Comparison of Output Volume Thresholds for Drain Removal After Selective Lateral Neck Dissection: A Randomized Clinical Trial

Key Points

Questions  Can the threshold for drain removal be safely raised to an output volume of less than 100 mL in a 24-hour period, and does it lead to a reduced hospital length of stay?

Findings  In this single-blind randomized clinical trial of 53 patients undergoing selective lateral neck dissection, the group with the 100-mL threshold for drain removal had a reduced mean hospital length of stay and mean duration of drain use with no increase in complications or seroma formation compared with the group with the 30-mL threshold.

Meaning  A 100-mL threshold for drain removal for certain selective lateral neck dissections appears to be safe and may significantly reduce time to drain removal and hospital length of stay.

Importance  Limited evidence is available to guide drain removal after selective lateral neck dissection (SLND). Patients may have drains left in longer than necessary, leading to patient discomfort, longer hospitalizations, and increased costs.

Objective  To compare 2 output volume thresholds for drain removal after SLND.

Design, Setting, and Participants  This single-blind randomized clinical trial included a consecutive sample of all adult patients undergoing unilateral or bilateral SLND of levels I to III, I to IV, II to III, or II to IV from March 1, 2015, to December 1, 2016, at a tertiary academic medical center. Eligible patients had at least 30 days of follow-up. Patients undergoing a parotidectomy, a level V lymphadenectomy, or an SLND that communicated with the upper aerodigestive tract or who had a suspected chylous fistula on the first postoperative day were excluded from enrollment. Sixty-five patients were offered enrollment and 12 refused. Fifty-three patients who underwent 67 SLNDs were included in the final analysis, with no patients lost to follow-up. Analysis was based on intention to treat.

Interventions  On the first postoperative day, patients were randomized to either a drain removal threshold of less than 30 mL or less than 100 mL during a 24-hour period.

Main Outcomes and Measures  Duration of drain use, hospital length of stay, and wound complications for both groups.

Results  Among the 53 patients with 67 SLNDs included in the analysis (45 men [85%] and 8 women [15%]; mean age, 58.5 years [95% CI, 53.2-64.5 years]), 32 SLNDs were randomized to the 100-mL group and 35 were randomized to the 30-mL group. No meaningful differences in preoperative characteristics were noted between groups. Two seromas occurred in the 100-mL group (2 of 32 [6.3%; 95% CI, 0%-13.5%]) and in the 30-mL group (2 of 35 [5.7%; 95% CI, 0%-14.6%]). No hematomas, chylous fistulas, or wound infections occurred. The 100-mL group had a 1.87-day reduction in mean hospital length of stay (95% CI, 0.66-3.10 days).

Conclusions and Relevance  A volume threshold for drain removal of 100 mL during a 24-hour period after SLNDs appears to be safe and may significantly reduce duration of drain use and hospital length of stay.

Trial Registration  clinicaltrials.gov Identifier: NCT03113526

Evidence regarding the need for and management of postsurgical closed-suction drains (CSDs) after selective lateral neck dissections (SLNDs) in head and neck surgical oncology is scarce. Closed-suction drains are routinely used by most head and neck surgeons for SLNDs performed for the treatment of various cancers. A recent survey of more than 360 head and neck surgeons found that 86% routinely place a single CSD after a selective neck dissection and more than 60% use a criterion of less than 30 mL of output for 24 hours for removal.¹ Although CSDs are routinely placed for many different procedures across surgical fields, recent studies have shown a lack of definite benefit. Quiz Ref IDFor example, after elective cholecystectomy, hepatectomy, and thyroidectomy with and without central neck dissections, the routine use of CSDs does not appear to prevent complications.^2-4 A prospective randomized trial by Lee et al⁵ showed no increased complications and a significant decrease in hospital stay in thyroidectomies with central neck dissections performed without drain placement. Quiz Ref IDA recent meta-analysis of randomized clinical trials and 2-armed studies comparing drain and no-drain groups after thyroidectomy⁶ found that the drain group had a higher infection rate (odds ratio, 2.94) and longer hospital length of stay (1.16 days) compared with the no-drain group. In 2010, Mekel et al⁷ demonstrated the possible safety of not using CSDs at all after SLNDs for thyroid cancer.

