This study has been reported in line with Consolidated Standards of Reporting Trials (CONSORT 2010) Guidelines. This single-center, double-blind, randomized controlled trial was approved by the Biomedical Ethics Committee of Sichuan University West China Hospital (date February 8, 2022/no. 2021–1699). All patients who were candidates for cemented TKA at our hospital from February 2022 to June 2022 were considered for inclusion. Patients were included if they underwent TKA for knee osteoarthritis in our hospital and showed a flexion-contracture deformity  of < 20°, varus or valgus deformity of < 20° [26]. Each patient provided written informed consent before surgery.

Patients were excluded if they had a history of knee infection, had a level of hemoglobin of < 100 g/L or coagulopathy, were using anticoagulants or antiplatelet drugs, had a body mass index (BMI) of > 40 kg/m2, or refused to participate in the study.

Prior to TKA, patients were randomly assigned 1:1 to a group that received a tourniquet during TKA or to a group that did not. Random numbers were generated using a computer algorithm and sealed in opaque envelopes. Each patient was asked to select an envelope, inside which their group allocation was indicated. Observers who collected data after surgery were not involved in the surgery and were unaware of the group allocation.

All patients in our study received general anesthesia involving induction with midazolam (0.02–0.03 mg/kg), propofol (1–2 mg/kg), sulfentanyl (0.3–0.5 μg/kg), and rocuronium (0.6–1.0 mg/kg), which were delivered by intravenous bolus injection. Exceptions were patients older than 60 years, who did not receive midazolam. Anesthesia was maintained through continuous intravenous infusion of remifentanil (0.1–0.2 μg/kg·min) and continuous inhalation of sevoflurane. Rocuronium was added every 40–60 min at 25–33% of the induction dose. Sulfentanyl (5ug) was added every hour.

Intraoperative blood pressure was recorded every 3 min using an electrocardiogram and an upper-arm sphygmomanometer. Intraoperative blood pressure was maintained at baseline in the tourniquet group, or at approximately 70% of baseline in the non-tourniquet group. The target blood pressure was achieved through intravenous injection of m-hydroxylamine, ephedrine, and nicardipine.

All patients received antibiotics at 0.5–2 h before surgery. At 10 min before surgery, all patients received intravenous dexamethasone (10 mg) and intravenous tranexamic acid (60 mg/kg). Previous studies have confirmed the efficacy and safety of a preoperative high-dose (60 mg/kg) combined with postoperative multiple-dose tranexamic acid sequential application regimen [27,28,29,30,31,32]. Immediately before surgery, a tourniquet was applied at the base of the thigh in the tourniquet group and inflated to 100 mmHg above baseline systolic pressure. All surgeries were conducted by the same team of surgeons at our hospital, who had more than 10 years of experience in total joint arthroplasty, and were performed using a standardized medial parapatellar approach. All patients received the same type of cemented posterior-stabilized prosthesis (DePuy Synthes, Johnson and Johnson, New Brunswick, USA). During surgery, intramedullary guides were used for femoral preparation and extramedullary guides for tibial preparation. No postoperative drain was used.

All patients stopped using antibiotics within 24 h after surgery, and all received intravenous tranexamic acid (1 g) at 3, 6, 12, and 24 h after surgery. All patients received intravenous dexamethasone (10 mg) on postoperative day 1 and intravenous dexamethasone (5 mg) on postoperative day 2, and they began to receive oral prednisone (10 mg) from postoperative day 2 onward in order to control pain and reduce inflammation. All patients were required to start functional exercise as soon as they had recovered from anesthesia and to begin walking under pain control from postoperative day 1. Every patient received a lower-extremity pump and a subcutaneous injection of low-molecular-weight heparin (2000 IU) to prevent deep vein thrombosis from postoperative day 1. The patients were discharged on the third day after the operation if they had no signs of complications and could walk independently.

All patients underwent Doppler ultrasonography either immediately if they showed any sign of deep vein thrombosis or otherwise on postoperative day 2 or 3 and 14. Patients were scheduled for computed tomography angiography if they suddenly experienced chest discomfort or breathing difficulties, if they coughed up pink foamy sputum, or if they exhibited other symptoms suggestive of pulmonary embolism.

