Postoperative autotransfusion drain after total hip arthroplasty: a meta-analysis of randomized controlled trials

The use of a postoperative autotransfusion drain (PATD) to reduce allogenic blood transfusions in total hip arthroplasty (THA) remains controversial. Therefore, we conducted a meta-analysis to evaluate the efficacy and safety of this technique. Randomized controlled trials (RCTs) were identified from PubMed, Embase, and the Cochrane Central Register of Controlled Trials (CENTRAL). Thirteen RCTs (1,424 participants) were included in our meta-analysis. The results showed that PATD reduced the rate of allogenic transfusions (RR = 0.56; 95% CI [0.40, 0.77]) and total blood loss (MD = −196.04; 95% CI [−311.01, −81.07]). Haemoglobin (Hb) levels were higher in the PATD group on postoperative day 1 (MD = 0.28; 95% CI [0.06, 0.49]), but no significant differences on postoperative days 2 or 3 (MD = 0.29; 95% CI [−0.02, 0.60]; MD = 0.26; 95% CI [−0.04, 0.56]; respectively). There were no differences in length of hospital stay (MD = −0.18; 95% CI [−0.61, 0.25]), febrile reaction (RR = 1.26; 95% CI [0.95, 1.67]), infection (RR = 0.95; 95% CI [0.54, 1.65]), wound problems (RR = 1.07; 95% CI [0.87, 1.33]), or serious adverse events (RR = 0.59; 95% CI [0.10, 3.58]). Our findings suggest that PATD is effective in reducing the rate of allogenic transfusion. However, the included studies are inadequately powered to conclusively determine the safety of this technique.


Serious adverse events.
Only three studies 16,26,27 reported serious adverse events, including one death in the PATD group and one pulmonary embolism and two deaths in the CSD group. No significant heterogeneity was observed (p = 0.43, I 2 = 0%). No significant difference was found between the two groups (RR = 0.59; 95% CI: 0.10 to 3.58; p = 0.57; Fig. 10). Sensitivity analysis. Sensitivity analysis was performed by removing each study individually to identify whether the pooled results changed. All results were stable except postoperative Hb levels. On postoperative day 1, the difference between groups became statistically insignificant after the removal of one study 16 ; in addition, the removal of another study 27 on postoperative day 3 reduced I 2 to 0% but resulted in a significant difference between groups. In addition, two studies 22,29 accounted for the main source of heterogeneity of the allogenic transfusion rate; removing these two studies resulted in a large reduction of heterogeneity (I 2 decreased to 0%); however, the results still suggested that PATD reduced the transfusion rate.
Publication bias. Publication bias was evaluated using Begg's test and Harbord's test (or Egger's test). There was no evidence for significant publication bias among most of the included studies. Details are shown in Table 3.

