Cardiac surgery, particularly coronary artery bypass grafting, has become one of the most common operations in the United States. Because of its success and an aging population, cardiac surgery is estimated to cost $27 billion annually in the United States (1). In an attempt to decrease the costs of cardiac surgery, fast-track programs have become popular. They rely on an anesthetic technique that permits prompt extubation instead of the traditional overnight mechanical ventilation. Early success was achieved with a propofol-based technique (2). However, another study has suggested that a fentanyl/isoflurane technique can achieve the same prompt extubation at a much lower anesthetic cost (3). Remifentanil, an ultra-short-acting opioid, was introduced to practice in 1993 (4). This drug, which is metabolized by plasma cholinesterases, permits more rapid emergence than either fentanyl or sufentanil and may permit a more prompt extubation. However, remifentanil is much more expensive than either fentanyl or sufentanil, and its putative advantages to achieve more rapid extubation and discharge need to be evaluated to determine whether use of remifentanil can produce sufficient savings to justify its higher acquisition cost. We hypothesized that remifentanil would produce a shorter time to tracheal extubation and therefore designed this study to compare the effects of three different opioid techniques for cardiac surgery on time to tracheal extubation, time to intensive care unit (ICU) discharge, time to hospital discharge, postoperative pain, and cost. Additionally, we sought to evaluate the relationships among time of mechanical ventilation, ICU length of stay, hospital length of stay, and cost.
This study was approved by our IRB, and written, informed consent was obtained from all patients before group assignment and surgery. Ninety adult patients undergoing cardiac surgery at our university-affiliated tertiary care center were randomized to a fentanyl-based, sufentanil-based, or remifentanil-based anesthetic as chosen by a computer-generated random number list designed to produce true randomization rather than balanced groups. Patients were excluded if they were having nonelective surgery, having combined carotid and cardiac surgery, did not speak English, were mechanically ventilated before surgery, or were not capable of understanding instructions.
All patients had standard monitoring, including electrocardiography, arterial and central venous manometry, pulse oximetry, and measurement of inspiratory and expiratory carbon dioxide and isoflurane concentrations. Pulmonary artery catheterization was at the discretion of the anesthesiologist and surgeon (and was used in <10% of the patients in each of the three groups), and the decision for this was made after enrollment in the study but before assignment to any of the three groups. Surgery was performed via a median sternotomy. Cardiopulmonary bypass, when used, was normothermic.
All patients were premedicated with lorazepam 2 mg sublingually (1 mg if the patient was <55 kg or >70 yr). All patients had anesthesia induced with diazepam 0–5 mg, thiopental 0–250 mg, and pancuronium 0.1 mg/kg (median and interquartile ranges are given in Table 1). Patients in the Fentanyl group received fentanyl 7–10 μg/kg for the induction and additional doses of 1–2 μg/kg as needed for intense stimulus. The Sufentanil group patients received sufentanil 1–4 μg/kg for the induction and 0.1–0.3 μg/kg as needed for intense stimulus. Although sufentanil is typically considered to be 5 to 10 times more potent than fentanyl, it has a shorter half-life than fentanyl—approximately half as long (5). Therefore, we front-loaded sufentanil at approximately one third the dose of the fentanyl to compensate for the shorter half-life. Remifentanil patients had anesthesia induced with a remifentanil infusion at 0.5–1.0 μg · kg−1 · min−1 (on the basis of ideal body weight), then maintained anesthesia by titrating the infusion between 0.05 and 1.0 μg · kg−1 · min−1. Boluses of 0.5–1.0 μg/kg could be given as needed for intense stimulus. On arrival in the ICU, the infusion was decreased to 0.025–0.2 μg · kg−1 · min−1. Then, 15 to 30 min later (after initial doses of ketorolac and morphine), the remifentanil infusion was discontinued. All Remifentanil group patients received fentanyl 250 μg as part of the induction. We believed that this would decrease the remifentanil dose needed for the induction, provide some initial postoperative analgesia (when the remifentanil was discontinued), and have minimal effects on the rest of the postoperative course. In addition, patients in all three groups received maintenance doses of diazepam 0–5 mg, pancuronium 0.01 mg/kg as needed to maintain one to three twitches of a train-of-four as measured by peripheral nerve stimulator, and isoflurane at an end-tidal concentration of 0.3%–0.7%.
