Perioperative Methadone and Ketamine for Postoperative Pain Control in Spinal Surgical Patients

A Randomized, Double-blind, Placebo-controlled Trial

Glenn S. Murphy, M.D.; Michael J. Avram, Ph.D.; Steven B. Greenberg, M.D.; Jessica Benson, B.S.; Sara Bilimoria, B.S.; Colleen E. Maher, B.S.; Kevin Teister, B.S.; Joseph W. Szokol, M.D.


Anesthesiology. 2021;134(5):697-708. 

In This Article

Materials and Methods

Study Population and Perioperative Management

The Institutional Review Board of NorthShore University HealthSystem (Evanston, Illinois) approved this randomized, double-blind, placebo-controlled clinical trial (registry,; registration number, NCT02827526; date of registration, July 1, 2016; principal investigator, Glenn Murphy). The investigation was conducted at a single tertiary medical center (NorthShore Evanston Hospital), and written informed consent was obtained from all subjects. Patients were approached by research assistants and enrolled on the day of surgery. Study staff evaluated eligibility, obtained informed consent, managed the conduct of the trial, and collected and managed the data.

A total of 130 patients, ages 18 to 80 yr, undergoing elective spinal fusion surgery of one or more sacral, lumbar, and/or thoracic levels were enrolled. Exclusion criteria included American Society of Anesthesiologists (Schaumburg, Illinois) Physical Status IV or V, preoperative renal insufficiency or failure (serum creatinine ≥ 2 mg/dl), pulmonary disease necessitating home oxygen therapy, significant liver disease (cirrhosis or hepatic failure), recent history of alcohol or opioid abuse, allergy to methadone, hydromorphone, or ketamine, poor comprehension of the English language, or inability to use a patient-controlled analgesia (PCA) pump (due to intellectual impairment, disorientation, or confusion).

Patients were assigned to one of two groups using a computer-generated randomization table (simple randomization without restrictions): a methadone group (intraoperative methadone and a dextrose 5% in water infusion) or a methadone/ketamine group (intraoperative methadone and a ketamine infusion). The allocation sequence was generated by one of the study investigators, who provided the randomization assignments to the operating room pharmacy that assigned patients to the study groups and prepared all of the study medications. Care providers, researchers, and patients were blinded to group assignment. Clinicians delivering intraoperative care were provided with a 3-ml syringe labeled "methadone" that contained 0.2 mg/kg of methadone (based on ideal body weight, up to a maximal dose of 20 mg). This dose was selected based on data establishing effectiveness and safety of methadone 0.2 mg/kg in patients undergoing spine surgery[13–15] and other procedures.[16] The contents of the syringe labeled "methadone" were given to all study subjects over 5 min at induction of anesthesia. In addition, the pharmacy prepared sequentially numbered, identical-appearing 500-ml bags of dextrose 5% in water. For patients assigned to the methadone/ketamine group, 250 mg of ketamine was added to the dextrose 5% in water bag (total volume 500 ml), whereas patients assigned to the methadone group received only an infusion of dextrose 5% in water (no drug added to the 500 ml volume). The 500 ml bags were connected to a pump that was programed to deliver an infusion of ketamine dosed at ideal body weight (or an equal volume of dextrose 5% in water) at a rate of 0.3 mg · kg−1 · h−1 from induction of anesthesia until surgical closure, at which time the infusion was decreased to 0.1 mg · kg−1 · h−1. The infusion was maintained at a rate of 0.1 mg · kg−1 · h−1 in the postanesthesia care unit (PACU) and for the next 48 postoperative hours. Dosing of ketamine was based on recommendations in the literature[17,18] and from clinical experience at our institution.

