Keywords
drainage - hematoma - substernal goiter - thyroid - thyroid surgery
Introduction
Postoperative hematoma is among the most clinically significant complications following total thyroidectomy, particularly in patients with substernal goiter.[1]
[2] In cases of substernal goiter, there is close contact between the thyroid gland and vital aerodigestive and neurovascular structures. This poses several challenges for operating surgeons, including the extremely high risk of hematoma, among other complications.[3]
[4] A substernal goiter occupies a large space in the neck and in the superior mediastinum, which requires greater dissection of surrounding structures and longer duration of surgery.
Drain placement following thyroidectomy has been a standard practice in many surgical departments worldwide.[5] However, this has been declining as evidence has shown no significant impact of drain placement on the development of postoperative neck hematoma.[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13] Contrarily, drain use has been associated with higher risks of other adverse outcomes.[5]
[7]
[14] Two recent meta-analyses[5]
[14] based on randomized controlled trials found that drain placement independently increased the likelihood of higher postoperative pain, prolonged hospital stay, and wound infections.
Despite the evidence against drain placement after thyroidectomy,[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14] there is a lack of consensus on drain use in patients with substernal goiter. This could be attributed to many comparative studies excluding cases with substernal goiter, as identified by a recent systematic review.[14] In addition, substernal goiter has a low incidence, which makes it difficult to conduct adequately powered analyses. Based on the higher risk of hematoma in this cohort, Herranz and Latorre[15] advocate drain use despite a paucity of evidence comparing hematoma development among patients with and without drain placement.
In the present study, we assessed the factors that increase the likelihood of drain placement in patients undergoing thyroidectomy for substernal goiter using a multi-institutional database. We further explored the impact of drain placement on postoperative hematoma and other complications in this cohort.
Methods
Data Source, Study Design, and Population
The present was a multicenter, retrospective cohort study based on data from the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP). The ACS-NSQIP is a multi-institutional, nationally-validated database that evaluates surgical outcomes, with more than 700 partnering hospitals. Data are collected prospectively by trained surgical clinical reviewers (SCRs) at each hospital using a standardized protocol. The details on the sampling technique, data collection methodology, and measures of the ACS-NSQIP are described elsewhere.[16]
The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines were followed in the present study, which was approved with exemption from consent by the Ethics Review Committee at Aga Khan University, Karachi, Pakistan (reference ID: 2022-7219-20725).
Adult patients (aged ≥18 years) who underwent elective thyroidectomy for substernal goiter from 2016 to 2020 were included. These cases were identified using the following current procedural terminology (CPT) codes: 60270 (thyroidectomy, including substernal thyroid; sternal split or transthoracic approach) and 60271 (thyroidectomy, including substernal thyroid; cervical approach). Cases with missing data in the ACS-NSQIP Thyroidectomy Targeted Dataset and those undergoing emergency or urgent surgeries were excluded. Since thyroidectomy is mostly performed by otolaryngologists or general surgeons, procedures performed by other principal surgical specialties were further excluded.
Measures
The perioperative variables for risk adjustment included sociodemographic characteristics, medical comorbidities, and operative details. Sociodemographic characteristics comprised age, gender, race, Hispanic ethnicity, and body mass index (BMI). Race was categorized as white, black or African American, and others (which included Asians, Native Americans, Alaska natives, native Hawaiians, and Pacific islanders).
Comorbidities comprised diabetes mellitus, smoking, hypertension, bleeding disorder, previous neck surgery, and the score on the American Society of Anesthesiologists (ASA) physical status classification. Smoking status was quantified as smoking within one year of surgery. Bleeding disorders included hemophilia, vitamin K deficiency, thrombocytopenia, chronic anticoagulation therapy, and similar disorders predisposing patients to severe bleeding.
The operative characteristics comprised surgical approach, indication, principal specialty, central neck dissection, use of vessel sealant device, wound classification, and operative time. Surgical indications were categorized as benign or malignant based on the codes of the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) and Tenth Revision (ICD-10-CM).
Drain and Non-Drain Groups
Patients were categorized into drain and non-drain groups. Cases with closed suction neck drains placed upon completion of surgery were included in the drain group. The remaining cases formed the nondrain group.
Outcomes
The primary outcome included 30-day postoperative neck hematoma. The ACS-NSQIP defines it as development of a hematoma or postoperative bleeding at the neck site. Ecchymosis or bruising alone does not qualify as hematoma.[17] Interventions required for hematoma management were also documented.
