Drains in Breast Reduction: How Good Is the Recommendation Not to Use Them?

Abstract

A clinical practice guideline (CPG) from the American Society of Plastic Surgeons recommended not to use drains in breast reduction. This CPG was based on 3 randomized controlled trials (RCTs). The objective of this review was to double-check the methodological quality of the 3 RCTs. These RCTs were critically appraised using: (1) the “User's Guide to the Surgical Literature” checklist to critically appraise the methodological quality, (2) the CONSORT guidelines for reporting quality, and (3) the Cochrane risk-of-bias tool 2 (RoB 2) for risk of bias. Weaknesses were identified in all assessments for the 3 RCTs. Items with the poorest adherence in the “User's Guide” included: “Were patients stratified?,” “Was follow-up complete?,” and “Were all clinically important outcomes considered?” The overall adherence to the CONSORT reporting checklist across all 3 studies was moderate with 40.0%, 62.1%, and 48.3% adherence. All 3 RCTs had a similar low to moderate risk of bias, with no areas with a high risk of bias. None of the studies took into consideration a single critical outcome (such as major hematoma) and the outcome's minimally important difference as the basis for the sample size and power calculation of the study. All 3 RCTs additionally lacked clear reporting of treatment effect sizes or precision of estimates. Our re-examination of the evidence questions the recommendation of the CPG. We believe that the recommendation should have been: “We remain uncertain whether drains in breast reduction have a salutary effect.” As such, we recommend that a methodologically robust RCT be conducted to resolve the question of whether drains should be used in breast reduction.

The use of drains in breast reduction remains a controversial topic in plastic surgery. A survey done in 2006 with 2665 respondents found that 58% of plastic surgeons use drains in breast reduction.¹ An evidence-based clinical practice guideline from the American Society of Plastic Surgeons (ASPS) strongly recommended that the use of drains is not beneficial.² This was later revised in 2022 with the same recommendation against the use of drains, but with moderate strength of recommendation instead, although the rationale for this change in recommendation was not explained.³ This difference was not based on the development of new evidence: both were based on the same 3 randomized controlled trials (RCTs).^4-6 In addition, a Cochrane review in 2015 of the same 3 RCTs reached a similar conclusion that drains are not necessary in breast reduction.⁷ As these 3 RCTs have been highly influential in dictating the use of drains in breast reduction, it is important to individually reassess their methodological quality. Thus, the primary objective of this study was to double-check the methodological quality of the 3 RCTs upon which the 2 guidelines^2,3 were based.

METHODS

This study looked specifically at the 3 RCTs referenced by the ASPS breast reduction guideline: Collis et al,⁴ Corion et al,⁵ and Wrye et al.⁶ The quality of the RCTs was assessed by the following three methods: (1) the “User's Guide to the Surgical Literature” checklist to critically appraise the methodological quality,⁸ (2) the CONSORT checklist to assess the reporting quality,⁹ and (3) the Cochrane risk-of-bias tool 2 (RoB 2) to determine the risk of bias.¹⁰ Additional information about the characteristics of the study will be extracted: first author, publication year, journal of publication, country, aim, sample size, participant characteristics, and outcomes studied, including if the RCTs declared critical outcomes.

The methodological quality was assessed by the “User's Guide to the Surgical Literature” checklist. This checklist provides a framework for critical appraisal for surgical RCTs.⁸ It includes answering the following questions: Did the 2 surgical groups begin the study with a similar prognosis? Did the surgical groups retain a similar prognosis after the study started? How large was the treatment effect? How precise was the estimate of the treatment effect?⁸ Nonapplicable questions were excluded. In addition, it recommends evaluation of appropriate sample size and power. Each study was assessed for adherence to individual items within the “Users' Guide” checklist, with “yes,” “no,” or “can’t tell” assigned to each item based on reporting adherence. The results are presented descriptively.

The reporting quality was determined according to CONSORT guidelines.⁹ CONSORT is a 25-item checklist used worldwide to improve the reporting of RCTs and is recommended by the EQUATOR (Enhancing the QUAlity and Transparency Of health Research) Network.⁹ For each study, a reporting quality percentage was calculated by dividing the number of checklist items that met criteria for applicable items, with a score of 1 given if a checklist item was reported and 0 if the item was not reported or was unclear.

The risk of bias was determined with the RoB 2 tool. This involves 6 critical domains in which to examine the risk of bias in RCTs. The results of the risk of bias are presented using the Cochrane RoB representation table.¹⁰

Two independent reviewers (C.H., T.B.) extracted data independently and in duplicate. All conflicts were settled by consensus.