Patients were given 2 mg of midazolam before entering the operating room. Standard intraoperative monitoring was applied, which included an automatic blood pressure cuff, electrocardiography, capnography, pulse oximetry, and bispectral index monitoring (BIS system, Aspect Medical Systems, USA). Anesthesia was induced with propofol 1 to 2 mg/kg, lidocaine 50 mg, fentanyl 100 μg, rocuronium 0.6 mg/kg, dexamethasone 8 mg, and methadone 0.2 mg/kg. The ketamine or dextrose 5% in water infusion was initiated after induction and dosed as described above. Anesthesia was maintained with sevoflurane 1%, remifentanil 0.1 μg · kg−1 · min−1, and a propofol infusion titrated to 50 and 150 μg · kg−1 · min−1 in order to achieve bispectral index values between 40 and 60 and mean arterial pressures within 20% of baseline measures (a combination of lower concentrations of propofol and sevoflurane are used at our institution during spinal surgery in order to allow for acceptable sensory evoked potential monitoring). Redosing of rocuronium was determined based on requirements for motor evoked potential monitoring. The administration of additional fentanyl (up to a total dose of 200 μg) or hydromorphone (1 mg), was at the discretion of the anesthesia care team. Hypotension was treated with phenylephrine 80 μg, ephedrine 5 mg, or a fluid bolus, as indicated. Episodes of hypertension were managed by increasing the propofol infusion rate. Blood glucose concentrations were monitored every hour in insulin-dependent diabetic patients and every 2 to 4 h in non–insulin-dependent diabetic patients, and patients with blood glucose concentrations greater than 180 to 200 mg/dl were treated with insulin. At the conclusion of the procedure, patients were given odansetron 4 mg, and neuromuscular blockade was reversed with neostigmine 50 to 70 μg/kg (with an appropriate dose of glycopyrrolate) or sugammadex 2 to 4 mg/kg. Patients were extubated in the operating room unless concerns about airway edema were raised by the care team.

Patients were assessed for pain by nurses at PACU admission and every 15 min thereafter. Moderate pain was treated with hydromorphone 0.25 mg, and severe pain with hydromorphone 0.5 mg, with the goal of reducing pain scores to less than or equal to 3 on a 0 to 10 numeric rating scale (0, no pain; 10, worst pain imaginable). Patients were transitioned to a hydromorphone PCA device when adequate pain control was achieved (initial programming: a demand dose of 0.2 mg with a 10 min lockout interval, a 1 h limit of 1.2 mg). Discharge to the surgical wards with continuous pulse oximetry monitoring occurred when Aldrete scores of greater than or equal to 8 out of 10 were achieved.

Pain was managed on the first 3 postoperative days using the hydromorphone PCA device, which was discontinued at the discretion of the managing surgical service. Patients were transitioned to hydrocodone 5 mg and acetaminophen 325 mg tablets when able to resume oral intake. Pain was assessed and treated by surgical ward nurses per standard protocols, with the goal of maintaining pain scores less than or equal to 4 on the 0 to 10 numerical rating scale.

Data Collection

In the preoperative holding area, a research team member recorded preoperative levels of pain at rest, with coughing, and with movement using a 0 to 10 numeric rating scale. Preoperative sedation was measured on a 0 to 3 scale (0, fully awake; 1, mildly sedated, seldom drowsy, and easy to awaken; 2, moderately sedated, often drowsy, and easy to awaken; and 3, severely sedated, somnolent, and difficult to awaken). The presence or absence of dizziness, nausea, vomiting, or itching was noted. The use of preoperative opioids was documented.

Anesthesia team members recorded heart rate and mean arterial blood pressures preinduction and then 5, 15, 30, 45, and 60 min postinduction. In addition, the times between the end of surgery and tracheal extubation were noted, as was the use of any additional hydromorphone in the operating room. PACU nurses documented the times from PACU admission until first request for pain medication and the total dose of hydromorphone (nurse-administered and via the PCA device). Nurses also recorded the times to meet discharge criteria and to achieve actual discharge. At the time of discharge from the PACU, overall satisfaction with pain management was measured using an 11-point numeric rating scale (0, worst possible; 10, best possible).

