
`IV group and 4 for the IN group (P = .08). Change in GCS was 4 for the IV group and 3 for the IN
`group (P = .19). Correlations between RR and GCS for initial, final, and change were significant at the
`0.01 level (ρ = 0.577, 0.462, 0.568, respectively).
`Conclusion: Intranasal naloxone is statistically as effective as IV naloxone at reversing the effects of
`opioid overdose. The IV and IN groups had similar average increases in RR and GCS. Based on our
`
`☆ This study received no grants or financial support. It has not been presented at any meeting.
`⁎ Corresponding author. Department of Emergency Medicine, UMDNJ-Robert Wood Johnson Medical School, New Brunswick, NJ 08901, USA. Tel.:+1
`732 235 8717.
`E-mail address: merlinma@umdnj.edu (M.A. Merlin).
`
`0735-6757/$ – see front matter © 2010 Elsevier Inc. All rights reserved.
`doi:10.1016/j.ajem.2008.12.009
`
`Nalox1040
`Nalox-1 Pharmaceuticals, LLC
`Page 1 of 8
`
`
`
`Intranasal naloxone delivery for opioid overdose
`
`297
`
`results, IN naloxone is a viable alternative to IV naloxone while posing less risk of needle stick injury.
`Additionally, we demonstrated that GCS is correlated with RR in opioid intoxication.
`© 2010 Elsevier Inc. All rights reserved.
`
`1. Introduction
`
`1.1. Background
`
`In 1991, the Occupational Health and Safety Administra-
`tion mandated the implementation of alternative drug
`delivery systems to minimize needle stick injuries and
`decrease the exposure of blood borne pathogens to
`emergency health workers [1]. The risk of exposure to
`blood borne pathogens is especially high in the emergency
`medical services (EMS) environment. The annual blood
`contact for individual EMS providers is estimated to be as
`high as 12.3 exposures per year in populations with more than
`90% of the HIV statuses unknown [2]. In high-risk
`populations, such as intravenous (IV) drug abusers, alter-
`native practices are vital in maintaining the safety of the EMS
`personnel while providing adequate care to the patients.
`Intranasal (IN) medication delivery is a safe and direct
`means to provide medication to patients without using
`needles. Some advantages compared with parenteral include
`avoidance of painful injection, avoidance of risks associated
`with IV access, rapid onset, and high levels of patient
`acceptability [3].
`Human studies elucidate naloxone pharmacokinetics [4-6].
`The onset of IV naloxone is 1 to 2 minutes; it has a clinical
`duration of 20 to 90 minutes that varies with dosage and
`administration route [7]. Intranasal administration of nalox-
`one bypasses hepatic first-pass metabolism because absorp-
`tion is direct via nasal mucosa, due to richly supplied
`vasculature and low barrier to drug permeation [8]. Intranasal
`drug delivery also has the potential to target brain delivery,
`bypassing the blood brain barrier [9].
`Pharmacokinetic data for IN administration in humans is
`lacking. Currently, data in rats describe 100% bioavailability
`for IN naloxone, with similar elimination half-life to IV
`naloxone [10]. In this animal study, peak plasma concentra-
`tions for
`IN naloxone occurred within 3 minutes of
`administration. This evidence corroborates supporting this
`route of administration. Clinical outcome data also support
`the use of IN naloxone in reversing opioid effects in both the
`overdose setting and for opioid dependency [11-13].
`Multiple articles suggest
`IN naloxone has a strong
`evidence base as a first-line therapy for people with
`suspected opioid overdose in the prehospital setting. The
`2006 Best Evidence Topic Report [14], published in the
`Emergency Medicine Journal, summarizes the findings in
`these articles published since 1992. The review concludes IN
`naloxone has minimal adverse side effects and is a safe route
`of administration.
