Analysis of the Need for Postoperative Drainage Application for Hip Arthroplasty: A Systematic Review and Meta-Analysis.

OBJECTIVE: To synthesize the evidence regarding the effect and safety of drainage after the hip arthroplasty in randomized control trials.
BACKGROUND: Although the standard of hip replacement has matured in recent years, the need for postoperative drainage is still controversial which also is a clinical problem that needs to be addressed.
DESIGN: A systematic review and meta-analysis based on the Cochrane methods and Prisma guideline. Data Resources. A systematic search of the Cochrane Library, PubMed, EMBASE, CINAHL, Ovid, Wan Fang database, CNKI, and CBM database was carried out from January 1, 2000, to December, 2021. Review Methods. The quality of included randomized controlled trials was assessed individually by two reviewers independently using criteria recommended in the Cochrane Handbook for Systematic Reviews of Interventions 5.1.0.
RESULTS: Nineteen randomized control trials involving 3354 participants were included in this analysis. From the above analysis, we can know that compared with nondrainage, there was a statistically significant difference in VAS score on the postoperative first day (SD = -0.6; 95% CI: -0.79, -0.41) and second day (SD = -0.38, 95% CI: -0.58, -0.18), hematocrit reduction (MD =2.89; 95% CI: 1.3, 4.48), blood transfusion rate (OR =1.47; 95% CI: 1.12, 1.92), change of thigh circumstance (SMD = -0.48; 95% CI: -0.66, -0.31), and hospital stay (MD = 1.06; 95% CI: 0.73, 1.39) in drainage. However, there were no statistically significant differences in hemoglobin and hematocrit level, hip function, total blood loss, transfusion volume, dressing use, and complications between them.
CONCLUSION: Drainage after hip arthroplasty can reduce swelling in the thigh and relieve pain while no drainage can bring down hematocrit reduction, decrease dressing uses, and shorten the hospital stay which promotes rapid recovery. This review provides a detailed theoretical reference for the proper clinical application of drains and improves the efficient use of resources.

1. Introduction

Hip arthroplasty (HA) is a procedure that entails replacing the femoral head and acetabulum destroyed by disease or trauma with an artificial hip joint to avoid the risk of pain, deteriorated joint function, and undesirable quality of life for patients [1]. In the past few decades, an aging society, high obesity rates due to poor lifestyle habits, and an improved social healthcare system have led to a surge in demand for HA [2]. It is estimated that the number of HA and revision surgeries in the United States was projected to increase from 343,095 in 2012 to 668,700 in 2030 [3]. By 2014, the total annual number of HA in China was approaching 240,000 [4]. The dramatic increase in demand is undoubtedly posing a huge challenge to the government healthcare system. Although HA has been widely used and developed at a mature level worldwide, there are still some complications including incisional infection, blood loss, deep vein thrombosis, exudation, and impaired wound healing, for which there are many causes but the controversial issue is the drainage tube. In order to reduce the postoperative financial burden of patients and improve their quality of life, there is an urgent need to find evidence to explore the effectiveness and safety of postoperative drainage tube use [5-7].

The use of drainage in surgery dates back to the Hippocrates era in 400 BC, when it was recorded as placing drains in surgical wounds for drainage, and to this day, drainage for wound effusion remains as a fundamental principle in the surgical treatment of such conditions [8]. Postoperative drainage reduces hematoma formation and prevents incisional infection, but it also carries the risk of infection [9]. Many scholars have also conducted literature researches to address this issue. Most of the early meta-analysis of different types of drainage solved a controversy that drainage had no statistical relation with infection and indicated no significant advantage of drainage after HA over nondrainage, mainly on account of increased blood loss and blood transfusions, and prolonged hospitalization caused by drainage [10-12]. However, Koyano et al. [13] performed negative pressure drainage (NPD) and nondrainage (ND) in 51 bilateral HA patients and showed that the NPD group was superior to the ND group in terms of pain, wound skin temperature, thigh swelling, and infection, suggesting that postoperative drainage is beneficial to the patient's rapid recovery. Furthermore, a retrospective experiment showed that drainage placement after HA was not significantly associated with infection rates but could reduce the risk of wound formation of hematoma, while relatively increasing the blood loss and blood transfusion [14]. Lychagin et al. [15] demonstrated that drainage after HA can slightly alleviate pain, but increase the requirement for blood transfusion through a rigorous, double-blind, controlled clinical trial. The above indicates that the advantages and disadvantages of drainage tube application after HA have not been unified; therefore, this study will analyze the recent literature through systematic review and meta-analysis to refine the drainage tube application for clinical.

