Etomidate, an imidazole-derived ultrashort-acting nonbarbiturate hypnotic, is frequently used to induce anesthesia in critically ill patients because of its favorable hemodynamic profile and rapid onset. Etomidate has the advantage of minimizing induction hypotension which can cause coronary hypoperfusion, dysrhythmia, and cardiac arrest. However, etomidate suppresses adrenocortical function by blocking 11β-hydroxylase. Even doses of etomidate as small as 0.04 mg/kg block the enzyme,1 2 3 4 , 5 6–10 11 , 12 13–17 18 19–24
The potential putative link between etomidate and worsened postoperative outcomes has yet to be studied in a large cohort of high-risk general surgical patients. We therefore evaluated the association between etomidate administration and adverse outcomes in ASA physical status III and IV adults having noncardiac surgery. Specifically, we tested the primary hypotheses that patients given etomidate rather than propofol for anesthetic induction have more 30-day postoperative mortality, more infectious complications, and a higher risk of cardiovascular complications. Our secondary hypothesis was that patients given etomidate rather than propofol at anesthetic induction have a prolonged duration of hospitalization.
METHODS
With IRB approval and informed consent waiver, we queried the Perioperative Health Documentation System (PHDS) Registry at the Cleveland Clinic for adults who had noncardiac surgery between January 6, 2005 and December 31, 2009. During this period artifact values including extreme values of arterial blood pressures, heart rates, electrocardiogram, processed electroencephalogram, and pulse oximeter oxygen saturation (SpO2 ) values caused by electrocautery, movements, and transducer failure for invasive arterial, central venous, and pulmonary artery pressure values were removed from the PHDS by a contracted device company (Aspect Medical, now part of Covidien). The PHDS contains data on all patients who had noncardiac surgery since May 2005 at Cleveland Clinic’s main campus, except for cases performed in the magnetic resonance imaging suite, computed tomography suite, and eye surgery cases, which were performed at separate operation sites where an electronic anesthesia record keeping system was not available. Therefore, we screened >95% of all noncardiac surgical cases for inclusion into our study. The system integrates preoperative variables (demographic and baseline characteristics), intraoperative variables (via our proprietary Anesthesia Record Keeping System), and postoperative outcomes (by linking to Cleveland Clinic billing and other systems).
The study population consisted of ASA physical status III and IV adults having noncardiac surgery under general anesthesia, with or without regional anesthesia, requiring at least 1 night of postoperative hospitalization. Anesthesia was induced with propofol, thiopental, etomidate, or ketamine. Patients given any of these drugs at any other time during anesthesia were excluded from our analysis. Anesthesia was maintained with a volatile anesthetic, supplemented per clinical preference with opioids and/or muscle relaxants.
Propensity Score Matching
Each patient who received etomidate was matched to a maximum of 3 patients who received propofol using propensity score matching.25 Table 1 . We then matched etomidate and propofol patients on the propensity score 1 to 3 (using a greedy distance matching algorithm [SAS macro]).a
) were within 0.2 standard deviations of one another (i.e., within 0.2 × 1.1887 = 0.2377)26 International Classification of Diseases, 9th Revision , Clinical Modification procedure codes (Table 2 ). Each analysis below used this subset of matched patients.
Table 1: Demographics Baseline and Intraoperative Characteristics Before and After the Propensity Score Matching
Table 2: Types of Surgery (AHRQ-CCS Categoriesa) Before and After the Propensity Score Matching
Assessment of balance on the covariables used for the propensity score matching was performed using standardized differences (i.e., difference in means or proportions divided by the pooled standard deviation). Imbalance was defined as a standardized difference >0.1 in absolute value; any imbalanced covariables after the propensity score matching were adjusted for in all analyses.
Primary Outcomes
Our primary outcomes were 30-day mortality, any major in-hospital cardiovascular morbidity, and any major in-hospital infectious morbidity (as defined in Table 3 ). We assessed the heterogeneity of the etomidate effect across the 3 primary outcomes by testing the etomidate-by-outcome interaction in a “distinct-effects” generalized estimating equation (GEE) model which enabled adjustment for the correlation among the 3 outcomes. The heterogeneity test thus compares the odds ratios for etomidate between the individual outcomes of interest. Since heterogeneity was found, we reported and tested the individual odds ratios (1 for each outcome) from the distinct effects GEE model, adjusting for any imbalanced baseline covariables after the propensity score matching. Bonferroni correction for multiple comparisons was used to control the type I error at 0.05 so that P < 0.017 was considered significant (i.e., 0.05/3 = 0.017).
