Predictive model for increased postoperative drainage volume in patients with papillary thyroid carcinoma: a retrospective cohort study

Introduction

Papillary thyroid carcinoma (PTC) is the most common pathological subtype of thyroid malignancy, accounting for 84% of all thyroid malignancies (1). PTC exhibits an excellent long-term prognosis, with a mortality rate substantially lower than that of other common malignancies. Following standardized treatment, the 5-year overall survival rate for patients with PTC exceeds 90% (2). Surgery is the preferred and most definitive treatment for PTC. The standard surgical approach for PTC involves resection of the thyroid lobe and isthmus, with selective central neck dissection (CND) or lateral neck dissection (LND) guided by the presence of lymph node (LN) metastasis (LNM) (3).

Thyroid surgery inevitably results in cervical soft tissue, vascular, and lymphatic vessel injuries. These injuries pose a risk of bleeding, exudate, lymphatic leakage, and even chyle leakage (4). To promptly drain fluid accumulated in the surgical field and to monitor the neck for active bleeding, postoperative placement of a drainage tube has become a safe and well-established practice. Recent studies have questioned whether routine placement of a drainage tube after thyroid surgery is necessary (5-7). Substantial geographic disparities exist in the routine placement of cervical drains. According to the largest international survey of thyroid surgeons, the highest usage rate was reported in Asia (48% of respondents routinely used drains), particularly in patients undergoing total thyroidectomy (8). In high-risk populations, most surgeons still prefer routine placement of a drainage tube. It must be acknowledged that prolonged placement of a drainage tube may expose patients to extended periods of pain, restricted mobility, and discomfort. Furthermore, as an invasive device, the presence of a drainage tube inherently creates a conduit for retrograde bacterial migration along the lumen into the surgical wound. Prolonged drainage tube retention significantly increases the risk of surgical site infection (9).

However, there is currently no unified standard for the timing of drainage tube removal after thyroid surgery. Clinicians typically remove the tube based on daily drainage output volume (DOV) and the characteristics of the drainage fluid. The timing of drainage tube removal is guided more from clinical experience and limited retrospective studies than from high-level prospective evidence (10-12). This lack of standardization creates a clinical dilemma. Discrepancies in the judgment of “acceptable DOV” exist among different surgeons or medical centers, preventing patients from receiving standardized drainage management protocols. The limitations of this drainage tube management exist not only in thyroid surgery but also in thoracic surgery and anterior cervical corpectomy and fusion (13,14).

Therefore, investigations into the factors influencing DOV are of particular importance. This study aimed to identify independent predictors of increased DOV after PTC surgery and to develop a predictive model. The model was comprehensively evaluated using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA), ultimately yielding an individualized risk assessment tool that is applicable for clinical use. We present this article in accordance with the TRIPOD reporting checklist (available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0358/rc).

Methods

Study design and participants

This single-center retrospective study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Yangzhou Hospital of Traditional Chinese Medicine (TCM) (No. 2026-23). Individual consent for this retrospective analysis was waived. This study included 178 patients with a confirmed diagnosis of PTC who were admitted to the Department of Thyroid and Breast Surgery at Yangzhou Hospital of TCM between July 2024 and May 2026. Eligible patients were randomly allocated into training and validation cohorts at a ratio of 7:3. The workflow of this study is presented in Figure 1.

Figure 1 Workflow of this study. The included patients were divided into the training and validation cohorts. The training cohort was used to construct the nomogram. ROC, calibration, and DCA curves were used to evaluate the accuracy of the model. DCA, decision curve analysis; PTC, papillary thyroid carcinoma; ROC, receiver operating characteristic.

The inclusion criteria were as follows: (I) postoperative pathological confirmation of PTC; (II) undergoing thyroidectomy (lobectomy or total thyroidectomy) for the first time; (III) receiving prelaryngeal lymph node (PLN) dissection (PLND) and central lymph node (CLN) dissection (CLND); (IV) routine placement of a cervical drainage tube after surgery; (V) complete clinical data. The exclusion criteria were as follows: (I) previous history of thyroid or neck surgery; (II) receipt of neoadjuvant therapies such as radiotherapy, chemotherapy, or targeted therapy before surgery; (III) presence of other malignant tumors; (IV) missing key data in medical records.

