Presented at the Society for Academic Emergency Medicine Annual Meeting, Chicago, IL, May 2007; Winner of Best Resident Presentation Award; Canadian Association of Emergency Physicians Annual Scientific Assembly, Victoria, BC, June 2007.
Subdissociative-dose Ketamine versus Fentanyl for Analgesia during Propofol Procedural Sedation: A Randomized Clinical Trial
Article first published online: 27 AUG 2008
© 2008 by the Society for Academic Emergency Medicine
Academic Emergency Medicine
Volume 15, Issue 10, pages 877–886, October 2008
How to Cite
Messenger, D. W., Murray, H. E., Dungey, P. E., Van Vlymen, J. and Sivilotti, M. L.A. (2008), Subdissociative-dose Ketamine versus Fentanyl for Analgesia during Propofol Procedural Sedation: A Randomized Clinical Trial. Academic Emergency Medicine, 15: 877–886. doi: 10.1111/j.1553-2712.2008.00219.x
This study was supported by the physicians of Ontario through a resident research grant from the Physicians’ Services Incorporated Foundation (Grant R04-43).
ClinicalTrials.gov Identifier: NCT00137085.
A related commentary appears on page 955.
- Issue published online: 1 OCT 2008
- Article first published online: 27 AUG 2008
- Received February 23, 2008; revision received June 9, 2008; accepted June 24, 2008.
- conscious sedation;
- hypnotics and sedatives;
- emergency medicine
Objectives: The authors sought to compare the safety and efficacy of subdissociative-dose ketamine versus fentanyl as adjunct analgesics for emergency department (ED) procedural sedation and analgesia (PSA) with propofol.
Methods: This double-blind, randomized trial enrolled American Society of Anesthesiology (ASA) Class I or II ED patients, aged 14–65 years, requiring PSA for orthopedic reduction or abscess drainage. Subjects received 0.3 mg/kg ketamine or 1.5 μg/kg fentanyl intravenously (IV), followed by IV propofol titrated to deep sedation. Supplemental oxygen was not routinely administered. The primary outcomes were the frequency and severity of cardiorespiratory events and interventions, rated using a composite intrasedation event rating scale. Secondary outcomes included the frequency of specific scale component events, propofol doses required to achieve and maintain sedation, times to sedation and recovery, and physician and patient satisfaction.
Results: Sixty-three patients were enrolled. Of patients who received fentanyl, 26/31 (83.9%) had an intrasedation event versus 15/32 (46.9%) of those who received ketamine. Events prospectively rated as moderate or severe were seen in 16/31 (51.6%) of fentanyl subjects versus 7/32 (21.9%) of ketamine subjects. Patients receiving fentanyl had 5.1 (95% confidence interval [CI] = 1.9 to 13.6; p < 0.001) times the odds of having a more serious intrasedation event rating than patients receiving ketamine. There were no significant differences in secondary outcomes, apart from higher propofol doses in the ketamine arm.
Conclusions: Subdissociative-dose ketamine is safer than fentanyl for ED PSA with propofol and appears to have similar efficacy.
Propofol has gained popularity for emergency department (ED) procedural sedation and analgesia (PSA) because of its rapid induction of deep sedation and brief duration of action.1–4 However, propofol produces dose-dependent respiratory depression and may cause transient hypotension in some patients.5,6 A potent amnestic, propofol lacks intrinsic analgesic properties. As a result, a rapidly acting opioid such as fentanyl (at doses between 1 and 2 μg/kg) is often coadministered for pain control during PSA.3,7–11 Unfortunately, fentanyl also possesses respiratory-depressant effects that may be additive with those of propofol when these agents are used in combination.12
At doses between 0.1 and 0.5 mg/kg, well below those used to induce its unique state of dissociative sedation,13 ketamine has well-documented analgesic effects14,15 and has been shown to be an effective adjunct during propofol sedation.16–19 The safety and efficacy of this combination have not previously been compared to fentanyl-propofol for PSA in the ED. We hypothesized that a subdissociative analgesic dose of ketamine might be safer than fentanyl for propofol PSA. We therefore sought to compare the frequency and severity of intrasedation cardiorespiratory events and interventions in ED patients undergoing propofol PSA for orthopedic reductions or minor surgical procedures, using either low-dose ketamine or fentanyl as the adjunct analgesic agent. We also planned to compare secondary outcomes of adequacy of sedation and analgesia, frequency of individual hemodynamic and respiratory events and interventions, and recovery times postsedation.