The secondary outcomes included any complication within 30 days, major morbidity, and prolonged length of stay. The complications were further divided into infectious and noninfectious. Infectious complications comprised surgical site infection (SSI), urinary tract infection (UTI), sepsis, septic shock, wound disruption, pneumonia, and clostridium difficile colitis. Noninfectious complications included cerebrovascular accident (CVA)/stroke, cardiac complications (myocardial infarction [MI] or cardiac arrest), pulmonary complications (ventilator dependence for > 48 hours or unplanned reintubation), acute renal failure, deep vein thrombosis (DVT), intra-/postoperative blood transfusion(s), postoperative hypocalcemia, and recurrent laryngeal nerve injury. Major morbidity included any of the following adverse outcomes: deep or organ space SSI, wound dehiscence, pulmonary embolism (PE), prolonged ventilation, unplanned reintubation, sepsis, septic shock, MI, cardiac arrest, or CVA. Prolonged length of hospital stay was defined as stay > 2 days during the index admission.
Statistical Analysis
Descriptive statistics were reported. The Shapiro-Wilk test was used to confirm the nonparametric distribution of continuous variables, including age, BMI, operative time, and length of stay (p < 0.001 for all). Accordingly, the continuous variables were reported using median and interquartile range (IQR) values and compared across the drain and non-drain groups using the Mann-Whitney U test. The categorical variables were expressed as frequencies and percentages, and the Chi-squared (χ2) test or the Fisher exact test were computed to compare their differences across the drain and non-drain groups. Missing data were reported in tables to keep the denominators consistent in the calculations.
Multivariable binary logistic regression models were computed to explore independent factors increasing the likelihood of drain placement and to assess the impact of drain placement on postoperative neck hematoma. To select covariates for adjustment, variance inflation factors (VIFs) were calculated for clinically-relevant baseline characteristics, comorbidities, and operative variables with p < 0.20 on the univariate analysis. The threshold for significant multicollinearity was set at VIF ≥10, and no significant multicollinearity was observed during the analysis.
Separate binary logistic regression models were also computed to assess the impact of drain placement on any 30-day postoperative complication, major morbidity, and prolonged length of hospital stay. After ruling out significant multicollinearity, these models were adjusted for age, gender, BMI, ASA score, surgical approach, principal specialty, use of vessel sealants, wound classification, and operative time.
Two-sided analyses were conducted. The threshold for statistical significance was set at p < 0.05. Adjusted odds ratios (AORs) and their corresponding 95% confidence intervals (95%CIs) were reported. IBM SPSS Statistics for Windows (IBM Corp., Armonk, NY, United States), software, version 26.0, was used to conduct the analyses.
Results
Patient Characteristics
A total of 1,229 participants were included in the present study ([Fig. 1]). Most patients were female (72.2%), had a median age of 59.0 ([IQR 22.0) years, and were operated on for benign indications (89.2%). Other baseline characteristics, comorbidities, and operative details are summarized in [Table 1].