Table 3: Description and Incidence of Individual in Hospital Cardiovascular and Infectious Morbidities for the Propensity Score–Matched Patients
In addition, we conducted sensitivity analyses to assess the robustness of the estimated associations between etomidate and the primary outcomes to an unmeasured binary covariate. We assumed various levels of association between the unmeasured covariate and both etomidate and outcome (see details in Table 1, Appendix ).
Secondary Outcomes
For etomidate patients, we assessed the relationship between amount of etomidate received scaling by weight and the primary outcomes, each using a multivariable logistic regression model to adjust for all potential confounders used for propensity matching. We adjusted for severity of procedure (in terms of risk of outcome) as a continuous covariable calculated for each Clinical Classifications Software category as the incidence of any major outcome (i.e., any of 30-day mortality, major in-hospital cardiovascular morbidity or major in-hospital infectious morbidity, versus none of these).
Etomidate and propofol propensity-matched patients were compared on intraoperative vasopressor use (i.e., dobutamine, dopamine, ephedrine, epinephrine, norepinephrine, phenylephrine, or vasopressin) and duration of hospitalization using multivariable logistic regression or Cox proportional hazard regression, as appropriate. For the duration of hospitalization outcome, we needed to account for patients who died before discharge; otherwise, an early death would be counted as a good thing in the analysis. We therefore used a survival analysis (Cox regression) in which the outcome event was “discharged alive,” and patients who died in the hospital were analyzed as never having the event by assigning them a follow-up time 1 day more than any of the observed discharged alive times.
Finally, we summarized and plotted the within-patient average and minimum of systolic and diastolic intraoperative blood pressures during each intraoperative period (i.e., start of case to induction, induction to intubation, intubation to incision, incision to closing, closing to emergence, and emergence to end of case).
Differences in intraoperative arterial blood pressure between the etomidate and propofol patients were summarized using the standardized difference and tested by student t test for normally distributed continuous measures and Wilcoxon rank sum test for nonnormal continuous measures. Intraoperative hemodynamic monitoring data were acquired from our electronic anesthesia record-keeping system that records data from the anesthesia monitor. Arterial blood pressure in patients with invasive arterial catheters was recorded each minute, and at 1- to 5-minute intervals in other patients. Average, minimum, and maximum of systolic and diastolic blood pressures were computed during each period (i.e., start case to induction, induction to intubation, intubation to incision, incision to closing, closing to emergence, and emergence to end of case) in each patient.
Sample Size and Power
With 7377 propensity score matched patients (approximate ratio of 1.0 etomidate to 2.5 propofol patients) and incidences (propofol group) of 2.3%, 4.9%, and 8.5% for 30-day mortality, cardiac morbidity, and infectious morbidity, respectively, we had 90% power to detect odds ratios (ORs) of 1.8, 1.5, and 1.4 or more for the above 3 primary outcomes at the overall 0.05 significance level (0.017 criterion adjusting for 3 primary outcomes). SAS software version 9.3 (SAS Institute) and R software version 2.12.0 (The R Foundation for Statistical Computing, Vienna, Austria) were used for all statistical analysis.
RESULTS
We considered the electronic records of 103,324 adult patients who had noncardiac surgery between 2005 and 2009. Among the ASA III/IV patients who had surgery under general anesthesia requiring at least 1 night of postoperative hospitalization, there were 2616 (8%) patients who received etomidate only, 28,532 (84%) patients who received propofol only, 1976 (6%) patients who received ketamine and/or thiopental, and 658 (2%) patients who received both etomidate and propofol. Among these, we successfully matched 2144 etomidate only patients (82% of 2616) with 5233 propofol only patients, for a total of 7377 patients who were used for analysis of the effect of etomidate on outcome. The observed median [quartiles] of the amount of etomidate and propofol received was 0.22 [0.19, 0.26] and 1.8 [1.4, 2.3] mg/kg, respectively.