`
`From 2002 to 2005, several case series were published on
`IN naloxone [15,16]. Limitations of these studies included
`small patient number, variable exclusion/inclusion criteria,
`differing route and timing of naloxone, and inconsistent
`methods of response measurement quantified. A 2005 article
`concluded that IN naloxone was a good first-line therapy for
`patients suspected of opioid overdose, with findings of rapid
`reversal of overdose in most patients and a limited risk of
`needle stick exposures [11]. Two additional studies [12,17]
`—a randomized control trial and a retrospective case review
`—both conclude that IN naloxone was effective, but time to
`onset was prolonged from IV and intramuscular naloxone.
`
`1.2. Purpose
`
`The intent of this study was to investigate whether IN
`naloxone was noninferior compared to IV naloxone in
`increasing respiratory rates (RRs) and mental status in
`patients presenting with suspected opioid overdose in the
`prehospital setting. Our primary outcome measures were
`changes in Glasgow Coma Scale (GCS) and unassisted RRs
`after administration of IN and IV naloxone. We also attempt
`to demonstrate that GCS is correlated with RR in opioid
`overdose.
`
`1.3. Hypothesis
`
`We hypothesize that in patients presenting with opioid
`overdoses, IN naloxone will be noninferior to IV naloxone in
`increasing RR and GCS.
`
`2. Methods
`
`2.1. Design
`
`The study is a retrospective cohort conducted by chart
`review.
`
`2.2. Setting
`
`The study was conducted at a university-based level I
`trauma center in an urban setting. The EMS system contain 6
`advanced life support (ALS) units that perform 6920 ALS
`treats per year within a context of approximately 30 000
`dispatches per year (including basic life support units). Only
`ALS may administer naloxone in our study's site. All ALS
`personnel
`received training in IN and IV naloxone
`Nalox1040
`Nalox-1 Pharmaceuticals, LLC
`Page 2 of 8
`
`
`
`298
`
`administration during paramedic class, and this procedure is
`frequently performed throughout our state.
`
`2.3. Selection of participants
`
`Testing the hypothesis requires determination of opioid
`intoxication. Criteria created to ensure acute opioid overdose
`includes documentation of one of the following: patient
`admission of
`illegal or nontherapeutic opioid use to
`paramedics or emergency department
`(ED) physician,
`witness testimony to paramedics or ED physician, evidence
`of opioid use observed by paramedics (eg, heroin, prescrip-
`tion narcotics, or used paraphernalia found on person), or
`positive urine toxicologic screen for opioids.
`Participant exclusion criteria included patients in cardiac
`arrest, intubation before naloxone administration, sedation
`by paramedics before naloxone administration, or patients
`with end point data missing from patient care reports (PCRs).
`
`2.4. Interventions
`
`M.A. Merlin et al.
`
`a Microsoft Excel spreadsheet. In addition, the investigators
`recorded patients' RR and GCS values documented
`immediately before and after administration of a single
`dose of naloxone. After enrolling qualified participants,
`patient records were cross-referenced with ED records in
`Emergency Department Information Management to obtain
`additional confirmation of opioid intoxication by ED
`physician progress notes or participant urine toxicologic
`screens. The admitting or discharge diagnosis was also
`obtained when available. Among the participants with
`confirmed opioid overdoses, PCRs and physician progress
`notes were reevaluated to determine any coingestion in
`addition to opioids. All data were entered into a standardized
`abstraction form. End points were reconfirmed 3 times for
`each patient by reinspection of PCRs by the investigators.
`The investigators met bimonthly to discuss progress and
`review discrepancies.
`
`2.7. Outcome measures
`
`From a database of ALS responses, patients who received
`naloxone between January 1, 2005, and December 31, 2007,
`were selected as participants. As per state standing orders,
`patients with altered mental status received IV naloxone at an
`initial dose between 0.4 and 2.0 mg or IN naloxone at 1 mg
`per nostril at the discretion of the paramedics.