Hemodynamic status was supported as previously described (3). Briefly, nitroglycerin and nitroprusside were used to maintain mean arterial blood pressure (MAP) <70 mm Hg during surgery and <90 mm Hg after surgery. Ephedrine and phenylephrine were used to maintain MAP >60 mm Hg during and after surgery. During surgery, esmolol and propranolol were used to keep heart rate <100 bpm. After surgery, propranolol 1 mg IV every 4 h was administered. Propranolol was withheld for MAP <70 mm Hg, systolic blood pressure <110 mm Hg, or heart rate <70 bpm. Dobutamine, epinephrine, and dopamine were used per the surgeon’s choice as inotropes.
All patients were weaned from mechanical ventilation by our standard protocol (3). Briefly, mechanical ventilation was started in synchronized intermittent mandatory ventilation mode with a rate of 10 breaths/min and a tidal volume of 10–12 mL/kg. Fraction of inspired oxygen was adjusted to maintain oxygen saturation by pulse oximetry ≥95%. Positive end-expiratory pressure was set at 5 cm H2O. End-tidal CO2 measurements were used to adjust tidal volume to maintain predicted arterial CO2 between 35 and 40 mm Hg. When patients were hemodynamically stable and rousable, they were changed to continuous positive airway pressure. If patients were awake but showed clinical signs of residual neuromuscular blockade, glycopyrrolate 0.4 mg and neostigmine 2.5 mg were given IV. If end-tidal CO2 remained within 5 mm Hg and the patient was alert, mechanics were checked, and if these were acceptable (respiratory rate 10–28 breaths/min, tidal volume >5 mL/kg, vital capacity >10 mL/kg, and negative inspiratory force ≤20 cm H2O), the patient was extubated. If end-tidal CO2 increased by more than 5 mm Hg, a synchronized intermittent mandatory ventilation rate was reinstituted until the patient was more awake and continuous positive airway pressure could be retried. Ventilator hours were defined as time from arrival in cardiovascular ICU (CVICU) until tracheal extubation. Prolonged ventilation was defined as continued mechanical ventilation at 6:30 am the day after surgery.
Ketorolac and morphine were used for analgesia, and the doses between arrival in the ICU and 6:30 the next morning were recorded. Ketorolac 15 mg IV was given on arrival in the CVICU, on first complaint of pain, and then every 6 h as needed. Ketorolac was not used if the patient had a history of aspirin intolerance, recent peptic ulcer disease, or an increased creatinine. Morphine 1–2 mg IV as needed was used if ketorolac was insufficient.
Thirty minutes after extubation and at 6:30 am on the first postoperative day, each patient was asked to rate his or her pain on the 101-point numeric rating pain scale: 0 as no pain and 100 as pain as bad as it could be (6).
Hospital cost was calculated as the sum of the direct variable cost for each item and service used by the patient from preoperative preparation through discharge or death and was obtained from the hospital’s internal accounting system.
The a priori power calculation was based on 90% power, 1.7% type I error (to control for multiple comparisons among groups), and an sd estimate of 202 min in time of mechanical ventilation from a previous study (3). This yielded 90 patients. For patient characteristics and primary outcome of the continuous type, groups were compared by using a nonparametric Kruskal-Wallis test and were presented as median and interquartile range. If there was evidence, with P < 0.05, for at least some group differences, then a Bonferroni multiple comparisons procedure was used to examine which groups differed. Groups were compared by using χ2 or Fisher’s exact tests for categoric-type data. Spearman correlation coefficients were used.