Surgical and drainage tube management

All operations were performed by a senior attending surgeon from the Department of Thyroid and Breast Surgery at Yangzhou Hospital of TCM. For patients with PTC whose tumors were confined to a single lobe, without extrathyroidal extension (ETE) or LNM, lobectomy + isthmusectomy was performed. Total thyroidectomy was performed in patients with tumor diameter >4 cm, definite ETE, bilateral multifocal carcinoma foci, or the presence of cervical LNM. All PTC patients underwent PLND and CLND. LND was performed for those with confirmed lateral cervical LNM. The limited sample size precluded the inclusion of any patient who had undergone LND. The strategy for cervical drainage tube placement involved inserting a silicone tube connected to a negative-pressure drainage bulb in all cases at the end of surgery. Total DOV referred to the cumulative drainage volume recorded from the end of surgery until drain removal. Total DOV and drainage duration were calculated based on records collected by nursing staff at intervals of no more than 24 hours. The drainage tube was removed upon the surgeon’s assessment when the daily DOV was less than 10 mL and the drained fluid was clear.

Grouping strategy

In the training cohort, patients were grouped based on postoperative DOV. In previous analogous studies, the median value has commonly been adopted as the cutoff for group stratification (15,16). We ultimately selected the median postoperative DOV as the internal cutoff within our cohort to maximize the balance between the two groups for exploratory analysis, which is a common practice in hypothesis-generating retrospective studies. Accordingly, patients were divided into high-output (HO) and low-output (LO) groups according to the median DOV, as shown in Figure 2.

Figure 2 Grouping strategy for the HO and LO groups. Each dot represents the DOV of an individual patient. Among patients with the same drain duration, the highest DOV is indicated by an orange dot and the lowest by a blue dot. Using the median as the cutoff threshold, patients with DOV >74.5 mL were divided into the HO group, and those with DOV <74.5 mL were divided into the LO group. DOV, drainage output volume; HO, high-output; LO, low-output.

Data collection

All data were extracted from the electronic medical record system and independently verified by two reviewers. Data collected included DOV, drainage duration, age, sex, history of underlying diseases, use of anticoagulant medications, and body mass index (BMI); surgical-related factors [scope of surgery (lobectomy + isthmusectomy or total thyroidectomy), scope of LNs dissected, surgical time, and intraoperative blood loss]; tumor characteristics [T and N stagings, LNM, maximum tumor diameter, number of tumors, vascular and neural invasion, thyroid capsular invasion, and presence of Hashimoto’s thyroiditis (HT)]; and laboratory parameters [preoperative thyroglobulin (Tg), thyroid-stimulating hormone (TSH), and postoperative parathyroid hormone (PTH)].

Statistical analysis

This study used SPSS Statistics version 26.0 and R software version 4.6.0 for data organization, statistical analysis, and model construction. All analyses were based on complete case data without any imputation. After dividing the patients with PTC randomly into training and validation groups in a 7:3 ratio, the variables underwent comparison. Variable coding and reference categories for categorical variables were shown in Table S1. Continuous variables were expressed as mean ± standard deviation (SD) and compared by Student’s t-test or Mann-Whitney U test. Categorical variables were presented as frequencies (percentages), and intergroup comparisons were conducted using the chi-squared test or Fisher’s exact test. In the training cohort, variables with P<0.05 in the univariate analysis were entered into multivariate logistic regression analysis to identify independent predictors, presented as odds ratios (ORs) with 95% confidence intervals (CIs). To mitigate the potential influence of extreme values, we conducted a sensitivity analysis by excluding the case with an exceptionally high DOV (1,080 mL) and re-evaluated the model. To account for the potential influence of drainage duration on total DOV, we further performed a stratified analysis by drainage duration. A nomogram was constructed using R version 4.6.0 and R Studio software based on the results of multivariate logistic regression analysis. The performance of the model was evaluated by the area under the ROC curve (AUC) and calibration curves. DCA was employed to assess the clinical utility of the nomogram. To reduce overfitting bias, the nomogram model was internally verified using bootstrap resampling, with 1,000 replications. P value <0.05 was considered statistically significant.

Results

Baseline characteristics of the training and validation cohorts

Between July 2024 and May 2026, a total of 178 patients were diagnosed with PTC at Yangzhou Hospital of TCM. After screening, 171 patients were ultimately enrolled in this study. In this study, the majority of patients were female (n=145, 84.8%), with a median age of 44 years. The mean BMI was 23.7 kg/m². Approximately 19.9% of patients had a history of hypertension, 4 patients had a history of diabetes, and 2.3% of patients had taken anticoagulant medication. Lobectomy + isthmusectomy was performed in 125 patients, whereas total thyroidectomy was performed in 46 patients. All patients underwent prophylactic PLND and CLND, among whom 23 were confirmed to develop PLN metastasis, and 84 were confirmed to develop CLN metastasis. All patients had a cervical drainage tube inserted in the neck after surgical treatment. Baseline characteristics are presented in Table 1.