Before the propensity score matching (Table 1 , left panel), patients given etomidate for induction were generally older and sicker (higher Charlson comorbidity score and higher ASA status). They were also more likely to be male, have a lower body mass index, have cardiovascular and/or cerebrovascular disease, have emergent surgery, and to receive general anesthesia supplemented with regional anesthesia. They were also less likely to have cancer or receive steroids intraoperatively although the interpretation of steroid use is difficult because it can be used for different indications (i.e. antiemesis, cerebral edemas, and hormone replacement) (standardized differences [STDs] >0.1 in absolute value). All potential confounding factors were much better balanced in the 7377 propensity score matched patients which were used to assess association with outcomes (Table 1 , right panel). However, ASA status, Charlson comorbidity score, and emergent surgery were still slightly imbalanced (STD: 0.21, 0.11, and 0.14, respectively) between the etomidate and propofol patients. To be conservative, we thus included ASA status, Charlson comorbidity score, and emergent surgery in the multivariable models when comparing the 2 groups on the outcomes.
Among the propensity-matched patients, receiving etomidate was significantly associated with increased odds of experiencing 30-day mortality (estimated OR [98.3% confidence interval {CI}]: 2.49 [1.85–3.35]; P < 0.001), and increased odds of having major cardiovascular morbidity (1.51 [1.14–1.94]; P < 0.001), after adjusting for ASA status, Charlson comorbidity score, emergent surgery, and within-patient correlation (Table 4 ). However, etomidate was not significantly associated with major infectious morbidity (1.00 [0.80–1.25]; P = 0.99; Table 4 ). Our sensitivity analysis (Appendix, Table 1 ) suggests that our conclusion on 30-day mortality is robust to a very strong unmeasured binary confounding variable. For example, if we assume that the patients having covariate “u” are 5 times as likely to receive etomidate and also 5 times more likely to have the outcome, and 50% of patients have u , then the OR (95% CI) of etomidate versus propofol on 30-day mortality would still be significant at 1.4 (1.1–1.8). Our conclusions on cardiovascular morbidity and infectious morbidity were robust to a less strong unmeasured binary covariate (i.e., OR of ≤4 for cardiovascular and about ≤3 for infectious morbidity). The effects of etomidate were not consistent across the outcomes (etomidate-by-outcome interaction P < 0.001). Furthermore, the relationship between etomidate and the outcomes did not depend on ASA status (interaction P = 0.24), Charlson comorbidity score (P = 0.46), or emergency procedure (P = 0.15).
Table 4: Associations Between Use of Etomidate (Versus Propofol) Intraoperatively and Outcomes Among the Propensity Score–Matched Patients
However, no “dose effect” of etomidate was found on any of the primary outcomes. The estimated ORs (98.3% CI) for a unit (i.e., 0.1 mg/kg) increase in the amount of etomidate were 1.03 (0.97–1.10) (P = 0.27), 0.98 (0.90–1.06) (P = 0.50), and 1.01 (0.96–1.07) (P = 0.65) for having 30-day mortality, major cardiovascular morbidity, and major infectious morbidity, respectively.
Etomidate was significantly associated with prolonged hospital stay (P < 0.001; Table 4 ). Etomidate patients were 18% less likely (HR [95% CI]: 0.82 [0.78–0.87]) to be discharged from hospital at any given time point postoperatively as compared with propofol patients.
The etomidate and propofol groups were statistically different on mean systolic and diastolic blood pressure at various phases of the surgery (Table 5 ), and the etomidate group was often higher. However, many of the observed differences were too small to be clinically important. Etomidate was not associated with intraoperative vasopressor use, including dobutamine, dopamine, ephedrine, epinephrine, norepinephrine, phenylephrine, or vasopressin (OR [95% CI]: 0.92 [0.82–1.03] for etomidate versus propofol; P = 0.16).