2. Materials and Methods

2.1. Search Strategy

We conducted a comprehensive literature search of the databases including CNKI, WanFang Database, PubMed, EMBASE, Cochrane Library, Ovid, and CINAHL, covering the period from 2000 to the present. Three separate search studies combining three sets of subject terms and text terms will be carried out: “hip arthroplasty” or “Arthroplasty, Replacement, Hip” or “Arthroplasties, Replacement, Hip,” “Arthroplasty, Hip Replacement” and “Drainage,” “drainage, surgical.” These terms were searched separately and grouped together using “and/or.” The detailed search strategy is shown in Table 1. The full text is presented in Chinese or English. Furthermore, we processed a supplementary search, chasing through back and forward citations and reference lists of previous systematic evaluations or similar reviews of related topics.

Table 1

Basic search strategy for PubMed, EMBASE, the Cochrane Library, and CBM.

Database	Search strategy
PubMed	(“Drainage”[Mesh]) AND ((“Arthroplasty, Replacement, Hip”[Mesh]) OR (((((((((((((((((((Arthroplasties, Replacement, Hip[Title/Abstract]) OR (Arthroplasty, Hip Replacement[Title/Abstract])) OR (Hip Prosthesis Implantation[Title/Abstract])) OR (Hip Prosthesis Implantations[Title/Abstract])) OR (Implantation, Hip Prosthesis[Title/Abstract])) OR (Implantations, Hip Prosthesis[Title/Abstract])) OR (Prosthesis Implantation, Hip[Title/Abstract])) OR (Prosthesis Implantations, Hip[Title/Abstract])) OR (Hip Replacement Arthroplasty[Title/Abstract])) OR (Replacement Arthroplasties, Hip[Title/Abstract])) OR (Replacement Arthroplasty, Hip[Title/Abstract])) OR (Arthroplasties, Hip Replacement[Title/Abstract])) OR (Hip Replacement Arthroplasties[Title/Abstract])) OR (Hip Replacement, Total[Title/Abstract])) OR (Replacement, Total Hip[Title/Abstract])) OR (Hip Replacements, Total[Title/Abstract])) OR (Replacements, Total Hip[Title/Abstract])) OR (Total Hip Replacements[Title/Abstract])) OR (Total Hip Replacement[Title/Abstract])))AND ((((randomized controlled trial[Publication Type])) OR (RCT [Title/Abstract])) OR (placebo[Title/Abstract])) OR (random[Title/Abstract]))
EMBASE	(“hip arthroplasty”/exp) OR (“arthroplasty, hip”:ab, ti OR “hip arthroplasties”:ab, ti OR “hip extracapsular arthroplasty”:ab, ti OR “hip joint alloplasty”:ab, ti OR “hip joint arthroplasty”:ab, ti OR “hip plasty”:ab, ti OR hiparthroplasty:ab, ti) AND (“drainage”/exp) OR (“drainage, surgical”:ab, ti)AND(“randomized controlled trial “/exp) OR (RCT:ab, ti OR random:ab, ti OR placebo:ab, ti)
The Cochrane	(“Total Hip Replacements” OR “Total Hip Replacement” OR “Hip Replacement, Total” OR “Replacement, Total Hip” OR “Hip Replacements, Total” OR “Replacements, Total Hip” OR “Replacement Arthroplasty, Hip” OR “Hip Replacement Arthroplasty” OR “Implantation, Hip Prosthesis” OR “Prosthesis Implantations, Hip” OR “Implantations, Hip Prosthesis” OR “Hip Replacement Arthroplasties” OR “Arthroplasty, Hip Replacement” OR “Arthroplasties, Hip Replacement” OR “Arthroplasties, Replacement, Hip” OR “Hip Prosthesis Implantation” OR “Prosthesis Implantation, Hip” OR “Replacement Arthroplasties, Hip” OR “Hip Prosthesis Implantations” OR “MeSH descriptor: [Arthroplasty, Replacement, Hip] explode all trees”) AND (MeSH descriptor: [Drainage] explode all trees)AND ((MeSH descriptor: [Randomized Controlled Trials] explode all trees) OR ((RCT OR placebo OR random):ti, ab, kw))
CBMa	(“Hip Replacement”(Chinese)[Common fields: smart] OR “Hip Replacement patients”(Chinese)[Common fields: smart] AND “Drainage tube”(Chinese)[Common fields: smart]) OR “Drainage”(Chinese)[Common fields: smart]AND 2000-2021[date]

aCBM: China Biology Medicine Database.