This was a double-blind, prospective, randomized controlled trial conducted in the ED of a 460-bed, university-affiliated tertiary care hospital between December 2004 and February 2006. This study was approved by our institutional research ethics board, and informed consent was provided by all participants.
Study Setting and Population
We enrolled ED patients presenting with a fracture or dislocation requiring reduction, or with an abscess requiring incision and drainage, and for whom administration of PSA was deemed appropriate by the treating physician. We attempted to enroll consecutive eligible patients. Potential subjects were identified by on-site research nurses, emergency physicians (EPs), or ED nurses. One of seven study physicians, either attending EPs or senior emergency medicine (EM) residents, agreed to be available on-call to enroll eligible patients and to administer PSA. All EPs and nurses were encouraged to contact a study physician whenever PSA was contemplated, and a brief summary of inclusion and exclusion criteria was posted near the procedure rooms. The opportunity to summon a second physician, as required by institutional protocol for all ED procedural sedations, was believed to serve as an incentive for subject identification.
Patients were excluded from the study for: age < 14 or > 65 years; American Society of Anesthesiology (ASA) Class III or greater; history of significant active cardiac, pulmonary, hepatic, or renal disease; weight > 130 kg; history of physician-diagnosed obstructive sleep apnea; chronic use of opioids; history of recent substance abuse or prior opioid dependence; acute intoxication with drugs or alcohol; history of psychotic disorder; or history of allergy or sensitivity to any study medication.
Patients were monitored using a Medtronic LifePak 12 monitor/defibrillator unit (Medtronic ERS, Redmond, WA) with continuous oral–nasal end-tidal carbon dioxide (ETCO2) sampling (Nellcor Smart Capnoline O2/CO2 oral–nasal cannula, Tyko Healthcare, Pleasanton, CA). By protocol, supplemental oxygen was administered to patients only if they developed oxygen desaturation. The oxygen limb of the nasal cannula was connected to a closed oxygen source prior to PSA initiation, such that nasal oxygen was immediately available if desired. Patients received 1:1 nursing care throughout their procedure and recovery. At least two physicians were present at the bedside during the procedure: one responsible for the procedure (the “operating physician”) and one responsible solely for administration of the sedation and analgesia (the “sedating physician”). Subjects received analgesic agents at presentation and prior to their sedation at the discretion of the treating physician; however, a minimum 30-minute washout period prior to initiation of PSA was prespecified for all patients who had received opioids.
Patients were randomized to receive a single intravenous (IV) bolus of either 0.3 mg/kg ketamine or 1.5 μg/kg fentanyl at Time 0. After 2 minutes, all patients received 0.4 mg/kg propofol IV, followed by additional 0.1 mg/kg boluses every 30 seconds until adequate sedation was achieved. Patients were deemed to be adequately sedated at a score of 1 on the Modified Observer’s Assessment of Alertness/Sedation (OAA/S) scale20 (i.e., respond only to painful stimulus). During the procedure, additional 0.1 mg/kg boluses of propofol were administered at 30-second intervals as required to maintain the desired depth of sedation. Indications and thresholds for specific airway, respiratory, and cardiovascular interventions were prespecified; however, sedating physicians were permitted to initiate any supportive or resuscitative measure at their discretion.
Following completion of the procedure, patients continued to be monitored until deemed recovered based on a score of 9/10 on a standardized, objective scoring tool (the Modified Post-Anesthesia Discharge Scoring System–Modified PADSS).21 The nurse monitoring patient recovery recorded each component of the score at 5-minute intervals during this recovery phase.
Subjects were randomized in consecutive, computer-generated, randomly permuted blocks of six. Consecutively numbered, prefilled 5-mL syringes of study medications prepared by the hospital pharmacy contained either 10 mg/mL ketamine or 50 μg/mL fentanyl diluted in saline, such that a volume of 0.03 mL/kg study medication represented the study dose of either ketamine (0.3 mg/kg) or fentanyl (1.5 μg/kg). Both ketamine and fentanyl are clear colorless solutions, indistinguishable based on appearance. Patients, physicians, nurses, and data entry personnel were blinded to the contents of the syringes and the randomization schedule. The statistician performing data analysis and the data monitoring committee were also blinded to the drug assignment of each arm of the study.
Relevant data were prospectively collected using a standardized data collection form. Heart rate, blood pressure, pulse-oxygen saturation, and ETCO2 were measured at baseline and recorded electronically at 3-minute intervals until return of appropriate verbal response. During the procedure, any respiratory or hemodynamic events encountered, as well as any airway, breathing, or cardiovascular interventions undertaken were recorded onto the data collection form, and captured with a time stamp and simultaneous vital sign documentation on the monitor printout.