Earlier reports of CSDs after radical neck dissections^8,9 used the criterion of 30 mL or less output during a 24-hour period without mention of evidence to support this practice. Urquhart and Berg¹⁰ demonstrated that SLND drains had a mean total output of 116 mL when removed when the volume was less than 25 mL during a 24-hour period, which took a mean of 4 days. A similar study⁵ found the mean total drain output for thyroidectomies with central neck dissections to be 130 mL when similar removal variables were used. These studies suggest that a mean postoperative drain output of 130 mL may not be clinically significant enough to lead to an increased rate of seroma or infection. Moreover, the additional hospital stay associated with prolonged drain placement after SLND may not necessarily be cost-effective.

This study aims to evaluate the criteria of volume-based CSD removal after SLNDs and thereby establish the possible benefits of earlier CSD removal. Thus, we developed, to our knowledge, the first prospective, single-blinded, randomized clinical trial comparing drain output of 100 mL or less with 30 mL or less during 24 hours as the main criterion for CSD removal after SLND.

Adult patients undergoing unilateral or bilateral SLNDs at levels I to III, I to IV, II to III, or II to IV for oral cavity, oropharynx, thyroid, nonparotid salivary, or skin carcinoma by four of us (C.H., J.R.G., S.J.W., and W.R.R.) from the Department of Otolaryngology–Head and Neck Surgery at the University of California San Francisco Medical Center at Mission Bay from March 1, 2015, through December 1, 2016, were offered enrollment. A copy of the trial protocol is available in the Supplement. The committee on human research of the University of California, San Francisco, approved this clinical trial. All patients provided written informed consent.

We excluded any patients who underwent a parotidectomy, a level V lymphadenectomy, SLNDs that communicated with the upper aerodigestive tract (including free-flap operations), revision neck dissection (previous surgery in the ipsilateral neck, including excisional lymph node biopsy), previous radiotherapy, sternocleidomastoid muscle excision, internal jugular vein excision, or pectoralis major flap reconstruction of a ipsilateral skin defect or who were taking anticoagulants other than routine deep venous thrombosis prophylaxis with weight-based subcutaneous heparin sodium or enoxaparin sodium within 8 postoperative days. Primary site resections (oral cavity resections, transoral oropharynx resections with robotic surgery, thyroidectomy, salivary gland excisions, and skin excisions) were performed concurrently with some SLNDs in this study. The surgeons were blinded to the treatment group in that they had no knowledge ahead of time of the randomization group or whether the patient would consent to participation.

Quiz Ref IDAfter completion of the SLND, all patients had a single 10-mm perforated Jackson-Pratt drain placed. After the operation, all eligible patients were offered enrollment by the surgical team. A randomized packet containing the patient’s research identification number was placed in the patient’s medical record. Randomization was performed at the initiation of the study with a block size of 8. An envelope containing the patient’s randomized group assignment was not opened until postoperative day 1. The patients were randomized to 1 of the following 2 groups: drains removed after the output is less than 30 mL or less than 100 mL during a 24-hour period. Standard and similar postoperative protocols were used for all patients in both groups. All patients received inpatient antibiotic treatment not to exceed 24 hours (unless an infection was suspected) and deep venous thrombosis prophylaxis consisting of weight-based subcutaneous heparin or enoxaparin. Frequent ambulation was encouraged. Patients were encouraged to go home with a drain in place if appropriate. For 30 days after the operation, we assessed the duration of drain use, hospital length of stay, wound infection, and hematoma and seroma formation. Medical records were reviewed to collect information on age, sex, American Society of Anesthesiologists physical status classification score, body mass index, cancer type, rate of elective neck dissection, number of lymph nodes removed according to the surgical pathologic report (in cases of bilateral SLND, the number of lymph nodes was recorded independently per neck side), levels of neck dissection performed, rate of isolated neck dissection (if no additional resection was performed at the time of neck dissection), unilateral vs bilateral SLND, and estimated blood loss for each patient.

Data were organized, evaluated, and statistically analyzed using Stata software (version 13; StataCorp). We estimated the necessary sample size to be 62 neck dissections with 31 in each group to detect a 33% increase in hematoma and seroma rates with a significance level of 0.05 and 0.8 power, assuming mean hematoma and seroma rates of 7% based on the current literature.^7,10 The unpaired 2-tailed t test was performed for all interval data. Fisher exact and χ² tests were used for categorical data as appropriate. The Cohen d statistic was used to estimate effect sizes. Multivariate linear regression analysis was performed using a stepwise model selection method as appropriate.