`
`Glasgow Coma Scale and RR values recorded on the PCR
`immediately before administration of the first dose of
`naloxone determined “initial measurement;” values recorded
`immediately following the first administration of naloxone
`defined “final measurement.” Naloxone redosing was
`defined as subsequent doses naloxone. The accepted scoring
`system was used to determine composite GCS values.
`
`2.5. Methods of measurement
`
`2.8. Data analysis
`
`Paramedics recorded data on standard ALS PCRs while
`treating their patients. Vital signs, including RR and GCS,
`were assessed and recorded upon initial evaluation and after
`any treatment or intervention. Any illegible handwritten
`values were confirmed with the paramedic who wrote
`the PCR.
`
`2.6. Data collection and processing
`
`The study was approved by our institutional review board.
`All data were collected by an investigator
`trained in
`Microsoft Access and the Emergency Department Informa-
`tion Management database. The investigators who collected
`data were 2 medical students. The investigators had to both
`agree independently if records were clear that the patient
`received IV as well as proper determination of opioid abuse.
`After students documented these findings,
`the principal
`investigator reviewed all material. From an Access database
`of EMS responses, a query was performed to list all patients
`who were administered naloxone between January 1, 2005,
`and December 31, 2007. Referring to the original PCRs, the
`investigators recorded date, destination hospital, route of
`naloxone delivery, dosage, time to reassessment, participant
`age and sex, and positive narrative identification of acute
`opioid intoxication. Data were extracted from the PCRs onto
`
`Our hypothesis tests the noninferiority of IN. A power
`calculation for RR improvement was calculated. We
`assumed the type I error rate to be less than 5%. From the
`confirmed group (IN, n = 38), RR mean change is 4.37 and
`Standard deviation (SD) is 4.58. The approximate power for
`detecting such a mean (μ) SD (σ) ratio (μ/σ = 0.95) is 100%
`with sample size, n = 38. We also find that n = 38 (IN,
`confirmed opioid) can detect a μ/σ ratio as small as 0.55 with
`95% power and type I error rate 5% or less.
`A power calculation for GCS improvement was com-
`pleted. The IN group GCS mean change is 4.29 and SD is
`4.61. The approximate power for detecting such a mean (μ)
`SD (σ) ratio (μ/σ = 0.93) is 100% with sample size n = 38.
`A sample size calculation for RR and/or GCS improve-
`ment was calculated. We use η to stand for the probability
`that sum of 2 independently and identically distributed
`random variables from a continuous symmetric distribution
`is greater than zero (η = 0.5 represents median = 0). For our
`retrospective study, the hypothesized comparison between η
`= 0.5 and η = 0.80 is reasonable, and sample size n = 38
`suffices for this specific test.
`A power calculation for RR improvement comparison (Δ)
`between IN and IV was completed. We tested H0: Δ = 0 vs
`Ha: Δ does not equal 0, where Δ represents the median shift
`between 2 improvement size distributions (IN and IV). We
`Nalox1040
`Nalox-1 Pharmaceuticals, LLC
`Page 3 of 8
`
`
`
`Intranasal naloxone delivery for opioid overdose
`
`assume the type I error rate to be less than 5%, from
`confirmed group (IN, n = 38; IV, n = 55). The RR change SD
`(σ) is around 4.6 for IN group; the approximate power for
`detecting location shift (Δ = 2) with ratio (Δ/σ = 0.44) to be
`66% with sample size n = 38, m = 55. We also found that n =
`38 and n = 55 (IN and IV, confirmed opioid) can detect a Δ/σ
`ratio as small as 0.70 with power 95% and type I error rate
`5% or less.
`A power calculation for GCS improvement comparison
`between IN and IV was completed. The GCS change SD (σ)
`is 4.6 for IN group; the approximate power for detecting
`location shift (Δ = 1) with ratio (Δ/σ = 0.22) is 27% with
`sample size n = 38, n = 55. The smaller GCS change
`difference is more difficult to detect compared with RR
`change difference, which are approximately 2.