On PACU arrival, a research team member assessed patients for pain at rest, with coughing, and with movement using a 0 to 10 numeric rating scale. Level of sedation was determined using the 0 to 3 scale described above. The presence or absence of nausea and vomiting was noted, the severity graded on a 3-point scale (1, mild; 2, moderate; 3, severe), and drugs used to treat these events were recorded. The presence or absence of episodes of itching, hypoxemia (peripheral oxygen saturation less than 90%), hypoventilation (respiratory rate less than 8), dizziness, or hallucinations was documented. The heart rate, respiratory rate, mean arterial pressure, and peripheral oxygen saturation at the time of assessment were noted. All of the outcome variables assessed at PACU arrival, with the addition of patient satisfaction with pain management scores, were again measured 1 h after PACU admission, as well as on the morning (between 8 AM and 10 AM) and late afternoon (between 3 PM and 4 PM) of postoperative days 1, 2, and 3. Total doses of intravenous hydromorphone and oral hydrocodone and acetaminophen tablets used during each of the first 3 postoperative days were recorded.

Data collected from the electronic anesthesia record included anesthesia duration, total volume of crystalloid solutions, blood loss, urine output, and total dose of fentanyl. The electronic medical record was reviewed for any cardiac, respiratory, gastrointestinal, renal, neurologic, or infection complications during the hospitalization. The duration of hospitalization was obtained from the electronic medical record.

Statistical Analysis

The primary endpoint of this investigation was hydromorphone use on postoperative day 1. In a study of patients undergoing spinal fusion surgery using a standard anesthetic, the average ± SD intravenous hydromorphone dose in the first 24 postoperative hours was 27 ± 10 mg in patients given standard postoperative pain management and 18.5 ± 14 mg in patients given ketamine as an adjunct to that management.[19] We anticipated hydromorphone consumption in the methadone group in the current study would be 19 mg and hypothesized it would be reduced by 40% in the methadone/ketamine group. Because expected group SDs of 14 mg were large relative to the means, sample size was estimated for two-sided Mann–Whitney test using the standard t test formulations with a simple adjustment to the sample sizes based on the assumption that data distribution is logistic. Using this approach, group sample sizes of 51 and 51 achieve 81% power to detect a difference of 7.6 mg between the null hypothesis that both group means are 19.0 mg and the alternative hypothesis that the mean of group 2 is 11.4 mg with a significance level (alpha) of 0.05 (PASS 2008 Number Cruncher Statistical System, USA). One hundred thirty patients were enrolled to ensure complete collection of data.

Data for the primary outcome variable, milligrams of intravenous hydromorphone in the first 24 h after the operation, are reported as the median (interquartile range) for both the methadone group and the methadone/ketamine group. These data were compared between groups using the Mann–Whiney U test (StatsDirect, United Kingdom). The median difference and its 95% CI were estimated using the Hodges–Lehmann estimator. The criterion for rejection of the null hypothesis was a two-tailed P < 0.05.

All other data, including secondary outcome data, are reported as mean ± SD, median (interquartile range), or number (percentage) of patients. Continuous data were checked for equality of variances; those data that did not meet the assumption of equal variances were reduced to ordinal data and are reported as median (interquartile range). Absolute standardized differences of the baseline characteristics of the patients in the two groups were determined to identify any possible imbalance between the groups. Other data reported as mean ± SD were compared with the unpaired, two-sample t test, data reported as the median (interquartile range) were compared with the Mann–Whitney U test, and data reported as the number of patients (%) were compared using the chi-square test (with Yates correction) unless at least one expected frequency was often less than 5, in which case the Fisher exact probability test was used. Because there was no interest in the change in the variables across time, primary and secondary outcome data were only compared between groups at each time. Given the large number of secondary comparisons, mean differences, median differences, and differences in proportions are reported with their 99% CIs. Median differences and their 99% CIs were estimated using the Hodges–Lehmann estimator. The robust approximation of Miettinen and Nurminen was used to construct the CIs for risk differences. The criterion for rejection of the null hypothesis was a two-tailed P < 0.01 throughout. All statistical analyses except calculation of absolute standardized differences were conducted with StatsDirect (United Kingdom).