Blood loss, complications, and the role of suction drains in shoulder arthroplasty: a systematic review and meta-analysis

Abstract

Background

Total shoulder replacement, including both anatomic and reverse procedures are one of the treatment options for shoulder arthritis and proximal humerus fractures, with consistently favorable outcomes in terms of pain relief and function improvement. During surgery, some surgeons prefer to place suction drains in the deep tissue to prevent hematomas or seromas, thereby reducing risks like wound dehiscence and infections. However, the routine use of suction drains in other orthopedic surgeries, such as hip and knee arthroplasty, has been questioned, with studies suggesting they may lead to extended hospital stays and increased need for transfusions without significantly reducing complications. Given the conflicting evidence in the literature about suction drains in shoulder arthroplasty, this study aims to evaluate their efficacy in reducing complications and in consequence the need for hospital stay and reoperations.

Methods

This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines. The inclusion criteria encompassed studies comparing outcomes of patients undergoing shoulder arthroplasty (anatomic or reverse) with and without suction drains. Databases such as PubMed, Lilacs, Cochrane Library, and Scopus were searched, with no restrictions on language or publication date. Data extraction was performed for duration of hospital stay, transfusions, infections, wound dehiscence, reoperations, and hemoglobin changes. Meta-analyses were conducted using a random-effects model for mean differences (MDs) and odds ratios (ORs).

Results

Six studies with a total of 22,710 patients were included. The mean patient age was 68.8 years, with a slight female predominance. Anatomic and reverse shoulder arthroplasties were both commonly performed. There was no significant difference in hospital stay (MD: 8.78 hours), infection rates (OR: 0.64; 95% confidence intervals (CI): 0.30-1.35), or reoperation rates (OR: 1.62; 95% CI: 0.16-16.23) between patients with drains and those without. However, hemoglobin loss was significantly greater in the drain group (MD: 0.31; 95% CI: 0.02-0.60).

Conclusion

The use of suction drains in shoulder arthroplasty does not significantly reduce infection or wound-related complications and is associated with increased hemoglobin loss. These findings suggest that the routine use of drains in shoulder arthroplasty may not provide tangible benefits and should be reconsidered to optimize patient outcomes.

Total shoulder replacement, including both anatomic (aTSA) and reverse (rTSA), are one of the surgical options for managing primary and secondary shoulder arthritis as well as proximal humerus fractures.²⁵ Those procedures have consistently good and predictable outcomes reported in the literature, offering significant pain relief and improved function.⁴ To further optimize surgical outcomes, it is a common practice by some surgeons to place suction drains in the deep tissue planes during joint arthroplasty.⁹ The primary rationale behind this practice is to prevent the accumulation of postoperative hematomas or seromas, which could otherwise lead to complications such as wound dehiscence and infections,⁷ thereby promoting better wound healing and postoperative recovery.

However, the routine use of suction drains in orthopedic surgeries, such as total knee arthroplasty (TKA) and total hip arthroplasty (THA), has been increasingly questioned. Studies on these procedures have suggested that drains do not significantly reduce the overall incidence of complications.^1,16,27 On the contrary, some reports indicate that suction drains may be associated with prolonged hospital stays and an increased need for blood transfusions, raising concerns about their overall efficacy and necessity.²⁶ This has led to a reevaluation of their role in arthroplasty. While some studies on shoulder arthroplasty have reported potential benefits of suction drains,^10,24 no comprehensive analysis of this topic has been conducted to date.

Therefore, this study aims to critically evaluate the existing literature on the use of suction drains in shoulder arthroplasty. The primary goal is to assess whether these devices offer tangible benefits in terms of reducing complications such as infections, revision surgeries, wound dehiscence, hospital stay duration, and the necessity for transfusions. We hypothesized that the use of suction drains in shoulder arthroplasty does not reduce infection rates or overall complications but is associated with increased blood loss and prolonged hospital stays, based on similar findings in THA and TKA.

Methods

This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA)¹⁸ guidelines.