2.2. Inclusion and Exclusion Criteria

The inclusion criteria were as follows: (1) randomized controlled trials (RCTs) of patients who underwent HA evaluated the use of various drainage types after HA; (2) there is a comparison of various drainage methods with nondrainage methods; and (3) the primary outcomes including hemoglobin level (hb), hematocrit level (hct), blood transfusion volume, transfusion rate, and the change of thigh circumstances after HA will be used as the main reference for this study, and secondary indicators, including pain, hip function, complications, dressing use, and hospital days, will serve as important auxiliary supports.

In addition, we will exclude literature that meets the following criteria: (1) duplicate literature and (2) inaccessible critical research data and 95% confidence interval (CI) even if we try to contact the author.

2.3. Search Outcomes and Data Extraction

Two independent investigators will extract and summarize literature-related data according to the Cochrane Handbook for Systematic Reviews of Interventions [16], including author name, year of publication, country, experimental group, control group, sample size, hip arthroplasty, surgical access, anesthesia modality, antibiotics, thrombus prophylactic, follow-up, and outcome measures. If possible, we will complete the missing data by contacting the authors. During this period, a third senior researcher may consult in case of disagreement.

2.4. Quality Appraisal

The quality assessment tool recommended by the Cochrane Handbook V.5.1.0 (Cochrane Collaboration, London, UK) was used to evaluate the included studies. The tool assessed the following six aspects: random sequence generation, assignment concealment, blinding, participant outcomes and data integrity, no selective reporting bias, and no other biases. Each item was judged with high risk, low risk, and unknown to divide the studies into three levels: high risk of bias, moderate risk of bias, and the low risk of bias. The above analysis will support the credibility of the results presented in our later outcome indicators. Risk of bias graph and summary will be generated by Review Manager (RevMan) software 5.3.

2.5. Statistical Analysis

Review Manager 5.3 software (Cochrane, London, UK) was adopted to perform our meta-analysis. Summary estimates were expressed as mean difference (MD) and 95% CI. For continuous variables, standardized mean differences (SMDs) and 95% confidence intervals (CI) were used to report when multiple outcome measures existed, with nonparametric data transformed to mean (SD). Statistical effects of categorical variables included the risk ratio (RR), odd ratio (OR), and risk difference (RD). OR was used for defining the rate of exposure between the two groups while RR represented the ratio of the incidence of outcomes between groups. And RD stood for the difference in outcome rates. A chi-square test and I2 were used for heterogeneity analysis of each test result. Considering the different heterogeneity, a random effects model was applied if the result shown as P < 0.10 or I2 > 50%. Otherwise, a fixed effects model was applied. For reporting purposes, we classify heterogeneities 25-50%, 50-75%, and 75-100% as low, moderate, and high, respectively [17]. Moreover, sensitivity analysis was performed and the studies with high heterogeneities were further analyzed by subgroup analysis. Eventually, to access the published bias, funnel plots were conducted by Stata 14.1.

3. Results

3.1. Characteristics of the Included Studies

A total of 821 literature were searched initially from the database. Duplicate studies were removed from the software, leaving 571 articles for title and abstract screening. There were 507 excluded after screening the title and abstract. 19 substandard articles were removed by scanning the full text. The reasons for excluding were as follows: incomplete research data, not meeting the required implementation time, not randomized control trials, lack of strict quality supervision, and substandard outcomes. Finally, 19 literatures were selected from 50 articles involving 3354 participants. The detailed selective process is shown in Figure 1. All included studies were RCT which were published from 2004 to 2020. The characteristics of the included studies are shown in Table 2.

Figure 1

Flowchart outlining the literature search process.