Discussion
Although restrictive blood management and several transfusion alternatives have been developed for minimizing exposure to allogenic blood [35][36][37][38][39][40][41] , an increasing rate of allogenic transfusion remains following THA due to a number of identifiable risk factors, such as female gender, older age, black race, medical insurance and previous anaemia 1,5,42,43 . Moreover, under certain circumstances, allogenic transfusion is not feasible, such as with Jehovah's Witnesses who refuse allogenic blood and patients with rare blood types. Optimizing the use of blood conservation potentially resolves such conditions; nevertheless, there is little evidence regarding the efficacy of PATD. Previous meta-analyses 12,20,24,44 have investigated the efficacy and safety of cell salvage in THA. However, the strength of these meta-analyses was weakened by poor methodological quality or other limitations, and the conclusions were inconsistent. The studies by Carless et al. and Haien et al. combined several types of orthopaedic surgery, which inevitably resulted in clinical heterogeneity because a tourniquet is commonly used in total knee  24 , PATD showed no effect on reducing the transfusion rate but appeared to be associated with less total blood loss and lower superficial infection. However, few studies were included to analyse the transfusion rate as well as certain other outcomes. In a more recent study 44 , inconsistent results were obtained; when all studies were pooled, the conclusion favoured cell salvage, but the pooled results of recent trials (2010 to 2012) showed no difference between groups. Because the authors subjectively considered that studies published after 2010 had a lower risk of bias, the subgroup analyses appeared to explain the clinical significance and the substantial heterogeneity in other subgroups with difficulty. Furthermore, the analysis might have neglected some high-quality trials published before 2010 or included recent    trials of poor quality. In addition, the authors only compared the transfusion rate; other data of clinical significance were not analysed. Given the defects found in previous studies, we performed the present meta-analysis to determine whether PATD could be of greater benefit to THA patients than CSD. We eliminated potential confounding     Table 3. Assessment of publication bias. N: number of studies; NA: data not available; * Assessment of publication bias could not be performed because the number of studies was less than 3.
factors from total knee arthroplasty and intraoperative cell salvage, which controlled for the clinical heterogeneity of the included studies. Moreover, the results of high-quality studies strengthened the conclusion. To our knowledge, the current study is the largest meta-analysis that has independently investigated the use of PATD after THA. Thirteen eligible RCTs, including 1,424 participants, were included in our meta-analysis to evaluate the efficacy and safety of PATD.
Overall, the most important findings were that PATD significantly reduces the rate of allogenic blood transfusion. PATD is associated with a 44% reduction in the exposure rate of allogenic blood. Moreover, the pooled results of high-quality studies showed a similar effect (a 41% reduction of RR) with no significant heterogeneity (I 2 = 0, p = 0.81). A recent large cohort study investigated 2,087,423 patients undergoing THA who received allogenic transfusion. The results showed that allogenic transfusion was associated with a longer hospital stay, increased costs, and worse surgical and medical outcomes 1 . Another study analysed data from more than 12,000 patients who underwent THA or total knee arthroplasty. That study also demonstrated that allogenic transfusion was significantly associated with a higher risk of infection 10 . Therefore, the reduction of allogenic transfusion could potentially decrease the risk of numerous comorbidities and total costs. Although different transfusion management strategies may affect the transfusion rate, in the high-quality studies, restrictive transfusion triggers were used and the transfusion rate was still reduced, indicating that PATD provided an independent effect; this conclusion is strengthened by the methodological quality.
Increased allogenic transfusion is associated with increased blood loss 45 . Total blood loss was shown to be lower in the PATD group, a finding that might account for the lower allogenic transfusion rate. Nonetheless, only two studies reported the calculated blood loss, which considers hidden blood loss 3,4 ; thus, it is difficult to draw a conclusion due to the small number of included studies. Other studies reporting estimated blood loss were not available to pool the data.
Although transfusion decisions should consider various factors, haemoglobin concentration remains an important indicator 46,47 . A higher postoperative Hb level is correlated with a lower transfusion rate, better early functional recovery, and higher patient satisfaction 16,48 . In this meta-analysis, we found that the PATD group maintained a higher Hb level on the first postoperative day, but no differences were present on the next two days. A potential explanation for this finding is that PATD was only used within 6 hours after surgery. However, this result changed when subjected to sensitivity analysis, suggesting that it was unstable. Munoz et al. indicated that no increase in a patient's Hb levels should be expected due to the lower haemoglobin concentration in postoperatively salvaged blood. Indeed, the retransfusion of shed blood is likely to maintain Hb levels above the transfusion trigger until bleeding stops 15 .
Regarding length of hospital stay, several studies reported a shorter length of hospital stay in the PATD group 14,49,50 . However, no difference was observed in our study. Given that the length of hospitalization might be correlated with a number of confounding factors, such as patient rehabilitation, comorbidities, and different discharge policies, we recommend that future studies describe the standard of discharge with more details and isolate the potential confounders.
Postoperative shed blood, particularly unwashed blood, may be contaminated with wound material and contains a variety of tissue materials and chemical debris, potentially causing complications, such as febrile reaction, infections, embolism, immune response and even death 25,51,52 . Washed shed blood is considered safer because most bioactive contaminants are removed [53][54][55] ; however, washing shed blood is expensive and complex. Nevertheless, studies have suggested that the incidence of adverse events is lower than theoretically predicted 15,51,56 . In addition, with recent improvements in techniques and practices, the use of cell salvage is safe, even in obstetrics or malignancy 13 . Similarly, in our study, adverse events showed a low incidence. However, no differences in adverse events between PATD and CSD were observed. This finding may be due to the low incidence; most of the included studies were underpowered to accurately reflect the incidence of adverse events. Therefore, future studies with larger sample sizes are urgently needed to determine the safety of PATD.
In spite of the rigorous protocol of this meta-analysis, several limitations should be taken into account. First, some of the included studies had one or more risks of bias, such as inappropriate randomization, no allocation concealment, lack of blinding, and other shortcomings, which limited the reliability of the outcomes. However, the pooled results of high-quality studies reached the same conclusion, which strengthens our conclusion. Second, the number of studies investigating several outcomes was relatively small because some data were not available; thus, these outcomes may be changed by the findings of future research. Third, there were several transfusion triggers in different studies, which might be a potential source of clinical heterogeneity that affects the results. In addition, the sample sizes of most of the included studies did not have sufficient power to draw a conclusion regarding adverse events, suggesting that further study on this topic is needed.
In conclusion, we found that PATD is effective in reducing the rate of allogenic transfusion in patients undergoing THA; a reduction of RR of more than 40% was found. Moreover, the use of PATD appears to be associated with a higher postoperative Hb level and less total blood loss, without any significant adverse events. Thus, PATD may reduce the exposure to allogenic blood and its underlying risks. The current evidence may guide clinicians in their decisions with regard to transfusion for THA patients. However, due to the limitations of this meta-analysis, we recommend more widely accepted transfusion guidelines, and additional well-designed RCTs with adequate sample sizes and a consolidated standard are needed.