Eligibility criteria

The inclusion criteria for this study encompassed all methodological designs that reported original clinical data on patients who underwent both rTSA and aTSA, with a specific focus on comparing the use of suction drains vs. nonuse of suction drains. Eligible studies included prospective clinical trials, cross-sectional studies and other relevant designs. We included all surgical techniques, indications (trauma or degenerative), types of suction devices and all shoulder replacement implant systems and manufacturers. Studies were excluded if they focused solely on other types of interventions during shoulder replacement surgery, unrelated to the use of suction drains or if they presented no comparison group. There were no restrictions on language or publication date. The inclusion and exclusion criteria for this study are summarized in Table I.

Table I.

Inclusion and exclusion criteria of the targeted studies.

Inclusion criteria	Exclusion criteria
Clinical trials and cross-sectional studies, reporting original data.	In vitro studies, systematic reviews, technical notes, and review articles.
Comparison between use and nonuse of suction drain.	Studies without comparison groups regarding use of suction drain.
Patients that underwent total shoulder arthroplasty (Anatomic and reverse).

Information source and search strategy

Before the initial search, registration of the systematic review with the International Prospective Register of Systematic Reviews was completed (CRD42024504811). A comprehensive literature search was developed by the authors and was run on May 8, 2024, in the following databases: Pubmed/MEDLINE, Lilacs, Cochrane Library and Scopus. Both controlled vocabularies (e.g. MeSH terms) and keywords were searched, a detailed search string is depicted in Supplementary Material S1. There were no restrictions on country in which the study was conducted, age of participants, or language of publication. Additionally, a hand search was conducted of the reference lists of selected articles. All identified studies were exported to reference management software (Rayyan, Cambridge MA, USA) to remove duplicates and organize the studies for screening. Two reviewers (V.L.B. and A.P.V.) independently screened the studies for eligibility based on the title and abstract, and full-text articles were retrieved for studies that met the eligibility criteria. Disagreements between reviewers were resolved by a third independent reviewer (G.V.L.), as suggested by the reporting guidelines for meta-analysis PRISMA.¹⁸

Data extraction

Specific information on publication origin, study design, type of shoulder implant evaluated, drain protocol implemented, follow-up and outcomes of interest were collected, accordingly to the Cochrane Handbook for Systematic Reviews of Interventions¹⁴ recommendations. Two authors (V.L.B. and A.P.V.) independently extracted data from the included studies using a standardized data extraction form. A third author (G.V.L.) reviewed the extracted data and decided on conflicts when present. The following outcomes data were considered of interest and extracted from each study: (1) hospital stay time; (2) postoperative blood transfusion; (3) superficial or deep wound infection; (4) wound dehiscence; (5) reoperations or readmissions; (6) hemoglobin change (from preoperative to postoperative). We opted not to include functional and clinical outcomes (eg, range of motion and patient-reported outcomes) as those are not usually reported in the early postoperative stage (6 weeks or less) and most likely are not related to the use of suction drains.

Study risk of bias

The Cochrane Risk of Bias Tool 2⁶ was employed for the assessment of the risk of bias in the included randomized clinical trials, the Risk Of Bias In Nonrandomized Studies—of Interventions²² tool was used for studies that did not use randomization to allocate interventions and the Risk Of Bias In Nonrandomized Studies—of Exposure¹³ tool was used to asses the risk of bias in observational epidemiological studies. The Risk Of Bias In Nonrandomized Studies tool evaluates 7 key domains, including confounding, participant selection, intervention classification, deviations from intended interventions, missing data, outcome measurement, and selective reporting.^13,22 Each domain is rated as having low, moderate, serious, or critical risk of bias. Two authors independently evaluated the included articles for risk of bias (A.P.V. and O.M.) and conflicts were decided by a third independent author (V.L.B.)

Effect measures

The length of hospital stay was expressed in hours and by means and mean differences (MD), while event incidences (Transfusions, infection, dehiscence, reoperations) were compared using risk ratios or odds ratios (ORs). Changes in Hb levels were reported as MDs in mm Hg.