Table 2

Demographic characteristics of included studies.

First author	Year	Country	Sample size	Surgical approach	Anesthesia	Antibiotics	Antithrombotics	Follow-up	Outcome
Chengnian Zhang	2019	China	CPD: 60ND: 60	Posterolateral	Joint anesthesia	Cefuroxime sodium	Enoxaparin	NR	(1) Hemoglobin values and its drop values(2) Hematocrit value and its drop values(3) Intraoperative blood loss(4) Postoperative drainage(5) Blood transfusion rate(6) Complications(7) VAS
Dong Yang	2008	China	CPD: 51ND: 51	Lateral	Epidural anesthesia	NR	Low molecular weight heparin sodium	1 year	(1) Thigh circumference change(2) Hospital stay(3) Complications(4) HHS
Dora, C.	2007	Switzerland	CPD: 50ND: 50	NR	Spinal/general anesthesia	Cephalosporin	Low molecular weight heparin	1 year	(1) Number of wound dressings(2) Time of wound discharge(3) Thigh circumference(4) Volume of hematoma(5) Total blood loss(6) Blood transfusions(7) Pain(8) HHS
Fagotti, L.	2018	Brazil	CPD: 42ND: 51	NR	General anesthesia	Cefuroxime	Enoxaparin	3 months	(1) Weight of dressings increases(2) Secretion area of dressings(3) Midthigh circumference(4) Increment need for reinforcement(5) Tenderness(6) VAS(7) Hematocrit and leucogram(8) CRP, VHS
Fichman	2015	Canada	CSD: 44ND: 44	NR	NR	Cefazolin	Low molecular weight heparin	6 weeks	(1) Hemoglobin levels(2) Transfusion rates(3) Length of hospital stay(4) Wound drainage(5) Pain management(6) Satisfaction(7) Hip function score
González	2004	USA	CSD: 53ND: 51	Posterolateral	Epidural anesthesia	NR	Heparin sodium	3 months	(1) Hematocrit(2) Transfusion(3) Change of thigh(4) Hospital stay
Horstmann, W. G.	2012	Netherlands	ABT: 50ND: 50	Anterolateral/posterolateral	Spinal/general anesthesia	Cefazolin	NR	3 months	(1) Perioperative blood loss(2) Autologous transfusion and Hb levels(3) Pain(4) HHS(5) SF-36 scores(6) Wound healing(7) Adverse events(8) Hospital stay
Horstmann, W. G.	2013	Netherlands	ABT: 102ND: 102	NR	NR	Cefazolin	Fondaparinux	3 months	(1) Perioperative blood loss(2) Autologous transfusion(3) Hemoglobin levels(4) Adverse events(5) Kidney function(6) Hospital stay
Kleinert, K.	2012	Switzerland	ND: 40CSD: 40ABT:40	Anterior	Joint/general anesthesia	Cefuroxime	Enoxaparin sodium	3 months	(1) Hemoglobin and Hkt values(2) Pain(3) Thigh swelling(4) Hematoma volume(5) Hospital stay(6) Homologous blood transfusions(7) HHS
Lychagin, A. V.	2020	Russia	CSD: 635ND: 527	Direct lateral	NR	Cefuroxime	Rivaroxaban	1 year	(1) Hemoglobin values(2) Hematocrit values(3) Blood transfusion rate(4) HHS(5) SF-36(6) Pain
Matsuda, K.	2007	Japan	CPD: 20ND: 20	NR	NR	NR	NR	3 months	(1) Hemoglobin values(2) Hematocrit values(3) Length of skin incision(4) Operative time(5) Intraoperative blood loss volume(6) Transfused autologous blood volume
Qiaoling Chen	2009	China	CPD: 35ND: 35	Modified Gibson incision	Spinal anesthesia combined with continuous epidural anesthesia	NR	NR	NR	(1) Blood transfusion volume(2) Off-bed time(3) Antifebrile time(4) Clearing time
Qingshan Li	2019	China	ND: 20CPD: 20CD: 20	NR	NR	NR	Rivaroxaban	1 year	(1) Temperature(2) HHS(3) Hemoglobin drop values(4) Blood transfusion volume
Tianwen Yang	2016	China	ABT: 25ND: 25	Anterior	NR	NR	Low molecular weight heparin calcium	NR	(1) Week of thigh root diameter(2) Hemoglobin and hematocrit(3) Coagulation function(4) D-2 polymers(5) Whether change dressings(6) Volume of drainage(7) Blood transfusion(8) Pain degree(9) Blood loss(10) Complications
Walmsley	2005	UK	CPD: 282ND: 295	Anterolateral	Spinal/general anesthesia	Cefuroxime	Low molecular weight heparin	3 years	(1) Hemoglobin level(2) Number of hips transfused(3) Length of stay in days(4) Complications
Weinan Zeng	2014	China	CPD: 83ND: 85	Posterolateral	General anesthesia	NR	Fraxiparine	3 months	(1) Blood loss(2) Volume of drainage(3) Hemoglobin levels(4) Blood transfusion cases(5) Blood transfusion volume(6) Complications(7) Hospital stay
Xiaofeng Han	2017	China	CPD: 26ND: 26	Lateral	General anesthesia	NR	NR	1 year	(1) Hematocrit(2) Total blood loss and blood(3) Transfusion(4) Serum protein(5) Complications(6) HAMA(7) Hip function score
Xiufeng Shi	2019	China	CSD: 36ND: 36	Lateral	Joint anesthesia	NR	NR	NR	(1) Blood loss

