A small difference in recovery between total knee arthroplasty with and without tourniquet use the first 3 months after surgery: a randomized controlled study

Patients eligible for recruitment were aged 50–80 years and were undergoing TKA for the treatment of primary osteoarthritis. Exclusion criteria were revision surgery, valgus deformity > 30°, one-stage bilateral procedures, rheumatoid arthritis, and BMI > 35. Furthermore, after the randomization but before the group allocation was revealed, the surgeon could exclude a patient in the operating room for medical reasons, primarily impaired blood circulation (as determined by the surgeon). This criterion was predetermined but is missing from the original ISRCTN registration; it was complemented in ISRCTN 12/11/2014. In addition, patients who underwent surgery on Thursdays were excluded because measurements on Sundays were not possible.

Eighty-eight patients were randomized (Fig. 1). Seven patients were excluded after randomization: Six were excluded due to impaired blood circulation, and 1 was excluded because the surgery started with a tourniquet and was finished without one due to the long duration (120 min) in a bloodless field. This left 81 patients for analysis at day 3. In addition, 4 patients dropped out, which left 77 patients for analysis at 3 months. In both of these samples, the nontourniquet and tourniquet groups were similar at baseline except in the use of a walking aid (Table 1) (data not shown for the 3-month sample).

An independent statistician used a computerized random number generator in R to create a random number table for group allocation prior to the study’s start. The research physiotherapist (MA) determined the patients’ preliminary eligibility based on their medical records and then recruited participants consecutively at the preoperative physiotherapy appointment. The independent nurse at the orthopedic clinic then used the computerized surgery planning system “Orbit” to inform the surgeon and operating team of the participant’s group allocation according to the random number table. This information was provided immediately prior to surgery. The participants were not informed of their group allocation and were not aware of our hypothesis. To help preserve blinding, the participants listened to music through headphones during surgery, and a curtain was used to prevent the participants from receiving visual cues. Furthermore, tourniquet use was not noted in the participants’ medical records.

Before 2011, the vast majority of TKAs were performed with a tourniquet; therefore, the surgeons had approximately 18 months of experience performing TKA without a tourniquet prior to the study’s start. Seven senior orthopedic surgeons with experience in both procedures were included and none of them was a member of the research team. One group underwent surgery with a tourniquet (34 in., single bladder, dual port, Zimmer) around the thigh that applied a pressure of 300 mmHg and the other group underwent surgery without a tourniquet. No femoral nerve block was used. A standard medial parapatellar incision was used. The cemented NexGen CR- or PS-Flex fixed bearing knee (Zimmer) prosthesis was used without patellar resurfacing. Infiltration with 150 ml of ropivacaine-supplemented ketorolac and adrenaline was applied during surgery. If a tourniquet was used, it was released after the bandages were applied. Tranexamic acid (1 g) was given intravenously, 10 min before surgery in the nontourniquet group, and 10 min before tourniquet release in the tourniquet group. 2 g of cloxacillin was administered intravenously just before and twice after the surgery. Low-molecular weight heparin (Fragmin, 5000 IE subcutaneously) was used for the first 14 postoperative days. Postoperative pain management included oxycodone 5–10 mg (controlled-release oral formulation) twice a day, paracetamol 1 g 4 times a day and oxycodone 5 mg when needed. Weight bearing was allowed on the evening of the day of the surgery. On weekdays, the patients performed rehabilitation training under the supervision of a physiotherapist, i.e., active-assisted and active range of motion exercises, strengthening exercises, and gait training. A home-based exercise program followed.

The primary outcome measure was active flexion of the knee at day 3. Outcome measures were collected presurgery and at day 1, day 3 and 3 months postsurgery. Outcome measures were collected by nine independent physiotherapists at the orthopedic clinic. The physiotherapists trained together at several meetings held before the start of the study and during each semester of the data collection period to improve the reliability of the measurements. Data were collected by the same physiotherapist preoperatively and at postoperative days 1–3 for 55 patients and by a different physiotherapist for the remaining 26 patients, of which 12 patients belonged to the nontourniquet group and 14 to the tourniquet group. By self-report, it was determined that the blinding of the physiotherapists who collected the data was preserved for all but 2 patients.

Active flexion and active extension of the knee were measured with a standard plastic goniometer with a scale marked in 1° increments and 30-cm arms with the patient lying supine on a gurney. The absolute intra- and interrater reliabilities, assessed as the standard deviation of flexion-of-the-knee measurements, have been shown to be 4.0° and 5.9°, respectively [24].

The Timed up and go (TUG) test was used to assess basic mobility [25, 26]. It is a timed test in which the patient is instructed to safely get up from a chair, walk 3 m, turn, walk back, and sit down again. If a walking aid is usually used, it is used during the test.

The visual analog scale (VAS) was used to assess pain before surgery and after 24 ± 2 h, 72 ± 2 h, and 3 months [27]. The patient answered the question “How painful is your leg?” using a 0–100 mm VAS. The question was asked prior to rehabilitation training, while the patient was at rest and the use of any additional analgesia was noted.

Knee circumference was measured to assess swelling. This value was measured with a tape measure at the superior border of the patella with the patient lying supine [28].

SLR was used to test quadriceps function. The patient was asked to perform an SLR to 45° flexion in the hip while lying supine with the other leg in flexion and the foot on the base of support. Performance was assessed as able/not able.

After the 3-month control period, the research physiotherapist examined the patients’ medical records to collect data on Hb-level, length of stay (LOS) and complications such as deep vein thrombosis and wound infections.

All the participants gave their written consent and were recruited in agreement with the Helsinki Declaration. The study was approved by the Regional Ethical Review Board in Stockholm (reference number 2011/1625-31/1 and 2014/1528-32).

A power calculation was performed for our primary outcome measure, flexion of the knee, at day 3 post-surgery, and the minimum sample size required was 37 patients in each group to detect a group difference of 10° (estimated as a clinically significant difference) with a power of 80% in a 2-tailed independent t test with an alpha level of 0.05. The standard deviation (SD), ± 15°, that was used in the calculation was the pooled SD of the 2 groups in another study [13].

The analyses were performed according to the modified intention-to-treat principle as exclusions were made after randomization, and there were losses to follow-up at 3 months [29]. Hb level and LOS were post hoc analyses.

Analysis of covariance (ANCOVA) was used to analyze possible between-group effects on the continuous outcome measures. The postintervention values of the outcome measures were entered as dependent variables. The selection of independent variables as possible confounding factors for adjustment of the ANCOVA was based on 3 criteria [30]. 1) The predetermined independent variables were surgeon and the preintervention value of the dependent variable (except in the analysis of LOS). Three of the surgeons performed ≤ 3 operations and were, therefore, pooled together. 2) The variables in Table 1 that showed a between-group difference preintervention (p < 0.05), i.e., use of a walking aid. 3) The variables in Table 1 for which the preintervention measure value correlated with any of the preplanned continuous outcome measure values postintervention with a strength of r > 0.3 (Pearson correlation coefficient), i.e., age and extension of the knee. The exception was outcome measures for which the preintervention value only correlated with the corresponding postintervention value.

The dispersion in the tourniquet group differed from the dispersion in the nontourniquet group in the analysis of TUG test and extension of the knee at 3 months, Levine’s test p < 0.05. Therefore, the Mann–Whitney U test was also performed for the continuous outcome measures, with the difference between pre- and postintervention as the dependent variable.

Dichotomous outcome measures were analyzed with the Chi-square test or Fisher’s exact test.

All statistical tests were performed in SPSS version 23 (IBM Corp., Armonk, NY, USA). The alpha level was set to 0.05.