Patients in the Fentanyl group were slightly younger (P = 0.04) than those in either the Sufentanil or Remifentanil groups. The other preoperative demographics and types of surgery were similar (Table 2). No patient had reoperative cardiac surgery. They received similar doses of diazepam, thiopental, and pancuronium (Table 1). We found no differences in the ability of fentanyl, sufentanil, and remifentanil to promote faster liberation from mechanical ventilation, shorter ICU stays, and shorter hospital stays (Table 3). However, patients who received sufentanil during surgery required less morphine in the ICU compared with fentanyl (P = 0.02) and remifentanil (P < 0.001). Patients in the Remifentanil group were more likely to require bolus doses of phenylephrine during surgery (83% vs 55% for the Fentanyl group and 43% for the Sufentanil group, P < 0.01). They were also more likely to receive bolus doses of nitroglycerin in the CVICU (41% vs 21% for the Fentanyl group and 14% for the Sufentanil group, P < 0.05). Use of all other hemodynamic drugs was similar among the three groups.
There were six patients with complications in the Fentanyl group: one patient required propofol for emergence delirium, one with reexploration for hemorrhage, and one with intraoperative insertion of an intraaortic balloon pump. The fourth patient was reintubated (3 days after extubation), had a partial bowel resection for ischemia, and needed a tracheostomy before recovering. The fifth patient had isolated acute respiratory distress syndrome. The sixth patient had an intraoperative myocardial infarction, progressed to multisystem organ dysfunction, and ultimately died 4 days after surgery (he was included in all analyses; excluding him did not change any results). Two patients in the Remifentanil group had complications: one required propofol for emergence delirium, and the other had a postoperative encephalopathy. Five patients in the Sufentanil group had complications: two patients needed to be reintubated (2 and 7 days after extubation, respectively), the third required propofol for emergence delirium, the fourth required a permanent pacemaker, and the fifth developed ischemic hepatitis and renal failure necessitating hemodialysis. Two patients in each of the Sufentanil and Remifentanil groups required reversal of residual neuromuscular block in the CVICU.
Although the highest opioid (P < 0.001) and anesthetic (P < 0.01) costs were in the Remifentanil group and the lowest in the Fentanyl group, total direct variable costs were similar (P = 0.3) among all three groups (Table 3). The study also found that although there were statistically significant correlations between time of mechanical ventilation and ICU length of stay (r = 0.50, P < 0.0001), hospital length of stay (r = 0.32, P = 0.002), and direct variable cost (r = 0.25, P = 0.02), the Spearman correlations, which made no assumptions about linearity, were fair to poor (Fig. 1). Also, ICU length of stay correlated poorly with hospital length of stay (r = 0.38, P = 0.0002) (Fig. 1, bottom right).
The use of more expensive but shorter-acting anesthetics may be justified if they permit faster extubation and shorter ICU and hospital stays and these shorter stays translate into lower total costs. However, this study found that the use of the shorter-acting opioids, sufentanil and remifentanil, was not associated with shorter ventilator times, ICU stays, or hospital stays. Fentanyl, sufentanil, and remifentanil all produced similar outcomes with similar direct variable costs. Prompt extubation has been touted as the first step in fast-track cardiac surgery. However, we found that time to extubation explained only 6.3% of the variance (r = 0.25) in cost and 10.2% of the variance (r = 0.32) in hospital length of stay. As Cheng (7) has pointed out and Arom et al. (8) have shown, process of care and intermediate outcomes (chest tube drainage, arrhythmias, and so on) may have a greater effect on cost and hospital length of stay than does choice of anesthetic. This suggests that further efforts to decrease already short ventilation time might have minimal effects on decreasing hospital direct variable costs. In addition, anesthetic cost was only a small fraction (0.6% for the Fentanyl group, 1.0% for the Sufentanil group, and 2.2% for the Remifentanil group) of the total cost.
A previous study has found that patients who receive remifentanil may have more postoperative pain (9). However, we found similar pain scores among the three groups, with the Remifentanil group receiving more morphine than the Sufentanil group. The similar pain scores may be related to the residual effects of the 250 μg fentanyl used as part of the induction in the Remifentanil group.