The cumulative dose of propofol required to achieve adequate sedation and the total dose administered during the procedure were recorded and corroborated against nursing records. Times of study drug administration, achievement of adequate sedation, procedure start and finish, recovery of appropriate verbal response by the patient, and readiness for discharge were recorded. Following the procedure, the adequacy of blinding was assessed by asking sedating physicians to guess which study medication the patient had received and to rate the certainty of their guess using a 10-point scale.
The following were also rated, using 10-point scales: the patient-reported pain at baseline and during procedure (10 = worst pain imaginable), recall of procedure (10 = full recall), overall satisfaction with sedation (10 = completely satisfied), the sedating physician’s impression of adequacy of sedation and of analgesia (10 = excellent), and the operating physician’s impression of adequacy of the sedation. Finally, sedating physicians, who were blinded to the primary outcome scale, were asked to rate the overall severity of adverse events that occurred during each sedation, using a 10 point scale (10 = very severe).
Data were entered into a secure, password-protected computer database (Microsoft Access 2000, Microsoft Corp., Redmond, WA) by data entry personnel not otherwise involved in the study and independently verified against the original data forms.
The primary outcome measure was the frequency of cardiorespiratory clinical events and interventions, graded by relative severity or invasiveness. These events and interventions were categorized a priori into a four-level ordinal scale (the intrasedation event rating scale) designed specifically for this study (Table 1).22 This rating scale was developed by an expert committee of four EPs and an anesthesiologist. This committee generated a list of clinical events and interventions potentially encountered during PSA and stratified them by their perceived clinical severity or invasiveness into one of four mutually exclusive levels (none, mild, moderate, or severe). The overall intrasedation event rating for each subject was assigned based on the highest level event or intervention encountered.
|SaO2 < 92% at any time|
|Administration of supplemental O2 by nasal cannula|
|sBP < 100 mmHg (if baseline ≥ 110 mm Hg)|
|Rise in ETCO2 > 10 mm Hg above baseline|
|SaO2 < 80% at any time|
|SaO2 < 90% for ≥ 1 minute despite supplemental oxygen|
|Administration of supplemental O2 by nonrebreather mask|
|Jaw thrust or chin lift required|
|Loss of ETCO2 waveform for ≥ 30 seconds or recurrent loss|
|sBP < 90 mm Hg|
|Cardiac dysrhythmia* with sBP > 100 mm Hg|
|SaO2 < 70% at any time|
|SaO2 < 85% for ≥ 1 minute despite supplemental oxygen|
|Assisted ventilations provided with bag valve mask|
|Artificial airway required|
|Vomiting prior to recovery of verbal response|
|Cardiac dysrhythmia* with sBP < 100 mm Hg|
|Vasoactive agent administered|
Secondary outcome measures were the frequency of prespecified intrasedation events and interventions; the dose of propofol required to achieve sedation; the cumulative dose of propofol administered to maintain sedation; the times from study drug administration to each of achievement of adequate sedation, end of procedure, recovery of appropriate verbal response, and discharge readiness; the sedating physician’s, operating physician’s, and patient’s ratings of sedation and analgesia adequacy and satisfaction; the frequency of any emergence phenomena (as observed and recorded by the sedating physician); and the sedating physician’s overall rating of the severity of intrasedation adverse events encountered.
At the time of protocol development, few published ED studies were available to estimate adverse event frequency with propofol PSA. Using a local administrative review, and data from a randomized trial of propofol sedation in ED patients,1 we estimated the relative frequency of each intrasedation event rating level in the fentanyl group as follows: 35% mild, 15% moderate, and 5% severe. We determined a priori that a sample size of 62 subjects per arm would have 90% power to detect a threefold reduction in these odds using an unadjusted proportional odds model at a two-sided alpha of 0.05.23
An interim data safety analysis, limited to comparison of moderate or severe events between study arms, was planned at 50% of target enrollment. The results of this interim analysis were reviewed by an independent data monitoring committee consisting of a statistician and two EPs. The interim analysis used the method of Lan and DeMets,24 with an O’Brien Fleming–type boundary for a first interim analysis at 50% data collection. This method, with a specific early termination condition at p = 0.003, adjusts the nominal alpha-value for the final analysis to 0.049, yet maintains an overall Type I error rate of 0.05 and has a negligible effect on the study’s power.