Table 1
Baseline characteristics, comorbidities, and operative variables stratified by drain status
|
Variable
|
No drain
n = 657
|
Drain
n = 572
|
p-value
|
|
Age in years*
|
59.0 (22.0)
|
59.0 (22.0)
|
0.278
|
|
< 65
≥ 65
|
440 (67.0%)
217 (33.0%)
|
368 (64.3%)
204 (35.7%)
|
0.332
|
|
Gender
Female
Male
|
498 (75.8%)
159 (24.2%)
|
389 (68.0%)
183 (32.0%)
|
0.002**
|
|
Race
White
Black or African American
Others
Missing
|
337 (56.4%)
242 (40.5%)
18 (3.0%)
60
|
286 (57.2%)
197 (39.4%)
17 (3.4%)
72
|
0.886
|
|
Hispanic ethnicity
Missing
|
39 (6.4%)
50
|
32 (6.2%)
55
|
0.872
|
|
BMI (kg/m2)*
|
31.8 (9.6)
|
34.3 (11.4)
|
< 0.001**
|
|
BMI (kg/m2
< 25.0
25.0–30.0
≥ 30.0
Missing
|
96 (14.7%)
155 (23.7%)
403 (61.6%)
3
|
54 (9.6%)
112 (19.8%)
399 (70.6%)
7
|
0.002**
|
|
ASA score
1–2
3–5
Missing
|
304 (46.3%)
352 (53.7%)
1
|
196 (34.4%)
374 (65.6%)
2
|
< 0.001**
|
|
Diabetes mellitus
|
121 (18.4%)
|
127 (22.2%)
|
0.099
|
|
Current smoker
|
88 (13.4%)
|
90 (15.7%)
|
0.245
|
|
Hypertension
|
351 (53.4%)
|
323 (56.5%)
|
0.285
|
|
Bleeding disorder
|
8 (1.2%)
|
17 (3.0%)
|
0.030**
|
|
Previous neck surgery
|
53 (8.1%)
|
48 (8.4%)
|
0.836
|
|
Surgical approach
Cervical
Sternal split or transthoracic
|
587 (89.3%)
70 (10.7%)
|
488 (85.3%)
84 (14.7%)
|
0.033**
|
|
Indication
Benign
Malignant
Missing
|
591 (90.1%)
65 (9.9%)
1
|
505 (88.3%)
67 (11.7%)
0
|
0.308
|
|
Specialty
General surgery
Otolaryngology
|
466 (70.9%)
191 (29.1%)
|
190 (33.2%)
382 (66.8%)
|
< 0.001**
|
|
Central neck dissection
Missing
|
70 (10.8%)
9
|
66 (11.6%)
4
|
0.652
|
|
Vessel sealant device
Missing
|
508 (78.8%);
12
|
379 (68.2%);
16
|
< 0.001**
|
|
Wound classification
Clean
Clean-contaminated or contaminated
|
624 (95.0%);
33 (5.0%)
|
564 (98.6%);
8 (1.4%)
|
< 0.001**
|
|
Operative time in minutes*
|
114.0 (73.0)
|
150.0 (100.0)
|
< 0.001**
|
|
Operative time/mins
< 90
90–120
120–150
> 150
|
211 (32.1%)
140 (21.3%)
125 (19.0%)
181 (27.5%)
|
80 (14.0%)
109 (19.1%)
95 (16.6%)
288 (50.3%)
|
< 0.001**
|
Abbreviations: ASA, American Society of Anesthesiologists; BMI, body mass index.
Notes: *Reported as median and interquartile range values; **statistically significant (p < 0.05); percentages are presented in columns.
Fig. 1 Details of the total cases included and the cases that were excluded based on CPT codes, missing data, emergency/ non-elective procedures, and ineligible surgeon subspecialties.
Drains were placed in 46.5% of the cases. Compared with the non-drain group, a greater proportion of patients in the drain group: were male (32.0% versus 24.2%), had a BMI ≥ 30 kg/m2 (70.6% versus 61.6%), presented ASA scores from 3 to 5 (65.6% versus 53.7%), presented bleeding disorders (3.0 versus 1.2%), had operative time ≥ 150 minutes (50.3% versus 27.5%), and underwent a sternal split or transthoracic approach (14.7% versus 10.7%). Similarly, having a clean wound classification was significantly more common among patients in the drain group compared to those in the non-drain group (98.6% versus 95.0%). Additionally, otolaryngologists were more likely to use drains (66.8%) in comparison with general surgeons (33.2%). Use of vessel sealant devices was more common in the non-drain group (78.8%) than in the drain group (68.2%).
Factors Increasing the Likelihood of Drain Placement
After adjustment for baseline characteristics, comorbidities, and operative details, factors increasing the likelihood of drain placement included BMI ≥ 30 kg/m2, ASA score from 3 to 5, sternal split/transthoracic approach, operative time ≥ 90 minutes, and surgery conducted by otolaryngologists ([Table 2]). Patients with clean-contaminated or contaminated wound classifications were less likely to undergo drain placement (AOR: 0.214; 95%CI: 0.089–0.514).
