No Benefits of Adding Dexmedetomidine, Ketamine, Dexamethasone, and Nerve Blocks to an Established Multimodal Analgesic Regimen after Total Knee Arthroplasty

This prospective, randomized, double-blinded study was registered on www.clinicaltrials.gov (NCT03954379; principal investigator, Vincent Chan, M.D.; registration date April 16, 2019) and received approval from the University Health Network Research and Ethics Board (18-5920, approval date May 28, 2019). This study was conducted at Toronto Western Hospital between June 2019 and December 2020 in accordance with the Declaration of Helsinki principles and followed the Consolidated Standards of Reporting Trials guidelines.⁹ 

Patients aged 18 to 85 yr, with a body mass index of 38 kg/m² or lower, who were having elective primary, unilateral knee arthroplasty were included. Exclusion criteria were revision/bilateral arthroplasty procedures, contraindications to spinal anesthesia, allergy to any of the study medications, neuropathy in the operative extremity, inability or refusal to provide informed consent, uncontrolled cardiac, blood pressure and respiratory diseases, and history of chronic pain unrelated to knee pathology requiring more than 50 mg of oral morphine equivalents per day. All patients were informed about the study procedures and provided written informed consent before randomization.

Patients were randomized into either the control group or study group with a 1:1 allocation ratio using a computer-generated block randomization technique (www.randomization.com) and a block size of 6 patients per block. The investigator who generated the random allocation sequence was the one who enrolled and assigned study patients to the study group per the randomization list. Group allocation was concealed until the day of surgery using numbered sealed opaque envelopes. Each envelope was opened by an attending anesthesiologist, who then prepared the study medications accordingly; this anesthesiologist did not participate further in the study or care of the patient. The patient, surgeon, physiotherapists, acute pain nurses, and investigators performing patient assessment after surgery were blinded to treatment group allocation.

Preoperatively, all patients were administered 650 to 1,000 mg of acetaminophen and 100 to 200 mg of celecoxib orally. Regional anesthesia procedures were performed by attending anesthesiologists or fellows and residents under supervision before surgery in a dedicated block room where noninvasive blood pressure, electrocardiogram, pulse oximetry, supplemental oxygen via face mask, and IV access were established (table 1). Titrated doses of 1 to 2 mg of IV midazolam and 25 to 50 μg of IV fentanyl were administered as needed to provide anxiolysis and analgesia during block performance.

An adductor canal block catheter was inserted in all patients at the level of the midthigh. The adductor canal was identified using a linear 5 to 12 MHz ultrasound probe (Sonosite Edge, USA). The probe was rotated to obtain an oblique view of the superficial femoral artery and adductor canal to allow for a more proximal needle insertion site. After skin sterilization and infiltration with 1 to 2 ml of 2% lidocaine, an 80-mm, 17-gauge Tuohy needle (Arrow StimuCath kit, Teleflex Medical, USA) was advanced in plane in an anterolateral to posteromedial direction until the needle tip was positioned within the adductor canal deep to the vastoadductor membrane using a hydrodissection technique with up to 10 ml of 5% dextrose solution. A 19-gauge catheter was advanced 2 to 3 cm into this fluid “pocket” within the adductor canal and then tunneled subcutaneously to exit the skin proximal to the site of the surgical thigh tourniquet and secured in place with adhesive dressings. After negative aspiration for blood, a bolus of 15 ml of 0.5% ropivacaine (75 mg) in 1:200,000 epinephrine was injected through the tunneled catheter for all patients preoperatively.

After insertion of the adductor canal block catheter, spinal anesthesia was administered with patients in the sitting position. All patients received 2 to 3 ml of 0.5% preservative-free isobaric bupivacaine (total 10 to 15 mg) and 100 μg of intrathecal morphine injected through a 25-gauge Whitacre needle (CHS MED-RX, Canada) at an appropriate space between the L2 and L5 vertebrae. The patients were then placed in a lateral decubitus position with the operative side uppermost, and adequacy of motor and sensory blockade was confirmed before surgery.