The prespecified primary outcome compared intrasedation event rating using an unadjusted proportional odds logistic regression model. The Cochran-Armitage test for trend was to be used instead if the proportional odds assumption was violated. The primary analysis was confirmed by an adjusted proportional odds model that included selected baseline covariates. Secondary efficacy outcomes were compared between groups by the Wilcoxon Mann-Whitney test. All tests are two-sided and were performed using SAS statistical software (Version 9.1, SAS Institute Inc., Cary, NC).
The trial enrolled 63 subjects (32 in the ketamine group and 31 in the fentanyl group) between December 2004 and April 2006. Following the planned interim safety analysis after 61 patients had been enrolled, the data monitoring committee directed that subject enrollment be terminated and that an early analysis of the complete study data be undertaken. Two additional patients had been enrolled into the study during the interim review and prior to the termination of study recruitment.
Figure 1 demonstrates subject enrollment and progress throughout the trial. Review of an electronic hospital administrative database suggested that, during the period of enrollment into this study, 438 patients received procedural sedation in our ED, of whom 197 were potentially eligible based on age and procedure type. Table 2 shows randomized subjects’ baseline characteristics at enrollment. There was moderate gender imbalance between groups, with more men in the ketamine group. There were no differences between groups with respect to the mean dose or proportion of patients who received parenteral opioids within 2 hours of enrollment.
|Ketamine (n = 32)||Fentanyl (n = 31)|
|Age, years (mean ± SD)||35.6 ± 17.0||43.2 ± 17.4|
|Female||12 (37.5)||20 (64.5)|
|Male||20 (62.5)||11 (35.5)|
|Weight, kg (mean ± sd)||74.5 ± 15.9||77.3 ± 14.5|
|Procedure type (%)|
|Orthopedic procedures||31 (96.9)||27 (87.1)|
|Upper extremity||18 (56.3)||15 (48.4)|
|Lower extremity||12 (37.5)||9 (29.0)|
|Other||1 (3.1)||3 (9.6)|
|Abscess I&D||1 (3.1)||4 (12.9)|
|Mean preprocedure pain score (mean ± SD)||4.6 ± 2.4||5.9 ± 2.5|
|IV opioid analgesia in 2 hours preceding PSA (%)||12 (37.5)||14 (45.2)|
|Total dose, mg of morphine (median [IQR])||6.2 [3.6, 10.0]||7.5 [5.0, 10.0]|
|Baseline vital signs (mean ± sd)|
|Heart rate (beats/min)||88.1 ± 14.8||82.9 ± 16.6|
|sBP (mm Hg)||141.9 ± 17.7||137.6 ± 19.4|
|dBP (mm Hg)||83.6 ± 10.7||84.4 ± 13.6|
|Mean arterial pressure (mm Hg)||105.1 ± 11.5||102.7 ± 14.9|
|Oxygen saturation (%)||98.4 ± 1.5||97.6 ± 2.7|
|ETCO2 (mm Hg)||39.4 ± 5.8||38.7 ± 4.7|
The frequency of intrasedation events between study groups is displayed in Table 3. Of patients who received ketamine, 15/32 (46.9%) had at least one event or intervention, versus 26/31 (83.9%) of those who received fentanyl. All five severe events were seen in subjects who received fentanyl. The unadjusted proportional odds model estimated that fentanyl subjects had 5.1 (95% confidence interval [CI] = 1.9 to 13.6; p < 0.001) times the odds of having a higher intrasedation event rating than ketamine subjects. The score test provided no suggestion that the proportional odds assumption was violated (p = 0.4). The treatment effect was similar after adjustment for gender, age, weight, preprocedure pain, and procedure type (odds ratio [OR] = 4.6; 95% CI = 1.4 to 15.3; p = 0.012). Patients who received presedation opioids displayed a trend toward developing a higher intrasedation event rating, but this effect was not statistically significant (OR = 1.9; 95% CI = 0.76 to 5.0), suggesting that the washout period was adequate. There was no evidence of an interaction between preenrollment opioid administration and study arm assignment on the primary outcome, and the main treatment effect was unchanged after controlling for this factor.
|Ketamine (n = 32)||Fentanyl (n = 31)||OR at threshold|
|None||17 (53.1%)||5 (16.1%)|
|Mild||8 (25.0%)||10 (32.3%)|
|Moderate||7 (21.9%)||11 (35.5%)|
|Severe||0 (0.0%)||5 (16.1%)|
Each observed component of the intrasedation event rating was encountered more frequently in the fentanyl group (Table 4). Oxygen desaturation events were frequent, especially within the first 10 minutes following study drug administration (Figure 2). Oxygen saturation fell below 92% in 12 ketamine subjects versus 24 fentanyl subjects (absolute risk reduction 39.9%; 95% CI = 17.6% to 62.2%). Oxygen desaturation events were, in fact, the principal contributor to the observed differences in intrasedation event rating between groups. Subjects randomized to fentanyl had a substantially higher risk of oxygen desaturation, with an estimated “number needed to harm” of 2.8 (95% CI = 1.9 to 5.7) for oxygen desaturation below 80%.