Only a few previous studies have compared remifentanil or sufentanil with fentanyl (10,11). In a small study of 27 patients (another 9 patients were excluded for death or prolonged hospital stay), the Remifentanil group had shorter (by an average of 2.5 hours) times to extubation and shorter (6.6 ± 0.26 vs 8.4 ± 0.5 days, mean ± sem, P = 0.015) hospital stays compared with patients receiving fentanyl (10). However, we achieved shorter hospital length of stays in all three of our groups; this probably reflects differences in process of care or criteria for discharge. Our results differ from those of Butterworth et al. (11), who found that sufentanil produced a quicker extubation than did fentanyl (geometric mean 12.2 vs 14.2 hours) but likewise found no association between time of mechanical ventilation and ICU or hospital length of stay. We expected a very rapid emergence and tracheal extubation in the Remifentanil group, but we did not find this. Although our median time to tracheal extubation in the Remifentanil group of 3.90 hours was longer than expected, it was similar to the 3.3 hours found by Cheng et al. (12) and the 5.3 hours found by Möllhoff et al. (13). Additionally, Zarate et al. (14) found a mean time to tracheal extubation of 5.1 ± 4.3 hours in patients receiving remifentanil and intrathecal morphine. Reasons for this delayed emergence and extubation found with remifentanil after cardiac surgery remain to be elucidated.
Similar to London et al. (15), we identified a sharp break in ICU length of stay (Fig. 1, bottom left) on the basis of whether patients were transferred out of the ICU before midnight the day of surgery or transfer was accomplished after 8:00 the next morning. Although this pattern of transfer before midnight or after 8:00 am occurs because of nursing staffing requirements, it also minimizes patient sleep disturbances, which can lead to ICU psychosis and delirium (16). However, this did not seem to affect hospital length of stay (Fig. 1, bottom right).
Prompt tracheal extubation is safe, and a variety of drugs (opioids, inhaled vapors, and propofol) can be used to achieve prompt extubation (2,3,17). However, although they have similar effects on mechanical ventilation, these drugs may have different effects on cost. Previous studies examining cost have been limited—they use only costs of anesthetic drugs (3) or use arbitrary assignment to different patient locations of fixed overhead costs (2). This is the first study to examine total direct variable cost for cardiac surgery. Study has shown that charge/cost ratios have a large variation from department to department (18). We chose to use direct variable costs as the outcome financial measure of interest (19). These are the costs that are actually spent by the hospital to perform a particular task. It ignores factors such as overhead that are not reduced or eliminated by changes in process of care. For example, if a patient is tracheally extubated and transferred from the ICU to the stepdown unit and the nurse/patient ratio is not changed, no direct variable cost is saved, although charges may be lower on the stepdown unit, reflecting an arbitrary assignment of more of the hospital’s overhead to the ICU. Unfortunately, the vast majority of costs in hospitals may be fixed overhead and are not lowered when the anesthetic technique is changed (20).
A major limitation of this study is that the costs may be applicable to only this hospital. Hospitals may pay different prices for disposable equipment and medications. Different nurse/patient staffing ratios and mixtures of nurses to paraprofessionals will affect the labor component of the costs. The determination of the labor associated with any laboratory test or procedure, such as a radiograph, may even differ between hospitals. Hospitals may need to develop their own cost models and evaluate the effectiveness of changes in the process of care or in drug use to determine the cost benefits of any change.
Another limitation was the use of normothermic cardiopulmonary bypass. Most centers use hypothermia, which may affect the pharmacokinetics and pharmacodynamics of the studied drugs and may limit the generalizability of this study. Also, low statistical power could account for the lack of statistically significant differences among groups for some of the outcome variables. Our sample size was chosen a priori to detect a large effect size of 1 between any two groups for time to extubation (202 minutes or 3.37 hours). The n was calculated on the basis of 90% power, 1.7% type I error (to control for multiple comparisons among groups), and an sd estimate of 3.37 hours from a previous study (3). However, the observed sd for extubation time was much larger than estimated. Among the three groups, the median sd was 5.37 hours. Therefore, the post hoc power to detect a difference of 3.37 hours (as chosen a priori) was only 46%. Yet the largest observed difference in extubation time among groups was only 1.97 hours, which is smaller than we had originally deemed clinically important. Data were not available to calculate a priori sample size requirements for the other outcome variables. However, similar post hoc analysis shows that we would have needed to study 942 patients to detect a $1500 difference in direct variable costs among the three groups or 2115 patients for a $1000 difference. The pain analysis was limited because several patients could not provide a numeric estimate of their pain. Our power to detect pain scale differences of 15 points, a clinically important difference, with α = 0.017 was only 35%.