The surgical procedures were performed by a group of four experienced surgeons with patients in the supine position, using a standard medial parapatellar surgical approach and a thigh tourniquet. All patients received IV antibiotics and 1 g of tranexamic acid before surgical incision. The patients were sedated using IV propofol infusion (25 to 75 μg kg^–1 min^–1) and supplemented with 1 to 2 mg of IV midazolam or 25 to 50 μg of fentanyl at the discretion of the anesthetic provider, in line with routine institutional practice. The surgeons, but not the anesthetic provider, were blinded to group allocation.

Periarticular local anesthetic infiltration was performed by the surgeon under direct vision using a mixture of 100 ml of 0.2% ropivacaine (total 200 mg) + 0.6 mg epinephrine + 30 mg ketorolac. The posterior capsule was infiltrated using half of the solution before placement of the prosthesis, and the periarticular and superficial soft tissues were infiltrated after the prosthesis was in place using the remaining volume. At wound closure, 3 g of topical tranexamic acid was applied to the surgical site, and all patients received 8 mg of IV dexamethasone and 4 mg of ondansetron for postoperative nausea and vomiting prophylaxis.

After surgery, the patients were managed in the postanesthetic care unit (PACU) by a nurse blinded to group allocation. Opioid analgesia was administered as required to treat pain scores of 4 or higher on an 11-point numerical rating scale (0 to 10 points) and 25 mg of IV dimenhydrinate and/or 4 mg of ondansetron for postoperative nausea and vomiting. The patients were discharged to the ward once they achieved an Aldrete score of 9. Multimodal analgesia on the ward comprised 650 to 1,000 mg of oral acetaminophen every 6 h and 100 to 200 mg of celecoxib every 12 h, supplemented by immediate-release oral oxycodone (5 to 10 mg) or hydromorphone (1 to 2 mg) every 2 h as needed. If oral analgesia was insufficient to control pain, IV patient-controlled analgesia with hydromorphone or morphine was offered for rescue. Patients were followed-up twice daily by the acute pain service team, who titrated opioid dose ranges and transitioned opioids from IV to oral when needed.

The study group received five additional analgesic interventions (table 1).

An ultrasound guided iPACK block in the lateral decubitus position was performed after spinal anesthesia. Thus, the sedated patient was unaware of the procedure. A bolus of 15 ml of 0.5% ropivacaine (75 mg) in 1/200,000 epinephrine was injected through an 80-mm 22-gauge block needle (SonoPlex, Pajunk, Germany) just proximal to the femoral intercondylar fossa.

Patients in the study group received an IV admixture containing 1 μg/kg dexmedetomidine (to a maximum dose of 100 μg) and 0.5 mg/kg ketamine (to a maximum dose of 50 mg) diluted with normal saline to a total volume of 20 ml. The solution was infused over 15 to 20 min without a loading bolus. Supplemental low-dose IV propofol infusion was given as necessary.

The study group received two additional 10-ml boluses of 0.5% ropivacaine in 1/200,000 epinephrine administered through the adductor canal block catheter by one of the study investigators. The first one was in the evening of the day of surgery (postoperative day 0) between 9 and 11 pm, and the second was in the morning of postoperative day 1 between 8 and 10 am before the first physiotherapy session. Patients in the control group received placebo injections of 2 ml of saline at similar times. The catheter was removed after the second postoperative injection.

The study group also received a second dose of 8 mg of IV dexamethasone at 8 am on postoperative day 1, while the control group received 2 ml of IV saline.

