Clinical Analysis of Suction Drainage in Cementless Hip Replacement: A Prospective Randomised Study

Background: Several studies have suggested that the use of drains did not improve surgical outcomes after hip replacement. There is still a lack of strict recommendations for drainage use. Methods: Prospective, randomised study was performed. The analysis included 100 patients. Inclusion criteria: idiopathic hip osteoarthritis. Exclusion criteria: secondary coxarthrosis, autoimmune disease, coagulopathy, venous/arterial thrombosis, hepatic/renal insuciency, cement, or hybrid endoprostheses. Results: We found smaller haematomas in the no-drainage group (9.76 mm vs. 10.33 mm, p = 0.653). The visual analogue scale score was lower in the no-drainage group (5 vs. 6). Less bloodloss in the no-drainage group (1,124 ml vs 1,224 ml, p = 0.59). Two patients had a deep joint infection in the no-drainage group, none in the drainage group. Conclusion: It is noteworthy that two cases of early infections were observed in the no-drainage group, whereas there were no such complications in the drainage group. Further research is warranted to validate our ndings.


Background
Hip replacement surgery has been called the most effective procedure of the century [1]. It is an orthopaedic procedure that necessarily entails soft tissue and bone damage. Intraoperative blood loss can be limited by haemostasis at several stages and by using anatomical, non-aggressive surgical techniques. The use of brinolysis inhibitors such as tranexamic acid, as well as topical use of vasoconstrictors, reduces both intra-and postoperative bleeding [2].
Signi cant intraoperative blood loss can affect haemostasis, leading to increased postoperative bleeding [3]. Suction drainage is used to drain the blood from the hip joint. It also serves to assess bleeding activity that may be an indication for urgent surgical intervention and repeated haemostasis. An excessive haematoma might also serve as a nidus for bacterial growth. Unfortunately, the drainage tube is also a signi cant point of access for infections in the hip. Several publications documented the risk of infection in the context of prolonged suction drainage [4,5].
Postoperative blood loss is di cult to assess. The basic diagnostic method used is physical examination supported by diagnostic imaging. An ultrasound is a method of choice to assess uid collection in soft tissues [6].
According to the literature, the reduction of postoperative haematomas might be affected by the proper use of thromboembolic prophylaxis, including low-molecular-weight heparin (LMWH) [7]. According to recommendations, in Poland, LMWH is administered 12 hours before surgery, and the therapy is continued for 35 days after the intervention [8].
Until recently, we did not have strict recommendations for suction drainage after primary total hip replacement. Therefore, we sought to assess a homogenous group of patients with extended thromboprophylaxis to determine whether forgoing the drain in this cohort was safe. The objective of our study was to measure joint uid collection, blood loss, and complications in patients undergoing hip replacement with and without drain placement.