In conclusion, this study found no differences in the outcomes from fentanyl-, sufentanil-, and remifentanil-based cardiac anesthetics. They all produce similar outcomes and have similar direct variable costs.
We would like to thank the CVICU nurses, the anesthesiologists, and nurse anesthetists for their help and participation in this study.
1. 2000 heart and stroke statistical update. Dallas: American Heart Association, 1999: 26.
2. Cheng DC, Karski J, Peniston C, et al. Early tracheal extubation after coronary artery bypass graft surgery reduces costs and improves resource use: a prospective, randomized, controlled trial. Anesthesiology 1996; 85: 1300–10.
3. Engoren MC, Kraras C, Garzia F. Propofol-based versus fentanyl-isoflurane-based anesthesia for cardiac surgery. J Cardiothorac Vasc Anesth 1998; 12: 177–81.
4. Glass PS, Hardman D, Kamiyama Y, et al. Preliminary pharmacokinetics and pharmacodynamics of an ultra-short-acting opioid: remifentanil (Gl87084B). Anesth Analg 1993; 77: 1031–40.
5. Bailey PL, Stanley TH. Pharmacology of intravenous narcotic anesthetics. In: Miller RD, ed. Anesthesia. 2nd ed. New York: Churchill Livingstone, 1986: 745–98.
6. Jensen MP, Karoly P, Braver S. The measurement of clinical pain intensity: a comparison of six methods. Pain 1986; 27: 117–26.
7. Cheng DC. Fast track cardiac surgery pathways: early extubation, process of care, and cost containment. Anesthesiology 1998; 88: 1429–33.
8. Arom KV, Emery RW, Petersen RJ, Schwartz M. Cost-effectiveness and predictors of early extubation. Ann Thorac Surg 1995; 60: 127–32.
9. Guignard B, Bossard AE, Coste C, et al. Acute opioid tolerance: intraoperative remifentanil increases postoperative pain and morphine requirement. Anesthesiology 2000; 93: 409–17.
10. Royston D. Remifentanil in cardiac surgery. Eur J Anaesthesiol Suppl 1995; 10: 77–9.
11. Butterworth J, James R, Prielipp RC, et al. Do shorter-acting neuromuscular blocking drugs or opioids associate with reduced intensive care unit or hospital lengths of stay after coronary artery bypass surgery? Anesthesiology 1988; 88: 1437–46.
12. Cheng DC, Newman MF, Duke P, et al. A prospective randomized, double-blind controlled trial of remifentanil and fentanyl in fast tract CABG surgery: efficacy and resource utilization [abstract]. Anesthesiology 1999; 91: A142.
13. Möllhoff T, Herregods L, Blake D, et al. Remifentanil versus fentanyl in patients undergoing CABG surgery [abstract]. Anesthesiology 1999; 91: A144.
14. Zarate E, Latham P, White PF, et al. Fast-track cardiac anesthesia: use of remifentanil combined with intrathecal morphine as an alternative to sufentanil during desflurane anesthesia. Anesth Analg 2000; 91: 283–7.
15. London MJ, Shroyer AL, Coll JR, et al. Early extubation following cardiac surgery in a veterans population. Anesthesiology 1998; 88: 1447–58.
16. Gelling L. Causes of ICU psychosis: the environmental factors. Nurs Crit Care 1999; 4: 22–6.
17. Quasha AL, Loeber N, Feeley TW, et al. Postoperative respiratory care: a controlled trial of early and late extubation following coronary artery bypass grafting. Anesthesiology 1980; 52: 135–41.
18. Macario A, Vitez TS, Dunn B, McDonald T. Where are the costs in perioperative care? Analysis of hospital costs and charges for inpatient surgical care. Anesthesiology 1995; 83: 1138–44.
19. Pronovost P, Angus DC. Cost reduction and quality improvement: it takes two to tango. Crit Care Med 2000; 28: 581–3.
20. Roberts RR, Frutos PW, Ciavarella GG, et al. Distribution of variable vs fixed costs of hospital care. JAMA 1999; 281: 644–9.