All outcomes were collected by blinded researchers. The primary outcome was cumulative opioid consumption in oral morphine equivalents at 24 h after PACU arrival. Secondary analgesic outcomes were 11-point numerical rating scale pain scores in the operative knee, where 0 indicates “no pain,” and 10 indicates “the worst pain imaginable.” These were measured at rest and during movement or physical therapy at the following time points: before surgery, in the PACU, and three times a day on postoperative days 1 and 2 during the patient’s hospital stay: (1) between 8 and 10 am, 2) during physical therapy, and 3) between 8 and 10 pm. After hospital discharge, daily rest and dynamic pain scores were also obtained by phone on postoperative day 2 and at 1, 2, and 6 weeks after surgery. Opioid consumption on days 7, 14, and 42 after discharge was also documented. Other outcomes included the time to first opioid analgesic request, time to reach physiotherapy criteria for hospital discharge (i.e., ability to ambulate independently from the bed to the bathroom, walk along a hallway unassisted with walker and climb stairs safely), and length of hospital stay (defined as the number of days from admission to discharge). Quality of recovery (QoR) was assessed using a validated QoR-15 tool¹⁰  immediately before surgery and 24 h, 48 h, and 2 weeks after surgery.

Rehabilitation milestones were evaluated with the Timed Up and Go test (time it takes a subject to stand up from a standard-height armchair, walk 3 m, walk back to the chair, and sit down),^11,12  the distance walked during physiotherapy, and both active and passive joint range of motion, defined as knee flexion from neutral (0°) to maximum flexion. The Timed Up and Go test, distance walked, and range of motion were measured before surgery and on postoperative days 1 and 2 (unless the patient was discharged earlier). Patient satisfaction was assessed at the time of discharge, and adverse events (nausea, vomiting, sedation, hypotension, urinary retention, hyperglycemia, foot/ankle muscle weakness, and symptoms of local anesthetic systemic toxicity) were recorded. Motor function at the ankle was categorized as follows: 0 indicates no power, 1 indicates decreased power (any movement without resistance), and 2 indicates normal power (complete movement against resistance). Hypotension was defined as a decline of systolic blood pressure of 25% or greater from baseline or blood pressure lower than 90 mmHg requiring treatment.

The primary outcome was cumulative postoperative opioid consumption in the first 24 h after surgery. Based on past institutional clinical data, we assumed that patients in the control group would require 80 ± 40 mg (mean ± SD; 95% CI, 67.1 to 92.9 mg) oral morphine equivalents in 24 h, and the new interventions would result in a 33% opioid reduction in the study group (i.e., minimum clinically important difference of 26.7 mg; 95% CI, 22.3 to 31.1 mg); 37 patients would be needed per group based on a power analysis using a 5% type I error estimate and 80% power within a two-tailed t test. To allow for a 5% drop out, we enrolled 39 patients per group (78 in total).

The data were analyzed with SPSS 23.0 for Mac (IBM, USA). Normality of data distribution was tested using the Shapiro–Wilk test. The data with normal distribution were reported as means (SD), and data that were skewed are described as medians (interquartile range). For data that are normally distributed, the independent Student’s t test was used to analyze for differences between groups, and the Mann–Whitney U test was used for analysis of differences between continuous variables with skewed distribution. The differences of the medians and 95% CI were estimated using the Hodges–Lehman method. Categorical variables were described as numbers (percentages) or proportions and were compared using the chi-square or Fisher’s exact test where appropriate. P < 0.05 was designated as statistically significant. All hypothesis testing was two-tailed.

A total of 273 patients were screened, of which 138 met inclusion criteria. Of these, 50 refused to consent, 4 withdrew consent on the day of surgery, and 6 cases were postponed due to the COVID-19 pandemic. In the end, 78 patients were enrolled and randomized into 2 study groups (39 patients per group; fig. 1), and primary analysis was conducted on the data of 78 patients.

All subjects had similar baseline demographics (table 2). One patient in the study group required general anesthesia due to inadequate block height of the spinal anesthetic. Sixty-two patients (31 in each group) were discharged home on postoperative day 1. Two patients (1 in each group) were lost to follow-up at 6 weeks (fig. 1).