Table 5: Summary of Intraoperative Systolic and Diastolic Blood Pressure Between the Propensity-Matched Groups
DISCUSSION
Our analysis indicates that induction of general anesthesia with etomidate is associated with more 30-day mortality and cardiovascular morbidity than when propofol is used for ASA III and IV patients undergoing noncardiac surgery, with highly significant ORs of 2.5 and 1.5, respectively. Use of etomidate is also associated with 1 day longer length of hospital stay compared with propofol. Our conclusions, especially on 30-day mortality, are robust to a strong unmeasured binary confounding variable.
Serious infectious complications in patients recovering from multiple traumas who are given long-term etomidate sedation have been attributed to suppression of cortisol synthesis.27 24
Induction drugs were not randomly assigned; instead, they were chosen at the discretion of attending anesthesiologists. It is likely that their choices were influenced by the individual patient’s physical status and perceived risk of hemodynamic instability. To remove the selection bias due to all the observed covariates, we used propensity score matching, which presumably improves validity of the analysis. We matched patients exactly on surgical types (Table 2 ) and used propensity score matching on other potential confounding factors (Table 1 ). The result was groups that were well balanced on many factors that potentially influence outcomes of interest. Furthermore, ASA physical status, Charlson comorbidity score, and emergency surgery which were still slightly imbalanced after propensity score matching were included in the multivariable models when comparing the 2 groups on the outcomes.
Surely there remains a degree of selection bias and confounding related to factors that are unavailable in our electronic records. Clinically perceived conditions of patients by anesthesia providers that was not explained by variables available in our electronic record could not be balanced in our analysis. However, our sensitivity analysis suggests that our conclusions, especially on 30-day mortality, are robust to a strong unmeasured binary confounding variable. It therefore seems unlikely that uncompensated bias and confounding account for all of the substantial association between etomidate use and adverse outcomes, although each surely contributes to some degree; the true OR of mortality resulting from anesthetic induction with etomidate may thus be considerably less than the 2.5 we observed.
For example, we were unable to adjust for the skill level of surgeons, the experience level of anesthesia providers, or the fact that various surgical approaches are characterized by the same surgical billing codes. Furthermore, we retrieved International Classification of Diseases, 9th Revision code–defined postoperative morbidity data for analysis which could potentially include an inaccurate diagnosis. Because data were not available electronically, we were also unable to account for a number of factors that potentially affect adrenal function such as chronic preoperative steroid use and preoperative use of medications that inhibit cortisol biosynthesis (i.e., ketoconazole, metyrapone, suramin) or those that increase steroid metabolism (i.e., carbamazepine, phenobarbital, phenytoin, rifampicin, mitotane). Preoperative cardiovascular and cerebrovascular morbidity were propensity matched, but specific variables that stratify severity of cardiac morbidity (i.e., left ventricular ejection fraction, brain natriuretic peptide, myocardial ischemia on cardiac stress test) and type of cardiac morbidity (i.e., coronary artery disease, valvular heart disease, cardiomyopathy) that affects anesthesiologists’ decision to choose etomidate were not propensity matched, again due to unavailability of information. As with all observational studies, we report statistical associations that may or may not indicate causal relationships between etomidate use and adverse outcomes. And this was a single-center study; results may differ in other settings and other populations. We speculate that the corticosteroid suppression effect of etomidate has some bearing on postoperative outcome but could not explain why mortality and cardiovascular morbidity and not infectious morbidity are associated with etomidate use, which is also a limitation of administrative database studies as opposed to prospective studies.
We chose propofol as a comparator induction drug, because it is by far the most commonly used induction drug. Propofol very slightly inhibits cortisol secretion from adrenal cells in a dose-related fashion in vitro.28 29 28 2
In summary, our analysis of ASA physical status III and IV patients undergoing noncardiac surgery with general anesthesia indicates that use of etomidate is associated with an increased odds of 30-day mortality and cardiovascular morbidity, although etomidate offers the advantage of minimizing induction hypotension that can cause coronary hypoperfusion, dysrhythmia, and cardiac arrest. Use of etomidate is also associated with prolonged duration of hospitalization. Randomized trials are necessary to determine whether there is a causal relationship between etomidate use and adverse outcomes and precisely define the treatment effect. In the meantime, etomidate should be used judiciously, considering that improved hemodynamic stability at induction may be accompanied by substantially worse longer-term outcomes.