Univariate analysis

In the training cohort, the median postoperative DOV was 74.5 mL. Based on this median, 60 patients were divided into the HO group and 60 to the LO group. Univariate analysis identified surgical time, number of tumors, number of metastases in CLN, total number of LNs dissected, total number of LNM, scope of surgery, thyroid capsular invasion, and HT as factors significantly associated with postoperative DOV (Table 2).

Multivariable logistic regression analysis

To further investigate the risk factors, we performed multivariate logistic regression analysis. We used the risk of elevated postoperative DOV as the dependent variable (Y: LO =0; HO =1) and included statistically significant variables from univariate analysis as independent variables. Multivariate logistic regression analysis revealed that the total number of LNs dissected (OR =1.125, 95% CI: 1.022–1.239, P=0.02), scope of surgery (OR =6.544, 95% CI: 1.445–29.64, P=0.02), and thyroid capsular invasion (OR =6.507, 95% CI: 1.63–25.975, P=0.008) were independent predictors of increased postoperative DOV (P<0.05), as shown in Table 3. The sensitivity analysis excluding the extreme outlier yielded similar results (Table S2), confirming that the three independent predictors were not disproportionately influenced by a single case.

Stratified analysis by drainage duration

In our study cohort, patients were stratified according to drainage duration, and the subgroup with the largest sample size (drainage duration =4 days) was selected as the primary validation subset. Within this subgroup, we refitted the multivariate logistic regression model to verify whether the predictive effects of each independent risk factor on the outcome of high drainage output remained consistent. The results demonstrated that, within this subgroup, total number of LN dissected (OR =1.11, P=0.048) and thyroid capsular invasion (OR =4.3, P=0.03) remained significantly associated with higher total DOV. Although the P value for scope of surgery was not statistically significant, the OR remained greater than 1 (OR =2.421, 95% CI: 0.829–7.071), indicating that the effect direction of total thyroidectomy on high DOV risk was consistent with the primary analysis and showed no reversal. We have summarized these findings in Table S3. In the subgroups with drainage duration ≤3 days (4 cases in the HO group) and ≥5 days (5 cases in the LO group), multivariate regression could not be performed due to the small sample size; only descriptive analyses were conducted (Table S4). In both subgroups, the HO group had higher rates of total thyroidectomy and thyroid capsular invasion, as well as a higher median number of LNs dissected, compared to the LO group. This trend was consistent with the main analysis and supports the robustness of the model. We believe that the stratified analysis demonstrates that the model’s predictive performance was independent of the drainage duration.

Development of the nomogram

We calculated the regression coefficients for total number of LNs dissected, scope of surgery, and thyroid capsular invasion, respectively. The intercept and all coefficients were derived from the training cohort. The complete multivariable logistic regression equation is as follows:

Logit(P)=−2.261+0.118×(TotalnumberofLNsdissected)+1.879×(Scopeofsurgery)+1.873×(Thyroidcapsularinvasion)[1]

Based on the three independent risk factors identified by multivariate logistic regression analysis, we further constructed a nomogram for visualization (Figure 3). In the nomogram, each independent risk factor was assigned a specific point value according to its relative contribution to the model. The total points were derived by summing the points for all independent risk factors. The probability of increased postoperative DOV could then be determined by projecting the total points onto the probability scale. A higher total number of points indicated a higher risk of increased postoperative DOV in patients with PTC.

Figure 3 Nomogram for predicting increased postoperative DOV in PTC patients. Each variable is assigned a point value on the top scale. The total points are summed and projected onto the bottom risk scale to estimate the individual probability of high-output drainage. DOV, drainage output volume; LN, lymph node; PTC, papillary thyroid carcinoma.