`The confirmed opioid overdose group is subdivided based
`on IV or IN administration. Subjects who received intramus-
`cular naloxone were excluded because of their limited
`number and irrelevance to study's purpose. Distributions of
`initial, final and change in RR, and GCS score were examined
`with graphical methods as well as by Shapiro-Wilk's W
`statistic for normality test. The nonnormal distribution of RR
`and GCS values necessitated nonparametric methods in the
`analysis [3,18,19]. Within IV and IN confirmed opioid
`overdose groups, the Wilcoxon signed rank test was used to
`compare initial and final values of RR and GCS. Associated
`with Wilcoxon signed rank test, medians are estimated by
`Hodges-Lehmann estimator [19] along with Tukey distribu-
`tion-free confidence interval (CI). Between the IV and IN-
`confirmed opioid overdose groups, the Wilcoxon rank sum
`test was used to compare initial, final, and average change in
`
`299
`
`RR and GCS. Associated with Wilcoxon rank sum test,
`median differences are estimated by Hodges-Lehmann
`estimator along with Moses' distribution-free CI [19].
`Spearman's rank correlation coefficient was used to measure
`the association between RR and GCS, initial and change in
`RR, and initial and change in GCS. Proportions were
`compared by the Pearson's χ2 test. All tests were 2-sided.
`Statistical analysis was carried out using SAS 9.1 TS level
`1M0, XP_PRO platform (SAS Institute Inc, Cary, NC) and
`Minitab 15 (Minitab Inc, State College, PA).
`
`3. Results
`
`3.1. Characteristics of study subjects
`
`From a database of advanced life support emergency
`medical calls, 344 patients received naloxone. These patients
`were assessed for eligibility for enrollment in the study.
`Patients excluded from the study were 23 in cardiac arrest, 8
`intubated before naloxone administration, and 3 sedated
`before naloxone administration. An additional 33 patients
`were excluded due to missing data on PCRs. Of these 33
`patients, 11 (3 IN, 8 IV) were confirmed acute opioid
`intoxications. The data points missing from the 11 PCRs
`were as follows: GCS (7 patients), RR (5), and route of
`administration (1). Two hundred seventy-seven patients
`remained for enrollment in the study.
`Participants were divided into 8 groups based on evidence
`of opioid overdose (confirmed, unknown) and route of
`
`Fig. 1
`
`Flow chart of study design.
`
`Nalox1040
`Nalox-1 Pharmaceuticals, LLC
`Page 4 of 8
`
`
`
`300
`
`Table 1 Baseline characteristics of subjects by confirmation of opioid overdose
`
`Opioid overdose, median (interquartile range)
`
`Confirmed (n = 96)
`
`Unknown (n = 181)
`
`Difference estimation
`(95% CI a)
`−12 (−17 to −6)
`51 (37.5-74.5)
`40 (29-50.8)
`Age, y
`8.8 (−3.2 to 20.9)
`99 (54.7)
`61 (63.5)
`Male sex, n (%)
`−6 (−8 to −6)
`16 (14-20)
`10 (6-16)
`Initial RR, per min
`−2 (−3 to 0)
`9 (4-13)
`3.5 (3-11)
`Initial GCS score
`0 (0 to 0)
`2 (1-2)
`2 (2-2)
`Naloxone dose, mg
`0 (−1 to 1)
`4 (2-7)
`5 (2-8)
`Reassessment time, min
`a The confidence interval for the median is slightly greater than 95%, as there is no assumption of distribution.
`b By Wilcoxon rank sum test unless otherwise noted.
`c By Pearson's χ2 test.
`
`M.A. Merlin et al.