APPENDIX Table 1: Sensitivity Analysis—Effects of an Unobserved 2-Category Covariate u on the Associations Between Etomidate and Primary Outcomes
DISCLOSURES
Name: Ryu Komatsu, MD.
Contribution: This author helped design the study, conduct the study, and write the manuscript.
Attestation: Ryu Komatsu has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.
Name: Jing You, MS.
Contribution: This author helped design the study, analyze the data, and write the manuscript.
Attestation: Jing You has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Edward J. Mascha, PhD.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Edward J. Mascha has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Daniel I. Sessler, MD.
Contribution: This author helped design the study, conduct the study, and write the manuscript.
Attestation: Daniel I. Sessler has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Yusuke Kasuya, MD.
Contribution: This author helped analyze the data and write the manuscript.
Attestation: Yusuke Kasuya has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.
Name: Alparslan Turan, MD.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Alparslan Turan has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
This manuscript was handled by: Sorin J. Brull, MD, FCARCSI (Hon).
FOOTNOTES
a Computerized matching of cases to controls using the greedy matching algorithm with a fixed number of controls per case. Controls may be matched to cases using one or more factors (X’s). Erik Bergstralh & Jon Kosanke. [10/2003] http://www.mayo.edu/research/documents/gmatchsas/DOC-10027248 Cited Here
REFERENCES
1. Diago MC, Amado JA, Otero M, Lopez-Cordovilla JJ. Anti-adrenal action of a subanaesthetic dose of etomidate. Anaesthesia. 1988;43:644–5
2. Fragen RJ, Weiss HW, Molteni A. The effect of propofol on adrenocortical steroidogenesis: a comparative study with etomidate and thiopental. Anesthesiology. 1987;66:839–42
3. Jameson P, Desborough JP, Bryant AE, Hall GM. The effect of cortisol suppression on interleukin-6 and white blood cell responses to surgery. Acta Anaesthesiol Scand. 1997;41:304–8
4. Absalom A, Pledger D, Kong A. Adrenocortical function in critically ill patients 24 h after a single dose of etomidate. Anaesthesia. 1999;54:861–7
5. Malerba G, Romano-Girard F, Cravoisy A, Dousset B, Nace L, Lévy B, Bollaert PE. Risk factors of relative adrenocortical deficiency in intensive care patients needing mechanical ventilation. Intensive Care Med. 2005;31:388–92
6. Annane D, Sébille V, Troché G, Raphaël JC, Gajdos P, Bellissant E. A 3-level prognostic classification in septic shock based on cortisol levels and cortisol response to corticotropin. JAMA. 2000;283:1038–45
7. Cooper MS, Stewart PM. Corticosteroid insufficiency in acutely ill patients. N Engl J Med. 2003;348:727–34
8. Marik PE, Zaloga GP. Adrenal insufficiency during septic shock. Crit Care Med. 2003;31:141–5
9. Pizarro CF, Troster EJ, Damiani D, Carcillo JA. Absolute and relative adrenal insufficiency in children with septic shock. Crit Care Med. 2005;33:855–9
10. Siraux V, De Backer D, Yalavatti G, Mélot C, Gervy C, Mockel J, Vincent JL. Relative adrenal insufficiency in patients with septic shock: comparison of low-dose and conventional corticotropin tests. Crit Care Med. 2005;33:2479–86
11. Bendel S, Koivisto T, Ruokonen E, Rinne J, Romppanen J, Vauhkonen I, Kiviniemi V, Uusaro A. Pituitary-adrenal function in patients with acute subarachnoid haemorrhage: a prospective cohort study. Crit Care. 2008;12:R126
12. Weant KA, Sasaki-Adams D, Dziedzic K, Ewend M. Acute relative adrenal insufficiency after aneurysmal subarachnoid hemorrhage. Neurosurgery. 2008;63:645–9
