Interactions between Midazolam and Remifentanil during Monitored Anesthesia Care

Eighty-one healthy (specified as American Society of Anesthesiologists physical status 1 to 2) consenting women scheduled for breast biopsy procedures under local anesthesia were randomly assigned to one of four sedative-analgesic treatment groups according to an institutional review board-approved double-blind, placebo-controlled protocol. Women of childbearing potential were included in the study only after obtaining a negative result of a urine pregnancy test (Abbott Test-Pack Plus hCG-urine; Abbott Laboratories, IL) on the morning of surgery. Exclusion criteria included excessive body weight (more than 100% over their ideal weight); a history of illegal drug use, hypersensitivity to opioid analgesics; and long-term use of sedatives, tricyclic antidepressants, beta-blockers, clonidine, or anticonvulsants.

All patients had nothing to eat or drink for at least 8 h before surgery except for their usual oral medications. No preanesthetic medication was administered. Before surgery, each patient was asked to assess her level of discomfort, pain, sleepiness, nervousness, and nausea using 100-mm visual analog scales anchored by word pairs (e.g., no discomfort to worst discomfort ever experienced, no pain to severe pain, not sleepy to deep sleep, not nervous to extremely nervous, no nausea to severe nausea). [3]When patients arrived in the operating room, an 18-gauge catheter was inserted to administer fluid and drugs. Standard monitors included a noninvasive blood pressure and heart rate device, electrocardiograph, pulse oximeter, and capnograph. A nasal cannula was positioned to deliver 3 to 4 l/min oxygen with a sampling catheter attached to estimate end-expiratory carbon dioxide values and to measure respiratory rate. [4]After obtaining baseline measurements of hemodynamic variables (systolic and diastolic blood pressures, mean arterial pressure, and heart rate) and respiratory variables (oxygen saturation, respiratory rate, end-tidal pressure of carbon dioxide [PET sub CO²]), the patients were asked to evaluate verbally their level of discomfort (on an 11-point numerical rating scale, with 0 = none to 10 = extreme) and pain (on a descriptive scale of 0 = none, 1 = mild, 2 = moderate, 3 = severe). The patient's level of sedation was assessed using the observer's assessment of alertness/sedation (OAA/S) scale [5](modified by reversing the order of the scores) by the same investigator (M.N.A.). The study medication (midazolam 2, 4, or 8 mg, or saline) was administered intravenously in four 2-ml increments at 1-min intervals. Hemodynamic and respiratory measurements were recorded at 1-min intervals for 5 min and thereafter at 1- to 5-min intervals. Five minutes after the administration of the last dose of the study medication, the degree of sedation was reassessed and the patient was shown a picture (a drawing of a cat). A continuous infusion of remifentanil (reconstituted to a concentration of 50 micro gram/ml) was initiated at a rate of 0.1 micro gram [centered dot] kg sup -1 [centered dot] min sup -1. Five minutes after the start of the remifentanil infusion, the surgeons infiltrated the operative field with a local anesthetic solution (1% lidocaine) and 1 or 2 min later they made the skin incision. The remifentanil infusion rate was subsequently adjusted in 0.025- to 0.05-micro gram [centered dot] kg sup -1 [centered dot] min sup -1 increments to achieve a discomfort score of less or equal to 4 and a pain score of less or equal to 1 while avoiding bradypnea (decrease in respiratory rate < 8 bpm), oxygen saturation values less than 90%, and oversedation (an OAA/S score of 5). Apnea was defined as the absence of an respiratory effort within a 15-s interval. Upon completion of the surgical procedure, the remifentanil infusion was discontinued. The number of changes in the remifentanil infusion rate, the maximum and minimum infusion rates, the time-weighted average infusion rate, and the total dose of remifentanil required during the procedure, and any side effects (such as bradypnea/apnea, pruritus, or nausea) were recorded.