Table 2
Univariate and multivariable binary logistic regression for drain placement among patients undergoing thyroidectomy for substernal goiter
|
Variable
|
Unadjusted odds ratio
|
p-value
|
Adjusted odds ratio
|
p-value
|
|
Age in years
< 65
≥ 65
|
Reference
1.124 [0.888–1.423]
|
0.332
|
-
|
-
|
|
Gender
Female
Male
|
Reference
1.473 [1.147–1.893]
|
0.002**
|
Reference
1.230 [0.910–1.663]
|
0.177
|
|
Race
White
Black or African American
Others
|
Reference
0.959 [0.751–1.226]
1.113 [0.563–2.200]
|
0.739
0.758
|
-
|
-
|
|
Hispanic ethnicity
|
0.961 [0.593–1.558]
|
0.872
|
−
|
−
|
|
BMI (kg/m2)
< 25.0
25.0–30.0
≥ 30.0
|
Reference
1.285 [0.850–1.941]
1.760 [1.227–2.525]
|
0.234
0.002**
|
Reference
1.287 [0.786–2.108]
1.675 [1.078–2.602]
|
0.316
0.022*
|
|
ASA score
1–2
3–5
|
Reference
1.648 [1.308–2.076]
|
< 0.001**
|
Reference
1.415 [1.060–1.887]
|
0.018*
|
|
Diabetes mellitus
|
1.264 [0.957–1.671]
|
0.099
|
1.240 [0.876–1.757]
|
0.225
|
|
Current smoker
|
1.207 [0.879–1.659]
|
0.245
|
−
|
−
|
|
Hypertension
|
1.131 [0.903–1.417]
|
0.285
|
−
|
−
|
|
Bleeding disorder
|
2.485 [1.064–5.802]
|
0.035*
|
2.321 [0.863–6.245]
|
0.095
|
|
Previous neck surgery
|
1.044 [0.694–1.569]
|
0.836
|
−
|
−
|
|
Surgical approach
Cervical
Sternal split or transthoracic
|
Reference
1.443 [1.028–2.026]
|
0.034*
|
Reference
1.756 [1.167–2.640]
|
0.007*
|
|
Indication
Benign
Malignant
|
Reference
1.206 [0.841–1.731]
|
0.309
|
-
|
-
|
|
Specialty
General surgery
Otolaryngology
|
Reference
4.905 [3.850–6.249]
|
< 0.001**
|
Reference
6.932 [5.179–9.278]
|
< 0.001**
|
|
Central neck dissection
|
1.086 [0.760–1.551]
|
0.652
|
−
|
−
|
|
Vessel sealant device
|
0.557 [0.445–0.749]
|
< 0.001**
|
0.803 [0.587–1.099]
|
0.171
|
|
Wound classification
Clean
Clean-contaminated or contaminated
|
Reference
0.268 [0.123–0.586]
|
0.001**
|
Reference
0.214 [0.089–0.514]
|
0.001**
|
|
Operative time in minutes
< 90
90–120
120–150
> 150
|
Reference
2.053 [1.434–2.940]
2.004 [1.384–2.904]
4.197 [3.055–5.765]
|
< 0.001**
< 0.001**
< 0.001**
|
Reference
2.495 [1.651–3.772]
2.637 [1.714–4.057]
6.323 [4.323–9.248]
|
< 0.001**
< 0.001**
< 0.001**
|
Abbreviations: ASA, American Society of Anesthesiologists; BMI, body mass index.
Notes: **Statistically significant (p < 0.05); only cases with complete data on all covariates were included (n = 1,189).
Postoperative Hematoma
A total of 2.2% of the patients (27 of 1,229) developed postoperative hematoma. Most hematoma cases were underwent additional observation during hospital admission (70.4%; 19 of 27) while 1 patient required tracheostomy (3.7%; 1 of 27). The remaining seven patients did not require additional observation.
On the univariate analysis, there was no significant difference in the incidence of neck hematoma in the drain and non-drain groups (3.0% versus 1.5% respectively; p = 0.083). This was consistent across the multivariable regression analysis after adjustment for clinically relevant characteristics (AOR: 1.289; 95%CI: 0.525–3.168). Instead, black or African American race was identified as a risk factor for postoperative hematoma, while use of vessel sealant devices was found to independently reduce the risk of hematoma development ([Table 3]).