The primary outcome of cumulative 24-h postoperative opioid consumption was similar between the two groups (means ± SD; study: 23.7 ± 18.0 mg vs. control: 29.3 ± 18.7 mg oral morphine equivalents; mean difference [95% CI], –5.6 mg [–2.7 to 13.9]; P = 0.189; table 3). There was also no difference between groups in opioid consumption at any other assessment time points, up to 6 weeks after surgery (fig. 2; table 3). No patient required rescue IV patient-controlled analgesia. Intraoperative fentanyl was given to 27 patients (69.2%) in the control group and 2 patients (5.1%) in the study group to complement sedation.

For secondary analgesic outcomes, there was no difference in the time to first opioid dose, (study: 635 ± 337 min vs. control: 574 ± 347 min; P = 0.437). In addition, there were no statistically significant differences in pain scores at rest at any time point up to 6 weeks after surgery (table 4). The lack of between-group differences in all primary and secondary analgesic outcomes persisted when adjusted for intraoperative fentanyl dose as a covariate. (A table showing the key analgesic outcomes analyzed by study group allocation and adjusted for intraoperative opioid dose can be found in the Supplemental Digital Content, https://links.lww.com/ALN/C880.)

After surgery, the range of movement, both active and passive, decreased from baseline similarly in both groups, but all patients managed to walk a mean distance of greater than 50 m on postoperative day 1 (table 5). No statistically significant differences were found between groups for Timed Up and Go test, range of movement, distance walked, or QoR-15 at baseline and on postoperative day 1. Neither was the time to reach discharge criteria/total in-hospital stay (table 5). Patient satisfaction was equally high in both groups.

The incidence of side effects was similar in both groups except for hypotension in the PACU (study: 64.1% vs. control: 33.3%, P = 0.010; table 6). Perioperative heart rate was comparable and remained within normal ranges (50 to 100 beats/min) for both groups. Plantar flexion was also more frequently impaired in the study group on postoperative day 0 (46.1% vs. 17.9%; table 6), although no falls were reported. Blood glucose level obtained in the morning of postoperative day 1 was not statistically different between groups and remained within the normal range (5 to 9 mM) at all time points despite a second IV dexamethasone dose (table 5). No patients experienced symptoms of local anesthetic systemic toxicity.

In this study, we sought to examine the opioid-sparing effect of a comprehensive multimodal analgesic regimen combining multiple systemic and regional modalities that have been shown to provide analgesic benefit in total knee arthroplasty. Our current institutional perioperative analgesic regimen comprises a single injection adductor canal block, low-dose intrathecal morphine (100 μg), intraoperative IV dexamethasone (8 mg), periarticular local anesthetic infiltration, and round-the-clock oral acetaminophen and celecoxib, with immediate-release opioids as needed. We studied the addition of five recently described analgesic modalities to this regimen: a preoperative iPACK block,¹³  an intraoperative IV infusion of low-dose dexmedetomidine^5,14–16  (1 μg/kg) and ketamine¹⁷  (0.5 mg/kg), a second dose of IV dexamethasone¹⁸  (8 mg) on postoperative day 1, and two additional adductor canal block¹⁹  bolus injections on postoperative days 0 and 1.

Contrary to expectations and to reports of analgesic benefits of each novel intervention when compared to placebo or no intervention, this additional bundle of interventions did not further reduce opioid consumption or pain scores in the first 24 to 48 h after total knee arthroplasty. Neither did they improve postoperative functional outcomes, quality of recovery, patient satisfaction, or longer-term pain and analgesic outcomes up to 6 weeks after surgery. Our results suggest that the standard multimodal analgesic regimen currently prescribed for the control group is rather robust. Thus, the therapeutic value of adding more analgesic interventions is limited with diminishing return.