RESULTS

The characteristics of the 3 included RCTs^4-6 are summarized in Table 1. The RCTs were performed in the United Kingdom,⁴ the Netherlands,⁴ and the United States,⁶ with a mean sample size of 102 (range, 49-150). Collis et al⁴ and Wrye et al⁶ used an intrapatient randomized study design, whereby patients were randomized to receive a drain in either the left or right breast. In contrast, Corion et al used an interpatient randomization.⁵ All studies compared the incidence of postoperative complications, for example, hematoma, seroma, and infection, between drained breasts with no drains as a comparator (Table 2). Across the 3 RCTs, there was no statistically significant difference in complication outcomes between the patient groups, but none of the studies provided a power calculation.

Table 1.

Characteristics of Included Studies

Study characteristics	Collis et al, 20054	Corion et al, 20095	Wrye et al, 20036
Journal	British Journal of Plastic Surgery	Journal of Plastic, Reconstructive & Aesthetic Surgery	Plastic and Reconstructive Surgery
Country	UK	Netherlands	USA
Aim	To compare complications in patients with and without drains following breast reduction surgery	To test the hypothesis that omission of drains does not result in significant increase in postoperative complications and could result in a shortening of hospital stay	To investigate if drains affect complications in breast reduction in a randomized fashion
Type of randomization	Intrapatient	Interpatient randomization (1:1 allocation ratio)	Intrapatient
Total participants (n)	150	107	49
Participants in control arm (n)	150	52	49
Participants in intervention arm (n)	150 (70 left breast, 80 right breast)	55	49
Population characteristics	Agea = 37 years BMIa = 29 kg/m2	Agea = 36 years (drain group), 35 years (no-drain group) Minimum age = 16 years BMIa = 26.87 kg/m2 (drain group), 26.01 kg/m2 (no-drain group) Patients were not on anticoagulant medications and did not have a history of coagulation disorders	Agea = 33
Type of drains used	Bellovac (Astratech, Molndal, Sweden) closed suction drain	NS	No. 10 Blake or Jackson-Pratt
Weight of tissue removed	799ga	1110 g (drain group) 1085 g (no-drain group)	675 ga (drain group), 620 ga (no-drain group)
Pedicle (n)	Inferior (141), superomedial pedicle (9)	Craniomedial pedicle (55)	Inferior (48), NS (1)
Timing of drain removal	When drainage was less than 30 mL in 24 hours	When drainage was less than 20 mL for at least 24 hours	On postoperative day 1 before discharge
Primary outcome measured	NS	Complication occurrence (hematoma, infection, edema, necrosis, wound dehiscence)	Complication occurrence
Additional outcomes investigated	Length and amount of postoperative wound drainage Wound healing and other postoperative complications	Hospital stay duration Discomfort Pain Satisfaction	Comfort
Timing of outcome assessment	3 months	1 day after surgery and each postoperative visit	3 months
Key results	No significant difference between the 2 groups across all outcomes investigated	No significant difference between the 2 groups across all outcomes investigated	No significant difference between the 2 groups across all outcomes investigated

NS, not specified. ^aMean.

Table 2.

Outcomes Listed for Each Randomized Controlled Trial

Collis et al, 20054	Corion et al, 20095	Wrye et al, 20036
Abscess drainage	Discomfort due to the drainsa	Aestheticsc
Fat necrosis	Duration of hospital stayb	Comfortb
Hematoma	Hematomaa	Difference in nipple sensation
Hypertrophic scarring	Infectiona	Fat necrosisa
Infection major	Necrosis	Hematomaa
Infection minor	Edemaa	Hypertrophic scarring
Major wound healing (>1 cm)	Paina	Long-term patient satisfaction
Minor wound healing (<1 cm)a	Satisfaction ratea	Partial nipple lossa
Nipple loss	Wound dehiscence	Seroma
Seroma aspiration		Significant asymmetry
Spontaneous hematoma drainage		Swelling and weepingc
		Total nipple loss
		Wound breakdowna

^aIndicates there was a definition provided in the paper. ^bStatistically significant difference. ^cIn the results the authors state that they observed more swelling and weeping within the first 24 hours of undrained breasts. Additionally, they state that they observed no differences in aesthetics between the drained and undrained breasts.

Study appraisal using the “Users' Guide to the Surgical Literature”⁸ is described in Table 3. Items with the poorest adherence included: “Were patients randomized?,” “Were patients stratified?,” “Was follow-up complete?”, and “Were all clinically important outcomes considered?” The 3 RCTs were similar with regard to the majority of items on the “Users' Guide.” However, with Corion et al it was unclear whether patients in their 2 groups had similar prognostic factors because this was not reported. A sample size calculation was only provided by Corion et al, but the reasoning behind why a 30% difference was chosen and a calculation for the 80% power was not given.⁵ However, none of the studies took into consideration a single critical outcome (such as major hematoma) and the outcome's minimally important difference as the basis for the sample size and power calculation of the study. All 3 RCTs additionally lacked clear reporting of treatment effect sizes or precision of estimates.