|Intrasedation Event Rating Scale Components||Ketamine (n = 32)||Fentanyl (n = 31)||Difference, % (95% CI)|
|SaO2 < 92% at any time||12 (37.5%)||24 (77.4%)||−39.9 (−62.2, −17.6)|
|Administration of supplemental O2 by nasal cannula||8 (25.0)||20 (64.5)||−39.5 (−62.1, −17.0)|
|sBP < 100 mmHg (if baseline ≥ 110 mm Hg)||0 (0.0%)||3 (9.7%)||−9.7 (−20.1, 0.7)|
|Rise in ETCO2 > 10 mm Hg above baseline||3 (9.4%)||4 (12.9%)||−3.5 (−19.1, 12.0)|
|SaO2 < 80% at any time||1 (3.1%)||12 (38.7%)||−35.6 (−53.8, −17.4)|
|SaO2 < 90% for ≥ 1 minute despite supplemental oxygen||2 (6.3%)||8 (25.8%)||−19.5 (−2.0, −37.1)|
|Administration of supplemental O2 by nonrebreather mask||0 (0.0%)||4 (12.9%)||−12.9 (−24.7, −1.1)|
|Jaw thrust or chin lift required||0 (0.0%)||3 (9.7%)||−9.7 (−20.1, 0.7)|
|Loss of ETCO2 waveform for ≥ 30 seconds or recurrent loss||5 (15.6%)||7 (22.6%)||−7.0 (−26.3, 12.4)|
|sBP < 90 mm Hg||0 (0.0%)||0 (0.0%)||0|
|Cardiac dysrhythmia* with sBP > 100 mm Hg||0 (0.0%)||0 (0.0%)||0|
|SaO2 < 70% at any time||0 (0.0%)||4 (12.9%)||−12.9 (−24.7, −1.1)|
|SaO2 < 85% for ≥ 1 minute despite supplemental oxygen||0 (0.0%)||3 (9.7%)||−9.7 (−20.1, 0.7)|
|Assisted ventilations provided with bag valve mask||0 (0.0%)||1 (3.2%)||−3.2 (−9.5, 3.0)|
|Artificial airway required||0 (0.0%)||0 (0.0%)||0|
|Vomiting prior to recovery of verbal response||0 (0.0%)||0 (0.0%)||0|
|Cardiac dysrhythmia* with sBP < 100 mm Hg||0 (0.0%)||0 (0.0%)||0|
|Vasoactive agent administered||0 (0.0%)||0 (0.0%)||0|
|Naloxone administered||0 (0.0%)||0 (0.0%)||0|
Three of 31 subjects in the fentanyl arm (9.7%) required an airway repositioning maneuver, and 1 subject briefly received assisted ventilation with a bag valve mask. All intrasedation events were transient, and no subject required prolonged observation nor hospitalization as a result of an adverse event related to PSA. Subsequent hospital record review identified zero unscheduled returns to the ED for respiratory complaints in the 7 days following PSA.
Additional secondary outcome measures are compared in Table 5. The cumulative propofol dose to maintain sedation was higher in the ketamine group (mean difference 0.4 mg/kg; 95% CI = 0.1 to 0.7 mg/kg). Indicators of sedation efficacy, such as time to hypnosis, procedure duration, recovery duration, physician satisfaction ratings, and patient pain, recall, and satisfaction ratings were similar between groups. Sedating physicians’ ratings of the overall severity of intrasedation events and interventions were significantly lower in ketamine patients. There were no emergence phenomena reported.