Validation of the model’s clinical efficacy

The discriminative ability of the model was evaluated using ROC curves. In the training cohort, the AUC was 0.823 (95% CI: 0.746–0.901), as shown in Figure 4A. The model was further validated in the validation cohort, yielding an AUC of 0.849 (95% CI: 0.737–0.960), as shown in Figure 4B. Generally, an AUC >0.8 represents a strong discriminative ability of the model (17). The performance of the training cohort in the ROC curves demonstrated that our model had good discriminatory capability, which was further confirmed in the validation cohort. The Bootstrap method was employed with 1000 repetitions, resulting in an AUC of 0.811, indicating that the model’s discriminative ability is robust and unlikely to be substantially affected by overfitting.

Figure 4 ROC curves of the prediction model for increased postoperative DOV. (A) Training cohort; (B) validation cohort. The dots on the curves mark the optimal cut-off values derived from the Youden index, with the corresponding data reported as ‘cut-off value (sensitivity, specificity).' The cut-off value is the risk probability threshold. Sensitivity is the proportion of high-DOV patients correctly identified, and specificity is the proportion of non-high-DOV patients correctly identified. AUC, area under the ROC curve; DOV, drainage output volume; ROC, receiver operating characteristic.

The calibration curves showed excellent agreement between predicted probabilities and observed outcomes. In the training cohort, the Hosmer-Lemeshow test yielded a P value of 0.147 (P>0.05). The calibration plots approached the ideal 45° reference line, indicating satisfactory model calibration and high consistency between predicted probabilities and actual incidence (Figure 5A). In the validation cohort, the Hosmer-Lemeshow test yielded a P value of 0.105 (P>0.05). Similarly, the calibration curve for the validation cohort also confirmed the reliability and validity of the model (Figure 5B).

Figure 5 Calibration curves of the prediction model for increased postoperative DOV. (A) Training cohort; (B) validation cohort. DOV, drainage output volume.

The ultimate goal of constructing this model is to aid clinical practice. Therefore, we employed DCA to evaluate the clinical utility of the model. The DCA demonstrated that the predictive model yielded a positive net benefit over a range of clinically relevant threshold probabilities in both the training and validation cohorts (Figure 6). Compared with the “remove-all” or “retain-all” strategies, the model provided a higher net benefit within this range, suggesting its potential clinical utility in predicting high postoperative DOV.

Figure 6 DCA of the prediction model for increased postoperative DOV. (A) Training cohort; (B) validation cohort. DCA, decision curve analysis; DOV, drainage output volume; ROC, receiver operating characteristic.

In the training cohort, the calibration intercept (calibration-in-the-large) and slope were −0.012 and 0.974, respectively; in the validation cohort, the corresponding values were 0.087 and 0.958. These estimates closely approximate the ideal benchmarks of 0 and 1, indicating excellent agreement between predicted probabilities and observed event rates. The Brier score, which reflects the overall prediction error, was 0.181 in the training cohort and 0.175 in the validation cohort, both well below the conventional threshold of 0.25 that is considered indicative of adequate predictive utility. Collectively, these indices further substantiate the favorable calibration and discriminative ability of our nomogram, consistent with the findings from the calibration plots and ROC analyses.

Discussion

Based on multivariate logistic regression analysis, this study integrated three independent risk factors (total number of LNs dissected, scope of surgery, and thyroid capsular invasion) to construct a nomogram for predicting increased cervical DOV after surgery. Using the nomogram, clinicians can assess a patient’s risk of high DOV and thereby formulate individualized drainage management. For high-risk patients, a longer period of drainage observation should be allowed, with enhanced infection prevention and communication with patients regarding recovery expectations. Furthermore, this study fills a gap in the literature regarding predictive models for high DOV in PTC patients following surgery. Previous studies have mostly focused on correlational analyses of influencing factors and have not developed a scoring system directly applicable in clinical practice. Leveraging well-established research methodologies from fields such as thoracic surgery and spinal surgery, this study applied a visualized prediction model to postoperative drainage management. Although the model achieved favorable performance in internal validation, external validation in independent cohorts is warranted to confirm its generalizability.

