Abstract
This study aimed to determine the incidence and independent factors of acute postoperative pain in oral cancer patients and develop an individualized predictive nomogram. A retrospective cohort study included 458 oral cancer patients who underwent tumor resection with free flap reconstruction at Wuhan University’s Hospital of Stomatology (January 2023–December 2024). Demographics, medical history, tumor features, surgical variables, and preoperative anxiety/depression were collected. Pain intensity was assessed using the Numerical Rating Scale (NRS) on postoperative days 1, 2, 3, and 7. Independent risk factors were identified via multivariate analysis, and a nomogram was validated using Receiver operating characteristic (ROC) curves, calibration curves, and Decision Curve Analysis ( DCA). The incidence of acute pain on postoperative day 3 was 74.67% (342/458), with 51.53% experiencing moderate pain (NRS 4–6) and 17.25% severe pain (NRS ≥ 7). Independent factors included education ≤junior high school (OR = 0.29), no history of immunotherapy (OR = 0.12), larger tumor volume (OR = 20.53), unilateral (OR = 3.46) or bilateral lymph node dissection(OR = 10.95), free flap type(OR = 10.17), and preoperative anxiety/depression (OR = 2.71; all P < 0.05).The nomogram showed excellent performance (AUC = 0.923; training/validation sets: 0.928/0.911), with high consistency between predicted and actual incidence (Hosmer-Lemeshow test, P > 0.05) and significant clinical net benefits (DCA). In conclusion, acute postoperative pain in oral cancer is common and multifactorial. The nomogram accurately predicts this risk, aiding individualized analgesia and optimizing pain management.
Keywords: Pain prediction, Risk stratification, Individualized analgesia, Perioperative pain management, Nomogram validation
Background
Oral cancer, a prevalent head and neck malignancy, is witnessing a rising global incidence. According to the International Agency for Research on Cancer (IARC) in the Global Cancer Statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 Countries, cancers of the lip and oral cavity account for approximately 389,485 new cases and 188,230 deaths annually worldwide [1]. Surgical resection, the cornerstone of curative treatment, often involves complex tissue removal and reconstruction; postoperative pain impairs swallowing, speech, and social function, and correlates closely with complications, rehabilitation trajectories, and long-term quality of life [2]. However, critical gaps persist in research on acute postoperative pain in oral cancer, particularly regarding multicenter data integration, stratified analysis of influencing factors, and optimization of individualized analgesic strategies—all requiring urgent advancement.
The intricate orofacial anatomy means that surgical trauma typically induces multidimensional pain, encompassing nociceptive (from tissue injury), neuropathic (from nerve damage), and inflammatory components [3, 4]. Clinically, 40–65% of oral cancer patients experience inadequately controlled acute postoperative pain, a rate significantly higher than for malignancies elsewhere [5]. Importantly, 15–30% of these patients may progress to chronic pain if acute pain is untreated, exacerbating functional and psychological burdens. A cohort study of 1,200 patients found that those with a Visual Analogue Scale (VAS) score > 4 within 7 days postoperatively had a 2.81-fold higher risk of chronic pain at 3 months (95% CI: 1.97–3.99) [6]. Despite postoperative pain management being central to Enhanced Recovery After Surgery (ERAS), oral cancer-specific research remains limited. First, reported acute pain incidence varies widely (20–70%) due to differences in assessment timing, surgical procedures, and tools [7, 8]. Second, existing studies on influencing factors focus primarily on single variables, lacking integrated multivariate analysis or predictive models [9–11]. Third, analgesic strategies rely on empirical regimens, with no consensus on prophylactic gabapentinoid timing/dosage for nerve damage, leading to adverse reactions or inadequate analgesia in ~ 20% of patients [4].
This study aims to systematically investigate acute postoperative pain incidence in oral cancer patients, clarify associated tumor clinical/histopathological characteristics, and identify risk factors. A nomogram model will be developed to predict acute pain risk, providing a tool for individualized analgesic planning to optimize pain management and improve prognosis.
Methods
This retrospective cohort study was conducted at the Hospital of Stomatology, Wuhan University, from January 2023 to December 2024. The protocol was reviewed and approved by the Ethics Committee of the Hospital of Stomatology, Wuhan University (Approval No.: [2023] Lun Shen Zi (B11) Hao). Due to the retrospective nature, written informed consent was waived, and all data were de-identified for privacy protection. The study complied with the Helsinki Declaration.
Patient eligibility criteria
Inclusion criteria
Adult patients (≥ 18 years) who underwent oral cancer surgery under general anesthesia (tumor resection combined with free flap reconstruction) at our institution between January 2023 and December 2024. Core surgical components: (1) Resection of oral cavity subsites lesions, including the buccal mucosa, tongue, gingiva, floor of the mouth, hard palate, and retromolar trigone; (2) Cervical lymph node dissection; (3) Reconstruction with anterolateral thigh, forearm, iliac bone, or fibular free flaps (individualized by tumor location/severity).
Exclusion criteria
(1) Chronic pain history; (2) Long-term preoperative use of analgesics or psychotropics; (3) Incomplete medical records; (4) Non-oral primary or metastatic malignancies; (5) Flap crisis requiring secondary surgical exploration; (6) Postoperative transfer due to severe systemic complications; (7) Prior oral cancer surgery with recurrence reoperation. Eligible patients were identified via the hospital’s electronic medical record system and surgical database.
