Spinal Versus General Anesthesia for the Outpatient Unicompartmental Knee Arthroplasty: A Randomized Controlled Trial

Purpose: To compare the early clinical outcomes between anesthetic techniques in outpatient unicompartmental knee arthroplasty (UKA) and investigate factors responsible for delayed discharge. Method: A randomized controlled trial was conducted on 64 consecutive and unselected patients who underwent primary, medial UKA and randomly divided into two groups:32 patients with spinal anaesthesia(Group A) and 32 with general anaesthesia(Group B). There were no signicant differences in gender, age, body mass index (BMI), the hospital for Special Surgery knee score (HSS), and American Society of Anesthesiology(ASA) class between the two groups (P >0.05). Results (cid:0) Totally 64 patients were involved, including 20 males and 42 females, with an average age of 60.4 ± 3.8 year-old. The length of stay(LOS) and total expense were signicantly less in group A than those in group B (P < 0.05). There was no signicant difference between two groups on the discharge day and the reasons for not discharge(P>0.05). Lack of mobilization (84%), pain (23%), and PONV (17%) were main reasons for patients not meeting the discharge criteria on DOS (Table 2). Patients with spinal anaesthesia underwent a longer surgery time than those with GA.(P<0.05); Patients in the Group A had a signicantly earlier oral uid intake and walking unaided(P<0.05); Pain occurred signicantly earlier in Group B(P<0.05). Conclusion: In patients who underwent outpatient UKA, spinal anaesthesia is signicantly superior to general anaesthesia with shorter discharge time and lower cost. With strict patient selection and safe anaesthesia, outpatient UKA could attain a good clinical outcome with both anaesthesia techniques.

remains controversial. Several studies suggested that SA reduces the rate of complications and healthcare resource utilization [13] [14]. But there were few studies speci cally examined those patients who would be eligible to be outpatient UKA candidates.
Thus, we conducted a randomized controlled trial in a single institution. The purpose of this study was to compare the clinical e cacy and early complications between spinal anaesthesia(SA) and general anaesthesia(GA) in patients who underwent outpatient unicompartmental knee replacement.

Methods
The inclusion criteria were as follows: (1)  June 2018, 64 consecutive and unselected patients who underwent primary, medial UKA in our institution were involved in our study. All patients were operated with cementless mobile bearing Oxford UKA system with a minimal invasive technique as described by the manufacturer. A single senior surgeon performed more than 200 primary UKAs, in which the outpatient procedures was employed in 26.7% of cases. After assessed by the senior surgeon, patients who met the inclusion criteria for day-time UKA operation could be involved in our study. 64 patients were equally divided into two groups using random number method:32 patients with spinal anaesthesia(group A) and 32 with general anaesthesia(group B).
All patients participating in the trial gave written informed consent.

Patients and procedures
Patients who met the inclusion criteria could be scheduled for the operation. Written informed consent was obtained before inclusion in the study.
Two weeks preoperatively: Comprehensive educational material would be provided by our surgeon, and the degree of preoperative pain was evaluated by visual analog score (VAS). Patients with VAS score ≥ 4 either at rest or in motion were prescribed with oral analgesia: NSAIDs (celecoxib), pregabalin, or buprenorphine for external use.
One week preoperatively: Senior surgeon and anaesthesiologist should reassess and con rm if patient met the inclusion/exclusion criteria. Patients were contacted by the protocol nursing, who would provide the introductions about the surgical procedure, anaesthesia techniques, risks and recovery procedures. Further preoperative education, including strict prohibition of smoking and alcohol, ankle pump exercise and quadriceps muscle contraction training, schedule of rehabilitation plan and daily goals of functional exercise after operation should also be informed to help them establish a reasonable psychological expectation, and make them fully understand and participate in recovery process.
Three days preoperatively: Chest X-ray, electrocardiogram and other auxiliary examinations should be implemented. Patients with anemia (HB: female < 110g/L, male < 120g/L) or hypoalbuminemia (< 30 g/L) should be delayed until their HB/albumin returned to normal level; For patients with hypertension and diabetes mellitus, the blood glucose and blood pressure should be adjusted to the normal range, then the operation could be scheduled again.
Day of operation: When arrived at surgical department, all patients were randomized arranged into one of two groups(Group A or Group B). After con rmed no absolute contraindication, related documents could be signed. Fasting solid food 6 hours before induction; 2 hours before induction, 600 ~ 800 ml high concentration carbohydrate were given orally. Prophylactic antibiotic and a single dose (1 g) of tranexamic acid was given intravenously 30 minutes before skin incision.