Immediate recovery was assessed using the Aldrete scoring system [6]at 1, 3, 5, and 10 min. Patients who had an Aldrete score of 9 or more while in the operating room were admitted directly from the operating room to the same-day surgery step-down unit (phase II recovery), bypassing the postanesthesia care unit (phase I recovery). During the recovery period, patients' assessments of discomfort and pain and their level of sedation (OAA/S) were evaluated at 15-min intervals. Patients were considered suitable for discharge when they were alert and oriented, could ambulate unassisted and tolerate oral fluids, and had no ongoing surgical or anesthetic complications. Postoperative side effects and the need for therapeutic interventions were noted. The recovery to "home readiness" and the time of actual discharge were recorded. Immediately before discharge, another set of visual analog scores was administered and intraoperative recall was assessed using a standard picture recall test. Recall was graded as "free" (patient could correctly name the picture shown), "cued" (patient could recognize the object in a group of objects), "visual" (patient could recognize the object on the picture itself), and "none" (patient could not recall the object even after being shown the picture).

In a telephone interview on the first postoperative day, the patients were asked about pain, nausea, vomiting, and other adverse reactions after their discharge from the day-surgery unit, and also about the amount of pain medication they were using at home. The outpatients were asked to assess their satisfaction with the analgesic management during the period of local anesthetic infiltration and with the overall surgical experience on a 7-point verbal rating scale (1 = extremely dissatisfied to 7 = extremely satisfied), as well as if they would prefer the same sedative-analgesic medications should they require a similar surgical procedure in the future.

Data from the four study groups were compared using one-way analysis of variance followed by paired tests when appropriate, with probability values less than 0.05 considered statistically significant. Data are presented using summary statistics (mean, standard deviation, minimum, maximum, median values, and percentages). The frequency distribution of scores before, during, and after the administration of the study drug and remifentanil were calculated. The Cochran-Mantel-Haenszel test was used to test differences in frequency distribution of responses among the four treatment groups. The pre- and postoperative visual analog scores for discomfort, pain, sedation, anxiety, and nausea were compared using nonparametric tests. Recovery data (such as time to ambulate unassisted, time to Aldrete score greater or equal to 9, time to being judged suitable for discharge, and time to actual phase II discharge) were compared among treatment groups using the Cox proportional hazard test. To evaluate the effect of the initial level of sedation on specific postoperative outcome measures (postoperative nausea and vomiting [PONV], discharge time), patients were also subdivided into two groups according to the level of sedation 5 min after administration of the study medication (midazolam or placebo): (1) "light" sedation (OAA/S scores of 1 to 3) and (2) "deep" sedation (OAA/S scores of 4 or 5). These two subgroups were compared using t test and chi-squared tests, with probability values less than 0.05 considered statistically significant.

The four study groups were similar with respect to demographic and baseline physiologic characteristics (Table 1). Midazolam produced a dose-dependent increase in the level of sedation (median composite OAA/S scores of 1 in the placebo to 4 in the 8-mg midazolam groups) that was not enhanced by the initial infusion of 0.1 micro gram [centered dot] kg sup -1 [centered dot] min sup -1 remifentanil (Figure 1). At the start of the surgical procedure, the median sedation scores varied from 2 in the saline (placebo) group to 4 in the 8 mg midazolam group. When used alone, remifentanil produced a modest increase in sedation (a median maximum sedation score of 2 in the placebo group). During the course of remifentanil administration, none of the patients in the placebo and 2-mg midazolam groups reached a maximum score of 5 (unresponsive), compared with 33% in the 4-mg midazolam group and 79% in the 8-mg midazolam group. Similarly, more patients in the 4-mg and 8-mg midazolam groups required decreases in the remifentanil infusion rate (P < 0.01). At the end of the procedure, the level of sedation was similar in all four treatment groups (median OAA/S scores of 2 or 3).

During the local anesthetic infiltration period, moderate-to-severe pain was present in only two patients, both in the placebo group. In one of the patients, the pain was accompanied by a positive discomfort response (score > 4). Mild pain was reported by two patients in the placebo group (11%), two patients (9%) in the 2-mg midazolam group, three patients (14%) in the 4-mg midazolam group, and by no patients in the 8-mg midazolam group. Three patients in the placebo group (16%) and two in the 4-mg midazolam (10%) versus none in the 2-mg and 8-mg midazolam groups reported mild discomfort responses (scores of 1 to 4) during the infiltration period (P < 0.05). However, during the intraoperative period, the overall incidence of any discomfort responses (that is, scores of greater or equal to 1) did not differ significantly among the treatment groups (43% in the placebo vs. 21% to 33% in the three midazolam groups). Similarly, there were no differences in groups with respect to the number of the patients requiring increases in the remifentanil infusion rate due to pain or discomfort.