`
`P of
`comparison b
`
`b.0001
`.16 c
`b.0001
`.0002
`.19
`.71
`
`intraoss-
`intramuscular,
`naloxone administration (IV, IN,
`eous) (Fig. 1). Table 1 shows the comparison of baseline
`characteristics between the confirmed (n = 96) and the
`unknown (n = 181) groups. Compared to the unknown
`group, the RR median rate was 10 vs 16 breaths per minute
`and the GCS median score was 3.5 vs 9. Further exploration
`of medical records was required to determine if subjects in
`the unknown opioid overdose group were unconfirmed
`opioid overdoses or if the patients presented with acute
`illnesses secondary to other medical conditions. Of the 181
`subjects in the unknown group, 97 were transported to our
`hospital and 89 diagnoses could be obtained. The 8 subjects
`who could not be accounted for probably left the ED before
`being registered. Of these patients with unconfirmed opioid
`overdoses, the treating physician gave only 3 (3%) patients a
`diagnosis of suspected (unconfirmed) opioid overdose,
`which indicates that most patients in the unknown group
`had a different acute illness. The remaining diagnoses were
`alcohol intoxication (18%), nonopioid drug overdose (18%),
`cerebrovascular accident/transient ischemic attack/intracra-
`nial bleed (15%), altered mental status of unknown etiology
`(10%), respiratory failure/asthma (7%), seizure (7%), sepsis
`
`(6%), trauma (6%), hypoglycemia (3%), dehydration (2%),
`syncope of unknown etiology (2%), anxiety (1%), dementia
`(1%), and hyperglycemia (1%). Considering all of the
`patients transported to our hospital, excluding the 8 in the
`unknown group who did not register (n = 158), 86 patients
`(54%) received naloxone with a medical condition, other
`than opioid intoxication, that potentially accounted for their
`acute presentations.
`Within the confirmed opioid overdose group, character-
`istics of subjects were compared by route of naloxone
`administration (Table 2). The 2 routes of administration were
`similar except for evidence of coingestions and dose of
`naloxone given. Subjects in the IV group had a higher
`percentage of coingestion confirmations than those in the IN
`group (median, 32% vs 13%; P = .02; 95% CI for proportion
`difference is 4% to 44%). Although the median naloxone
`dose for both groups was 2 mg, subjects receiving IN
`naloxone received a higher dose than those receiving
`naloxone intravenously (mean, 1.95 vs 1.71 mg; P = .01).
`This is because of EMS protocols, where IV naloxone may
`be titrated to effect from 0.4 to 2 mg, and IN naloxone is
`usually given as 2 mg, 1 mg in each nostril.
`
`Table 2 Baseline characteristics of subjects with confirmed opioid overdoses by route of naloxone administration
`
`Route of administration,
`median (interquartile range)
`
`Difference estimation
`(95% CI a)
`
`P of
`comparison b
`
`IV (n = 55)
`
`IN (n = 38)
`
`3 (−4 to 9)
`38 (27-54)
`42 (31-47)
`Age, y
`6.8 (−13.1 to 26.6)
`23 (60.5)
`37 (67.3)
`Male sex, n (%)
`0 (−2 to 4)
`10 (4-14.5)
`10 (6-16)
`Initial RR, per min
`0 (0 to 1)
`3 (3-9.25)
`4 (3-11)
`Initial GCS score
`0 (n/a)
`2 (2-2)
`2 (1-2)
`Naloxone dose, mg
`0 (−2 to 1)
`5 (2.8-7.3)
`4 (2-8)
`Reassessment time, min
`24.0 (4.0 to 43.9)
`13 (34.2)
`32 (58.2)
`Coingestion evidence, n (%)
`−2.0 (−11.9 to 7.9)
`36 (94.7)
`51 (92.7)
`Narrative evidence of opioid overdose, n (%)
`6.6 (−13.0 to 26.2)
`12 (31.6)
`21 (38.2)
`Toxicologic screen evidence of opioid overdose, n (%)
`a The confidence interval for the median is slightly greater than 95%, as there is no assumption of distribution.
`b By Wilcoxon rank sum test unless otherwise noted.
`c By Pearson's χ2 test.