LN dissection is an integral component of thyroid cancer surgery, aimed at eradicating potential regional LN metastases. However, an extended scope of LNs dissected inevitably entails more extensive transection of lymphatic networks, blood vessels, and surrounding tissues (18). Our data showed that the total number of LNs dissected was significantly positively correlated with postoperative DOV. The primary mechanism underlying the increase in DOV induced by LN dissection is the structural disruption of the lymphatic network. Lymphatic capillaries absorb proteins, cellular debris, and excess fluid from the interstitial space, which eventually drain into the thoracic duct or the lymphatic duct and return to the venous circulation (19). CLND removes en bloc tissue containing lymphatic vessels and LNs, resulting in the exposure of lymphatic vessel stumps. If stumps cannot be sealed immediately during or shortly after surgery, lymph will continuously leak and become a major component of the drainage fluid. Unlike extravasated blood, lymph typically lacks clotting factors. This implies that lymph lacks the capacity for spontaneous coagulation to seal the stumps, making it difficult to promptly halt persistent lymph leakage simply by relying on pressure dressings or negative-pressure suction (20). In addition to direct lymphatic leakage, LN dissection may also indirectly increase DOV by inducing a local inflammatory response. Surgical trauma activates immune cells such as neutrophils and macrophages, leading to the release of proinflammatory cytokines, including IL-6 and TNF-α (21). These factors can increase capillary permeability and promote vascular leakiness (22). Therefore, the increase in DOV induced by LN dissection is generally the result of a combination of lymph leakage and inflammatory vascular leakiness. Tailored drainage management strategies should be adopted in clinical practice for patients with varying extents of LN dissection. For patients with a limited number of LNs dissected, the drain duration may be appropriately shortened when the expected DOV is relatively controllable. For patients undergoing LND or those with a larger number of LNs dissected, a longer drain duration should be reserved, with close monitoring of the characteristics of the drainage fluid and vigilance for the occurrence of chyle leakage.

Our findings demonstrated that patients who underwent total thyroidectomy had significantly higher postoperative DOV than those who underwent lobectomy + isthmusectomy. The thyroid gland has an extremely rich blood supply, primarily derived from the superior and inferior thyroid arteries (23). Thyroidectomy requires transection of both the superior and inferior thyroid arteries, resulting in multiple vascular stumps. Although we employed bipolar electrocoagulation, ultrasonic scalpel, and vascular ligation for hemostasis, oozing from local capillaries and small veins remained unavoidable (24). Oozing blood contributes to the component of the drainage fluid. Moreover, the abundant lymphatic network surrounding the thyroid gland also serves as a major source of drainage fluid. The thyroid parenchyma contains an extensive plexus of lymphatic capillaries, and thyroidectomy disrupts this lymphatic network, thereby increasing lymphatic exudate. Lobectomy involves resection of only one thyroid lobe, with the surgical wound confined to a single side and the contralateral lobe and its surrounding structures preserved intact. In contrast, total thyroidectomy entails complete resection of thyroid lobes (25). The total wound surface area following total thyroidectomy is approximately twice that of lobectomy, which inevitably results in increased oozing blood and lymphatic exudate from the cervical wound. After a thyroidectomy, the thyroid bed becomes a dead space that requires time for the surrounding soft tissues to re-appose and heal. Before this dead space is obliterated, a considerable amount of exudate may accumulate in the tissue interstitial space. Compared with lobectomy, total thyroidectomy results in a larger “reservoir” capacity for drainage fluid, with corresponding increases in both DOV and drain duration. The burden of drainage management should also be considered as a factor when selecting the surgical scope. For low‑risk patients with unilateral PTC, current guidelines explicitly recommend lobectomy as the preferred option. This approach not only preserves contralateral thyroid function and reduces the risk of hypoparathyroidism but also lessens the postoperative drainage burden and shortens hospital stay. For high-risk patients who have to undergo total thyroidectomy, clinicians should fully inform them preoperatively about the potential increases in drain duration and hospitalization costs.

This study identified thyroid capsular invasion as an independent predictor of increased postoperative DOV in patients with PTC. The thyroid capsule is a fibrous sheath composed of dense connective tissue that envelops the thyroid parenchyma. The capsule serves as a natural barrier restricting tumor ETE (26). When tumors invade and breach the capsule, tumor cells have penetrated this barrier, leading to dense adhesion to or direct infiltration of the perithyroidal soft tissues. Among ETE in PTC, adhesion between the tumor and the strap muscles is the most frequently encountered (27). Compared with dissection along non-adherent anatomical planes, the process of separating adhesions entails transection of more small blood vessels, lymphatic vessels, and connective tissues, resulting in increased wound exudation. Moreover, thyroid capsular invasion is accompanied by an intensified local inflammatory response. Tumor cells secrete various cytokines (e.g., ILs and MMPs) that can induce peritumoral angiogenesis. These newly formed vessels are structurally abnormal and highly permeable. When transected during surgery, the vessels exhibit a greater tendency for postoperative oozing (28). For patients with confirmed thyroid capsular invasion in the perioperative period, surgeons should anticipate a potential increase in postoperative drainage burden and appropriately relax the indications for tube removal.