NR: not reported; ND: no drainage; CPD: common pressure drainage; CSD: closed-suction drainage; ABT: autogenous blood transfusion.

3.2. Methodological Quality and Risk of Bias for Included Trials

Risk of bias assessment is shown in Figures 2(a) and 2(b). As for random sequence generation, three studies [15, 18, 19] were completing using the computer-generated random number list, five studies [20-24] using the random number table, and two studies [25, 26] detailing the use of random cards. Besides, to avoid the assignment concealment, nine literatures [18, 23, 25–31] reported the use of sealed envelopes for distribution. Only two studies [18, 19] were blinding the outcome assessment, six studies [15, 18, 25, 26, 28, 31] were blinding the participants and personnel, and the remaining thirteen studies were completely open-label. Literature data are complete in terms of the literature currently included in the study. Four of the articles were not of high quality due to design issues [32-35].

Figure 2

Risk of bias graph (a) and risk of bias summary (b).

3.3. Effect of Drainage after HA on Primary Outcomes

Five studies with 428 participants reported hb levels. There was significant heterogeneity between studies (I2 = 88%, p = 0.94), and meta-analysis was performed using a random effects model. The results indicated that there is no statistically significant difference in hb level between drainage and nondrainage after HA (SMD = 0.02; 95% CI: -0.55, 0.59; p  > 0.05). Four literatures involving 1372 participants reported a decrease in hb levels. The results of the random effects model (I2 = 99%, p  = 0.17) analysis showed no statistically significant difference in decreased hb level between the drainage and nondrainage groups after HA (SMD = 1.55; 95% CI: -0.64, 3.74; p  < 0.00001) (Figure 3(a)).

Figure 3

Forest plot of comparison. (a) Hb levels in the drainage and non-drainage groups. (b) Hct levels in the drainage and nondrainage groups. Hb: hemoglobin; Hct: hematocrit.

Three studies with 1386 participants reported hct levels. Using a random effects model, we found no statistically significant difference in hct level between drainage and nondrainage after HA (MD = −1.86; 95% CI: -5.67, 1.95; p  = 0.34) with a low heterogeneity (I2 = 97%, p < 0.00001). However, a statistically significant difference (MD = 2.89; 95% CI: 1.3, 4.48; p  = 0.0004) in the meta-analysis applying the random effects model for decreased hct level with a high heterogeneity (I2 = 96%, p  < 0.00001) could be seen (Figure 3(b)).

In addition, there was no statistically significant difference (MD = −25.57; 95% CI: -84.76, 33.62; p >0.05) (Figure 4(a)) in the meta-analysis with fixed effects model between two groups occurring together with a moderate heterogeneity (I2 = 61%, p  = 0.4). Four studies for blood transfusion with the random effects model revealed that no statistically significant difference existed between drainage and nondrainage after HA (SMD = 0.48; 95% CI: -0.08, 1.04; p  > 0.05) (Figure 4(b)), but it also had a high heterogeneity (I2 = 85%, p  = 0.09) simultaneously. And a statistically significant difference (OR = 1.47; 95% CI: 1.12, 1.92; p  = 0.005) (Figure 4(c)) was obtained from the meta-analysis applying the fixed effects model for blood transfusion rate with a low heterogeneity (I2 = 0%, p = 0.43).