`
`.44
`.50 c
`.60
`.37
`.02
`.66
`.02 c
`.70 c
`.51 c
`
`Nalox1040
`Nalox-1 Pharmaceuticals, LLC
`Page 5 of 8
`
`
`
`Intranasal naloxone delivery for opioid overdose
`
`301
`
`3.2. Main results
`
`Within the IV and IN-confirmed opioid overdose groups,
`the paired Wilcoxon signed rank test was used to compare
`initial and final median values of RR and GCS. Naloxone
`was successful in elevating the RR and GCS in each of the 4
`comparisons, as the final values were significantly higher
`than the initial values (Table 3). For the IV group, RR
`increased from 10 to 18 (P b .0001), and the GCS score
`increased from 4 to 15 (P b .0001). Similarly, for the IN
`group, RR increased from 10 to 16 (P b .0001), and the GCS
`score increased from 3 to 12 (P b .0001).
`Naloxone redosing, defined as at least one additional
`naloxone dose, occurred in 11 (20%) of the IV patients and
`16 (42%) of the IN patients. Of the 16 IN patients, 9 received
`the repeat dose IV at the decision of the paramedic. One
`patient in the IV group received 3 doses, and one patient in
`the IN group received 3 doses.
`Neither the median initial RR (IV, 10 vs IN, 10) nor the
`median initial GCS scores (IV, 4 vs IN, 3) were significantly
`different between the 2 route groups. The median final RR
`was higher for the IV group than the IN group (18 vs 16; P =
`.001). The median final GCS score was also higher in the IV
`group than the IN group (15 vs 12; P = .01).
`Statistically significant differences in final RR and GCS, as
`well as differences in the change in RR or GCS between the IV
`and IN groups are noted (Table 3). The median change in RR
`for the IV group was 6 breaths per minute (95% CI, 4-10) and
`for the IN group was 4 (95% CI, 2-6). Hodges-Lehmann
`estimation of the median difference in these changes was 2
`(95% CI, −0.001 to 5). The median change in GCS score was 4
`for the IV group (95% CI, 3-8) and 3 for the IN group (95% CI,
`0-5). Hodges-Lehmann estimation for change median differ-
`ence was 1 (95% CI, −0.001 to 3). This inconsistency between
`final scores and improvement may be because the nonpara-
`metric Wilcoxon test is used to test the location shift between 2
`continuous distributions of identical shapes, and GCS has an
`upper bound of 15, which is an assumption limitation.
`For sample size (n) justification for Wilcoxon signed rank
`test, we denote η = P (Z1 + Z2 N 0) η), that is, the probability
`that sum of 2 independently and identically distributed
`
`random variables from a continuous symmetric distribution
`is greater than zero (η = 0.5 represents median = 0) [20].
`Sample size discussion is based on using probability
`difference as effect size other than distribution locations
`that have been extensively estimated. Because we are not
`interested in testing the null hypothesis vs a specific
`alternative hypothesis in location shift or probability
`difference, sample size determination and power analysis
`are not the focus in this article.
`Correlations examined the effectiveness of naloxone
`depending on the initial RR and GCS score. Because the
`data are not normally distributed and GCS is an ordinal
`categorical variable, Spearman's rank correlation coefficient
`was used. Among confirmed opioid overdoses, the correla-
`tion between initial RR and change in respiratory rate was ρ
`= −0.749. The correlation between initial GCS score and
`change in GCS score was ρ = −0.558. These values,
`significant at the 0.01 level, indicate that the lower the initial
`RR and GCS score, the larger the increase will be in response
`to naloxone.
`When comparing RR and GCS score, the correlations for
`initial, final, and change were ρ = 0.577, 0.462, and 0.568,
`respectively, and were each significant at the 0.01 level. The
`correlation between final values is expected to be lower
`because as the GCS gets higher, it approaches its maximum
`value. In a healthy population, the GCS score is constant and
`therefore not possible to be correlated with another variable.
