Study design and patients
Prior to conduction of the study, institutional Clinical Trials and Biomedical Ethics Committee approval was obtained. All patients knew the use of cell salvage was relatively contraindicated in cases of hip revisions for PJI and gave their written informed consent before reimplantation. We performed a single-center, retrospective, cohort study that enrolled consecutive patients who were treated for chronic hip PJI in a two-stage revision from 1 November 2012 to 30 April 2019, in which availability of second-stage culture results was mandatory. PJI was diagnosed according to the criteria published by the Musculoskeletal Infection Society and modified by the International Consensus Group as of 2014 , and chronic PJI was defined as any PJI present for more than 4 weeks from the index surgery . Exclusion criteria comprised patients (1) received intraoperative allogeneic blood transfusion or allograft bone transplantation during the second-stage reimplantation because these may increase the risk of reinfection and confound the analysis of reinfection sources; (2) had a known allergy to tranexamic acid (TXA), a history of arterial or venous thromboembolic event (e.g. stroke, pulmonary embolism, deep venous thromboembolism), coronary artery disease (placement of an arterial stent or myocardial infarction within the past six months), renal failure (serum creatinine > 200 mmol/l, creatinine clearance < 50 ml/min, dialysis), disseminated intravascular coagulation, hepatic failure, severe pulmonary disease; (3) had undergone preoperative anticoagulation therapy (excluding aspirin) or refusal of allogeneic blood transfusion; (4) treated with TXA in a way that deviated from the following standard practices in our institution from November 2012: TXA was routinely administrated a bolus (15mg/kg) 5 minutes before incision, and administrated again if the duration of reimplantation exceeded 2 hours. This study was conducted and reported in line with STROCSS 2019 criteria .
Two-Stage Revision Procedure
All revision procedures were performed using a classical posterolateral approach by 4 senior surgeons who specialized in total joint arthroplasty. During the first-stage revision procedure, thorough debridement was completed to excise sinus tracts, all infected tissues, and devitalized bone after removal of all foreign materials. Then, the wounds were repeatedly immersed in 3% hydrogen peroxide for 4-5 minutes, irrigated with 0.9% saline, and finally immersed in 2.5% povidone iodine for 4-5 minutes. This is the standard procedure in cases of PJI in our institution. Broad-spectrum antibiotics were administrated intravenously after the surgical fields were thoroughly irrigated with saline using a low-pressure system. Five to eight solid samples, both bone and soft-tissue, and a sample of synovial fluid were routinely obtained in the procedure and sent for microbiological culture and histological examination. All hips underwent implantation of an antibiotic-impregnated, handmade polymethylmethacrylate bone cement spacer containing 2 g of vancomycin and 1 g gentamicin per pack of Palacos cement (Biomet, Warsaw, IN).
Postoperative antibiotic administration protocol was under the guidance of infectious diseases specialists according to the antimicrobial susceptibility test. In all cases of PJI, medical therapy was initiated using broad-spectrum antibiotics (typically vancomycin and cefuroxime). If the tissue sample cultures were positive, organism-specific antibiotics were administered; otherwise, the empirical antibiotics strategy was vancomycin and cefuroxime. Patients received intravenous antibiotics for 6 weeks and oral antibiotics for 6 weeks following the first-stage revision. Second-stage reimplantation was considered when the general status of the patient was suitable and there were no signs of infection, which was confirmed by the return of C-reactive protein and erythrocyte sedimentation rate to normal or near normal levels.
The second-stage reimplantation was performed under targeted prophylactic-guided antibiotics. Synovial fluid was routinely taken for bacterial culture before antibiotic administration. Frozen sections were also routinely performed to exclude residual infection. Frozen sections with greater than 10 white blood cells per high-power field were considered to have persistent infection; 5 to 9 white blood cells per high-power field were considered to have suspicious infection; and less than 5 white blood cells per high-power field were considered to eradicate infection . After removal of the bone cement spacer, the joint cavity was immersed in hydrogen peroxide and povidone iodine and irrigated with saline which was consistent with the procedure of the first-stage revision. All patients had cementless, porous-coated, diaphyseal engaging nonmodular femoral stems.
According to the policy of our institution, we employed intraoperative cell salvage collection (3000P, Jingjing Medical Equipment, Beijing, China) for all second-stage revision hip procedures. Any fluid visibly contaminated by infection, metallic debris, and cement was not collected. All blood from operative fields and gauze was centrifuged with a leucocyte depletion filter (40 µm, Nanjing Shuangwei, Nanjing, China) and washed with 1000 mL of 0.9% sodium chloride, and then transferred to a sterile collecting bag. The volume of salvaged blood was standardized to haematocrit 55%. As per the result of Haemoglobin and haematocrit measured intraoperatively using blood gas analysis, salvaged blood was processed to transfuse once the estimated blood loss exceeded 500 ml.