Data collection
Demographics extracted
Age, gender, BMI, education (≤ junior high/≥high school), smoking, alcohol consumption, betel nut chewing history.Medical history: Comorbidities, chemotherapy/immunotherapy history.Tumor characteristics: Size, primary oral subsite, Tumor node metastasis(TNM) stage, pathological type, lymph node size.Surgical variables: Free flap type, lymph node dissection method, total operative duration.
Anesthesia protocol
Standardized endotracheal intubation with mechanical ventilation and intravenous-inhalational anesthesia (opioids: alfentanil, sufentanil, remifentanil; anesthetics: sevoflurane, propofol; muscle relaxant: cisatracurium). Intraoperative monitoring included heart rate, blood pressure, SpO₂, central venous pressure, and invasive arterial pressure. Postoperative analgesia used a standardized Patient-Controlled Intravenous Analgesia (PCIA) pump (hydromorphone 0.1 mg/kg + ketamine 0.1 mg/kg + sodium chloride 100 mL at 2 mL/h).
Preoperative assessments
14-item Hospital Anxiety and Depression Scale (HADS) (1 week preoperatively; HADS-A/HADS-D subscales, 0–3 scoring; total score ≥ 8 defined as anxiety/depression) and NRS pain score (24 h preoperatively; 0–10 scale).Pain evaluation: Trained nurses recorded resting NRS scores preoperatively (day before) and on Postoperative Day (POD)1, POD2, POD3, POD7. The presence of postoperative pain is defined as an NRS score ≥ 1 (including mild [NRS 1–3], moderate [NRS 4–6], and severe [NRS ≥ 7] pain).A score of 0 indicates no pain.
Primary outcome
Development and validation of a nomogram to predict postoperative acute pain risk. Secondary outcomes: Incidence of postoperative acute pain and identification of independent risk factors.
Statistical analysis
Statistical analyses were performed using SPSS v30.0. Data were presented as counts/percentages or mean ± SD. Group comparisons used t-test, Mann-Whitney U test, or chi-square test. Binary univariate and multivariate logistic regression evaluated risk factors. ROC curves assessed predictive value of continuous variables. Nomograms were constructed and evaluated using RStudio v2025.05.0 + 496. A two-tailed P < 0.05 was statistically significant.
Results
Screening process of study participants
A total of 650 medical records were initially retrieved, with 458 patients meeting inclusion criteria and completing the study after rigorous screening. The remaining 192 patients were excluded for the following reasons: 77 with a history of prior oral cancer resection, 34 with postoperative flap crisis requiring secondary surgical exploration, 26 transferred to other institutions due to severe systemic diseases, 39 with incomplete medical documentation, and 16 with chronic pain history or preoperative use of analgesics/psychotropic medications. The study flow can be found in Fig. 1.
Fig. 1.
Development of a nomogram model to predict the risk of postoperative pain in oral and maxillofacial tumors surgery patients
Demographic, clinical, tumor, and surgical characteristics of the study cohort
Demographic, Clinical, and Tumor-Surgical Characteristics of the Study Cohort The study cohort included 458 patients with a median age of 52 years (IQR: 42.00–61.00 years) and median body weight of 65 kg (IQR: 58.00–72.63 kg); 77.95% were male, and 45.85% had a high school or higher education. Preoperatively, 22.71% had betel nut exposure history, 58.08% received chemotherapy, and 29.69% underwent immunotherapy. Common comorbidities included hypertension (34.28%), diabetes mellitus (17.69%), cardiovascular diseases (24.24%), urinary system diseases (24.45%), thyroid diseases (18.78%), and respiratory diseases (25.76%), with 57.64% reporting significant preoperative pain. Regarding tumors, the most prevalent size was 2–4 cm, and 36.46% were at TNM stage Ⅳ; over half had preoperative lymphadenopathy, predominantly 3–6 cm in size. Squamous cell carcinoma (91.70%) was the main histopathological type, with primary sites mostly the tongue (49.34%), buccal mucosa (22.49%), and alveolar region (16.16%). Surgically, 24.02% underwent unilateral lymph node dissection (ULND), 47.16% bilateral lymph node dissection(BLND), and the rest no dissection. Free flaps primarily included anterolateral thigh free flaps (45.20%), anterobrachial free flaps (36.24%), iliac bone free flaps (13.32%), and fibular free flaps (5.24%), with a median operative duration of 370 min (interquartile range [IQR]: 309–435 min). Detailed characteristics are presented in Table 1.
Table 1.