Figure 4

Forest plot of comparison. (a) Total blood loss. (b) Blood transfusion volume. (c) Blood transfusion rate. (d) Change of thigh circumstance.

Five studies with 436 participants reported the change of thigh circumstances. Using a fixed effects model, we found a statistically significant difference in change of thigh circumstances between drainage and non-drainage after HA (MD = −0.48; 95% CI: -0.66, -0.31; p  < 0.01) (Figure 4(d)) with a moderate heterogeneity (I2 = 61%, p  < 0.001).

3.4. Effect of Drainage after HA on Secondary Outcome

There were seven literatures involving 615 participants reporting the evaluation of pain alleviation after HA. The meta-analysis for visual analogue scale (VAS) with the fixed effects model shown in Figure 5(a) revealed that there was a statistically significant difference in both postoperative one day (POD1) (MD = −0.6; 95% CI: -0.79, -0.41; p  < 0.001) and postoperative two days (POD2) (MD = −0.38; 95% CI: -0.58, -0.18; p  < 0.001). But they were both moderately heterogeneous (I2 = 73%, p  < 0.00001; I2 = 73%, p = 0.0002), respectively. However, no statistically significant difference in VAS scores over postoperative three days (POD3) between drainage and nondrainage (MD = −0.15; 95% CI: -0.3, 0.01; p  > 0.05) with moderate heterogeneity (I2 = 73%, p  = 0.06).

Figure 5

Forest plot of the Effect of drainage after HA on secondary outcome. (a) VAS score. (b) Hip function score. (c) Hospital stay. (d) Uses of dressing. (e) Complications.

For the hip function score, four references including 294 participants showed that there was no statistically significant difference in hip function in three months (SMD = −0.23; 95% CI: -1.07, 0.6; p  > 0.05), six months (SMD = 0.04; 95% CI: -0.54, 0.62; p  > 0.05), and one year (SMD = −0.02; 95% CI: -0.32, 0.27; p  > 0.05) after HA between drainage and nondrainage which were lowly heterogeneous (I2 = 0%, p  = 0.59; I2 = 23%, p  = 0.89; and I2 = 60%, p  = 0.06) (Figure 5(b)).

The result of meta-analysis of hospital day including five studies is shown in Figure 5(c) which indicated a statistically significant difference in hospital day with postoperative drainage compared to non-drainage (MD = 1.06; 95% CI: 0.73, 1.39; p  < 0.001) under a fixed effects model with no heterogeneity (I2 = 0%, p  < 0.001).

Only three studies reported the number of dressings after HA in random effects model which marked no statistically significant difference between two groups (SMD = −0.91; 95% CI: -3.12, 1.3; p  > 0.05), with a high heterogeneity in it (I2 = 99%, p = 0.42) (Figure 5(d)).

Eventually, fourteen papers reported the occurrence of complications including superficial infections, DVT, osmosis, and hematoma, to evaluate its safety. The meta-analysis of this binary variable indicated that there was no statistically significant difference in complications (OR = 1.03; 95% CI: 0.75, 1.42; p  > 0.05) between two groups with a moderate heterogeneity (I2 = 36%, p = 0.843) (Figure 5(e)).

3.5. Sensitivity Analysis

The result of sensitivity analysis is shown in Table 3. Focusing on the hb level on POD3, since the study of Zeng et al. [31] was removed, heterogeneity was down from 88% to 0%. Regarding the meta-analysis for hct level, after the removal of the study of Zhang et al. [21], heterogeneity was dropped down from 78% to 0%. Similarly, we also resolved all the heterogeneity of the total blood loss index by removing according to Horstmann et al. [26]. However, in the study of blood transfusion volume, we successively removed the studies of Zeng et al. (2014) and Li et al. [22] resulting in a decrease in heterogeneity from 85% to 0%. Regarding the meta-analysis for change of the thigh circumstance, we tried to remove the study written by Yang et al. [29], and the heterogeneity was decreased from 61% to 0%. Pertaining to the meta-analysis for VAS on POD1, we attempted to reduce heterogeneity by removing Zhang et al. [21] and Horstmann et al. [25] and heterogeneity became 0%. Additionally, removing the study of Yang [23] can bring down the heterogeneity from 73% to 18% in the meta-analysis for VAS on POD2. Similarly, in the meta-analysis for VAS on POD3, we also tried to reduce the heterogeneity by removing the articles of Yang [23], and finally, we got no heterogeneities. But it is a pity that we were never able to address the heterogeneity of dressing use of up to 99%, which was only reduced to 92% by removing Zeng et al. [31].