Table 3.

Study Appraisal Based on the “Users' Guide to the Surgical Literature”⁷

Items on the “Users' Guide”	Collis et al, 20054	Corion et al, 20095	Wrye et al, 20036
Are the results valid?
Did the 2 surgical groups begin the study with a similar prognosis?
Were patients randomized?	Can’t tell	Can’t tell	Can’t tell
Were patients stratified?	No	No	No
Were patients' data analyzed in the surgical groups to which they were randomized?	Yes	Yes	Yes
Were patients in the 2 surgical groups similar with respect to known prognostic factors?	Yes	Can’t tell	Yes
Did the surgical groups retain a similar prognosis after the study started?
Was follow-up complete?	Can’t tell	Can’t tell	Can’t tell
What are the results?
How large was the treatment effect?	Can’t tell	Can’t tell	Can’t tell
How precise was the estimate of the treatment effect?	Can’t tell	Can’t tell	Can’t tell
How can I apply the results to patient care?
Were all clinically important outcomes considered?	Can’t tell	Can’t tell	Can’t tell

The following questions were excluded from the “Users' Guide” as they were not deemed to be applicable for the included studies: Did the investigators take into consideration the learning curve? Was randomization concealed (blinded and masked)? Were patients aware of group allocation? Were surgeons aware of group allocation? Were outcome assessors aware of the surgical group allocation? Were the study patients similar to my patients? Are the likely treatment benefits worth the potential harm and costs?

The overall adherence to the CONSORT reporting checklist across all studies was moderate: 40.0% (12/30 items in Collis et al⁴), 62.1% (18/29 items in Corion et al,⁵), and 48.3% (14/29 items in Wrye et al,⁶) (Supplemental Table 1).

Overall, all of the RCTs had a moderate risk of bias, with no areas with a high risk of bias (Figure 1). The domains of “Bias arising from the randomization process” and “Bias due to missing outcome data” appeared satisfactory (Figure 2), although there were some concerns in the domains of “Bias due to deviations from intended interventions,” “Bias in measurement of the outcome,” and “Bias in selection of the reported result” in all studies.

Figure 1.

Risk-of-bias “traffic light” plot.

Figure 2.

Risk-of-bias weighted bar plot.

DISCUSSION

This review aimed to critically appraise the quality of the RCTs influencing current practice guidelines for the use drains in breast reduction. Although all 3 RCTs provide consistent evidence that drains do not improve the rate of postoperative complications, including hematomas, seromas, and infection, weaknesses were identified in all 3 assessments that raise doubts on the veracity of the statement to not use drains.

Producing a credible RCT investigating the effectiveness of drains in breast reduction requires adherence to robust methodology, conferring validity, treatment effect, and applicability.⁸ One such item is participant and investigator blinding. This is impossible given that drains are inherently visible to the outcome assessor. Thus, unintentional bias from both groups cannot be excluded.⁸ The lack of blinding caused some concerns in the RoB 2 for the domains of “Bias due to deviation from intended intervention” and “Bias in measuring the outcome.” Although this increased the risk of bias, it also could not have been avoided. Yet, an area that could have been improved was the domain of “Bias in selection of the reported result.” The analysis in all 3 papers lacked sufficient detail in which to evaluate this domain, leaving some concerns. To further improve upon this, prespecified analysis plans would need to be available to see if there were any changes after the availability of unblinded outcome data.

The state of outcome research in breast reductions is also an inherent limitation, specifically with regards to critical outcomes. Currently, there is no core outcome set for breast reductions, resulting in marked heterogeneity in the literature.¹¹ As such, it is not possible to discern whether all clinically relevant outcomes were considered as well as critical complications. A critical complication, such as a hematoma requiring valuable healthcare resources for surgical drainage in the operating rooms, should be emphasized over less critical outcomes, such as minor seromas that can be drained in the surgeon's office. Although all studies generally referenced “complications” as a combination of multiple events in the description of their results, they also individually described each outcome for readers to have an appropriate understanding regarding the risk of each. Yet, the definitions of the outcomes used between the studies varied. For example, Corion et al only included hematomas if the breast became more swollen and firmer within the first 24 hours, whereas Wyre et al had a hematoma occur 2 weeks postoperatively which was drained surgically. Collis et al lacked definitions for most outcomes. As outcomes research in breast reduction becomes more robust, it is important to perform future RCTs that include all relevant critical outcomes in a standardized manner and indicate what outcome and effect size they included in the sample size calculation.¹² Thus, an RCT investigating the role of drains in breast reduction should have identified hematoma as a critical outcome. Furthermore the 3 RCTs should have defined what a critical hematoma was—for example, “a hematoma that needed to be evacuated in the main operating room.”