Summary of the demographic and diagnostic characteristics of the study population
| Study characteristics | Category | Descriptive statistic |
|---|---|---|
| Age | Years | 52[42.00 ，61.00] |
| Weight | Kg | 65[58.00,72.63] |
| Education level | Higher education | 210(45.85%) |
| Sex | Male | 357(77.95%) |
| Betel quid or areca nut chewing | 104(22.71%) | |
| Hypertension | 157(34.28%) | |
| Diabetes | 81(17.69%) | |
| Diseases of the cardiovascular system | 111(24.24%) | |
| Diseases of the urological system | 112(24.45%) | |
| Thyroid disease | 86(18.78%) | |
| Diseases of the respiratory system | 118(25.76%) | |
| Preoperative anxiety and depression state | 205(44.76%) | |
| History of chemotherapy | 266(58.08%) | |
| History of immunotherapy | 136(29.69%) | |
| Preoperative pain | pain | 264(57.64%) |
| Lymph node size | 0cm | 178(38.86%) |
| <3cm | 111(24.24%) | |
| 3-6cm | 161(35.15%) | |
| >6cm | 8(1.75%) | |
| Size of tumor | <2cm | 116(25.33%) |
| 2-4cm | 240(52.40%) | |
| >4cm | 102(22.27%) | |
| TNM classification | Stage I | 64(13.97.%) |
| Stage II | 80(17.47%) | |
| Stage III | 147(32.10%) | |
| Stage IV | 167(36.46%) | |
| Type of Cancer | Squamous cell carcinoma | 420(91.70%) |
| Location of primarydisease | Tongue | 226(49.34%) |
| Alveolar region | 74(16.16%) | |
| Floor of the mouth | 32(7.00%) | |
| Buccal mucosa | 103(22.49%) | |
| Palate | 16(3.49%) | |
| Retro molar trigonearea | 7(1.52%) | |
| ULND | 110(24.02%) | |
| BLND | 216(47.16%) | |
| Site of flap transplantation | Anterobrachial free flap | 166(36.24%) |
| Anterolateral thigh free flap | 207(45.20%) | |
| Fibular free flap | 24(5.24%) | |
| Iliac bone free flap | 61(13.32%) | |
| Surgery time | minute | 370[309,435] |
TNM Tumor Node Metastasis, ULND Unilateral lymph node dissection, BLND Bilateral lymph node dissection
Time-series analysis of pain severity
This study analyzed the dynamic evolution of pain severity across time points. Over half of the patients reported preoperative pain. In the early postoperative period (POD1–POD2), over 80% of patients experienced no significant pain, with low rates of other pain levels, indicating effective initial pain control. On POD3, 342 patients (74.67%) developed significant pain, stratified as mild (27 patients, 5.90%), moderate (236 patients, 51.53%), and severe (79 patients, 17.25%). By POD7, 38.86% of patients had moderate to severe pain, reflecting the dynamic progression of postoperative pain over time. Detailed data are in Fig. 2.
Fig. 2.
Pain severity at different time points. The presence of postoperative pain is defined as an NRS score ≥ 1 (including mild [NRS 1–3], moderate [NRS 4–6], and severe [NRS ≥ 7] pain)
Analysis of factors associated with postoperative acute pain in oral cancer patients
The presence of pain was significantly correlated with educational level (p < 0.001), history of diabetes mellitus (p = 0.035), history of chemotherapy (p = 0.007), history of immunotherapy (p = 0.007), preoperative anxiety and depression (p < 0.001), and clinical lymph node involvement (p < 0.001). Further analysis indicated that the occurrence of postoperative acute pain was also significantly associated with tumor size (p < 0.001), TNM classification (p < 0.001), primary tumor site (p < 0.001), approach to cervical lymph node dissection (p < 0.001), and type of free flap (p < 0.001). Detailed characteristics are presented in Table 2..
Table 2.
Demographic and clinical characteristics associated with acute postoperative pain in oral cancer patients
| Variables | Acute Postoperative Pain | p-value | |
|---|---|---|---|
| Yes(n=342) | No(n=116) | ||
| Age(years) | 51.64[50.28 ，53.00] | 51.22[48.95,53.50] | 0.725 |
| Weight(kg) | 65.93[64.66,67.21] | 66.60[64.50,68.70] | 0.493 |
| Education level,n(%) | - | - | <0.001* |
| <higher education | 209(61.11%) | 39(33.62%) | |
| >=higher education | 133(38.89%) | 77(66.38%) | |
| Sex,n(%) | - | - | 0.880 |
| male | 266(77.78%) | 91(78.45%) | |
| female | 76(22.22%) | 25(21.56%) | |
| Betel quid or areca nut chewing,n(%) | 82(24.00%) | 22(19.00%) | 0.266 |
| Hypertension,n(%) | 119(34.80%) | 38(32.76%) | 0.690 |
| Diabetes,n(%) | 53(15.50%) | 28(24.14%) | 0.035* |
| Diseases of the cardiovascular system,n (%) | 82(24.00%) | 29(25.00%) | 0.824 |
| Diseases of the urological system,n(%) | 84(24.56%) | 28(24.14%) | 0.927 |
| Thyroid disease,n(%) | 66(19.30%) | 20(17.24%) | 0.624 |
| Diseases of the respiratory system,n(%) | 90(26.32%) | 28(24.14%) | 0.643 |
| Preoperative anxiety and depression state, n(%) | 174(50.88%) | 31(26.72%) | <0.001* |
| History of chemotherapy,n(%) | 211(61.70%) | 55(47.41%) | 0.007* |
| History of immunotherapy,n(%) | 90(26.32%) | 46(39.66%) | 0.007* |
| Preoperative pain | 205(59.94%) | 59(50.86%) | 0.087 |
| Lymph node size,n(%) | - | - | <0.001* |
| 0cm | 113(33.04%) | 65(56.03%) | |
| <3cm | 92(26.90%) | 19(16.38%) | |
| 3-6cm | 130(38.01%) | 31(26.72%) | |
| >6cm | 7(2.05%) | 1(0.86%) | |
| Size of tumor,n(%) | - | - | <0.001* |
| <2cm | 41(11.99%) | 75(64.66%) | |
| 2-4cm | 208(60.82%) | 32(27.59%) | |
| >4cm | 93(27.19%) | 9(7.76%) | |
| TNM classification,n(%) | - | - | <0.001* |
| Stage I | 19(5.56%) | 45(38.79%) | |
| Stage II | 64(18.71%) | 16(13.79%) | |
| Stage III | 124(36.26%) | 23(19.83%) | |