Table 3
Univariate and multivariable binary logistic regression analyses for postoperative hematoma
|
Variable
|
Unadjusted odds ratio
|
p-value
|
Adjusted odds ratio
|
p-value
|
|
Age in years
< 65
≥ 65
|
Reference
1.336 [0.614–2.906]
|
0.465
|
-
|
-
|
|
Gender
|
|
|
|
|
|
Female
Male
|
Reference
1.304 [0.580–2.932]
|
0.520
|
−
|
−
|
|
Race
White
Black or African American
Others
|
Reference
2.536 [1.055–6.099]
2.254 [0.274–18.540]
|
0.038**
0.450
|
Reference
3.239 [1.223–8.574]
2.142 [0.229–20.008]
|
0.018**
0.504
|
|
Hispanic ethnicity
|
2.787 [0.934–8.319]
|
0.066
|
3.572 [0.684–18.652]
|
0.131
|
|
BMI (kg/m2)
< 25.0
25.0–30.0
≥ 30.0
|
Reference
0.738 [0.163–3.342]
1.238 [0.363–4.219]
|
0.693
0.733
|
−
|
−
|
|
ASA score
1–2
3–5
|
Reference
3.101 [1.166–8.244]
|
0.023**
|
Reference
2.480 [0.888–6.929]
|
0.083
|
|
Diabetes mellitus
|
1.681 [0.727–3.886]
|
0.225
|
−
|
−
|
|
Current smoker
|
1.716 [0.683–4.313]
|
0.251
|
−
|
−
|
|
Hypertension
|
1.664 [0.741–3.733]
|
0.217
|
−
|
−
|
|
Bleeding disorder
|
1.881 [0.245–14.437]
|
0.543
|
−
|
−
|
|
Previous neck surgery
|
1.996 [0.677–5.891]
|
0.210
|
−
|
−
|
|
Surgical approach
Cervical
Sternal split or transthoracic
|
Reference
0.551 [0.129–2.348]
|
0.420
|
−
|
−
|
|
Indication
Benign
Malignant
|
Reference
0.313 [0.042–2.325]
|
0.256
|
−
|
−
|
|
Specialty
General surgery
Otolaryngology
|
Reference
1.439 [0.668–3.099]
|
0.353
|
−
|
−
|
|
Central neck dissection
|
1.464 [0.497–4.315]
|
0.489
|
−
|
−
|
|
Vessel sealant device
|
0.249 [0.113–0.549]
|
0.001**
|
0.167 [0.070–0.401]
|
< 0.001**
|
|
Wound classification
Clean
Clean-contaminated or contaminated
|
Reference
Could not be computed
|
−
|
−
|
−
|
|
Operative time in minutes
< 90;
90–120;
120–150;
> 150
|
Reference;
2.375 [0.706–7.984];
1.319 [0.326–5.335];
1.715 [0.541-5.437]
|
0.162
0.697
0.360
|
Reference;
2.141 [0.530–8.649];
1.242 [0.264–5.836];
0.906 [0.216–3.790]
|
0.285
0.784
0.892
|
|
Drain placement
|
1.983 [0.901–4.366]
|
0.089
|
1.289 [0.525–3.168]
|
0.580
|
Abbreviations: ASA, American Society of Anesthesiologists; BMI, body mass index.
Notes: **Statistically significant (p < 0.05); only cases with complete data on all covariates were included (n = 1,052).
Complications
The univariate analyses demonstrated no significant differences in terms of major morbidity, infectious complications, sepsis, wound disruption, pneumonia, cardiac complications, acute renal failure, postoperative hypocalcemia, recurrent laryngeal nerve injury, unplanned readmission, and mortality between the both groups. However, compared with the non-drain group, patients with drain placement presented significantly higher occurrences of surgical site infections (0.7% versus 0.0%; p = 0.047), unplanned reintubation (1.6% versus 0.5%; p = 0.047), blood transfusions (1.9% versus 0.3%; p = 0.006), unplanned reoperations (3.0 versus 0.9%; p = 0.008), and prolonged length of hospital stay (15.9% versus 5.0%; p < 0.001) ([Table 4]).