Synthesis methods

Initially, means, data on events, total sample sizes, and standard deviations were extracted from each study for both the drain and control groups. Ninety-five percent confidence intervals (CIs) for the MDs were calculated for each study and for the overall effect size (MD or OR). For the MDs meta-analysis, the DerSimonian-Laird random-effects model was chosen to account for variability both within and between studies. The overall effect size was calculated by combining the MDs from each study using inverse variance weighting. The Mantel–Haenszel random-effects model was employed to pool the ORs from individual studies. The overall OR was calculated by combining the individual ORs from each study using inverse variance weighting. Heterogeneity was quantified using the DerSimonian-Laird method (τ), I2 statistic, and Chi-squared statistic in accordance with the guidelines delineated in the Cochrane manual. Chi² test was interpreted as statistically significant if < 0.05. I2 statistics followed Cochrane Handbook for interpretation: 0%-40% might not be important; 30%-60%, may represent moderate heterogeneity; 50%-90% may represent substantial heterogeneity; and 75%-100% represents considerable heterogeneity.

All data analysis was performed using R software (version 4.1.2; R Foundation for Statistical Analysis, Vienna, Austria).

Results

Study selection

We identified 304 records on database searching (Pubmed/MEDLINE, Lilacs, Cochrane Library and Scopus). After the removal of duplicates, 157 articles were screened through their titles and/or abstracts. Considering the eligibility criteria, 144 studies were excluded and 13 were retrieved for full-text screening in more detail and assessed for eligibility. Of those, 7 studies were excluded due to: 2 were systematic and narrative reviews and 5 did not presented a comparison group (control group). Six studies were included for analysis. A PRISMA flow diagram was used to document the study selection process (Fig. 1).

Figure 1.

PRISMA flowchart illustrating the study selection process. Reasons for exclusion at various stages are noted. PRISMA, Preferred reporting items for systematic reviews and meta-analyses.

Study characteristics

Studies characteristics including groups distribution and follow-up is summarized in Table II.

Table II.

Summary of included studies characteristics.

Study	Year	Country	Study design	Diagnosis	n = (drain use)	n = (no drain)	Type of shoulder prothesis	Manufacturer	Time to remove drains
Gartsman et al10	1997	United States	Prospective clinical trial	Osteoarthritis, rheumatoid arthritis, avascular necrosis, cuff tear arthropathy	51	49	Anatomic total and hemi shoulder arthroplasty	n/a	24 h
Erickson et al5	2016	United States	Retrospective cohort	n/a	300	64	Anatomic and reverse total shoulder arthroplasty	n/a	Up to 24 h
Chan et al2	2019	United States	Retrospective cohort	n/a	5515	16,545	Anatomic total, reverse total and hemi shoulder arthroplasty.	n/a	n/a
Frye et al8	2019	United States	Retrospective cohort	Osteoarthritis (revision cases excluded)	111	267	Anatomic and reverse total shoulder arthroplasty	n/a	n/a
Trofa et al24	2019	United States	Prospective clinical trial	Osteoarthritis, inflammatory arthritis, posttraumatic arthritis, or proximal humeral malunion.	50	50	Anatomic and reverse total shoulder arthroplasty	Arthrex, Zimmer, DePuy, Stryker, Tornier	30.2 h (mean)
Garcia-Maya et al11	2023	Spain	Retrospective cohort	Degenerative shoulder conditions and acute trauma	44	59	Reverse total shoulder arthroplasty	n/a	Up to 24 h

Patients

A total of 22,710 patients were included across the analyzed studies. For the studies that provided age data^2,5,10,11,24—the pooled mean age was approximately 68.8 (±9.2) years. Gender distribution was reported in several studies: a total of 12,853 females and 9479 males across the studies that reported gender distribution, with an overall distribution of 57.6% female and 42.4% male. The surgical indications varied, with osteoarthritis being the most commonly cited, found in 3 of the included studies.^10,11,24 Proximal humeral fracture were mentioned by Trofa et al and Garcia-Maya et al, while inflammatory arthritis was only included in Trofa et al. Anticoagulation use was reported in 8 patients in Trofa et al. Garcia-Maya et al was the only study mentioning a percentage of cases with Tranexamic acid use (67.6%). Other studies did not provide details on anticoagulation or specific surgical information.