Fifty-three patients who underwent 67 SLNDs (45 men [85%] and 8 women [15%]; mean age, 58.5 years [95% CI, 53.2-64.5 years]) met inclusion criteria, consented to the study, and were enrolled and followed up during the study period. Thirty-two SLNDs were randomized to the group with the 100-mL threshold for drain removal (100-mL group), and 35 SLNDs were randomized to the group with the 30-mL threshold (30-mL group). No patients were lost to follow-up (Figure).

Between the groups, we found no significant differences in preoperative characteristics except for significantly more patients with oropharyngeal cancer (7 of 35 [20.0%; 95% CI, 6.7%-33.3%] vs 15 of 32 [46.9%; 95% CI, 29.6%-64.1%]) and significantly fewer patients with skin cancer (8 of 35 [22.9%; 95% CI, 8.9%-36.7%] vs 1 of 32 [3.1%; 95% CI, 0%-9.2%]) as the indications for SLND in the 100-mL group. No statistically significant differences were found between groups in the mean number of lymph nodes removed, rate of isolated SLNDs, rate of elective SLNDs, levels of SLND performed, number of bilateral SLNDs, or estimated blood loss (Table 1).

Two neck seromas each developed postoperatively in the 100-mL group (2 of 32 [6.3%; 95% CI, 0%-13.5%]) and in the 30-mL group (2 of 35 [5.7%; 95% CI, 0%-14.6%]; Cohen d, 0.02; 95% CI, 0.45-0.50) (Table 2). Quiz Ref IDAll seromas were found during the 1-week postoperative clinic follow-up and treated with needle aspiration once with no further complications. None of the patients with seroma in either group had any similar clinicopathologic characteristics to suggest a risk factor. No hematomas, chylous fistulas, or wound infections occurred in either group.

The 100-mL group had a reduction in mean drain duration of 1.90 days (95% CI, 0.95-2.70 days) and a reduction in mean hospital length of stay of 1.87 days (95% CI, 0.66-3.10 days) when compared with the 30-mL group (Table 2). Randomization to the 30-mL group (Cohen d, 0.75; 95% CI, 0.26 to 1.25) and skin cancer as the indication for SLND (Cohen d, −0.86; 95% CI, −1.58 to −0.15) were the only independent variables with a large effect sizes for hospital length stay (Table 3). However, after linear regression was performed, only randomization to the 30-mL group remained a statistically significant risk factor for increased hospital length of stay (Table 3).

The clinical decision for drain removal after SLND often varies with the preference and training of the surgeon, and until the present study, little evidence has been available to guide the optimal drain removal strategy.¹ In this randomized clinical trial, both patient groups had no major differences in preoperative characteristics, had similar general extents of neck dissection as indicated by the number of lymph nodes removed and the estimated blood loss (Table 1), and were placed in standard care pathways with the only altered variable being the criterion for drain removal. Quiz Ref IDThe 100-mL group had reduced mean duration of drain use and reduced mean hospital length of stay with no increase in wound complications or seroma formation. These results highlight the possibly underestimated effect that drain removal criteria may have on hospital length of stay and thus overall cost.

Before the initiation of this study, the common practice among head and neck surgical oncology faculty at our institution had been to remove CSDs after SLNDs only after the output was less than 30 mL during a 24-hour period. However, we found that achieving this output volume criterion often took several days. Patients were, at times, otherwise ready to be discharged before their drain output met criteria for removal. Given the choice to be discharged home with a drain or to continue hospitalization until drain removal, many patients were not interested in being discharged home with a drain. This general patient preference may be one that is regionally specific to Northern California or specific to large referral centers, where patients commonly live several hours away, making the arrangement of additional follow-up for drain removal challenging. Although we routinely encourage patients to be discharged with a drain in place if appropriate, we acknowledge that other practices may be more aggressive in discharging patients home with drains. The results of the present study are still relevant to these practices because the need for discharge with a drain could be reduced with a policy for removal at a discharge threshold of 100 mL during a 24-hour period, thus reducing the need for an additional visit for drain removal and its related time and costs to the patient and clinician.