| Stage IV | 135(39.47%) | 32(27.59%) | |
| Type of Cancer,n(%) | - | - | 0.072 |
| Squamous cell carcinoma | 309(90.35%) | 111(95.69%) | |
| Others | 33(9.65%) | 5(4.31%) | |
| Location of primarydisease,n(%) | - | - | <0.001* |
| Tongue | 152(44.44%) | 74(63.79%) | |
| Alveolar region | 70(20.47%) | 4(3.45%) | |
| Floor of the mouth | 24(7.02%) | 8(6.90%) | |
| Buccal mucosa | 77(22.51%) | 26(22.41%) | |
| Palate | 13(3.80%) | 3(2.59%) | |
| Retro molar trigonearea | 6(1.75%) | 1(0.86%) | |
| ULND,n(%) | 65(19.01%) | 45(38.79%) | <0.001* |
| BLND,n(%) | 192(56.14%) | 24(20.69%) | <0.001* |
| Site of flap transplantation,n(%) | - | - | <0.001* |
| antebrachium | 98(28.65%) | 68(58.62%) | |
| Anterolateral femoral skin flap,n(%) | 166(48.92%) | 41(35.34%) | |
| Fibular flap | 21(6.14%) | 3(2.59%) | |
| cutaneous iliac flap | 57(16.67%) | 4(3.45%) | |
| Surgery time(min) | 371.65[360.46,382.84] | 3449.66[329.04,370.27] | 0.207 |
TNM Tumor Node Metastasis, ULND Unilateral lymph node dissection, BLND Bilateral lymph node dissection
*P-value < 0.05 by chi-square test, Fisher exact test, t-test, or Mann-Whitney U-test
Multivariate analysis of independent factors for postoperative acute pain in oral cancer patients
A total of 12 associated factors with p < 0.05 identified in Table 2 were included in univariate and multivariate logistic regression analyses. Multivariate results identified the following independent factors for acute pain on postoperative day 3: lower educational level (OR = 0.29, 95% CI: 0.15–0.58, p < 0.001), no prior immunotherapy (OR = 0.12, 95% CI: 0.04–0.33, p < 0.001), larger tumor size (OR = 20.53, 95% CI: 6.9–61.12, p < 0.001), unilateral lymph node dissection (OR = 3.46, 95% CI: 1.29–9.26, p = 0.014), bilateral lymph node dissection (OR = 10.95, 95% CI: 3.25–36.97, p < 0.001), flap type (vs. antebrachium flap: anterolateral femoral skin flap, OR = 2.88, 95% CI: 1.43–5.79, p = 0.003; fibular flap, OR = 8.28, 95% CI: 1.53–44.91, p = 0.014; cutaneous iliac flap, OR = 10.17, 95% CI: 2.52–41.00, p = 0.001), and preoperative anxiety/depression (OR = 2.71, 95% CI: 1.42–5.20, p = 0.003). (Table 3).
Table 3.
Univariate and multivariate logistic regression analysis of postoperative acute pain in patients undergoing oral and maxillofacial tumor surgery
| Variables | Univariate | Multivariate | ||
|---|---|---|---|---|
| OR（ 95%CI） | P-value | OR（ 95%CI） | P-value | |
| Education level | 0.32[0.21 ，0.50] | <0.001* | 0.29[0.15,0.58] | <0.001* |
| Diabetes | 0.50[0.34,0.97] | 0.036* | 1.36[0.56,3.32] | 0.497 |
| History of chemotherapy | 1.79[1.17,2.73] | 0.007* | - | - |
| History of immunotherapy | 0.54[0.35,0.85] | 0.007* | 0.12[0.04,0.33] | <0.001* |
| lymphadenopathy | 2.58[1.68,3.97] | <0.001* | 0.93[0.16,5.48] | 0.939 |
| Size of tumor（ vs <=2cm） | - | - | - | |
| >2cm | 13.43[8.13,22.17] | <0.001* | 20.53[6.9,61.12] | <0.001* |
| TNM classification(vs Stage I) | <0.001* | 0.932 | ||
| Stage II | 9.47[4.4,20.39] | <0.001* | 1.475[0.298,7,288] | 0.658 |
| Stage III | 12.77[6.36,25.63] | <0.001* | - | 0.983 |
| Stage IV | 9.99[5.16,19.34] | <0.001* | - | 0.983 |
| Location of primary disease（tongue） | 2.20[1.43,3.4] | <0.001* | 1.27[0.63,2.56] | 0.496 |
| ULND | 0.37[0.23,0.59]] | <0.001* | 3.46[1.29,9.26] | 0.014* |
| BLND | 4.91[2.98,8.07] | <0.001* | 10.95[3.25,36.97] | <0.001* |
| Site of flap transplantation(vs antebrachium) | <0.001* | 0.039* | ||
| Anterolateral femoral skin flap | 2.81[1.77,4.45] | <0.001* | 2.88[1.43,5.79] | 0.003 |
| Fibular flap | 4.86[1.39,16.93] | 0.013* | 8.28[1.53,44.91] | 0.014 |
| Cutaneous iliac flap | 9.89[3.43,28.54] | <0.001* | 10.17[2.52,41.00] | 0.001 |
| Preoperative anxiety and depression state | 2.84[1.79,4.51] | <0.001* | 2.71[1.42,5.20] | 0.003* |
TNM Tumor Node Metastasis, ULND Unilateral lymph node dissection, BLND Bilateral lymph node dissection, OR Odds ratio, CI Confdence interval
*P-value < 0.05
Construction of a predictive model for postoperative acute pain in oral cancer patients
A nomogram model was developed to individually predict postoperative day 3 acute pain risk in oral cancer patients, incorporating independent risk factors identified via multivariate analysis: educational level, immunotherapy history, tumor size, lymph node dissection approach, flap type, and preoperative anxiety/depression (Fig. 3A). Receiver operating characteristic (ROC) curve analysis evaluated the model’s discriminative performance, revealing a prediction cutoff value of 0.79 (sensitivity = 0.87; specificity = 0.86). The area under the curve (AUC) was 0.923 (95% CI: 0.895–0.951), indicating excellent ability to distinguish between patients with and without postoperative acute pain. The narrow 95% CI confirmed result stability and reliability.The narrow 95% confidence interval indicates precise and reliable estimates (Fig. 3B and C). In addition, we use the Bootstrap method (B = 1000 resampling) to internally verify and correct the optimism of the model. The results show that the original discriminant ability of the model on the training set is 0.92 (calculation method: 0.5 + index.orig[‘Dxy’]/2), and the C-index corrected by Bootstrap is 0.91(calculation method: 0.5 + index.corrected[‘Dxy’]/2), indicating that the model has robust discrimination ability.