13. Agha A, Phillips J, O’Kelly P, Tormey W, Thompson CJ. The natural history of post-traumatic hypopituitarism: implications for assessment and treatment. Am J Med. 2005;118:1416
14. Benvenga S, Campenní A, Ruggeri RM, Trimarchi F. Clinical review 113: Hypopituitarism secondary to head trauma. J Clin Endocrinol Metab. 2000;85:1353–61
15. Cohan P, Wang C, McArthur DL, Cook SW, Dusick JR, Armin B, Swerdloff R, Vespa P, Muizelaar JP, Cryer HG, Christenson PD, Kelly DF. Acute secondary adrenal insufficiency after traumatic brain injury: a prospective study. Crit Care Med. 2005;33:2358–66
16. Dimopoulou I, Tsagarakis S, Douka E, Zervou M, Kouyialis AT, Dafni U, Thalassinos N, Roussos C. The low-dose corticotropin stimulation test in acute traumatic and non-traumatic brain injury: incidence of hypo-responsiveness and relationship to outcome. Intensive Care Med. 2004;30:1216–9
17. Dimopoulou I, Tsagarakis S, Theodorakopoulou M, Douka E, Zervou M, Kouyialis AT, Thalassinos N, Roussos C. Endocrine abnormalities in critical care patients with moderate-to-severe head trauma: incidence, pattern and predisposing factors. Intensive Care Med. 2004;30:1051–7
18. Hoen S, Asehnoune K, Brailly-Tabard S, Mazoit JX, Benhamou D, Moine P, Edouard AR. Cortisol response to corticotropin stimulation in trauma patients: influence of hemorrhagic shock. Anesthesiology. 2002;97:807–13
19. den Brinker M, Joosten KF, Liem O, de Jong FH, Hop WC, Hazelzet JA, van Dijk M, Hokken-Koelega AC. Adrenal insufficiency in meningococcal sepsis: bioavailable cortisol levels and impact of interleukin-6 levels and intubation with etomidate on adrenal function and mortality. J Clin Endocrinol Metab. 2005;90:5110–7
20. Hildreth AN, Mejia VA, Maxwell RA, Smith PW, Dart BW, Barker DE. Adrenal suppression following a single dose of etomidate for rapid sequence induction: a prospective randomized study. J Trauma. 2008;65:573–9
21. Lipiner-Friedman D, Sprung CL, Laterre PF, Weiss Y, Goodman SV, Vogeser M, Briegel J, Keh D, Singer M, Moreno R, Bellissant E, Annane DCorticus Study Group. . Adrenal function in sepsis: the retrospective Corticus cohort study. Crit Care Med. 2007;35:1012–8
22. Sprung CL, Annane D, Keh D, Moreno R, Singer M, Freivogel K, Weiss YG, Benbenishty J, Kalenka A, Forst H, Laterre PF, Reinhart K, Cuthbertson BH, Payen D, Briegel JCORTICUS Study Group. . Hydrocortisone therapy for patients with septic shock. N Engl J Med. 2008;358:111–24
23. Sunshine J, Deem S, Weiss N, Yanez ND, Daniel S, Keech K, Brown M, Treggiari M. Etomidate, adrenal function and mortality in critically ill patients. Respir Care. 2013;58:639–46
24. Warner KJ, Cuschieri J, Jurkovich GJ, Bulger EM. Single-dose etomidate for rapid sequence intubation may impact outcome after severe injury. J Trauma. 2009;67:45–50
25. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;70:41–55
26. Austin PC. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Stat Med. 2009;28:3083–107
27. Watt I, Ledingham IM. Mortality amongst multiple trauma patients admitted to an intensive therapy unit. Anaesthesia. 1984;39:973–81
28. Robertson WR, Reader SC, Davison B, Frost J, Mitchell R, Kayte R, Lambert A. On the biopotency and site of action of drugs affecting endocrine tissues with special reference to the anti-steroidogenic effect of anaesthetic agents. Postgrad Med J. 1985(61 Suppl 3):145–51
29. Newman LH, McDonald JC, Wallace PG, Ledingham IM. Propofol infusion for sedation in intensive care. Anaesthesia. 1987;42:929–37
30. Rosenbaum PR, Rubin DB. Assessing sensitivity to an unobserved binary covariate in an observational study with binary outcome. J R Statist Soc B. 1983;45:212–8