Type of shoulder replacement

Across the included studies, aTSA and rTSA were the most common procedures. aTSA was performed in 4 studies,^2,8,11,24 rTSA in 4 studies^2,8,10,24 and hemi shoulder arthroplasty in 2 studies.^2,11

Risk of bias in individual studies

The risk of bias in the included studies was assessed using the Risk Of Bias In Nonrandomized Studies—of Interventions and Risk Of Bias In Nonrandomized Studies—of Exposure tools, as the included articles are nonrandomized clinical trials^11,24 and observational studies.^2,5,8,10 The overall risk of bias is summarized in Figures 2 and 3. Most studies exhibited a moderate level of risk due to confounding, with only 1 study classified as having a high risk of bias in this domain. Similarly, the risk of bias due to missing data was moderate or not reported in most studies, while 1 study was graded with a high risk. The majority of studies were assessed as having a moderate or low overall risk of bias, with only 1 study rated as having a high risk.⁸

Figure 2.

Risk of bias assessment using the ROBINS-I tool. Summary of risk of bias across the included studies, showing the proportion of studies at low, moderate, serious, or critical risk of bias in each domain. ROBINS-I, risk of bias in nonrandomized studies—of interventions.

Figure 3.

Risk of bias assessment using the ROBINS-E tool. Summary of risk of bias across the included studies, showing the proportion of studies at low, moderate, serious, or critical risk of bias in each domain. ROBINS-E, risk of bias in nonrandomized studies—of exposure.

Synthesis of results

A meta-analysis was performed by pooling the data from 6 studies (7 cohorts). Garcia-Maya et al presented data of 2 cohorts of patients who underwent shoulder replacement, 1 for degenerative and another for traumatic (fractures) causes and so, those were examined independently. In those studies, we examined 5 outcomes, including hospital stay, hemoglobin change, infection rates, wound dehiscence rate, and reoperations rate.^{2,5,8,10,11,24}

Hospital stay

Regarding hospital stay, the combined data from the 5 included studies (6 cohorts) yielded a pooled MD of 8.78, with a CI of −2.81 to 20.37 (Fig. 4). The heterogeneity across studies was considerable (I² = 95%%; τ² = 170.3; P < .01), suggesting inconsistent findings between the studies. This analysis does not demonstrate a significant difference between the drain group and the control group.

Figure 4.

Forest plot summarizing the effect estimates for the duration of hospital stay across included studies. SD, standard deviation; MD, mean difference; CI, confidence interval.

Hemoglobin change

For hemoglobin change, the combined data from the 2 included studies (3 cohorts) yielded a pooled MD of 0.31, with a CI of 0.02-0.60 (Fig. 5). The heterogeneity across studies was not important (I² = 29%; τ² = 0.019; P < .01), suggesting consistent findings between the studies. This analysis demonstrated a significant difference between the drain group and the control group, showing that the control group had smaller hemoglobin changes compared to the drain group.

Figure 5.

Forest plot summarizing the effect estimates for the Hb change preoperatively and postoperatively across included studies. Hb, hemoglobin level; SD, standard deviation; MD, mean difference; CI, confidence interval.

Transfusion

In terms of transfusion rates, the combined data from the 4 included studies (4 cohorts) yielded an OR of 1.16, with a CI of 0.54-2.50 (Fig. 6). The heterogeneity across studies was not important (I² = 14%; τ² = 0.23; P = .32), suggesting consistent findings between the studies. This analysis does not demonstrate a significant difference between the drain group and the control group.

Figure 6.

Forest plot summarizing the effect estimates for transfusion events across included studies. OR, odds ratio; CI, confidence interval.

Infection

Examining infection rates, the combined data from the 6 included studies (6 cohorts) yielded an OR of 0.64, with a CI of 0.30-1.35 (Fig. 7). The heterogeneity across studies was not important (I² = 0%; τ² = 0.0; P = .79), suggesting consistent findings between the studies. This analysis does not demonstrate a significant difference between the drain group and the control group.

Figure 7.

Forest plot summarizing the effect estimates for infection events across included studies. OR, odds ratio; CI, confidence interval.

Wound dehiscence

Regarding wound dehiscence rates, the data from the 6 included studies (6 cohorts) had no dehiscence events, therefore OR and heterogeneity statistics were not calculated.

Reoperation

When looking at reoperation rates, the combined data from the 3 included studies (3 cohorts) yielded an OR of 1.62, with a CI of 0.16-16.23 (Fig. 8). The heterogeneity across studies was not important (I² = 14%; τ² = 0.38; P = .28), suggesting consistent findings between the studies. This analysis does not demonstrate a significant difference between the drain group and the control group.