Types of anaesthesia
The senior surgeon performed THA on all patients by using the same surgical technique and perioperative protocol for both groups. And all anesthetic procedures(SA and GA) were under the management of an experienced anesthetist. All patients were given femoral nerve block (FNB) preoperatively.

Spinal anaesthesia
Under aseptic conditions, spinal anaesthesia was performed in a midline approach. The anaesthetist identi ed the L3 to L4 interspace in the lateral decubitus position. After local in ltration anaesthesia(LIA), the subarachnoid space was sought. Free-owing cerebrospinal uid(CSF) and no re ux of blood indicated correct needle position, then 2 mL 0.5% hyperbaric bupivacaine was injected in the absence of paraesthesia. Blood pressure and heart rate should be carefully observed. After induction patients were immediately transferred to the operating room and positioned. Optionally, by request, patients were given propofol until a mild level of sedation was achieved (Richmond Agitation Sedation Scale 1 to 2). Sensory block and motor block were evaluated 3 min after injection. Sensory block was tested using a toothpick to elicit pinprick starting caudally and moving cephalad.

General anaesthesia
After pre-oxygenation, sufentanil (0.2 ~ 0.5mg/kg) and propofol (2 to 3 mg/kg) were intravenously injected. Laryngeal mask (LM) was provided according to body weight. For di cult or impossible ventilation that could not be resolved by replacement or re-insertion of the LM, an endotracheal tube would be provided with an inner diameter of 7.0 mm for women and 7.5 mm for men. The nondepolarising neuromuscular blocking agent atracurium was administered for orotracheal intubation.
Ventilation was managed by pressure-controlled mode at a frequency to achieve an end tidal PCO 2 of 35 40 mmHg. Anaesthesia was maintained by continuous propofol infusion through a perfusor pump, with the infusion rate set 8 ~ 12 mg kg/body weight/h, and could be adjusted according to the individual condition. To ensure prompt recovery, the infusion of propofol was stopped 5 min before completion of skin closure. If necessary, the anaesthetist would administer repetitive sufentanil doses according to clinical need. Indicated by the Apfel score, patients at risk of PONV received granisetron (1 mg) and dexamethasone (4 mg) as prophylaxis.

Surgical technique
After a satisfactory anaesthesia, all patients were put in hanging leg (HL) position, then an iodised incision drape (Ioban 3M®) was applied after conventional draping. All the surgical incisions were performed by minimally invasive surgical incision near the patella. Oxford UKA prostheses were utilized in all patients. The senior doctor with a xed surgical team performed operation according to the Oxford III UKA system operation manufacturer. Before skin suture, all patients received local in ltration analgesia (LIA) with 20ml "cocktail" (ropivacaine 200mg, epinephrine 5mg, triamcinolone acetonide 5mg) injected in the posterior capsule, periarticular tissues, and subcutaneous tissue. No intra-articular suction drain or urinary catheter was placed in all patients. The uoroscopy with a C-arm X-ray machine should be conducted to ascertain that the prostheses were satisfactory. Once the procedure was nished, all patients were immediately transferred into the postanesthesia care unit(PACU). After pain and PONV were controlled, patients were orally provided with 200ml high concentration carbohydrate and sent back to the ward after active bleeding disappeared.