`In comparison, the unknown opioid overdose group had
`correlations between RR and GCS score of ρ = 0.288, 0.248,
`and 0.246 for initial, final, and change, respectively, each
`significant at the 0.01 level. These lower values indicate that
`in a population of mixed medical conditions, there is only a
`small correlation between RR and GCS score.
`
`3.3. Study limitations
`
`A limitation of our study is that it is a nonrandomized,
`nonblinded, retrospective chart review from EMS PCRs.
`Although our initial intention was to randomize patients,
`certain legal informed consent laws in our state made this
`unobtainable.
`
`Table 3
`
`Comparison of response to naloxone by route of administration
`
`Route of administration, median
`(95% CI a)
`
`IV (n = 55)
`
`IN (n = 38)
`
`Difference estimation
`(95% CI a)
`
`P of
`comparison b
`
`GCS score
`
`RR, per min
`
`0 (−2 to 4)
`10 (6-12)
`10 (6-12)
`Initial
`4 (2 to 6)
`16 (12-16)
`18 (16-18)
`Final
`2 (−0.001 to 5)
`4 (2-6)
`6 (4-10)
`Change
`0 (0 to 1)
`3 (3-6)
`4 (3-9)
`Initial
`1 (0 to 3)
`12 (8-14)
`15 (14-15)
`Final
`1 (−0.001 to 3)
`3 (0-5)
`4 (3-8)
`Change
`a The confidence interval for the median is slightly greater than 95%, as there is no assumption of distribution.
`b By Wilcoxon rank sum test.
`
`.60
`.001
`.08
`.37
`.01
`.19
`
`Nalox1040
`Nalox-1 Pharmaceuticals, LLC
`Page 6 of 8
`
`
`
`302
`
`Another limitation is that the ED records could only be
`obtained for patients who were transported to our hospital.
`Had records from other hospitals been obtained, it is possible
`that more patients in the unknown group could have been
`confirmed by either narrative evidence or urinary drug
`screen. In addition, physician discharge/admitting diagnoses
`would have provided a greater sample to determine other
`reasons for the patients' acute illness. Despite this, the results
`of the study have not been diluted, as only confirmed opioid
`overdoses were analyzed. This limitation prevented us from
`having a larger analyzable sample size.
`Relying on paramedics' subjective decisions of how to
`treat a patient introduces possible biases in choosing dose
`and route of naloxone administration. In addition,
`time
`constraints in the prehospital environment may complicate
`the accuracy of reporting. For example, there is no standard
`time to reassess a patient after administration of naloxone,
`and this study assumes that paramedics recorded patients'
`actual initial GCS and RR exactly at the time of naloxone
`administration. This in fact
`is usually impossible, and
`therefore, there is a degree of inaccuracy in setting the
`beginning of intervention at time zero. However, we believe
`that this represents “real-life” scenarios where paramedics
`decide when a patient is altered and when to reassess.
`The use of RR by prehospital personnel is subjective.
`Experience suggests that EMS professionals do not always
`document RR correctly. Oftentimes RR are just thought of as
`hypoventilating or hyperventilating. Our system requires
`both paramedics to agree on the RR before documentation
`evidenced by both signing the PCR. We believe that much of
`the inherent bias of this poorly documented vital sign is
`eliminated by confirmation of a second ALS provider.
`Previous studies indicate naloxone dose should be 1 mg
`per nostril (total of 2 mg) [9]. It is difficult to determine
`whether our primary outcomes measures would have changed
`if 2 mg per nostril was used and if this change would be dose-
`dependent. Limitations to the use of IN naloxone focus on
`barriers to absorption. Many factors such as nasal mucociliary
`clearance [21], metabolic degradation in the nasal cavity, IN
`use of vasoactive agents, nasal trauma, epistaxis, ambient
`temperature, and mucosal inflammation influence systemic
`absorption of nasally administered drugs [5].
`Despite an existing policy encouraging paramedics to
`attempt IN delivery initially, sicker patients may have
`received IV naloxone rather
`than IN. Our subjective
`experience in dealing with paramedics suggests biases
`toward one method or another based on personal experience
`and not degree of patient intoxication.