Compared with placebo, perioperative ketamine administration decreases pain scores²⁰  and opioid consumption after total knee arthroplasty.⁶  The reported effective analgesic dose for IV ketamine varies widely—an initial 0.05 to 1 mg/kg bolus followed by an infusion of 1 to 16.7 μg/kg/min during and after surgery. Similarly, intraoperative sedation with low-dose IV dexmedetomidine (0.1 to 1 μg/kg bolus followed by 0.1 to 0.7 μg kg^–1 h^–1 infusion) has been shown to successfully reduce pain, postoperative nausea and vomiting, and delirium⁵  compared with placebo after total knee arthroplasty.^15,16 

Dexamethasone is another drug that has analgesic properties in addition to its antiemetic effect for a myriad of surgical procedures including joint arthroplasties.²¹  In the total knee arthroplasty population, a single perioperative IV dose (greater than 0.1 mg/kg) can be opioid-sparing and pain-relieving, and a second 10-mg dose after surgery may further improve postoperative nausea and vomiting, range of motion, and patient satisfaction.²²  Although the optimal dose and dosing interval remain unknown, perioperative IV dexamethasone (total of less than 20 mg) significantly reduces total opioid consumption and postoperative pain after total knee arthroplasty.¹⁸  Furthermore, perioperative dexamethasone administration limited to one or two doses appears safe with no reported increase in surgical site infection or sustained hyperglycemia.

Among regional analgesic modalities, adductor canal block has become a motor-sparing alternative to femoral nerve block to control anterior knee pain after total knee arthroplasty. Several randomized controlled trials reported that continuous adductor canal block could be superior to single-shot adductor canal block by further improving pain scores and time to rescue analgesia.²³  However, whether continuous adductor canal block can further improve postoperative rehabilitation and other functional outcomes is not clear.²⁴  Similarly, an iPACK block has been shown to reduce the incidence of posterior knee pain after total knee arthroplasty compared to a sham block.¹³  However, its analgesic contribution in the presence of periarticular local anesthesia infiltration remains debatable because conceivably, local anesthesia infiltration into the posterior capsule of the knee likely overlaps with the analgesic coverage of the iPACK block (i.e., terminal branches of the genicular nerves and popliteal plexus). Thus, the addition of an iPACK block to periarticular local anesthesia infiltration may be redundant, as suggested by recent randomized controlled trial data,¹⁹  a meta-analysis,²⁵  and the results of our current study. Consistent with previous studies, we also failed to show that the addition of an iPACK block improves functional outcomes or quality of recovery after total knee arthroplasty, irrespective of pain score differences in the first 24 h.¹³ 

Both the study and control groups in the current study received a single IV dexamethasone dose, single-shot adductor canal block, intrathecal morphine (100 μg), and periarticular local anesthesia infiltration for postoperative analgesia (table 1). The synergistic analgesic effect of combining periarticular local anesthesia infiltration and single-dose adductor canal block can significantly delay rescue analgesia,²⁶  reduce cumulative opioid requirements,²⁷  and improve range of motion²⁸  and early discharge after total knee arthroplasty.²⁹  The addition of intrathecal morphine to adductor canal block and periarticular local anesthesia infiltration can further improve analgesia and reduce postoperative opioid requirements.³⁰  The intrathecal morphine dose selected for this study (100 μg) appears optimal for elderly patients undergoing total knee arthroplasty, balancing its analgesic effect with potential adverse effects³¹  with no increase in the incidence of urinary retention, nausea, vomiting, or pruritus.³⁰  Furthermore, opioid-related respiratory complications are infrequent even in patients with obstructive sleep apnea.³²  Findings of the current study suggest that analgesia resulting from a combination of these interventions is rather robust, with little benefit derived from additional analgesic interventions.

Interestingly, we found that opioid requirement and pain scores were significantly lower in the control group of the current study as compared to an almost identical treatment group (single-shot adductor canal block plus 100 μg of intrathecal morphine and periarticular local anesthesia infiltration) in an earlier randomized controlled trial we conducted 4 years ago. For the past study,³⁰  the 24-h IV morphine equivalent requirement was 34 ± 21 mg (mean ± SD), which equates to 102 ± 63 mg oral morphine equivalents. For the current study, the 24 h oral morphine equivalent consumption was 29.3 ± 18.7 mg (mean ± SD), representing a ~70% reduction, despite no significant change in the total knee arthroplasty perioperative care pathway, surgical/anesthetic technique, or medical staff. The median 24-h pain score at rest was likewise significantly lower, 1 (0 to 3) in the current study versus 5 (3 to 7) in the earlier study, corresponding to a five-fold reduction.