Midazolam produced a nonsignificant decrease in the mean remifentanil infusion rate during the operation, which ranged from 0.11 +/- 0.03 micro gram [centered dot] kg sup -1 [centered dot] min sup -1 in the saline group to 0.06 +/- 0.02 micro gram [centered dot] kg sup -1 [centered dot] min sup -1 in the 8-mg midazolam group (Table 1). In the saline group, more patients required rate increases than decreases, whereas in the 4-mg and 8-mg midazolam groups more patients required decreases in the infusion rate (P < 0.01). At the end of the surgical procedure, the mean remifentanil infusion rate in the 8-mg midazolam group (0.05 +/- 0.03 micro gram [centered dot] kg sup -1 [centered dot] min sup -1) was significantly less than that in the placebo group (0.11 +/- 0.05 micro gram [centered dot] kg sup -1 [centered dot] min sup -1; P < 0.001).

Increasing the midazolam dose from 2 mg to 4 mg and 8 mg increased the incidence of bradypnea (decreases in respiratory rate < 8 bpm) from 16% in the placebo and 46% in the 2-mg midazolam groups to 62% and 69% in the 4-mg and 8-mg midazolam groups, respectively (P < 0.01). Decreases in respiratory rate progressed to transient apnea in four patients (one in the 2-mg and three in the 4-mg midazolam groups). The patients with these episodes of respiratory depression responded to decreases in the remifentanil infusion rate. Episodes of bradypnea resolved within a median time of 4 min (in the 2-mg midazolam group) to 15 min (in the 8-mg midazolam group). The hemodynamic variables remained similar in all four treatment groups.

After discontinuation of the remifentanil infusion, recovery was rapid, with a median time to reach an Aldrete score of 9 or more of 1 min in all four groups. No significant difference was observed among the groups with respect to the median times to ambulate unassisted (47 min to 67 min, for the placebo to 8-mg midazolam groups, respectively), to "home readiness" (55 to 71 min), and to actual discharge (70 to 108 min) from the phase II step-down unit. The quantitative psychological assessments of discomfort, pain, sedation, anxiety, and nausea immediately before discharge did not reveal significant differences from their respective preoperative values. With the exception of pruritus, there were no significant differences in the incidence of minor adverse events (Table 2).

When analyzed based on the level of sedation 5 min after the study medication, the incidence of PONV was higher in the "light" sedation group (n = 56; 18% had nausea and 11% vomited in the postanesthesia care unit) than in the "deep" sedation group (n = 25; 4% had nausea and none vomited before discharge; P < 0.05). Similarly, there was a significantly greater incidence of intraoperative pruritus in the "light" (vs. "deep") sedation group (27% vs. 8%, respectively; P < 0.05).

Midazolam produced dose-dependent decreases in patient recall at the start of the operation (Table 3). However, most patients were highly satisfied with the treatment they received, with more than 90% of them expressing a willingness to receive the same sedative-analgesic technique again in the future. There were no differences in the severity of pain or in requirements for analgesic medication after discharge home.

Use of remifentanil in combination with midazolam was well tolerated by women having breast biopsy surgery with local anesthesia. Patients who received only remifentanil had lower levels of sedation and remained alert throughout the procedure in contrast to those who received 4 to 8 mg midazolam intravenously. However, midazolam provided significant intraoperative amnesia and a higher degree of patient comfort compared with the placebo treatment. A dose-dependent increase occurred in the incidence of bradypnea (respiratory rate less than 8 bpm) and "oversedation" in the groups that received higher doses of midazolam. Although midazolam produced minimal opioid-sparing effects during the operation, the remifentanil infusion rate at the end of the case was nearly 50% less in the 8-mg midazolam group than in the saline group. However, despite the lower dose of remifentanil required in the midazolam groups, three-to-four times as many patients experienced bradypnea compared with the saline group. A comparative study of remifentanil and alfentanil in volunteers found similar incidences of respiratory depression with these two potent, rapidly acting opioid analgesics. [7]The lower incidence of clinically significant respiratory effects (as defined by episodes of bradypnea [or apnea] and hemoglobin oxygen desaturation) with remifentanil alone in these surgical patients compared with the earlier study in healthy volunteers may be attributed to the stimulation caused by the surgical procedure itself.