Although our predictive model was derived from patients with routine drainage, its findings may still inform the ongoing debate on selective drain omission. Specifically, patients with a low predicted probability of high DOV—for example, those undergoing lobectomy with isthmusectomy, limited LNs dissected, and no thyroid capsular invasion—might be suitable for drain‑free management. Nevertheless, this hypothesis requires confirmation in well‑designed prospective studies, and clinical decisions should remain individualized based on comprehensive intraoperative findings.

Several limitations should be acknowledged. First, this was a single-center retrospective study with a relatively small sample size, and all surgeries were performed by a single senior surgeon. While this minimized inter-operator variability, it inevitably limits the generalizability of our findings to other centers or surgeons with differing levels of experience. Second, due to limitations in sample size, this study did not include patients who underwent LND. However, we recognize that these patients constitute a high-risk group for increased postoperative drainage in clinical practice. Consequently, the current nomogram is directly applicable only to patients undergoing CLND; its utility in the LND setting remains unproven. Third, the characteristics of the drainage fluid also constitute an important clinical indicator for assessing drainage status. However, we lacked a quantitative method for this parameter and therefore did not include the characteristics in the present analysis. The generalizability of our nomogram is limited, and independent external validation in large-scale, multicenter cohorts is an essential prerequisite before this model can be safely adopted in routine clinical practice.

Conclusions

This study confirmed that the total number of LNs dissected, scope of surgery, and thyroid capsular invasion are independent risk factors for increased cervical DOV in PTC patients following surgery. The nomogram constructed based on these three factors demonstrated satisfactory discrimination and calibration and can effectively identify patients at high risk of elevated DOV. Large-scale, prospective multicenter external validation is urgently warranted to confirm the robustness and facilitate the clinical translation of this preliminary predictive model.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0358/rc

Data Sharing Statement: Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0358/dss

