Symptom Clusters and Quality of Life in Cervical Cancer Patients During the Perioperative Period: A Longitudinal Study Authors: Juan Yang, Yuting Zhang, Yaru Wu, Yuting Wang, Ling Chen, Chunyan Xu Published: 2026 ABSTRACT BackgroundThis longitudinal observational study aimed to investigate symptom clusters and quality of life (QOL) in patients with cervical cancer (CC) at multiple time points during the perioperative period. It also sought to analyze the correlation between symptom clusters and QOL, thereby providing evidence for dynamic symptom cluster management and strategies to improve patients’ QOL.MethodsFrom August to December 2025, we conducted a longitudinal observational study among 183 cervical cancer patients during the perioperative period. Participants were recruited from a gynecology center within a tertiary hospital in Northwestern China. Patients completed a demographic questionnaire, the Chinese version of the MD Anderson Symptom Inventory-perioperative symptom in patients with gynecologic cancer (MDASI-PeriOp-GYN), and the Functional Assessment of Cancer Therapy-General (FACT-G). Assessments were performed on the first day of admission (T1, preoperative baseline), 1–3 days after surgery (T2), and 7–10 days after surgery (T3). Symptom clusters were extracted using exploratory factor analysis (EFA). Longitudinal trends in symptom clusters were examined using latent class mixed models (LCMM). Trends in QOL were assessed with generalized estimating equations (GEE). Correlations between symptom clusters and QOL were analyzed using Spearman correlation analysis.ResultsFive symptom clusters were identified in cervical cancer patients during the perioperative period: pain-fatigue, psychological, hot flashes–neurological, digestive tract, and energy-deficit. The energy-deficit cluster was detected only at T2, whereas the digestive tract cluster was present at T2 and T3. LCMM analyses revealed that the pain-fatigue, psychological, and hot flashes–neurological clusters rose from T1 to T2 and subsequently declined after T2. Overall QOL followed a V-shaped trajectory: both domain and total scores declined significantly from T1 to T2, partially recovered at T3, but remained below preoperative levels. Moreover, the severity scores of all symptom clusters were negatively correlated with the total QOL score (p < 0.05).ConclusionPatients undergoing cervical cancer surgery often experience various symptom clusters. High symptom severity is typically associated with lower QOL. Therefore, clinical staff should prioritize the dynamic management of symptom clusters in perioperative cervical cancer patients. This strategy can help reduce symptom burden and enhance patients’ QOL. Introduction Cervical cancer (CC) is the most common malignancy of the female reproductive system. In 2022, approximately 661,000 new cases and 348,000 deaths were reported globally, ranking fourth in incidence among women.1 China reported approximately 151,000 new cases and 56,000 deaths, accounting for approximately 22.7% of the global total and 16.0% of global deaths. Cervical cancer ranks fifth in incidence among malignant tumors in Chinese women.2 Although incidence rates have declined in some high-income countries due to HPV vaccination and screening programs, the disease burden remains substantial in low- and middle-income regions.3,4 Advances in early screening have enabled timely detection of cervical cancer. Surgery remains the primary treatment for early-stage disease. Common procedures include radical hysterectomy and pelvic lymph node dissection.5 For young patients wishing to preserve fertility, conservative surgical options may be considered. A meta-analysis reported that for stage IA–IB1 cervical cancer, oncologic outcomes following conization and radical cervical resection were comparable, with recurrence rates of 0.4% and 2.3%, respectively. However, obstetric outcomes favored conization, with preterm birth rates of 6.8% versus 26.6%.6 A systematic review concluded that no standard fertility-preservation protocol currently exists for early-stage cervical cancer with tumors ≥ 2 cm. When patients strongly request fertility preservation, abdominal radical trachelectomy (ART) should be chosen over minimally invasive or vaginal approaches.7 Although fertility-sparing procedures do not appear to compromise oncologic safety, the surgical trauma can still induce a range of physical and psychological symptoms, including fatigue, pain, sadness, distress, and sleep disturbance. These