Table 4
Thirty-day postoperative complications, stratified by drain status
|
Variable
|
No drain
n = 657
|
Drain
n = 572
|
p-value
|
|
Neck hematoma
Missing
|
10 (1.5%)
2
|
17 (3.0%)
2
|
0.083
|
|
Any complication
|
87 (13.2%)
|
78 (13.6%)
|
0.840
|
|
Major morbidity
|
11 (1.7%)
|
11 (1.9%)
|
0.743
|
|
Infectious complications
|
8 (1.2%)
|
7 (1.2%)
|
0.992
|
|
Surgical site infection
|
0 (0.0%)
|
4 (0.7%)
|
0.047**
|
|
Urinary tract infection
|
3 (0.5%)
|
0 (0.0%)
|
0.253
|
|
Sepsis
|
1 (0.2%)
|
0 (0.0%)
|
> 0.999
|
|
Septic shock
|
1 (0.2%)
|
0 (0.0%)
|
> 0.999
|
|
Wound disruption
|
2 (0.3%)
|
1 (0.2%)
|
> 0.999
|
|
Pneumonia
|
3 (0.5%)
|
1 (0.2%)
|
0.628
|
|
Clostridium difficile colitis
|
0 (0.0%)
|
1 (0.2%)
|
0.465
|
|
Noninfectious complications
|
83 (12.6%)
|
74 (12.9%)
|
0.874
|
|
CVA/stroke with neurological deficit
|
1 (0.2%)
|
0 (0.0%)
|
> 0.999
|
|
Cardiac complications
Myocardial infarction
Cardiac arrest requiring CPR
|
2 (0.3%)
2 (0.3%)
0 (0.0%)
|
2 (0.3%)
0 (0.0%)
2 (0.3%)
|
> 0.999
0.502
0.216
|
|
Pulmonary complications
Ventilator > 48 hours
Unplanned intubation
|
5 (0.8%)
3 (0.5%)
3 (0.5%)
|
11 (1.9%)
4 (0.7%)
9 (1.6%)
|
0.073
0.711
0.047**
|
|
Acute renal failure
|
1 (0.2%)
|
0 (0.0%)
|
> 0.999
|
|
DVT/thrombophlebitis
|
1 (0.2%)
|
0 (0.00%)
|
> 0.999
|
|
Blood transfusion
|
2 (0.3%)
|
11 (1.9%)
|
0.006**
|
|
Postoperative hypocalcemia
Missing
|
31 (4.9%)
21
|
24 (4.4%)
28
|
0.707
|
|
Recurrent laryngeal nerve injury
Missing
|
46 (7.0%)
3
|
39 (6.9%)
5
|
0.915
|
|
Unplanned reoperation
|
6 (0.9%)
|
17 (3.0%)
|
0.008**
|
|
Unplanned readmission
|
18 (2.7%)
|
17 (3.0%)
|
0.807
|
|
Length of stay*
|
1.0 (1.0)
|
1.0 (1.0)
|
< 0.001**
|
|
Prolonged length of stay
Missing
|
33 (5.0%)
0
|
91 (15.9%)
1
|
< 0.001**
|
|
Mortality
|
2 (0.3%)
|
0 (0.0%)
|
0.502
|
Abbreviations: CPR, cardiopulmonary resuscitation; CVA, cerebrovascular accident; DVT, deep vein thrombosis.
Notes: *Reported as median and interquartile range values; **statistically significant (p < 0.05); percentages are presented in columns.
The results of the regression analysis of the association of drain placement with postoperative outcomes are shown in [Table 5]. After adjustment for clinically relevant covariates, drain use in patients with thyroidectomy for substernal goiter independently increased the likelihood of prolonged length of hospital stay (AOR: 2.047; 95%CI: 1.245–3.368). Drain placement was not independently associated with any postoperative complication or major morbidity ([Table 5]).
Table 5
Multivariable binary logistic regression analyses for other postoperative outcomes with drain status as the main explanatory covariate
|
Outcome
|
Adjusted odds ratio
|
p-value
|
|
Any complication
|
0.698 [0.469-1.037]
|
0.075
|
|
Major morbidity
|
0.518 [0.183-1.467]
|
0.215
|
|
Prolonged length of stay
|
2.047 [1.245-3.368]
|
0.005**
|
Notes: No drain group was reference; regression adjusted for age, gender, body mass index (BMI), score on the American Society of Anesthesiology (ASA) physical status classification, surgical approach, principal specialty, use of vessel sealants, wound classification, and operative time; **statistically significant (p < 0.05); only cases with complete data on all covariates were included (n = 1,188 for prolonged length of stay and 1,189 for all other outcomes).