`Furthermore, our study used urinary drug screens, when
`available, to confirm opioid abuse. Certain synthetic opioids
`and opioid-like substances, such as tramadol and propox-
`yphene, have a dose-dependent response to naloxone and
`escape detection from our standard urine toxicologic screens
`[21]. Conversely, 6 patients had positive urinary drug screens
`but no narrative evidence on the prehospital patient care
`report or ED progress notes. These patients may have been
`
`M.A. Merlin et al.
`
`on opioids in therapeutic doses but were altered secondarily
`to other causes, such as sepsis or cerebrovascular accident.
`However, most initial vital signs were consistent with central
`nervous system and respiratory depression.
`
`4. Discussion
`
`Among subjects with confirmed opioid overdoses, IN
`naloxone is as effective as IV naloxone at reversing the
`central nervous system-depressing effects caused by opioids.
`Subjects were compared by route of naloxone administration,
`using RR and GCS score as indicators of opioid intoxication.
`In addition to having similar baseline characteristics, both the
`IV and IN groups had significant increases in RR and GCS
`score. Furthermore, the data shows both IV and IN naloxone
`significantly increases both the RR and GCS of patients with
`confirmed opioid intoxications.
`Previous studies have been criticized for using GCS to
`quantify the degree of improvement in opioid-intoxicated
`patients after naloxone administration [9]. Although GCS
`has been questioned in nontrauma patients [22-24]. Opioids
`are central nervous system depressants that, among other
`actions,
`lower patients' RRs. Therefore, we sought
`to
`correlate RR and GCS scores to validate the use of GCS
`for quantification of patient
`improvement. In confirmed
`opioid overdose patients, we found correlations between RR
`and GCS, indicating evidence of a relationship between RR
`and GCS. For comparison, correlations between RR and
`GCS were performed in subjects who received naloxone but
`were not confirmed opioid overdoses (the unknown group).
`In this group of subjects with a heterogeneous mixture of
`medical problems, we showed very weak correlations
`between RR and GCS. The strong correlations between RR
`and GCS in the confirmed group together with the large
`difference in degree of correlations between the confirmed
`opioid abuse group and the unknown group indicate that
`increases in GCS is a sign of opioid overdose reversal. This
`demonstrates the prognostic value of GCS for evaluation of
`opioid overdoses.
`Among confirmed opioid intoxications, there was a strong
`negative correlation between initial RR and change in RR, as
`well as a similar negative correlation between initial GCS
`and change in GCS. This indicates that the lower the initial
`RR and GCS, the larger the increase will be in response to
`naloxone. This leads us to believe that there is a physiologic
`ceiling on both values among the entire study population. We
`conclude that this occurs for 1 of 2 reasons. Portions of the
`study population may have demonstrated a maximal
`response to naloxone or rather a maximal elevation of RR
`and GCS in the context of an unknown quantity of systemic
`opioid. On the contrary, there may be an average physiologic
`ceiling of RR and GCS values among the entire population
`that is unrelated to drug administration but rather is a portrait
`of the population's basal RR.
`
`Nalox1040
`Nalox-1 Pharmaceuticals, LLC
`Page 7 of 8
`
`
`
`Intranasal naloxone delivery for opioid overdose
`
`Naloxone redosing was at the discretion of the paramedics
`under physician order. Twice as many patients in the IN group
`were given second doses, as compared with the IV group. The
`decision to redose is very subjective and may represent the
`inability to wait for the full desired clinical response.
`Alternately, this finding may represent the need for a higher
`naloxone dose when given IN, rather than IV, to achieve
`similar increases in RR and GCS. However, the 1 mg per
`nostril dose may be adequate because the ultimate prehospital
`goal is to avoid hypoventilation-induced hypoxemia. In our
`study, paramedics were using naloxone for decreased GCS
`without hypoventilation, so perhaps the typical IV dose
`should b