Methods
Search strategy. A comprehensive search was performed in the PubMed, Embase, and Cochrane Central Register of Controlled Trials databases up to March 2016. The search terms included "autologous blood transfusion", "operative blood salvage", "autotransfusion", "blood salvage", "retransfusion", "arthroplasty, replacement, hip", "total hip arthroplast* ", "total hip replacement* ", and "total hip prosthes* ". The related references in the identified studies were manually searched.
Scientific RepoRts | 6:27461 | DOI: 10.1038/srep27461 Selection criteria. The inclusion criteria were as follows: (1) randomized controlled trial; (2) patients treated with primary or revision total hip arthroplasty; (3) PATD compared with CSD; (4) at least one of the key data available, including allogenic transfusion rate, total blood loss, Hb level, length of hospital stay, infection, febrile reaction, wound problems or serious adverse events.
The exclusion criteria were as follows: (1) duplicate articles; (2) cohort studies, case reports, editorials, letters, reviews, and animal experimental studies; (3) data that could not be extracted. Data extraction. Two reviewers (HX and JKP) independently extracted the following data from the included studies: authors' names, date of publication, sample size, patients' age and gender, surgery type, allogenic transfusion rate, total blood loss, preoperative and postoperative Hb levels, length of hospital stay, febrile reaction, infection rate, wound problems (wound leakage, haematoma, delayed healing) and severe adverse events (life-threatening events and death). In the event of missing data, we attempted to contact the corresponding authors for details. Quality assessment. The methodological quality of the included studies was independently evaluated by two reviewers (HX and HKH) using the Cochrane Collaboration's tool for assessing the risk of bias 34 . These domains were selection bias (random sequence generation and allocation concealment), performance bias (blinding of participants and personnel), detection bias (blinding of outcome assessments), attrition bias (incomplete outcome data), reporting bias (selective reporting) and other bias (other sources of bias). Any disagreements were resolved by discussion or were arbitrated by the corresponding author (JL).

Statistical analysis.
Risk ratio (RR) and mean difference (MD) were used to pool dichotomous and continuous data, respectively. The meta-analysis was performed using Review Manager 5.3.5 (Cochrane Collaboration, Oxford, UK). For continuous data presented as the mean with quartile/range, the standard deviations were estimated according to the Cochrane Handbook 34 or the method described by Hozo et al. 57 . Heterogeneity was assessed using the Cochrane Q test and I-square statistic. A sensitivity analysis was performed to identify the source of the heterogeneity. All data were pooled using the random-effects model. Begg's test and Harbord's test (or Egger's test) were used to estimate potential publication bias.