|Ketamine (n = 32)||Fentanyl (n = 31)||Difference (95% CI)|
|Propofol dose required to achieve adequate sedation, mg/kg (mean ± SD)||1.5 ± 0.9||1.1 ± 0.6||0.4 (0.0, 0.7)|
|Supplemental propofol administered after sedation achieved, mg/kg (mean ± SD)||0.74 ± 0.64||0.36 ± 0.42||0.38 (0.46, 0.66)|
|Time from study drug administration to adequate sedation, minutes (median [IQR])||6.7 [5.9, 8.4]||5.4 [3.8, 8.9]|
|Length of procedure, minutes (median [IQR])||7.9 [4.1, 10.8]||5.9 [2.7, 10.0]|
|Recovery time,* minutes (median [IQR])||28.0 [10.0, 52.5]||37.0 [19.5, 40.3]|
|Sedating physician’s opinion (1–10 scale) of the adequacy of sedation (mean ± SD)||7.2 ± 2.2||7.6 ± 1.9||−0.4 (−1.4, 0.6)|
|Sedating physician’s opinion (1–10 scale) of the adequacy of analgesia (mean ± SD)||6.6 ± 2.3||7.3 ± 2.2||−0.7 (−1.8, 2.3)|
|Operating physician’s opinion (1–10 scale) of the adequacy of sedation (mean ± SD)||7.4 ± 2.2||8.0 ± 2.0||−0.6 (−1.7, 0.4)|
|Patient’s recall (1–10 scale) of procedure (mean ± SD)||3.2 ± 3.0||4.1 ± 3.7||−0.9 (−2.6, 3.4)|
|Patient’s pain remembered (1–10 scale) during procedure (mean ± SD)||2.1 ± 2.2||2.3 ± 2.0||−0.3 (−1.3, 0.8)|
|Patient’s overall satisfaction (1–10 scale) with sedation (mean ± SD)||9.4 ± 1 .4||9.4 ± 1.4||0 (−0.7, 1.4)|
|Sedating physician’s rating (1–10 scale) of overall adverse event severity (mean ± SD)||1.6 ± 1.0||3.2 ± 1.8||−1.6 (−2.3, −0.9)|
There were two protocol violations. An 81-year-old patient was enrolled and randomized to fentanyl despite exceeding the upper age limit and experienced a moderate adverse event (transient arterial oxygen saturation [SaO2] < 80%). A second patient, randomized to ketamine, was withdrawn 19 minutes after study drug administration, and three failed attempts at distal radius fracture reduction and adjunct medications outside of the study protocol were administered prior to successful completion of the procedure. No intrasedation events occurred during this patient’s sedation. Incomplete electronic vital sign data were recorded for one patient (randomized to fentanyl) due to monitor malfunction; however, an episode of oxygen desaturation below 92% was documented on the ED sedation record and included in the analysis. All patients enrolled were included in the final analysis according to the intention-to-treat principle.
Sedating physicians were able to guess the correct treatment arm in 78% of cases, which is significantly better than expected by chance (p < 0.001). A series of unplanned analyses were used to explore this observation. Physicians were more likely to guess fentanyl when the intrasedation event rating was moderate or severe. Both logistic regression and linear regression (incorporating certainty of guess as the dependent variable) indicated that physician guess was related to the actual treatment even after adjusting for intrasedation event rating (p < 0.001). The event rating was not significantly related to the prediction after adjusting for treatment (p > 0.2). Certainty of prediction (0–10) was not significantly higher in physicians who guessed the study drug correctly versus incorrectly (mean ± standard deviation [SD] 7.3 ± 2.0 vs. 6.4 ± 1.9; p = 0.18 by Wilcoxon Mann-Whitney test).
To the best of our knowledge, this randomized clinical trial is the first to investigate subdissociative-dose ketamine as an analgesic adjunct for ED PSA with propofol.25,26 When compared to 1.5 μg/kg fentanyl, 0.3 mg/kg ketamine administered IV 2 minutes before titrated propofol caused substantially fewer adverse events, particularly oxygen desaturation. This difference is noteworthy in light of the comparable efficacy in both groups, and the higher mean doses of propofol administered to ketamine subjects. Recovery times and physician and patient satisfaction ratings were similar between groups.
Ketamine–propofol combinations, attractive because of the opposing hemodynamic and respiratory effects of these two agents, and ketamine’s documented analgesic properties at subdissociative doses have previously been used outside of the ED for a variety of procedures.11,17–19,27–30 A prospective cohort study of a same-syringe mixture of propofol and ketamine (so-called “ketofol”), titrated to effect in 114 ED patients, found this combination to be safe and effective when used for PSA.16 However, without a direct comparison group, this study does not provide sufficient evidence to adopt ketofol over propofol–opioid combinations or propofol alone.26,31 Furthermore, simultaneous same-syringe titration of two drugs whose duration of action differ by an order of magnitude is difficult to rationalize.31 Our study assessed the safety of a fixed, subdissociative dose of ketamine as an adjunct to titrated propofol. We have demonstrated a large, clinically relevant benefit to this novel combination when directly compared to fentanyl–propofol for ED PSA.