Table 3

The result of sensitivity analysis.

Outcomes	Removed studies	Result of meta-analysis		Effect-model	Results of heterogeneity
		MD (SMD)	95% CI		I 2 (%)	p
Hb level on POD3	Weinan Zeng, 2014	-0.71	[-1.02, -0.39]	Random effects model	0%	0.03
Hct level	Lychagin, AV, 2020	-1.86	[-5.67, 1.95]	Random effects model	0%	0.73
Total blood loss	Horstmann, WG, 2013	-190	[-335.33, -44.67]	Fixed effects model	0%	0.83
Blood transfusion volume	Weinan Zeng, 2014	0.97	[0.65, 1.29]	Random effects model	67%	0.29
	Qingshan Li, 2019	0.9	[0.24, 1.55]	Random effects model	0%	0.88
Change of thigh circumstance	Dong Yang, 2008	0.1	[-0.35, 0.55]	Fixed effects model	0%	0.48
VAS on POD1	Chengnian Zhang, 2019	-0.22	[-0.58, 0.14]	Fixed effects model	68%	0.01
	Horstmann, WG, 2012	0.2	[-0.41, 0.81]	Fixed effects model	64%	0.02
VAS on POD2	Tianwen Yang, 2016	-0.38	[-0.58, -0.18]	Fixed effects model	18%	0.3
VAS after POD3	Tianwen Yang, 2016	-1	[-1.43, -0.57]	Fixed effects model	0%	0.81
Uses of dressing	Weinan Zeng, 2014	-3.24	[-3.7, 2.77]	Random effects model	92%	0.62

3.6. Publication Bias and Quality Control of Evidence

Due to the limited number of publications included, this study only showed publication risk bias for complications between drainage and nondrainage (Figure 6). The funnel plot revealed clear distributional anomalies which meant that the publication bias did not exist.

Figure 6

Funnel plot comparison of complications in the drainage and nondrainage groups.

4. Discussion

This review, through a comprehensive analysis of 19 publications, indicates that nondrainage after hip surgery decreases hct reduction, transfusion rates, dressing use, and hospital days. However, drainage plays a role in relieving early postoperative pain and thigh swelling.

According to Zhang et al. [6], total blood loss consists of both dominant and hidden blood losses. In the absence of wound drainage, an accumulation of the exudate in the joint cavity is higher than the surrounding intravascular pressure, which acts as a compressive hemostat for the site of the exudate. In addition, water and plasma seeping from the wound into the soft tissue cavity can be partially reabsorbed by the tissue, and the coagulation factors in the exudate also act as a hemostatic agent. In this study, nondrainage did not significantly reduce blood loss, and there was no significant effect on hb, hct, or transfusion, but there was a statistically significant difference in the hb and hct reduction and the transfusion rate. We considered that it was due to the fact that the experimental group in our included study was mainly based on normobaric drainage and ABT and blood loss was closely related to the type and duration of drainage and the amount of drainage pressure [12]. According to previous studies [36, 37], those causing more severe blood loss were NPD and CSD. Therefore, the effect of drainage on this part of blood loss may be smaller in this study, but there is still a clear advantage of nondrainage in reducing perioperative blood loss.