Postoperative management
All patients were treated by multimodal analgesia once sent back to ward, in combination with PCIA (Patient control intravenous anesthesia) pump. 2 hours after surgery, patients were orally given celecoxib (200mg/12h), Tylenol (650mg/8h) and gabapentin (300mg/8h). If the pain symptoms were still obvious (VAS score ≥ 5, and the pain affected postoperative functional exercise), tramadol and buprenorphine were given additionally. 1 h, 3 h and 24 h postoperatively, single dose of tranexamic acid (1g) was given separately.
Postoperative functional exercise was initiated and evaluated by the therapists in day-time operation team. After the effect of anesthesia subsided, the functional exercise should be started, including CPM, ankle pump, knee exion and extension exercise. 2 hours after operation, all patients were required to sit and stand up. Then with the walking aid, they were required to try to walk independently with adjustable lower limb brace. The therapists ensured that all patients can walk safely and, meanwhile, assessed their gait and pain level. 4 hours after the operation, they walked on the oor again and tried to walk up and down stairs independently.
The discharge criteria were as follows: 1) Mobilizes in and out of bed independently; 2) walk 30 m independently; 3) walk up and down stairs independently; 4) obvious relief of pain; 5) has minimal or no wound wound exudation; 6) HB ≥ 90; 7) patient feel comfortable and con dent to go home.
Only patients met the discharge criteria can they be discharged. Anticoagulation with low molecular weight heparin was given every day from 12 hours after operation, and rivaroxaban was given orally after discharge until 3 weeks after operation; All discharged patients were maintained contact via Wechat, a social networking application (Tencent Inc., China) and telephone. They were followed up by a doctor on regular basis.
The primary outcome measures included the length of stay (LOS), the total hospitalization cost. Early post-operative complications and reasons for delayed discharge were recorded. And the operation time, intraoperative blood loss and VAS at the time of discharge were recorded. Patients were followed up regularly1, 3, 6, 12 month(s) and then annually after the operation, to record the complications, including periprosthetic infection, thromboembolic event, prosthesis dislocation and aseptic loosening. And the hospital for Special Surgery knee score (HSS) and Oxford knee score (OKS) were used to assess the knee function.

Study size and statistical methods
The sample size was calculated according to the primary outcome (the length of stay) using the SAS software package (version 9.0, SAS Institute Inc., Iowa, IA, USA). We presumed that a difference of 2h between the two groups would be clinically important. Assuming a SD of 4h, a 5% signi cance level and a 90% power using a two-sided test, we obtained a total sample size of n=22 for each group. As we were not able to ensure that these data would be normally distributed, we enhanced the sample size to n= 32 patients per group. SPSS 22.0 statistical software was employed for statistical analysis. Measurement data were expressed as mean±standard deviation (SD). Repeated analysis of variance was used for comparison between time points before and after surgery in the groups. Paired t-test was used for pairwise comparison. Independent sample t test was utilized for comparison between groups. Comparison of count data between groups was made by using χ 2 test. Inspection level was set at α = 0.05.

Results
A total of 64 patients were involved, consisting of 20 males and 42 females, with an average age of 60.4 ± 3.8 year-old at the time of surgery. The demographics are shown in Table 1. There existed no signi cant differences in demographics between the two groups, including sex (P = 0.59), age (P = 0.96), BMI (P = 0.83), ASA (P = 0.74), and preoperative HSS (P = 0.47). The length of stay(LOS) and total cost were signi cantly less in group A than in group B (P < 0.05). The patients in group A were followed up for 6-16 months (mean time: 9.7 months) and 6-18 months (mean time: 9.3 months) in group B. In terms of postoperative satisfaction, the numeric rating scales (NRS) of group A was (8.87 ± 0.83), which was slightly higher than that of group B (8.59 ± 0.84), and the difference was not statistically signi cant (P > 0.05) (see Table 2). Lack of mobilization (84%), pain (23%), and PONV (17%) were the main reasons for patients not meeting the discharge criteria on DOS (Table 3). There was no signi cant difference between two groups on the reasons for not discharge on DOS (P>0.05).  Table 4). Time spans marked with an M are measured from the end of induction; median (range) are given. Figure 1 showed the comparison of HSS score and OKS score between the two groups preoperatively and post-operatively: the HSS score and OKS score of both groups were signi cantly improved at 1, 3 and 6 months after operation, and the differences between two groups were not statistically signi cant (P > 0.05).