Fig. 3.
Nomogram model (A), ROC curve (B) and calibration curve (C) of acute pain after oral and maxillofacial tumor. edu: education, immunT: history of immunotherapy, ULND: unilateral lymph node dissection, BLND: bilateral lymph node dissection, surgtype: flap surgery type, depressed: preoperative anxiety and depression state
Validation of the nomogram model for predicting postoperative acute pain in oral cancer patients
A total of 458 patients were randomised in a seven-to-three ratio by using computer-generated random numbers.To verify the model’s stability and generalizability, training set (n = 320) and validation set (n = 138) were performed via R software. ROC curve analysis showed the training set yielded an AUC of 0.928 (95% CI: 0.898–0.958; Fig. 4A) and the validation set an AUC of 0.911 (95% CI: 0.849–0.973; Fig. 4B), demonstrating the model stably stratified postoperative acute pain risk across datasets with superior discriminative efficacy, and its AUC improved by > 0.2 vs. single-indicator prediction. Calibration curves (Fig. 4C and D) assessed by the Hosmer-Lemeshow test confirmed consistency with the ideal model in both sets (training set: P = 0.885; validation set: P = 0.804; both P > 0.05), validating predictive accuracy and stable calibration. Decision curve analysis(DCA) (Fig. 4E and F) showed the nomogram outperformed “no model” and “AI-only” strategies across sets; at high-risk thresholds (0.1–0.9), its net benefit curve was higher, and within 0.3–0.7 (critical for decisions), it identified 8–12 more true pain cases per 100 patients while avoiding 5–7 over-analgesia instances. Subgroup analyses by tumor sites (tongue, buccal, gingival) showed the tongue subgroup’s AUC of 0.902 (95% CI: 0.800–1; P = 0.534 vs. overall) confirmed generalizability, and age-stratified analysis found the elderly subgroup (> 60 years) had an AUC of 0.996 (95% CI: 0.986–1) with no age-related difference (P = 0.332), demonstrating stability. Thus, the nomogram exhibits excellent discriminative ability, accuracy, and clinical utility, validated across datasets and subgroups to guide individualized analgesic regimens and optimize perioperative outcomes for oral cancer patients.
Fig. 4.
ROC curve analysis was performed to evaluate the nomogram model of postoperative acute pain in oral and maxillofacial tumors in the training set (A) and validation set (B); Nomogram model evaluation of postoperative acute pain of oral and maxillofacial tumors in the training set (C) and validation set (D) based on calibration curve analysis; Nomogram model evaluation of postoperative acute pain of oral and maxillofacial tumors in the training set (E) and validation set (F) by DCA curve analysis.ROC, Receiver operating characteristic curve; AUC, Area under the curve; DCA, Decision Curve Analysis
Discussion
This retrospective analysis of 458 oral cancer patients undergoing free flap reconstruction represents the first systematic investigation to delineate the prevalence of postoperative acute pain, identify multidimensional independent risk factors, and develop a nomogram model with robust predictive performance (AUC = 0.923).The decision curve analysis of this study showed that when clinicians set the intervention threshold within the predicted probability range of 20% to 80%, the nomogram model was used to screen patients and perform preventive intervention, which was able to obtain a higher clinical net benefit than the ‘all intervention’ or ‘no intervention’ strategy. Combined with the incidence of pain and clinical practice experience in this study cohort, we recommend 30% as the reference threshold for initiating an intensive postoperative pain management protocol. This means that for patients whose model predicts a over 30% chance of developing moderate-to-severe acute postoperative pain, targeted multimodal prophylactic analgesia should be considered. This threshold is within the effective interval shown by the decision curve, and has good clinical operability in balancing intervention resources with patient benefits. These findings offer novel insights into the pathophysiological mechanisms of postoperative pain in oral cancer and establish a foundation for individualized pain management strategies.