Opioid analgesics decrease the slope of the ventilatory response to carbon dioxide (i.e., hypercapnic ventilatory drive), [1]and the hypoxic ventilatory response (i.e., hypoxic drive) provides protection from hypoventilation-induced hypoxia and apnea. Although sedative doses of midazolam depress the hypoxic ventilatory response, [8]apnea and hypoxemia do not usually occur when midazolam is administered alone. [1]However, when benzodiazepines and opioid analgesics are combined, they produce a potent synergistic interaction resulting in clinically significant respiratory depression with hypoxia and even apnea in healthy volunteers. [1]The greater incidence of bradypnea with increasing doses of midazolam in our study suggests a similar interaction between midazolam and remifentanil. However, the decreases in respiratory rate responded rapidly to reductions in the remifentanil infusion rate, and opioid antagonists were not required in any of these patients. Yet, the degree of respiratory depression is difficult to assess because these otherwise healthy patients were receiving supplemental oxygen (via nasal prongs) and blood gas values were not measured.

All patients who received remifentanil alone could clearly recall the intraoperative experience. Not surprisingly, the degree of intraoperative amnesia depended on the dose of midazolam administered. This finding corresponds with previous reports describing the amnestic effect of midazolam when administered in combination with ketamine [9]or propofol [10]during local anesthesia.

Regardless of differences in the early intraoperative levels of sedation and subsequent remifentanil infusion rates, early recovery was equally prompt in all four treatment groups. All patients qualified for admission to the step-down unit directly from the operating room. Similarly, times to ambulation, home-readiness, and discharge did not differ significantly among the four study groups. The dose-independent recovery after the termination of the remifentanil infusion reflects its unique pharmacokinetic profile, with a context-sensitive half-time of only 5 min. [11]The concern regarding inadequate postoperative analgesia due to the rapid clearance of remifentanil did not manifest itself in this study, probably because the local anesthetic used during the operation provided residual analgesia in the early postoperative period. Thus remifentanil should be a valuable adjuvant during outpatient procedures performed under local anesthesia when brief periods of intense opioid effect are required. However, the potent depressant effects of the midazolam-remifentanil combination on respiratory rate emphasizes the importance of carefully monitoring ventilation during MAC.

The overall incidence of postoperative nausea (32%) and emesis (21%) associated with remifentanil alone was similar to that reported in earlier studies involving alfentanil. [12,13]However, patients who reported PONV in the placebo and 2-mg midazolam groups did so in the early postoperative period, whereas patients in the 4- and 8-mg midazolam groups reported more nausea and emesis after discharge. Because it has been suggested that the level of sedation per se may influence the incidence of PONV, [14]a post hoc analysis of the incidence of PONV versus sedation level was performed. This analysis revealed that a greater depth of sedation may have protected against PONV in patients given remifentanil. These findings are also consistent with a recent report suggesting that sedative doses of midazolam (and propofol) possess antiemetic activity. [15]Given the potential beneficial effect of combining opioid analgesic and sedative-hypnotic drugs during MAC, the combination of remifentanil and propofol should be investigated with respect to their ability to provide precise control of the intraoperative analgesia and sedation and a prompt recovery without PONV.

When administered to healthy outpatients, 2 mg midazolam given intravenously in combination with a 0.05-to 0.1-micro gram [centered dot] kg sup -1 [centered dot] min sup -1 remifentanil infusion provided adequate sedation, amnesia, and analgesia during MAC. Although remifentanil is easily titrated to the desired analgesic end point, the potent interaction between remifentanil and midazolam can produce excessive sedation and significant respiratory depression, emphasizing the importance of careful administration of remifentanil by appropriately trained clinicians. Increased levels of sedation during the early intraoperative period decrease opioid-induced PONV and pruritus in the perioperative period without significantly delaying recovery.

The authors thank the anesthesia residents and certified registered nurse-anesthetists at Parkland Memorial Hospital who assisted with this study. In addition, the support of Randall Batenhorst, Pharm. D., Carl Roland, Pharm. D., and Brenda Jamerson, Pharm. D. from Glaxo-Wellcome (Research Triangle Park, NC) during this clinical investigation is greatly appreciated.