Tab.4 Duration of procedures and recovery times from start of intrathecal injection ()
There was no signi cant difference in VAS scores between the two groups at the time of discharge (P > 0.05), suggesting that while the LOS was signi cantly reduced, the pain symptoms could still be well controlled by both types of analgesia. At latest follow-up, no periprosthetic infection, aseptic loosening of prosthesis, deep vein thrombosis (DVT), pulmonary infection and other complications were found in all patients.( see Fig.2)

Discussion
The purpose of this randomized controlled trial was to compare the early clinical outcomes between anesthetic techniques in outpatient unicompartmental knee arthroplasty (UKA) and investigate factors responsible for delayed discharged. The results of our study showed that both spinal and general anesthesia can attain satisfactory short-term outcomes in outpatient UKA based on ERAS pathway and multimodal analgesia. Meanwhile, spinal anesthesia might be a more cost-effective than general anesthesia, which could reduce the LOS as well.
Compared to total knee arthroplasty (TKA), unicompartmental knee arthroplasty (UKA) is associated with an earlier recovery, less postoperative pain, lower morbidity and mortality, and a greater "feel" of a normal knee, which made UKA a better surgical option for patients with end-stage medial compartment osteoarthritis [15] [16]. Because of not increasing postoperative complications and risk of re-admission, outpatient joint arthroplasty have been widely performed in recent years [17]. A shorter LOS could effectively save the cost of hospitalization and reduce the risk of nosocomial infection [18]. Current studies have con rmed the safety and e cacy of outpatient UKA [7][9] [13]. Meanwhile, excellent anesthesia and intraoperative monitoring were necessary for ensuring the safety and e cacy of the operation. Though both spinal and general anaesthesia were appropriate techniques for outpatient UKA, it remained controversial which type of anaesthesia would attain a better clinical outcome.
A randomised controlled study conducted by Gebhardt et al [19] indicated that in patients of outpatient knee surgery, spinal anaesthesia can lead to a signi cantly earlier discharge and reduce total cost compared with general anaesthesia. A large national sample study by Memtsoudis et al [20] also showed that the utilization of neuraxial versus general anesthesia for primary joint arthroplasty was associated with superior perioperative outcomes. In our study, compare with GA, patients with SA did have a short mean LOS and lower cost of hospitalization. And both types of anaesthesia did not increase the incidence of post-operative complications. As for 30-day mortality, all patients in our study did not suffer any major post-operative complications at the latest follow-up.
Limited evidence suggested that SA could reduce total blood loss. A study of Perlas et al [21] suggested that a strong association between spinal anesthesia and lower 30-day mortality, as well as a less likely to experience major blood loss after elective joint replacement surgery. Poeran et al [22] provided incremental evidence of the potential effectiveness and safety of tranexamic acid in patients requiring orthopedic surgery, which indicated that tranexamic acid can effectively reduce the need for blood transfusions while not increasing the risk of complications, including thromboembolic events and renal failure. We also believed that the application of anti brinolytics might effectively eliminate the differences. In our study, tranexamic acid were routinely used before and after the surgical procedure. And the result showed no signi cant difference between two groups in term of total blood loss, with routine use of anticoagulants. And at latest follow-up, all patients did not suffer venous thromboembolic events (VTEs), which also indicated the safety of tranexamic acid.
In this study, a total of 13 patients (20%) completed DOS discharge, which was slightly lower when compared with most of current studies [9][10] [13]. This may because of the criteria for discharge might be relatively conservative considering the safety of patients and inadequate experience of outpatient joint arthroplasty in our institution. Among all 51 patients who could not be discharged on DOS, lack of mobilization (84%), pain (23%), and PONV (17%) were the main reasons for delayed discharge. A randomized controlled trial by Hoorntje et al [23] suggested that patients who lack or afraid of doing selfdirected exercise should not discharge on DOS. In our study, 43 patients were not discharged on the DOS due to the lack of mobilization, and the incidence in group A (24/32, 75.0%) was slightly higher than those in group B (19/32, 59.4%) without statistically signi cant difference (χ2 =0.741, P=0.245). And after recovery guidance by our therapists and application of CPM and other rehabilitation devices, most of them (37 cases) were successfully discharged in POD 1.
In this study, a total of 12 patients could not be discharged from hospital on DOS because of obvious pain, and the overall incidence was not signi cantly between group A (7/32, 21.9%) and group B (5/32, 15.6%)(χ2 =0.355, P=0.384). We believed that while both anesthetic techniques had no difference in postoperative pain management, attention should be paid to achieve a good pain control for safe discharge of day-time surgery. Many studies have shown that adequate preoperative education, preemptive oral analgesia, nerve block and multimodal analgesia can effectively ensure the safety of day-time surgery [27]. A case-controlled study of hoorntje et al [25] suggested that effective pain management of outpatient UKA can reduce the anxiety and depression of patients during the perioperative period, enable patients to actively participate in rehabilitation training, achieve better clinical e cacy, and effectively improve patients' surgical satisfaction. On the other hand, a study of Kort et al [26] showed that the application of multimodal analgesia in UKA surgery can not only improve the perioperative analgesic e ciency, but also effectively reduce the use of opioids. In our study, at the time of discharge, the VAS score of both groups had no signi cant difference (P>0.05). However, when considering rst occurrence of pain after surgery, patients with general anaesthesia were signi cant earlier than those with spinal anaesthesia (P>0.05), which might affect the e cacy of early function exercise. And patients in Group A also had a shorter LOS than Group A resulted from a better early function exercise. All patients were managed by multimodal analgesia, and only 5 patients (2 in Group A, 3 in Group B) were given tramadol and buprenorphine additionally due to obvious pain. Meanwhile, Turner et al. [27] indicated that compared with femoral nerve block(FNB), adductor canal block can achieve the same analgesic effect and has less impact on strength quadriceps, which is bene cial for patients to start early mobilization and recovery on DOS, and thus more suitable for outpatient UKA.
Poor quality of recovery and post-operative complications might increase the LOS and has an obvious impact on total cost. A cohort study by Baldawi et al [28] suggested that neuraxial anesthesia had reduced postoperative complications and LOS compared to patients who underwent general anesthesia. Chaturvedi et al [32] also indicated that in patients with a very high comorbidity burden, neuraxial anesthesia is a reasonable alternative to general anesthesia. However, it is uncertain whether those results are pertinent to outpatient total joint replacements considering that they were inpatient surgical procedures with total joint replacement. Our study speci cally investigated the association between anesthetic techniques and early outcomes after outpatient UKA. In our study, all patients in two group did not suffer any major complication in their latest follow-up. Meanwhile,there was no signi cant difference in HSS and OKS score between two groups at 1, 3 and 6 months after operation, suggested that both anaesthesia techniques could attain satisfactory clinical outcomes.
There are still some limitations of this study. First, due to certain inherent features, the sample of our study may relatively be small. But the researchers, outcome assessors and patients were all blind to treatment groups, which might eliminated some biases. Second, all patients selected were mainly chosen by the surgeon, and this might result in some selection biases even if the senior surgeon did not know which group those patient were eventually divided into. Additionally, the sample size of the study was relatively small and both groups had very low complication rates, which rendered it hard to identify the association of the procedure with complications. These limitations could be avoided in future larger and long-term follow-up well-designed prospective randomized controlled studies.