Among outcomes that were presented in these 3 RCTs, the size of the treatment effect was not communicated because absolute risk reduction, relative risk, and the numbers needed to treat were not reported. In addition, the studies lacked a confidence interval and therefore did not report on the precision of the treatment effect. The CONSORT guidelines recommend reporting these numerical values to better allow readers to interpret the application of the results.^8,9

Although 1 RCT described a sample size calculation, the authors did not provide the reasoning behind why a 30% difference was chosen or provide a calculation for the 80% power. None of the studies reported consideration of a critical outcome, such as major hematoma incidence, when choosing their sample size. When designating a study sample size, there should be evident consideration of the minimal important difference for a specific critical outcome. Without an appropriate sample size calculation, it is difficult to discern whether the study is adequately powered to answer the research question. On one hand, a Type I error (α) may arise, in which a statistically significant difference may be found between groups when, in reality, the difference is insignificant.^13,14 On the other hand, a Type II error (β) may arise, in which a statistically insignificant difference is found, despite there being an actual significant difference between groups.^13,14 As such, without a rigorous sample size, it is difficult for readers to trust the significance or lack of significance of results reported in these 3 RCTs.

The lack of methodological transparency in the 3 RCTs on which the ASPS guidelines were based makes it challenging for authors to make a definitive conclusion on a treatment arm's superiority and for readers to confidently interpret the study results. Furthermore, while all 3 RCTs were randomized, minimal detail was provided surrounding the actual randomization process, such as sequence generation, randomization type (and inclusion of any restriction), allocation concealment, and steps taken to implement the randomization scheme.⁹ Overall, the RCTs had moderate adherence to the CONSORT guidelines and failed to report on items such as randomization details and study limitations, which highlights a lack of robustness in their reporting quality.

Furthermore, although there is no consensus on the recommended postoperative duration of drains following breast reduction, each RCT had a different protocol for the removal of drains.¹ In one sense, the variety of removal times may indicate that any style in which drains are used will not result in improved complication rates. On the other hand, it is possible that the studies failed to implement drains to their best potential. In addition, the procedural steps were missing in Collis et al and Wyre et al, but Corion et al state that no epinephrine was administered in their breast reductions. As techniques for breast reduction can vary greatly depending on the surgeon, the applicability of the results in these RCTs can be questioned.

The use of an intrapatient randomization methodology in 2 RCTs was beneficial in accounting for potential patient-specific factors that may influence complication rates. However, a key limitation across all 3 studies was the limited reporting of demographic factors for the patient population, which makes it difficult to elicit the effect of drains on high-risk populations, such as patients who smoke, have diabetes, have a longer suprasternal notch to nipple distance, or have a greater resected specimen weight.¹⁵ A subgroup analysis or risk stratification in the randomization may have accounted for this variability in patient complication risk.

CONCLUSIONS

Although the 3 RCTs on which the ASPS clinical practice guideline is based reported cohesive and statistically significant results that drains do not prevent complications in breast reductions, the moderate reporting quality and lack of methodological rigor raises suspicion around the implications that “drains are not useful/necessary.” We do not wish to disregard the work that has been done in this area and recognize that these trials have been instrumental in guiding clinical practice to date. Yet, in the existing literature, there is insufficient evidence. Our analysis suggests that a recommendation on the use of drains in breast reduction should be: “We are uncertain at this time.” As such, we recommend that further robust RCTs be conducted to answer this ongoing controversial clinical question. Future trials should follow the 6 preconditions of credible research: (1) statement of the clinical research question, (2) selection of the appropriate study design, (3) measurement of critical outcomes, (4) use of correct scales to measure such outcomes, (5) inclusion of economic evaluations, and (6) adherence to the EQUATOR guidelines.¹⁶ The development of a core outcome set for breast reduction will additionally ensure that all critical outcomes are uniformly reported across future trials.

Supplemental Material

This article contains supplemental material located online at www.aestheticsurgeryjournal.com.

Disclosures

Dr. Thoma is an editor of the book Evidence-Based Surgery: A Guide for Understanding and Interpreting the Surgical Literature and receives royalties from Springer (New York). The other authors have no conflicts of interest to disclose.

Funding

The authors received no financial support for the research, authorship, and publication of this article.