Discussion
The present study used a large, multi-institutional dataset to compare the incidence of postoperative hematoma and other complications among patients undergoing thyroidectomy for substernal goiter with and without drain use. Our results demonstrate that drain use was associated with 28.9% higher odds of hematoma, but this was not statistically significant. We speculate that this may be due to a higher probability of detecting clinically-insignificant minor hematomas during hospital admission. This hypothesis is supported by the longer length of hospital stay observed among patients with drain compared with the non-drain cohort, which provided more time to detect insignificant hematomas. In contrast, an alternative explanation for the higher likelihood of hematoma formation could be the presence of inherent selection biases for drain placement in high-risk cases. Our analysis demonstrated that ASA scores from 3 to 5, BMI ≥ 30 kg/m2, and more complex surgical approaches were independently associated with drain use, further supporting this hypothesis. More appropriate variables to assess the risk of hematoma in this population would include size of resection, intraoperative hemorrhage, surgeon experience, and use of preoperative anticoagulation.[7] However, our regression analyses could not be adjusted for these variables, as they are not captured in the ACS-NSQIP, limiting the conclusiveness of our findings. Nevertheless, these results are in accordance with the existing literature on the impact of drain use on postoperative hematoma in thyroidectomies.[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
In the present study, drain use was independently associated with prolonged length of stay. This is consistent among studies on cases of non-substernal goiter undergoing thyroidectomy.[7]
[14] This could be attributed to surgeon reluctance to discharge patients until the drain output thresholds are met. While beyond the scope of the current study, prolonged length of stay can impose higher monetary costs on patients, lead to patient discomfort, and burden the limited hospital resources available.[18]
[19] In addition, prolonged hospital stay may predispose patients to nosocomial infections. Even though drain use did not independently increase the risk of developing infectious complications, the univariate analysis revealed a higher likelihood of developing infectious complications among patients with prolonged versus non-prolonged stay (4.0% versus 0.8% respectively; p = 0.009).
We found that otolaryngologists were more likely to use drains in thyroidectomy for substernal goiter compared with general surgeons (AOR: 6.932; 95%CI: 5.179–9.278), even after adjustment for relevant baseline, patient, and surgery characteristics. However, there was no significant difference in the rate of postoperative hematoma among their cases (1.8% in general surgery cases versus 2.6% in otolaryngology cases; p = 0.351). In addition, patients operated on by otolaryngologists were more likely to have prolonged hospital stay compared with general surgery patients (12.9% versus 7.6%, respectively; p = 0.002). Chung et al.[20] report similar findings in their comparison of thyroidectomies performed by general surgeons versus otolaryngologists. These apparent differences in length of hospital stay might be a surrogate measure for differences in drain placement practices among these two specialties.
Surgeons were less likely to place drains in clean-contaminated or contaminated wounds. Considering thyroidectomy is a clean procedure, this seems to have occurred in cases in which concomitant procedures such as tracheostomy were performed. We speculate that this would be when tracheomalacia or bilateral cord paralysis was suspected.
Several implications can be inferred from the present study. First, thyroidectomy without drain placement might be safe among patients with substernal goiter. However, this decision should be individualized for each patient, while also accounting for surgeon preference. Specifically considering the similar rates of postoperative hematoma observed among surgeries conducted by otolaryngologists and gender surgeons, otolaryngologists may better optimize their practice regarding patient selection for drain placement. The conclusiveness of these recommendations, however, is limited to patients without preoperative anticoagulation therapy or significant intraoperative bleeding, considering these factors could not be explored in the current study.
Second, a shorter hospital stay might be feasible with proper patient education related to drain management in certain cases in which drains are still indicated.[21] Familiarizing surgeons with short-stay protocols might also help avoid unnecessary hospital stays. However, this requires further evaluation, as patient education and short-stay protocols were beyond the scope of the current study.
The present study has multiple limitations which should be considered while interpreting its results. Since we used an existing database, our analyses are limited to variables included within the dataset. The ACS-NSQIP does not include the extent of hematoma, development of seroma, surgeon experience, and hospital volume. In addition, preoperative laboratory assessments could not be included in the statistical analyses because of the substantial number of missing values. These factors might be associated with hematoma formation in this cohort. Lastly, the database might be susceptible to coding and data collection errors.
Conclusion
Thyroidectomy without drain placement might be safe for substernal goiter, particularly in patients without preoperative anticoagulation use or intraoperative hemorrhage. However, this decision should be individualized for each patient, while accounting for surgeon preference. Otolaryngologists may better optimize their patient selection practice for drain placement. In addition, patient education regarding drain management and short-stay protocols should be further evaluated to help reduce unnecessary hospital stays.