Conclusion
Overall, our results suggest that, while both anaesthesia techniques for outpatient unicompartmental knee arthroplasty (UKA) are well tolerated and reliable, spinal anaesthesia is signi cantly superior to general anaesthesia with shorter time to discharge and lower cost of hospitalization. With strict patient selection and safe application of anaesthesia technique, outpatient UKA could attain a good clinical outcome. Tumor cell uptake and cytotoxicity a Fluorescence images and b ow cytometry analysis of K7M2 osteosarcoma cells after incubation with DOX or DDPNPDOX&R848 for 1 and 6 h.   Antitumor effect in vivo a Experimental schedule for tumor induction and drug treatments. b Tumor volume, c body weight, and d survival rate of mice after various treatments in seven groups. e 3D reconstructed images of the tibia in the orthotopic osteosarcoma inhibition experiment performed using micro-CT and h BV/TV and i BMD in the ROIs of the tibia in the orthotopic osteosarcoma inhibition experiment. f H&E staining analysis of lung metastasis in mice in different groups. g TIR of DOX + DDP, DOX + DDP + R848, DDPNPDOX, NPR848, and DDPNPDOX&R848 in this experiment. Data are presented as the mean ± SD (n = 8 for b -d and g; n = 3 for h -i; *P < 0.05, **P < 0.01, ***P < 0.001).

Figure 5
Immune response in vivo a Immuno uorescence (CD86, iNOS, and CD8) analyses and b -d the relative optical densities of tumor tissue sections after different treatments. Scale bars: 50 µm. Data are presented as the mean ± SD (n = 3 for b -d; **P < 0.01).