Strict transfusion criterions were implemented in keeping with the clinical practice guidelines recommended from the American Association of Blood Bank , Allogenic blood transfusion was only allowed for patients with a haemoglobin level < 70 g/l or with a haemoglobin level between 80 g/l and 100 g/l in the setting of pre-existing cardiovascular disease, active bleeding, arterial thromboembolic event or sepsis, persistent symptoms despite adequate volume resuscitation. Plasma was allowed to transfuse if more than four blood units were transfused or if coagulation parameters were out of acceptable ranges. When allogeneic blood transfusion was indicated, 1 unit of packed red blood cells was transfused to increase Hb levels to 8.0 g/dl.
Systemic intravenous antibiotics against the microorganism isolated at the first-stage revision were maintained until microbiological results were available. In general, if synovial cultures were positive, organism-specific antibiotics were administered for 4 weeks and then orally for 4 weeks. If no microorganisms were identified, intravenous antibiotic strategy was the same as that for the first-stage revision for one week and then orally for an additional 2 weeks. If a new microorganism was identified in more than 2 samples, organism-specific systemic antibiotics were administrated for 8 weeks.
We obtained demographic data as well as comorbidities, antibiotic administration, and culture results at the time point of reimplantation from all patients. McPherson’s host classification , patient-related risk factors (diabetes mellitus, inflammatory arthritis, chronic hepatitis, chronic anemia, tobacco use, alcoholism, resistant organism), component exchange, and positive culture are associated with an increase in reinfection rate. Hence, we obtained and compared these results between the 2 groups (Table 1, Table 4).
Clinical follow-up was conducted routinely at 3 weeks, 3 months and 6 months after two-stage revision and annually thereafter until the final follow-up. The definition of a successful treated infection was based on the modified Delphi-based international multidisciplinary criteria : (1) a healed wound, without a fistula, drainage, or pain and without recurrence caused by the same organism, (2) no subsequent surgical intervention for persistent or peri-operative infection, (3) no occurrence of PJI-related mortality; (4) no requirement for long-term (> 6 months) suppressive antibiotic treatment. If one of these criteria was not met, infection eradication failure was considered as established.
Furthermore, we analyzed whether the use of cell salvage during reimplantation led to a decreased rate or volume in postoperative allogeneic blood transfusion and identified the risk factors for allogeneic blood transfusion. We recorded preoperative and postoperative haemoglobin levels, preoperative and postoperative hematocrit levels, postoperative allogeneic blood transfusion (rates and units) within 7 days of reimplantation, and the amount of blood from cell salvage reinfused during reimplantation. The calculated blood loss = the patient’s blood volume (PBV) * (Hctpre - Hctpost)/Hctave (Hctpre refers to the preoperative hematocrit level. Hctpost refers to the minimum hematocrit level before transfusion. Hctave refers to the average of Hctpre and Hctpost) . PBV = k1 * height (m3) + k2 * weight (kg) +k3 (k1 = 0.3669, k2 = 0.03219, and k3 = 0.6041 for men; and k1 =0.3561, k2 = 0.03308, and k3 = 0.1833 for women)  . All haemoglobin and hematocrit levels were measured routinely up to day 5 after the procedure.
All statistical analyses were carried out using SPSS Version 26 (IBM Corporation, Armonk, NY, USA). The normality of the distributions was tested for all scale variables by Kolmogorov-Smirnov’s test. Differences between groups were analyzed using Fisher’s exact test for categorical variables while continuous variables were compared by Mann-Whitney U-test. Categorial variables are presented using counts and percentages, while continuous variables are summarized using the median with 25th and 75th percentiles. The primary analysis was to evaluate the influence of cell salvage during reimplantation by comparing the rate of infection eradication failure by Kaplan-Meier curve and log-rank test. Next, a multiple logistic regression was performed to identify independent variables associated with the need for allogeneic red blood transfusion. Covariates with a P value < .05 by univariate analysis were integrated stepwise into the multivariate regression model. For all comparisons, the level of statistical significance was set at P < .05. A post hoc power analysis (PASS, version 19.0) considered an alpha error probability of .05, a study power of 80%, a proportion of 30% of patients who did not receive intraoperative cell salvage reinfusion, and a relative risk of 2.5. The required sample size was 57 patients per group.