Blood transfusion was performed according to the guidelines of the Chinese Ministry of Health [33]: transfusions were given to patients whose hemoglobin level was lower than 70 g/L and who did not present clinical symptoms or to those whose hemoglobin level was lower than 100 g/L and who presented anemia-related organ dysfunction, intolerable anemia symptoms, or ongoing hidden blood loss. Albumin was used for patients whose albumin level was less than 35 g/L for 2 consecutive days after surgery. Twenty grams of albumin were used each time and albumin levels were rechecked. Whether to use additional albumin was determined according to the results of the re-examination.

Two investigators independently collected the following information from each patient: (1) basic information such as age, sex, height, weight, BMI, and comorbidities; (2) postoperative length of stay and overall length of hospitalization; (3) perioperative laboratory values, including pre- and postoperative hematocrit (Hct), hemoglobin (Hb), C-reactive protein (CRP), interleukin-6 (IL-6), fibrin degradation product (FDP), and D-dimer; (4) perioperative range of knee motion (ROM), knee circumference, and knee swelling rate; (5) intraoperative systolic blood pressure and blood loss; (6) postoperative VAS pain score; and (7) complications.

The primary outcome was perioperative blood loss, comprising total blood loss (TBL), intraoperative blood loss (IBL), and hidden blood loss (HBL). TBL was calculated using the Gross formula [34]:

$$} = } \times (}_}}} \, - \,\,}_}}} )} \mathord} \times (}_}}} \, - \,\,}_}}} )} }_}}} }}} \right. \kern-\nulldelimiterspace} }_}}} }},$$

where PBV is the predictive blood volume; Hctpre is the preoperative Hct level; Hctpost is the lowest postoperative Hct level, which usually occurred on postoperative day 2 or 3; and Hctavg is the average of Hctpre and Hctpost. PBV was calculated using the formula [35]

$$}_ \times }^ ] + [}_ \times }$$

where k1 = 0.3669, k2 = 0.03219, k3 = 0.6041 for men, or k1 = 0.3561, k2 = 0.03308, and k3 = 0.1833 for women. HBL was defined as the difference between TBL and IBL.

Secondary outcomes were surgery duration, postoperative laboratory indices of inflammation and fibrinolysis, range of knee motion, VAS pain score, knee circumference, knee swelling rate, homologous transfusion, albumin use, and complications. Complications included postoperative hypertension, deep vein thrombosis, pulmonary embolism, calf muscular venous thrombosis, aseptic or septic wound complications, periprosthetic joint infection, 30-day mortality, and 90-day readmission. Postoperative hypertension was defined as a systolic pressure  of > 160 mmHg within 2 h after surgery as determined by electrocardiography and an upper-arm sphygmomanometer in the ward [36]. Systolic pressure was also recorded postoperatively using electrocardiography in the ward. Wound complications were defined as the need for intervention, such as superficial surgical debridement, re-suture, or a longer hospital stay [37]. If the wound showed secretion, at least two samples were cultured to test for the presence of bacteria. If two cultures were positive for homogeneous bacteria, wound complications were classified as septic [37]. Otherwise, wound complications were classified as aseptic [37]. The knee circumference was measured at the thigh at a position 10 cm above the upper edge of the patella when the patient was in a supine position with the knee straight. Knee swelling rate was calculated using the formula

$$}}C_}}} \, - \,C_}}} )} \mathord}C_}}} \, - \,C_}}} )} C}} \right. \kern-\nulldelimiterspace} C}_}}} ,$$

where Cpre is the preoperative knee circumference and Cpost is the postoperative knee circumference measured on postoperative days 1, 2, and 3.

The minimum sample size was estimated based on a previous study in our institute [38]. In previous studies, the average duration of surgery without a tourniquet was 84.9 min with a standard deviation of 20.1 min. We assume that 10 min is the least clinically significant reduction in duration of surgery due to tourniquet application. The test power (1 − β) was 0.8 and the alpha error rate was 0.05. The lost-to-follow-up rate was set at 0.05. Calculations indicated that at least 68 patients were required for each group.

Statistical analysis was performed using SPSS 22.0 (IBM, Armonk, NY, USA). Continuous data with a normal distribution were expressed as mean ± standard deviation (SD), while categorical data were expressed as frequencies. Inter-group differences were analyzed for significance using the Mann–Whitney U test in the case of continuous data that were skewed or showed unequal variance, or using the independent samples t-test in the case of normally distributed continuous data. Inter-group differences in categorical data were assessed using the chi-squared test or Fisher’s exact test as appropriate. Differences with P < 0.05 were considered significant.