The prevalence of pain on the 3rd postoperative day in oral cancer patients in this study was 74.6%, which is consistent with the high prevalence of postoperative pain reported in existing literature. Koilada et al. pointed out in their study on patients undergoing head and neck cancer surgery that the incidence of postoperative acute pain can reach 50%. All patients in that study received multimodal postoperative analgesic intervention, with specific regimens including acetaminophen, non-steroidal anti-inflammatory drugs, gabapentinoids, etc., and the duration of analgesia covered the perioperative period to prevent the transition to chronic pain [12]. Khawaja et al. conducted a study on 100 patients with head and neck cancer, using the NRS (0 = no pain, 10 = worst pain imaginable) to quantify pain intensity. The results showed that the average pain intensity of patients was 5.88 ± 2.53, and 93% of the pain was persistent [13]. The high incidence of pain in oral cancer may stem from dual orofacial surgical trauma and flap donor-site injury, compounded by the oral cavity’s dense trigeminal innervation (3–5-fold higher nerve density in tongue/gingiva than trunk skin) [7]. Surgical trauma damages nerve fibers and activates TRPV1/P2 × 3 receptors [14], inducing primary afferent sensitization, while postoperative functional activities (swallowing, speech) perpetuate mechanical stimulation, amplifying pain distinct from non-oral malignancies [15].In addition, recent studies suggest that acute pain after extensive oral surgery may be related not only to surgery-related factors but also to mechanisms such as neuroinflammation/neuronal stress [16].
Patients received standardized patient-controlled intravenous analgesia (PCIA: hydromorphone 0.1 mg/kg + ketamine 0.1 mg/kg + 0.9% sodium chloride 100 mL at 2 mL/h) for 48 postoperative hours, with marked efficacy: pain incidence was 16.8% (median NRS 0, IQR 0–2) on day 1 and 19.7% (median NRS 0, IQR 0–2) on day 2. Due to logistical and cost constraints, PCIA was not continued beyond 48 h, contributing to a sharp escalation in pain incidence to 74.6% on day 3. This corroborates Orgill et al.’s finding that “only 35% of head and neck cancer patients achieve adequate pain control,” [17]underscoring a critical management challenge. Among patients with NRS ≥ 1 on day 3, only a subset required Non-steroidal Anti-inflammatory Drug intervention, while others declined medication. Given potential bias from variable medication utilization, later NRS scores were excluded from incidence analyses.
Tumor size emerged as a prominent independent risk factor. Larger tumors involve broader infiltration, heightened metastatic potential, and extensive invasion of adjacent tissues, nerves, and bone. Resection induces greater iatrogenic damage, with postoperative inflammatory responses (e.g., 3–5-fold elevations in IL-6 and TNF-α) correlating positively with tumor diameter [18]. TNM staging analyses revealed stepwise pain incidence increases (stage I: 14.0% vs. stage IV: 36.5%), validating tumor burden’s impact. Surgical approach strongly associates with pain via resection extent, tissue trauma, and nerve exposure risk.
Patients undergoing bilateral cervical lymph node dissection faced higher pain risk (OR = 10.95) than unilateral dissection (OR = 3.46), attributable to bilateral spinal accessory nerve and cervical plexus (C2–C4) injury: spinal accessory damage induces trapezius denervation and shoulder pain, while cervical plexus injury causes anterior neck dysesthesia. Nerve regeneration in levator scapulae is 40% slower after bilateral dissection, prolonging pain [1]. Flap type analyses showed iliac bone flaps conferred higher risk (OR = 10.17) than other free flaps, reflecting greater donor-site trauma (93% pain incidence vs. 59% with forearm flaps) [19]. Iliac flap harvest transects gluteus maximus attachments and superior gluteal nerve deep branches, with osseous resection inducing more intense pain than soft-tissue dissection. Preoperative anxiety and depression reinforce the “psycho-neuro-immune axis” in pain modulation [20]. They upregulate hypothalamic-pituitary-adrenal axis activity, elevating cortisol and lowering pain thresholds, while promoting inflammatory mediator release to perpetuate a “pain-anxiety” loop [21, 22].
Patients with abnormal preoperative HADS scores had nearly 3-fold increased pain risk, highlighting preoperative psychological intervention as a potential adjunct. Educational level associations likely reflect health literacy disparities: patients with ≤junior high education demonstrated poorer comprehension of pain assessment tools and communication with providers, potentially delaying intervention [23]. This subgroup’s higher pain incidence may also relate to delayed presentation (larger surgical defects), economic constraints, and inadequate insurance, leading to passive pain tolerance. Higher-educated patients may better utilize cognitive-behavioral strategies for pain management [24, 25].
Notably, immunotherapy history emerged as a protective factor. Immune checkpoint inhibitors may modulate pain via neuro-immune crosstalk, suppressing macrophage secretion of nerve injury factors and enhancing T-cell-mediated clearance of damaged nerves, mitigating neuropathic pain [26, 27]. The gut microbiome-immune axis also influences pain through immunomodulation [28], while adequate pain control can reverse immunosuppression, improving immune status. However, limited sample size necessitates multicenter studies to clarify pathways and long-term effects.
This study’s innovations are threefold: first, granular pain risk stratification across demographic, tumor-related, surgical, and psychological domains. Second, advanced multivariate predictive modeling via a nomogram (AUC = 0.923) enabling “preoperative-intraoperative-postoperative” risk prediction, aligned with ERAS “preventive analgesia.” Third, identification of a critical gap: 48-hour PCIA insufficiently controls complex pain (51.5% moderate, 17.2% severe on day 3), necessitating optimized multimodal analgesia and extended duration.