We observed a high frequency of transient hypoxemia during sedation in this trial (57% of patients overall), compared with other studies of propofol sedation that report rates between 5.5 and 31.7%.1–3,32 Oxygen desaturation was particularly common in the fentanyl arm with three of four patients developing hypoxemia, emphasizing the ability of opioids to compound the respiratory-depressant effects of propofol, and the potential hazard of this widely used combination.7,10 The fentanyl dose administered in this trial (1.5 μg/kg), which has been used in other PSA trials of short-acting sedatives,3,33 would appear to cause unacceptably frequent hypoxemia and as such, should be avoided both in future trials and in clinical practice.
Our decision to withhold routine oxygen unless the oxygen saturation fell below 92% likely contributed substantially to the frequency and rapidity of oxygen desaturation. The merits of routine supplemental oxygen administration during PSA of selected low-risk patients are debatable.34 Published trials have not had a standardized approach to its use.1,33 It can be argued that oxygen desaturation in patients breathing room air is an early and readily detected marker of respiratory depression, helping sedating physicians recognize an otherwise subtle event. Moreover, room air desaturation typically responds rapidly to administration of oxygen, patient stimulation, or procedure initiation and interruption of propofol administration. We monitored our subjects closely and measured ETCO2 throughout each procedure. In fact, an independent research hypothesis embedded within this study was to compare continuous capnography to pulse oximetry for the identification of early respiratory depression. These observations are reported elsewhere.35
We did not observe any emergence phenomena, which is not surprising given our low ketamine dose and small sample size. In a recent observational trial of ED PSA using a median dose of 0.75 mg/kg ketamine coadministered with propofol, 3 of 114 subjects manifested emergence reactions, of which 1 required medical intervention.16 Emergence phenomena following 0.3 mg/kg ketamine have been reported to be uncommon in an ambulatory surgery setting.36
Administering an adjunct analgesic during propofol PSA is not uniform practice. In fact, a recent clinical practice advisory recommends that propofol be administered as a solo agent following achievement of adequate analgesia with an opioid.37 In patients with fractures, dislocations, or abscesses, however, the greatest pain stimulus occurs at the time of reduction or drainage. As such, patients’ presedation pain may not reflect the adequacy of analgesia during the procedure. Although sedated patients may not clearly recall procedural pain, painful stimuli can sensitize the nervous system of clinically unresponsive patients and may lead to increased postoperative pain and hyperalgesia.38 The use of IV analgesics, including opioids and ketamine, has been advocated to prevent this phenomenon.14,39 We believe that it is important to distinguish between analgesia and sedation and to treat anticipated intraprocedure pain adequately during PSA. Other studies of ED PSA with propofol3,8 and etomidate33 (another short-acting sedative without analgesic properties) support both the concept and the timing of fentanyl administration just before the sedating agent.
Because this trial did not include a placebo arm that received propofol as a solo agent, we are unable to comment on any safety benefit of ketamine–propofol over propofol alone. The results of our trial do demonstrate, however, that the use of fentanyl–propofol for ED PSA may be inadvisable. Although opioids reduce the propofol dose necessary to prevent purposeful movement in response to surgical stimulus,40–42 this “propofol-sparing” effect must be balanced against the propensity of opioid–propofol combinations to cause respiratory depression and apnea.41–44
Our propofol dose and titration technique differ from those recommended in a recent clinical practice advisory for propofol PSA that suggests an initial dose of 1.0 mg/kg followed by 0.5 mg/kg at 3-minute intervals as required.37 Propofol kinetics are complex and only partly characterized in the first minutes following bolus dosing.42,43,45–47 It is recognized that slower infusion of the same bolus dose results in lower peak serum concentrations and reduced target organ effects.48 Moreover, a recent prospective observational study demonstrated a very low rate of adverse events when propofol was administered for PSA using smaller and more frequent doses than suggested by the Advisory.7 We therefore consider it unlikely that our propofol dosing protocol contributed to our high rate of intrasedation events and interventions. Our mean hypnotic doses and cumulative total doses of propofol are comparable to those administered in studies on which the Advisory is based.49,50 We are not aware of any randomized studies directly comparing different propofol dosing strategies for ED PSA.
Finally, none of our population had any short- or long-term morbidity from the intrasedation events observed during our study. Although serious adverse outcomes with ED PSA are exceedingly rare, patient safety during sedation remains the paramount concern. EPs should rigorously evaluate new PSA protocols and drug combinations to target optimal sedation and analgesia while minimizing patient risk.