Methods
The study was planned and conducted using a prospective randomised design. We performed simple randomisation using closed envelopes designating group allocation. The study was successfully retrospectively registered in Clinicaltrial.gov with identi cation number: NCT04333264 (03/04/2020).
The study began on 9 March 2016, after obtaining approval from our Bioethical Review Board. All methods provided in these study was carried out in accordance to approval from Bioethical Committee of Centre of Postgraduate Medical Education in Warsaw with approval number 13/PB/2016. Patients successively admitted to the ward for hip replacement were eligible for inclusion. All patients signed informed consent. Patients were prepared for surgery in a typical manner. LMWH was used for thromboprophylaxis in doses adjusted for body weight and risk factors. The rst heparin dose was administered in the evening of the day before surgery. Thromboprophylaxis was continued for 35 days after surgery. All patients received tranexamic acid (Exacyl) intravenously, at a dose of 15 mg per kilogram of body weight, 10 minutes before skin incision. Experienced surgeons performed surgeries through a posterolateral approach with cementless hip replacement implantation. In all patients, wound healing was monitored during the postoperative period.
Over the 72 hours after surgery, all patients underwent ultrasound scans of their hip joints and the postsurgical wound using the aseptic technique. The scans were performed with patients in the supine position. The ultrasound scan assessed the uid level at the level of the endoprosthesis neck and detected uid accumulation in the soft tissues.
Blood loss volume, along with the occult bleeding, was analysed using the Gross formula: EBV represented the estimated patient's blood volume, Ht (0) represented haematocrit before surgery, Ht (1) represented haematocrit 24 hours after surgery, and Ht av represented estimated pre-and postoperative haematocrit value. Despite limitations such as intraoperative uid transfusion or renal insu ciency, the Gross formula is thought to estimate interoperative blood loss credibly [9].
Before surgery and within 72 hours afterwards, the range of hip joint motion was assessed for exion, abduction, adduction, and exion contracture. Pain levels were assessed using the visual analogue scale (VAS) in all patients. Physical examination and ultrasound scans were conducted 72 hours after surgery by the same physician. Patients were discharged depending on their general condition and progress in rehabilitation, usually from the third to seventh day after surgery. The follow-up period was 30 days.
A total of 134 patients were successively operated on from 14 March 2016 to 16 May 2018, meeting the study inclusion and exclusion criteria.
The inclusion criteria were primary hip osteoarthritis and age between 30 and 80 years. The exclusion criteria were secondary degenerative hip joint disease, autoimmune disease, congenital/secondary coagulopathy, history of venous/arterial thrombosis, hepatic/renal insu ciency, cement or hybrid endoprosthesis, and lack of patient consent.
Patients were allocated to two groups depending on the presence or absence of drains; 100 patients quali ed for the nal analysis, of which 20 patients later withdrew consent for participation in the study. In eight patients, the surgeon decided to use suction drainage, regardless of randomisation results; and in six, a cemented implant was used because of intraoperative conditions. The allocation process was performed with simple randomisation use. Closed envelopes, with information on drainage usage, was opened in the operating theatre at the end of surgery. If the surgeon, considering the procedure and local conditions, decided that suction drainage was necessary, the patient was excluded from the study.
The investigated group included 55 females (55%) and 45 males (45%). The mean age was 62.  Calculations were made using Statistical Analysis System (version 9.4) software. Quantitative data were examined using the Student's t-test, and distribution was assessed using the Shapiro-Wilk test. To examine qualitative characteristics, contingency analysis was used; for these analyses, X2 with Yates's correction and Fisher's F-tests were applied.

Results
At the nal follow up was assessed 100 patients, 50 drainage group and 50 no-drainage group. The statistical analysis did not reveal any statistically signi cant differences between groups in terms of age, sex, body weight, degenerative disease severity on the Kellgren-Lawrence scale, and blood clotting factors (activated partial thromboplastin time [APTT], prothrombin time [PT], and international normalised ratio [INR]) ( Table 1).  Overall mean perioperative blood loss was 1,126 ml (range, 370-2,384 ml). In the no-drainage group, mean blood loss was 1,124 ml (400-2,281 ml), and in the drainage group, it was 1,224 ml (370-2,384 ml) (p = 0.59). Overall, the need for blood transfusion was rare (the median was 0). In both groups, blood transfusion was necessary for ve patients (p = 0.247).
The risk of deep vein thrombosis during the 30 days after hip replacement surgery was also assessed. No patient in either group experienced postoperative deep vein thrombosis.
There were no signi cant differences between groups in terms of wound exudation on the third day after surgery (p = 1). In the no-drainage group, an early infection was observed in two patients. Infection was detected as a result of prolonged wound leakage (over 5 days) and was con rmed intraoperatively with two positive bacteriological cultures. These patients were subjected to the Debridement, Antibiotics, Implant Retention (DAIR) procedure with good outcomes. There were no signi cant differences between groups regarding the incidence of infection (p = 0.47).
Clinical status was assessed using the VAS, evaluating pain before surgery and 72 hours after surgery. In the no-drainage group, the median VAS score before surgery in the no-drainage group was 6; in the drainage group, it was 7. After surgery, the median in the no-drainage group was 5, and it was 6 in the drainage group (p = 0.71) ( Table 2).  The mean hospitalisation duration in both groups was 7 days: 4-21 days without drainage vs. 4-19 days with drainage (p = 0.60). Because of prolonged wound leakage, two patients were readmitted 30 days after the surgery and quali ed for the DAIR procedure. These patients were diagnosed with early infection (p = 0.49).