Limitations include retrospective bias with unrecorded variables (e.g., intraoperative opioids, infections, occupational status and socioeconomic factors such as personal income and medical insurance type). Notably, occupational status and socioeconomic factors are closely correlated with educational level, potentially influencing patients’ health literacy, pain expression, and access to medical resources—factors that may indirectly affect postoperative pain perception and management outcomes. The absence of these variables in our analysis may limit the external generalizability of the study results. Pain assessment was restricted to 7 days, lacking chronic pain evaluation; prospective studies are needed for long-term outcomes. Mechanistic insights are limited by absent biomarker analyses. The model in this study is based on single-center data, which has been internally validated, but its generalizability still needs to be further verified by multicenter, prospective external data. In the future, we plan to collect [or collaborate with peers to acquire] multicenter data, systematically incorporating occupational status, socioeconomic factors, and other previously unrecorded variables to validate and refine this model and drive its translational application to clinical use. Future research should prioritize replication, mechanistic investigations, and the integration of these confounding factors to enhance the robustness and generalizability of the findings.
Conclusion
This study enrolled 458 oral cancer patients, and the results showed that the incidence of acute postoperative pain on postoperative day 3 was 74.67%. Seven independent risk factors were identified, and the constructed nomogram exhibited excellent predictive performance. This nomogram provides a reliable tool for risk stratification and individualized management, facilitating the development of personalized analgesic regimens and optimizing perioperative management. Future research will focus on multicenter validation and mechanistic exploration.
Acknowledgements
This study is a retrospective analysis of medical record data. No external funding was received during the research design, data collection, analysis, or manuscript preparation process. There was no specific institutional or departmental funding support for this study.
Abbreviations
- NRS
Numerical rating scale
- DCA
Decision curve analysis
- OR
Odds ratio
- ROC
Receiver operating characteristic
- AUC
Area under the curve
- ERAS
Enhanced recovery after surgery
- POD
Postoperative day
- TNM
Tumor node metastasis
- ULND
Unilateral lymph node dissection
- BLND
Bilateral lymph node dissection
- HADS
Hospital anxiety and depression scale
- PCIA
Patient-Controlled Intravenous Analgesia
Authors’ contributions
Fang Qi: contributed to the design, and acquisition, analysis, and interpretation of data; drafted the manuscript and critically revised manuscript. Yiqun Kang: contributed to the design and acquisition of data, critically revised manuscript. Long Zuo: contributed to acquisition and interpretation of data, critically revised manuscript. LiangYu Chen: contributed to conception and acquisition of data, critically revised manuscript. Yuan Yu: contributed to conception and acquisition of datacritically revised manuscript. Tuo Ji: contributed to conception and acquisition of data, critically revised manuscript.Wei Peng: contributed to interpretation of data, critically revised manuscript. Tiejun Zhan: Contributed to interpretation of data, critically revised manuscript. Li Wang: contributed to conception and design, and acquisition, analysis, and interpretation of data; drafted the manuscript and critically revised it for important intellectual content. All authors gave final approval and agreed to be accountable for all aspects of work ensuring integrity and accuracy.
Funding
This study received no funding.
Data availability
All data generated or analyzed during this study are included in this published article.
Declarations
Ethics approval and consent to participate
The study was reviewed and approved by the Ethics Committee of the Hospital of Stomatology, Wuhan University (Approval No.: [2023] Lun Shen Zi (B11) Hao) on April 12, 2023. Due to the retrospective nature of the study, written informed consent was waived by the Ethics Committee, and all data were de-identified to protect patient privacy.The study complied with the Helsinki Declaration.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, Jemal A. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74:229–63. [DOI] [PubMed] [Google Scholar]
- 2.Gao L, Sun J, Xu M, Chen L, Zhang Y, Wang Y, Chen Y, Chen X. Effect of the intraoperative head hypsokinesis on sore throat after thyroid surgery: a randomized controlled trial. Eur Arch Otorhinolaryngol. 2025;282(1):427–34. [DOI] [PubMed] [Google Scholar]
- 3.Burnstock G. Purinergic mechanisms and pain. Adv Pharmacol. 2016;75:91–137. [DOI] [PubMed] [Google Scholar]
- 4.Macionis V. Nociplastic pain: controversy of the concept. Korean J Pain. 2025;38(1):4–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Inchingolo AM, Dipalma G, Inchingolo AD, Palumbo I, Guglielmo M, Morolla R, Mancini A, Inchingolo F. Advancing postoperative pain management in oral cancer patients: A systematic review. Pharmaceuticals (Basel). 2024;17(4):542. [DOI] [PMC free article] [PubMed]