Also, this article differs from other studies in that drainage significantly relieves patients from early postoperative pain and thigh swelling. According to Jeon [38], when blood accumulating at the wound site cannot be drained, a hematoma is likely to form, and as the pressure of the hematoma increases, the accumulated blood infiltrates the tissues around the joint, causing peripheral edema and pain. Drainage can facilitate the drawn of the accumulated blood from the wound site and prevent the pressure of the hematoma on the surrounding tissues. Pain relief is often the primary goal of orthopedic surgery [39]. In the most recent year, perioperative pain was shown to be significantly associated with functional outcomes two years after total knee arthroplasty [40]. Although this has not been validated in HA, it is still an important entry point for pain management. In our study, drainage was not shown to be significantly associated with hip function, proposing that drainage after HA can significantly relieve pain in the early postoperative period, which may help physicians to intervene effectively in patients at a high risk for postoperative chronic pain. Most of the research literature in this study followed up in the later stages of recovery, but little attention was paid to chronic pain, which could also be attempted to be followed up in the later stages.

In contrast to previous studies on this issue [11, 12, 41, 42], the present study failed to agree on pain, total blood loss, hospital days, and complications. This may be related to the type of drainage included in the study, the specific procedure, demographic differences, anesthetic modality, and the material used to place the patient. Li et al. [11] suggested that drainage significantly increased total blood loss, and Chuang et al. [12] also suggested that drainage significantly increased blood loss. However, the former study focused on ABT, and its control group included closed negative pressure drainage, so the disparity between groups may be affected and the validity of the positive results obtained is open to further debate. The latter study only showed blood loss, which means that only explicit blood loss was considered, but total blood loss also included hidden blood loss, so the results are not directly comparable. Two studies concluded that drainage was not significant in relieving postoperative pain, but both studies did not specify the duration of pain evaluation [11, 12]. Mixing short-term and long-term evaluations would affect the accuracy of the results, while the evaluation of pain in this study was focused on three days postoperatively. In addition, only Li et al. [11] concluded that drainage can significantly increase the incidence of complications, and his experimental group was ABT. ABT loses some protein in the process of blood recovery, leading to hypoproteinemia in some patients, which can increase the rate of trauma infection [43, 44]. Finally, talking about hospital days, Li et al. [11] and Xue et al. [41] similarly concluded that drainage or not does not affect the number of hospital days. Their inclusion of patients with initial HA excluded surgical bipolarity, but our inclusion in the study did not make too many claims.

The heterogeneity of this article was high, but the absence of some basic information and the small literature on the study of each indicator prevented us from conducting subgroup analysis or more advanced meta-regression, so we attempted to find sources of heterogeneity through sensitivity analysis. We found several research indicators in which Zeng et al. [31], Zhang et al. [21], Yang [23], and Lychagin Alexey et al. [15] were the main sources of heterogeneity. Firstly, Zeng's surgical access included minimally invasive, which is likely to be important in reducing blood loss and dressing use [31]. We are also concerned about his transfusion criteria, which are higher than those known in other research literatures, which would be one of the potential factors contributing to the high heterogeneity in the volume of blood transfused as a study indicator. Unlike the above, Zhang et al. [21] had a low homogeneity with other research literatures in terms of pain evaluation on the first postoperative day, and we consider that the reason may be related to the effect of implementation of multimodal analgesic protocols. Yang [23] is indeed the article with high heterogeneity in pain scoring studies on the second and third postoperative days, but its study design did not differ significantly from other studies, in which case we would consider whether this is due to demographic differences or measurement error. And the reason why Lychagin brings 99% heterogeneity in hct level results may be due to the use of CSD, which exacerbates blood loss and therefore has a significant effect on this indicator [45].

5. Limitations

Firstly, this study only included articles published in eight major databases in both Chinese and English languages, which may have missed some grey literature searches resulting in the lack of comprehensiveness of the current findings. Secondly, the overall quality of the literatures we included in the study is not very high, and the basic information of many of the documents is incomplete, which has no way to analyze the impact on the heterogeneity of the results, although we have also done a corresponding sensitivity analysis according to the literatures. Moreover, the literature included in this study did not explore hidden blood loss, which leaves us in doubt as to whether drainage actually affects the amount of blood loss.

6. Conclusion

This study provides a comprehensive picture of the need for drainage in clinical applications. While drainage can relieve early postoperative pain and leg swelling, nondrainage can decrease hct reduction, transfusion rates, dressing use, and hospital days, facilitating rapid patient recovery. Combined with the current medical environment, we believe that drainage does not have more practical value and that it is more reasonable not to drain after HA. Given the paucity of literature on indicators, we look forward to the development of more high quality clinical studies in the future with large samples, multicenter, and rigorous design.