Discussion
We assessed a homogenous group of patients undergoing hip replacement. For legal reasons, all patients received extended thromboprophylaxis. The safety of suction drainage was not evaluated in this group of patients. The objective parameter that might provide information about the impact of preoperative heparin dose was the amount of blood in the hip postoperatively. In our analysis, we checked uid levels above the endoprosthesis neck using ultrasonography in the supine position. We found lower uid levels in the no-drainage group (9.76 vs. 10.33 mm, p = 0.653); from this, we conclude that refraining from placing suction drainage has no effect on patients with extended thromboembolic prophylaxis who undergo a primary cementless hip replacement.
In the no-drainage group, we observed two patients with deep infections, whereas, in the drainage group, there were no infections. We did not recognise surgical site infections (SSIs). Only two patients had prolonged wound leakage (above 5 days); however, in these cases, we recognised deep infections. Zimmerli questioned the diagnosis of SSI in patients with implants because it cannot be reliably clinically differentiated from deep infections [10]. Despite the absence of statistically signi cant differences, this raised our doubts about the wisdom of not using suction drainage.
Historically, suction drainage was justi ed by the need to reduce hip joint haematoma and was consequently used to reduce the risk of periprosthetic infection. The rst papers to question the use of suction drainage were published in the 1990s [11,12]. Hou et al., based on 27 randomised studies, did not demonstrate a higher incidence of infections in no-drainage patients [13]. Similar conclusions were drawn by Chen et al., based on 16 papers [14]. Fagotti et al. reported two SSIs in the drainage group but no deep periprosthetic infections; in the no-drainage group, there were no SSIs, but one patient was diagnosed with a deep infection. However, there were no signi cant differences between groups [15]. Walmsley et al., in a study of 552 patients (577 hip joints), reported a higher incidence of SSIs (48% vs 2.9%) and deep infections (0.7% vs 0.4%) in no-drainage patients [16].
Kelly et al., in their meta-analysis, demonstrated that patients from the suction drainage group required blood transfusion signi cantly more frequently and had greater postoperative blood loss [17]. Hou et al., in their meta-analysis, also demonstrated signi cantly more frequent blood transfusions in the drainage group. Blood loss volume in our paper was insigni cantly smaller in the no-drainage group; however, there was no difference between groups in blood transfusion rates.
Because of the short patient observation period, the VAS pain scale and the range of hip motion were used. The VAS scale is commonly used to assess pain experienced by patients undergoing hip replacement surgery [18]. In their meta-analysis, Hou et al. did not demonstrate any differences between groups in the VAS score. Fagotti et al. found signi cantly higher VAS scores among patients without suction drainage than among those with drainage. In our study, 72 hours postoperatively, a nonsigni cant difference between groups in favour of the no-drainage group was identi ed.
A hip joint haematoma can reduce the range of motion [19]. Zeng et al. demonstrated a reduced range of hip joint mobility following surgery in patients who did not have suction drainage [20]. Nevertheless, Chen et al., in their meta-analysis, did not nd any differences in ROM between the drainage and no-drainage groups, as was con rmed in our paper.
In most relevant publications, the majority of parameters did not differ signi cantly between groups [15][16][17]21]. All these studies included heterogeneous groups; patients with primary and secondary osteoarthritis, coxarthrosis in the course of rheumatoid arthritis, and others. Some of these conditions could affect perioperative blood loss and the need for blood transfusion [22]. These conditions illustrate why we need more well-prepared multi-centre prospective studies.
There are several limitations to this study. First, it was a single-centre study and therefore may be subject to selection bias. For this reason, we instituted strict inclusion and exclusion criteria. Multi-centre studies are needed to validate our ndings. A short observation period was appropriate for the intervention we investigated. For this reason, we used proper scales to assess short-term clinical outcomes. A large number of patients were excluded from the primary cohort, possibly in uencing outcomes; however, all exclusions occurred before we opened the sealed envelopes, which ultimately did not impact the randomisation process signi cantly.

Conclusions
Hip replacement without suction drainage after surgery is a recognised therapeutic method. No superiority of either method was demonstrated; nevertheless, it is worth noticing that there were two cases of early infection in the no-drainage group compared to none in the drainage group. This suggests the need for further research in this respect.