In the present study, randomization to the 100-mL group had a large effect size on reducing hospital length of stay, with mean reduction of 1.87 days (Table 2). Because a single additional day in the hospital can cost several thousand US dollars, we believe that any change in practice patterns that can reduce more than 1 hospital day is clinically meaningful. Each drain removal group had nearly identical rates of seroma and no hematomas, chyle leaks, or wound infections. Although there were no wound infections in this study, other studies have suggested that prolonged drainage, especially when the patient is discharged home with a drain, may lead to increased risk of surgical site infection.^6,11,12

We acknowledge several limitations of this study. First, we enrolled patients with heterogeneous indications for SLND. Thus, without intention, the 100-mL group had significantly fewer patients with skin cancer and significantly more patients with oropharyngeal cancer. Because of the lack of physical connection between the primary site resections for cases of skin and oropharyngeal cancer and the neck defect after lymphadenectomy, we believe that these primary site resections likely were not contributory to the fluid output within the neck cavity. Of interest, skin cancer as an indication for SLND had a large negative effect size for hospital length of stay, although nearly all these patients were in the 30-mL group. Thus, this confounder would only be expected to minimize the reduction in hospital stay in the 100-mL group, and the true effect of earlier drain removal may be larger. Also, the multivariate analysis did not show an independent association of the primary cancer site with hospital length of stay, and only randomization to the 30-mL group remained a statistically significant risk factor for increased hospital length of stay. With increased patient numbers, we would expect this variance of primary site resection between groups to disappear. Second, the surgeons were blinded to study group during the surgery and enrollment, and we attempted to limit potential bias; however, a decision bias for earlier discharge in the 100-mL group may remain. Third, this study compared 2 set volume criteria for drain removal; thus, any large daily outputs greater than 100 mL likely would not affect seroma formation but only duration of drain use and hospital length of stay. Fourth, 100 mL is a somewhat arbitrary number chosen by two of us (M.L.T. and W.R.R.). We chose this volume criterion based on the evidence that thyroidectomies have a mean serous fluid output of 130 mL and that the lack of use of drains for thyroidectomy, in general, appears to have no bearing on seroma formation. Other, more ideal volume threshold criteria for drains may not have been evaluated in the present study.

A higher-volume criterion for drain removal of 100 mL during a 24-hour period for certain SLNDs (levels I-III, II-III, II-IV, and I-IV without parotidectomy, connection to the upper aerodigestive tract, previous surgery or radiotherapy, or high-intensity anticoagulant therapy) appears to be safe and may significantly reduce time to drain removal and hospital length of stay. Earlier drain removal may reduce patient discomfort, improve patient satisfaction, and save the time and cost associated with an additional visit for drain removal. The potential reduction in hospital length of stay associated with earlier drain removal may significantly reduce overall cost. Use of this drain removal criterion should be considered in the described select patient situations.

Corresponding Author: William R. Ryan, MD, Division of Head and Neck Oncologic and Endocrine Surgery, Department of Otolaryngology–Head and Neck Surgery, Bakar Cancer Hospital, Helen Diller Comprehensive Cancer Center, University of California San Francisco Medical Center at Mission Bay, 1825 Fourth St, Fourth Floor, PO Box 1703, San Francisco, CA 94158 (william.ryan@ucsf.edu).

Accepted for Publication: June 10, 2017.

Published Online: August 24, 2017. doi:10.1001/jamaoto.2017.1414

Author Contributions: Drs M. Tamplen and Ryan had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: M. Tamplen, Heaton, George, Ryan.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: M. Tamplen, Heaton, George, Ryan.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: M. Tamplen, J. Tamplen, George.

Obtained funding: M. Tamplen.

Administrative, technical, or material support: M. Tamplen, J. Tamplen, Shuman, George, Ryan.

Study supervision: M. Tamplen, Heaton, George, Ryan.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Ryan reports serving on the scientific advisory board for Medtronic and as a consultant for Omni-guide Surgical and Ziteo. No other disclosures were reported.

Meeting Presentation: This study was presented at the Annual Meeting of the American Head and Neck Society at the Combined Otolaryngology Spring Meeting; April 27, 2017; San Diego, California.