Certain limitations of this trial warrant discussion. Sixty-three subjects were enrolled, representing only about one-third of patients who may have been eligible based on age and procedure type. Reduced subject identification outside of research and study personnel hours and need to call in a study physician within a clinically acceptable time frame likely accounted for this underenrollment. Because initial patient screening was often performed by nonresearch personnel, the exact numbers of patients who refused enrollment, were excluded and were missed are not known. While the enrolled population included an appropriate range of ages and procedure types, we cannot exclude some unmeasured enrollment bias.
By design, this trial compared a single, fixed dose of either ketamine or fentanyl administered prior to titrated propofol. These results do not necessarily apply to other doses, timing, or routes of administration, nor to other opioids. Although, to the best of our knowledge, this is the first randomized trial of low-dose ketamine in the ED, the dose of 0.3 mg/kg is comparable to other studies outside of the ED setting and is well within the documented analgesic dose range for this drug (0.1–0.5 mg/kg).17,26,29 The comparable physician ratings of analgesia adequacy between groups further support that this dose of ketamine provides appropriate analgesia. Further studies with different dosing schedules are warranted.
Our primary outcome measure, a composite intrasedation event rating, is an untested tool for reporting adverse events during ED PSA. There have been a variety of outcome measures proposed for PSA research, all of which have limitations.4 Many trials, both in the ED setting and elsewhere, have used adequacy of sedation, rather than safety, as the primary outcome measure.1,17,18,29,51,52 Adverse events in those trials are reported as the frequency of individual events in each group. It is often not possible to evaluate how many patients in those trials experienced more than one adverse event or to obtain a summary measure of the safety profile of the agents studied. One group has defined and used “subclinical respiratory depression”1,4,51,53 as a research outcome, but acknowledge that this outcome is not intended to be a measure of clinically relevant adverse events.4 Our composite intrasedation event rating scale not only includes subclinical markers of respiratory depression, but also captures and ranks clinical events and interventions directly relevant to the safety of PSA. Derived by an expert panel, this scale possesses substantial face validity. Its use to measure safety is also supported by the strong correlation with physicians’ perception of the severity of individual adverse events encountered and with the overall frequency of adverse events and clinical interventions observed between the two groups.22 Further research should independently validate and refine our intrasedation event rating scale. For example, as capnometry becomes more widely studied and used for detection of respiratory depression during ED PSA, criteria for abnormal ETCO2 values may change.32,35,53
Finally, there is evidence that blinding of sedating physicians was compromised. We had anticipated that the nystagmus that results from ketamine administration might weaken blinding, but deemed it potentially unsafe to prevent sedating physicians from observing subjects’ ocular responses during sedation. Any potential bias introduced was mitigated by using objective, quantitative outcome measures,12 including independent verification of the data collection forms against electronically recorded monitor data.
In patients breathing room air, ketamine (0.3 mg/kg) is safer than fentanyl (1.5 μg/kg) when used as an adjunct analgesic to ED procedural sedation with titrated propofol. When compared to fentanyl–propofol, ketamine–propofol causes substantially fewer intrasedation events at all levels of severity, particularly oxygen desaturation, while having an apparently similar efficacy and recovery profile. Coadministration of fentanyl should be avoided when propofol is used for procedural sedation.
The authors thank Mr. Andrew Day and Ms. Xuran Jiang for performing statistical analysis for this study. We are also grateful for the assistance of the Kingston General Hospital Clinical Research Centre, particularly Dr. Robert Brison and Ms. Catherine Isaacs. We thank Dr. Gordon Jones for serving on the Data Monitoring Committee with Dr. Brison and Mr. Day. We also acknowledge the assistance of Drs. Paul Tourigny, Damon Dagnone, and Bruce Cload as sedating physicians and thank Dr. John Ross for guidance with protocol development.
- 21Assessment of “home readiness”: discharge criteria and postdischarge complications. Curr Opin Anesthesiol. 1997; 10:445–50., .
- 22Development of a novel adverse events scale for clinical trials of ED procedural sedation [abstract]. Acad Emerg Med. 2007; 14:S59a., , , , .
- 25Short-Acting Agents for Procedural Sedation and Analgesia in Canadian Emergency Departments: A Review of Clinical Outcomes and Economic Evaluation [Technology Report Number 109]. Ottawa: Canadian Agency for Drugs and Technologies in Health, 2008., , , et al.
- 35Which alarms first during procedural sedation: the pulse oximeter or the capnograph? [abstract]. Can J Emerg Med. 2007; 9:186., , , , .