Significant improvement in pain scores and opioid consumption observed in the current study results possibly from two major changes in patient management made over the past 4 yr: intraoperative tranexamic acid and IV dexamethasone (8 mg). All patients received 1 g of IV tranexamic acid in the beginning and 3 g topical at the end of surgery. Conceivably, not only does tranexamic acid reduce major bleeding and transfusion requirements,³³  it can also potentially decrease pain and opioid consumption through a reduction in inflammatory surgical response,^34,35  articular swelling,³⁶  and hematoma within the wound. Blood conservation with tranexamic acid also prevents postoperative anemia and associated fatigue, resulting in expedited rehabilitation after total knee arthroplasty.^30,34,35,37  Some other contributing factors to pain relief and enhanced recovery were more consistent use of cryotherapy and early initiation of postoperative ambulation.

More patients in the study group developed hypotension (i.e., systolic blood pressure of 90 mmHg or lower) in the PACU (table 6). This is likely the effect of IV dexmedetomidine through its central and peripheral presynaptic α-2 adrenoceptor–mediated sympatholysis and vasodilation.¹⁴  Hypotension was, however, transient and quickly responded to IV fluid and phenylephrine rescue doses with no delay in PACU discharge. In addition, hypotension happened in the PACU and not during surgery. This is consistent with previous reports for timing of dexmedetomidine induced hypotension, which typically occurs 60 to 330 min after IV administration.^38,39 

Our study has several limitations. First, our control group received intrathecal morphine and periarticular local anesthesia infiltration. These interventions may not be possible in other institutions due to nursing monitoring policy and surgeon’s preference, which limits the extrapolation of our results to these scenarios. Second, the optimal analgesic dose and duration of administration for IV dexmedetomidine⁵  and ketamine¹⁷  have not been established. Thus, the single doses administered during surgery in our study may be suboptimal. Similarly, our study was not powered to detect the impact of these multimodal analgesic components on the incidence of chronic postsurgical pain beyond 6 weeks. Third, we have only assessed muscle function qualitatively without using dynamometry. Currently, unintended local anesthetic spread toward the sciatic nerve or its branches with periarticular local anesthesia infiltration and/or iPACK cannot be ruled out. Fourth, we used repeated local anesthetic boluses for continuous adductor canal block rather than an infusion and have only extended adductor canal block to the morning of postoperative day 1 while many patients were discharged home.

In the presence of periarticular local anesthesia infiltration, intrathecal morphine, single-shot adductor canal block, and dexamethasone, the addition of iPACK block, IV dexmedetomidine, IV ketamine, an additional IV dexamethasone dose and repeated adductor canal block injections failed to further reduce opioid consumption or pain scores or to improve functional outcomes after total knee arthroplasty.

The authors thank Mehdi Soheili, M.B.B.S. (Toronto Western Hospital, University Health Network, Toronto, Ontario, Canada) for assistance with data collection.

Supported by a grant from the Alternate Funding Plan Innovation Fund, Ontario Ministry of Health and Long-Term Care (Toronto, Ontario, Canada); merit award support from the Department of Anesthesiology and Pain Medicine, University of Toronto (Toronto, Ontario, Canada; to Drs. Chan and Perlas); research time support from the Department of Anesthesia and Pain Management, Toronto Western Hospital, University Health Network (Toronto, Ontario, Canada; to Drs. Chan, Chin, Perlas, and Bhatia); and divisional research support from Smith and Nephew and DePuy Synthes (Ontario, Canada; to Dr. Gandhi).

The authors declare no competing interests.

Full protocol available at: vincent.chan@uhn.ca. Raw data available at: vincent.chan@uhn.ca.

Supplemental table, https://links.lww.com/ALN/C880