Figure 8.

Forest plot summarizing the effect estimates for reoperations across included studies. OR, odds ratio; CI, confidence interval.

Risk of bias across studies

Funnel plots for all the analyzed outcomes were assessed. Evidence of publication bias among the included studies remains inconclusive. Although no clear asymmetry on funnel plots was identified, the limited number of studies hinders definitive conclusions regarding publication bias. Publication date and sample variation in individual studies are reported in Table II.

Discussion

This systematic review and meta-analysis found that the routine use of suction drains in shoulder arthroplasty does not significantly reduce infection rates or wound-related complications, while being associated with increased hemoglobin loss. Although suction drains have been traditionally used to minimize hematoma formation, with the goal of reducing infection and wound complications,⁹ their actual clinical benefits remain uncertain. Previous studies in TKA¹⁷ and THA^19,23 have already questioned their efficacy, and while some recent investigations have explored their role in shoulder arthroplasty,¹⁰ a comprehensive synthesis of the evidence was lacking. This study addresses that gap by providing an updated evaluation of their impact on key clinical outcomes.

Our findings reveal that patients who had suction drains following shoulder replacement exhibited a significantly greater decrease in hemoglobin levels compared to those without drains. This suggests that the use of drains may be associated with a greater depletion of blood count which has been also observed by others on the use of suction drains for knee arthroplasty¹ and spine surgery.¹⁵

A potential explanation for this could be the continuous drainage of blood clots, which may hinder the coagulation of small deep vessels, thereby contributing to increased blood loss.

Interestingly, no statistically significant differences were found in the rates of infection, wound dehiscence, reoperation, or the need for blood transfusions between the 2 groups. These results imply that suction drains may not provide the intended protective effects against postoperative complications and that their use could, in fact, contribute to increased blood loss without any additional benefits in terms of reducing infection or other wound-related issues.

Similar trends have been observed in prior research focused on hip and knee arthroplasty: Si et al²¹ in a meta-analysis found no differences in hematoma formation or infection rates between closed drainage or non drainage in TKA. For THA, Zhou et al²⁷ found no increase in the incidence of infection or other complication with the use of suction drain but an increase in the need of transfusion in the suction group. The parallels between shoulder, hip, and knee replacement surgeries provide a basis for reconsidering the routine use of suction drains in joint arthroplasties more broadly.

Despite no pooled MDs in hospital stay between the drain and control groups, high heterogeneity was observed across studies for this outcome. This variability could be explained by differences in surgical setting (inpatient vs. outpatient), institutional discharge policies, and patient baseline comorbidities.

This study represents a significant contribution to the literature as it is to the best of our knowledge the first to systematically review and conduct a meta-analysis on the effects of suction drains specifically for shoulder replacement surgery being able to draw robust conclusions regarding the impact of suction drains on postoperative outcomes in these procedures.

Despite its strengths, this study is not without limitations. The analysis relies heavily on the quality and availability of existing studies, which may introduce biases related to study design and methodology. It is important to note that most of the studies were graded as having a moderate or low risk of bias, with one study classified as having a high risk of bias. Additionally, several factors that could influence postoperative outcomes were not consistently controlled across studies, such as the use of tranexamic acid,³ patient comorbidities at baseline,¹² and variations in surgical technique.²⁰ Furthermore, the inclusion of both aTSA and rTSA procedures introduces variability, as these implant types have distinct biomechanical properties, surgical exposures, and perioperative considerations²⁰ that may impact blood loss and complication rates. Differences in inpatient vs. outpatient surgical settings across studies may have also influenced factors such as hospital stay and postoperative care protocols. Collectively, these variables may have affected the outcomes assessed and should be considered when interpreting the results.

Conclusion

In this study, no significant benefits from using suction drains in shoulder replacement surgery regarding infection, wound complications, reoperation, or transfusion needs were observed. Moreover, patients with drains experienced greater hemoglobin loss than those without. Our findings suggest that suction drains may not be necessary or benefitial in shoulder arthroplasties.

Disclaimers

Funding: No funding was disclosed by the authors.

Conflicts of interest: The authors, their immediate families, and any research foundations with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.