- 6.Chen D, Yang H, Yang L, Tang Y, Zeng H, He J, et al. Preoperative psychological symptoms and chronic postsurgical pain: analysis of the prospective China Surgery and Anaesthesia Cohort study. Br J Anaesth. 2024;132(2):359–71. [DOI] [PubMed] [Google Scholar]
- 7.Ing JW. Head and neck cancer pain. Otolaryngol Clin North Am. 2017;50(4):793–806. [DOI] [PubMed] [Google Scholar]
- 8.Hinther A, Nakoneshny SC, Chandarana SP, Wayne Matthews T, Dort JC. Efficacy of postoperative pain management in head and neck cancer patients. J Otolaryngol Head Neck Surg. 2018;47(1):29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Hoofwijk DMN, van Reij RRI, Rutten BPF, Kenis G, Theunissen M, Joosten EA, Buhre WF, van den Hoogen NJ. Genetic polymorphisms and prediction of chronic post-surgical pain after hysterectomy-a subgroup analysis of a multicenter cohort study. Acta Anaesthesiol Scand. 2019;63(8):1063–73. [DOI] [PubMed] [Google Scholar]
- 10.Mi X, Zou B, Rashidi P, Baharloo R, Fillingim RB, Wallace MR, Crispen PL, Parvataneni HK, Prieto HA, Gray CF, et al. Effects of patient and surgery characteristics on persistent postoperative pain: A mediation analysis. Clin J Pain. 2021;37(11):803–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Armstrong RA, Fayaz A, Manning GLP, Moonesinghe SR, Peri-operative Quality Improvement Programme delivery t, Oliver CM. collaborative P: Predicting severe pain after major surgery: a secondary analysis of the Peri-operative Quality Improvement Programme (PQIP) dataset. Anaesthesia 2023, 78(7):840–852. [DOI] [PubMed]
- 12.Kumar KS, Kiran SR. Pain after head-and-neck cancer surgery: the need for multidisciplinary approach. J Precis Oncol. 2023;3:61–2. [Google Scholar]
- 13.Khawaja S, Bavarian R, Abdul Rehman S, Hafeez H. Head and neck Cancer-Related pain: A descriptive analysis of the pain phenotypes. J Pain Res. 2023;16:2919–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Zhong W, Ma X, Xing Y, Kim DK, Wang A, Jiang W, Xu T. Blockade of peripheral nociceptive signal input relieves the formation of spinal central sensitization and retains morphine efficacy in a neuropathic pain rat model. Neurosci Lett. 2020;716:134643. [DOI] [PubMed] [Google Scholar]
- 15.Ye Y, Jensen DD, Viet CT, Pan HL, Campana WM, Amit M, Boada MD. Advances in head and neck cancer pain. J Dent Res. 2022;101(9):1025–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Zhang X, Xing L, Zhang X, Ai-Gumaei W, Wang Q, Li M, Long H, Lai W. Time-Series transcriptomics identifies Xanthine dehydrogenase in injured neurons correlated with orthodontic pain. Orthod Craniofac Res. 2025;28(6):1048–58. [DOI] [PubMed] [Google Scholar]
- 17.Orgill R, Krempl GA, Medina JE. Acute pain management following laryngectomy. Arch Otolaryngol Head Neck Surg. 2002;128(7):829–32. [DOI] [PubMed] [Google Scholar]
- 18.Khawaja SN, Jamshed A, Hussain RT. Prevalence of pain in oral cancer: A retrospective study. Oral Dis. 2021;27(7):1806–12. [DOI] [PubMed] [Google Scholar]
- 19.Ekstein MP, Weinbroum AA. Immediate postoperative pain in orthopedic patients is more intense and requires more analgesia than in post-laparotomy patients. Pain Med. 2011;12(2):308–13. [DOI] [PubMed] [Google Scholar]
- 20.Yang MMH, Hartley RL, Leung AA, Ronksley PE, Jette N, Casha S, Riva-Cambrin J. Preoperative predictors of poor acute postoperative pain control: a systematic review and meta-analysis. BMJ Open. 2019;9(4):e025091. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Franqueiro AR, Wilson JM, He J, Azizoddin DR, Karamnov S, Rathmell JP, Soens M, Schreiber KL. Prospective Study of Preoperative Negative Affect and Postoperative Pain in Patients Undergoing Thoracic Surgery: The Moderating Role of Sex. J Clin Med 2024, 13(19). [DOI] [PMC free article] [PubMed]
- 22.Niklasson A, Finan PH, Smith MT, Forsberg A, Dietz N, Kander T, Werner MU, Irwin MR, Kosek E, Bjurstrom MF. The relationship between preoperative sleep disturbance and acute postoperative pain control: A systematic review and meta-analysis. Sleep Med Rev. 2025;79:102014. [DOI] [PubMed] [Google Scholar]
- 23.Doan LV, Yoon J, Chun J, Perez R, Wang J. Pain associated with breast cancer: etiologies and therapies. Front Pain Res (Lausanne). 2023;4:1182488. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.van Lankveld W, Afram B, Staal JB, van der Sande R. The IASP pain curriculum for undergraduate allied health professionals: educators defining competence level using Dublin descriptors. BMC Med Educ. 2020;20(1):60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Wideman TH, Miller J, Bostick G, Thomas A, Bussieres A, Wickens RH. The current state of pain education within Canadian physiotherapy programs: a National survey of pain educators. Disabil Rehabil. 2020;42(9):1332–8. [DOI] [PubMed] [Google Scholar]
- 26.Zhang C, Li Y, Yu Y, Li Z, Xu X, Talifu Z, Liu W, Yang D, Gao F, Wei S, et al. Impact of inflammation and Treg cell regulation on neuropathic pain in spinal cord injury: mechanisms and therapeutic prospects. Front Immunol. 2024;15:1334828. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Ho MFS, Farkas O, Faria AV, Plemel JR, Kerr BJ. A recent history of immune cell sex differences in the peripheral nervous system in persistent pain States. Brain Behav Immun. 2025;128:766–75. [DOI] [PubMed] [Google Scholar]
- 28.Santoni M, Miccini F, Battelli N. Gut microbiota, immunity and pain. Immunol Lett. 2021;229:44–7. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
All data generated or analyzed during this study are included in this published article.