Peer Review File: Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0358/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0358/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This single-center retrospective study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Yangzhou Hospital of Traditional Chinese Medicine (TCM) (No. 2026-23). Individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Boucai L, Zafereo M, Cabanillas ME. Thyroid Cancer: A Review. JAMA 2024;331:425-35. [Crossref] [PubMed]
2. Carnazza M, Quaranto D, DeSouza N, et al. The Current Understanding of the Molecular Pathogenesis of Papillary Thyroid Cancer. Int J Mol Sci 2025;26:4646. [Crossref] [PubMed]
3. Health Commission Of The People's Republic Of China N. National guidelines for diagnosis and treatment of thyroid cancer 2022 in China (English version). Chin J Cancer Res 2022;34:131-50. [Crossref] [PubMed]
4. Zheng W, Zhu S, Zhang Y, et al. Novel application of microdissection tungsten needle in total thyroidectomy with central neck dissection for papillary thyroid carcinoma. Front Surg 2022;9:896275. [Crossref] [PubMed]
5. Manjunatha HA, Prashanth KB, Ranjani SK, et al. A clinical comparative study of thyroid surgeries with and without drain. Indian J Otolaryngol Head Neck Surg 2023;75:1681-6. [Crossref] [PubMed]
6. George NM, Chitrambalam TG, Christopher PJ, et al. To drain or not to drain following thyroidectomy.: A prospective, randomized study. Saudi Med J 2023;44:518-21. [Crossref] [PubMed]
7. Chen Y, Wang C, Bai B, et al. Drainage Tube Placement May Not Be Necessary During Endoscopic Thyroidectomy Bilateral Areola Approach: A Preliminary Report. Front Surg 2022;9:860130. [Crossref] [PubMed]
8. Maniakas A, Christopoulos A, Bissada E, et al. Perioperative practices in thyroid surgery: An international survey. Head Neck 2017;39:1296-305. [Crossref] [PubMed]
9. Telias A, Abdelbary H, Grammatopoulos G, et al. Prolonged wound drainage past two weeks is associated with increased treatment failure following prosthetic joint infection surgery of the hip. Int Orthop 2026;50:963-71. [Crossref] [PubMed]
10. Muranushi R, Harimoto N, Seki T, et al. Early drain removal after hepatectomy based on bile leakage prediction using drainage fluid volume and direct bilirubin level. Hepatol Res 2024;54:1070-7. [Crossref] [PubMed]
11. Mori Y, Yamashita H, Sato S, et al. Usefulness of preoperative ice cream consumption and novel postoperative drainage management in patients undergoing left-sided neck dissection for thyroid cancer: a nonrandomized prospective study. Surg Today 2024;54:642-50. [Crossref] [PubMed]
12. Moghib K, Ahmed MT, Ghanm TIE, et al. Optimal duration of postoperative drainage following burr hole surgery for chronic subdural hematoma: A systematic review and network meta-analysis. Langenbecks Arch Surg 2025;410:278. [Crossref] [PubMed]
13. Yuan H, Yu H, Liu L, et al. Risk Factors for Predicting Increased Surgical Drain Output in Patients After Anterior Cervical Decompression and Fusion. World Neurosurg 2022;164:e980-91. [Crossref] [PubMed]
14. Kosugi S, Kanda T, Yajima K, et al. Risk factors influencing the pleural drainage volume after transthoracic oesophagectomy. Eur J Cardiothorac Surg 2013;43:1116-20. [Crossref] [PubMed]
15. Zhang Z, Wang Z, Zhang Y, et al. Risk factors for increased postoperative drainage of calcaneal fractures after open reduction and internal fixation: An observational study. Medicine (Baltimore) 2018;97:e11818. [Crossref] [PubMed]
16. Lenze NR, Ghodke A, Shah RN, et al. Predictors of Increased Surgical Drain Output following Type I Thyroplasty for Glottic Insufficiency. Laryngoscope 2021;131:1557-60. [Crossref] [PubMed]
17. Nahm FS. Receiver operating characteristic curve: overview and practical use for clinicians. Korean J Anesthesiol 2022;75:25-36. [Crossref] [PubMed]
18. Yen TW, Fan X, Sparapani R, et al. A contemporary, population-based study of lymphedema risk factors in older women with breast cancer. Ann Surg Oncol 2009;16:979-88. [Crossref] [PubMed]
19. Null M, Arbor TC, Agarwal M. Anatomy, Lymphatic System. 2026. In: StatPearls [Internet]. Treasure Island (FL): StatPearls Publishing; 2026.
20. Ratnayake CBB, Escott ABJ, Phillips ARJ, et al. The anatomy and physiology of the terminal thoracic duct and ostial valve in health and disease: potential implications for intervention. J Anat 2018;233:1-14. [Crossref] [PubMed]
21. Wang J, Li J, Sheng X, et al. Beta-adrenoceptor mediated surgery-induced production of pro-inflammatory cytokines in rat microglia cells. J Neuroimmunol 2010;223:77-83. [Crossref] [PubMed]
22. Ng CT, Fong LY, Tan JJ, et al. Endothelial barrier disruptive effect of IFN-Ƴ and TNF-α: Synergism of pro-inflammatory cytokines. Cytokine 2025;190:156922. [Crossref] [PubMed]
23. Shaari AL, Spaulding SL, Xing MH, et al. The anatomical basis for preserving the blood supply to the parathyroids during thyroid surgery, and a review of current technologic advances. Am J Otolaryngol 2022;43:103161. [Crossref] [PubMed]
24. Zhang S, Huang R, Qiu Y, et al. Comparison of thyroid surgery techniques: a retrospective cohort study and meta-analysis of traditional electric knife vs. straight bipolar electrocoagulation forceps. Gland Surg 2024;13:1178-87.
25. Yan S, Cao Z, Zhang Q, et al. Survival outcomes following isthmusectomy versus lobectomy and total thyroidectomy for papillary thyroid carcinoma: A propensity score-matched study. Eur Arch Otorhinolaryngol 2026; Epub ahead of print. [Crossref]
26. Jin BJ, Kim MK, Ji YB, et al. Characteristics and significance of minimal and maximal extrathyroidal extension in papillary thyroid carcinoma. Oral Oncol 2015;51:759-63. [Crossref] [PubMed]
27. Park J, An S, Bae JS, et al. Impact of Tumor Size on Prognosis in Differentiated Thyroid Cancer with Gross Extrathyroidal Extension to Strap Muscles: Redefining T3b. Cancers (Basel) 2024;16:2577. [Crossref] [PubMed]
28. Lipka U, Orywal K, Gudowska-Sawczuk M. Matrix Metalloproteinases and Pro-Inflammatory Cytokines in Bladder Cancer: Diagnostic and Prognostic Perspectives: Narrative Review. Int J Mol Sci 2026;27:3218. [Crossref] [PubMed]