symptoms often co-occur and interact, forming symptom clusters. The internal interactions within these symptom clusters exacerbate patients’ physical and psychological burden and significantly impair QOL.8,9 Symptom clusters are defined as two or more interrelated symptoms that occur concurrently.10 This concept was first proposed by Dodd et al in 2001.11 Subsequent research has highlighted their synergistic effects,12 dynamic nature,13 and impact on patients’ QOL.14 An integrative review emphasizes that cancer patients’ symptom experiences change dynamically over time. Consequently, symptom clusters reflect the overall symptom burden and independently predict patient outcomes, making them key targets for symptom management and QOL improvement.15 A systematic review and meta-analysis further confirmed that effective symptom management reduces symptom burden, improves QOL, and prolongs overall survival.16 These findings suggest that systematic identification and management of symptom clusters can alleviate patients’ symptom burden and significantly enhance their overall QOL. Currently, most studies on symptom clusters and QOL in cervical cancer have focused on patients receiving radiotherapy or chemotherapy and used cross-sectional designs.17,18 For instance, one cross-sectional study of patients undergoing concurrent chemoradiotherapy identified four symptom clusters and found that illness perception mediated their interrelationships.19 However, cross-sectional designs cannot capture changes in symptom clusters over the treatment course. Another cross-sectional survey of 250 patients assessed 40 symptoms and extracted nine symptom clusters. The authors noted that the number and composition of clusters were highly dependent on the selected assessment tool.18 In the longitudinal context, a two-year prospective study tracked postoperative emotional distress and QOL. It identified persistent anxiety, lymphedema, and menopausal symptoms, but analyzed only individual symptoms rather than symptom clusters.20 A large longitudinal study of 1,280 heterogeneous cancer patients examined symptom clusters and their stability. Although five clusters remained highly stable over two months, cervical cancer patients accounted for only 3.4% of the sample, limiting the generalizability of these findings to this population.21 In conclusion, existing research on cervical cancer symptom clusters exhibits gaps in population specificity. Most studies have focused on patients receiving radiotherapy or chemotherapy, or heterogeneous mixed cohorts. Additionally, they lack longitudinal, full-course tracking from preoperative baseline through postoperative rehabilitation. Accordingly, this study aims to identify perioperative symptom clusters in cervical cancer patients, map their dynamic trajectories from preoperative baseline to postoperative rehabilitation, and examine the relationships between these symptom clusters and QOL. Materials and Methods Study Design This longitudinal observational study was reported in accordance with the “Strengthening the Reporting of Observational Studies (STROBE)” guidelines.22 The study used a single-group repeated-measures design without a separate control group; all changes were evaluated relative to each patient’s preoperative baseline. Study Setting This single-center investigation was conducted at the Gynecology Center of the Affiliated Cancer Hospital of Xinjiang Medical University. Sampling Technique Convenience sampling was used. We recruited patients with cervical cancer scheduled for surgery between August and December 2025 who met the eligibility criteria. Inclusion criteria were histologically confirmed cervical cancer, planned primary surgery, age ≥ 18 years, clear understanding of the cancer diagnosis (verified via face-to-face communication during recruitment), sufficient ability to communicate and comprehend study information, and signed written informed consent. Exclusion criteria included mental illness, altered mental status or impaired consciousness, other life-threatening malignancies or severe diseases, concurrent enrollment in other clinical trials, and inability to complete all three follow-up assessments. All patients undergoing primary surgery were eligible regardless of FIGO stage. Postoperative tumor stage was recorded and included as a covariate in the GEE models analyzing quality of life trajectories. This study was approved by the Medical Ethics Committee of the Affiliated Cancer Hospital of Xinjiang Medical University (approval number: K-2024218) and conducted in accordance with the Declaration of Helsinki. All participants provided written informed consent prior to enrollment. Sample Size Determination The sample size was determined based on a maximum item-to-participant ratio of 1:5 to 1:10, with an anticipated dropout rate of 10%–20%.23 The Chinese version of the MDASI-PeriOp-GYN used in this study includes 22 items for symptom cluster extraction (13 core module items and 9 gynecologic oncology-specific items). With a total of 22 items, a 5:1 item-to-participant ratio, and an anticipated 10% dropout rate, the minimum required sample size was estimated as 22×5/0.9 ≈ 122. For practical feasibility, the target sample size was set at 200 cases. A total of 200 questionnaires were distributed at T1. By T2, 5 patients had withdrawn due to severe postoperative complications, and 7 declined further participation. At T3, 5 patients were lost to follow-up (they could not be contacted or did not return on time). The final effective sample comprised 183 patients who completed all three surveys, yielding an effective response rate of 91.5% (183/200). Follow-Up Protocol for the Repeated Measurements This observational study adopted a longitudinal design with three assessment points. The first assessment was conducted on the day of admission (T1, preoperative baseline), the second on postoperative days 1–3 (T2), and the third on postoperative days 7–10 (T3). Instruments Demographic and Clinical Characteristics Sociodemographic data included age, marital status, residence, employment status, and educational level. Disease-related data comprised histologic type, postoperative tumor stage, and surgical approach. Chinese Version of the MD Anderson Symptom Inventory-Perioperative Symptom in Patients with Gynecologic Cancer (MDASI-PeriOp-GYN) The MDASI-PeriOp-GYN was developed by the MD Anderson Cancer Center.24 It was translated into Chinese by Zheng.25 The scale consists of 28 items across three sections: a gynecologic malignancy–specific symptom module (9 items), a core symptom module for general malignancies (13 items), and a symptom interference subscale (6 items). Each item is rated on a 0–10 scale ranging from “no symptom” to “worst imaginable severity,” with a total score range of 0–280. Higher scores indicate greater perioperative symptom burden. The Chinese version has demonstrated good reliability (Cronbach’s α = 0.826) and validity. In the present study, Cronbach’s α for the Chinese version of the MDASI-PeriOp-GYN ranged from 0.754 to 0.775. Functional Assessment of Cancer Therapy-General, FACT-G The FACT-G was developed by Cella et al in 1993.26 It was later localized and revised by Wan et al in 2006.27 It evaluates patients’ QOL over the preceding week. The scale contains 27 items across four domains: physical well-being, social/family well-being, emotional well-being, and functional well-being. Each item is rated on a 5-point Likert scale (0–4). Positively worded items are scored directly, while negatively worded items are reverse-scored. Higher total scores reflect better QOL. The FACT-G exhibits robust reliability and validity, with domain-specific Cronbach’s α coefficients exceeding 0.81, and is widely used to assess QOL in cancer populations. In the present study, Cronbach’s α values for the FACT-G domains ranged from 0.803 to 0.950. Data Collection All researchers involved in this study completed standardized training. Prior to questionnaire administration, patients were fully informed of the study’s objectives, significance, and precautions. Written informed consent was obtained before enrollment. Statistical Analysis Data were analyzed using SPSS 26.0 and R 4.3. Categorical variables were presented as frequencies and percentages. Normally distributed continuous variables were presented as the mean ± standard deviation (SD). Non-normally distributed continuous variables were reported as median (interquartile range) [M (P25, P75)]. Exploratory factor analysis (EFA) was used to identify symptom clusters. To minimize bias from low-incidence symptoms that do not reflect core cluster components, we only included symptoms with an incidence > 20%.28 The Kaiser–Meyer–Olkin (KMO) test and Bartlett’s test of sphericity confirmed that the data were appropriate for factor analysis, with thresholds of KMO ≥ 0.6 and Bartlett’s test p < 0.05. Principal component analysis (PCA) with varimax rotation was applied for factor extraction. PCA was chosen due to its common application in symptom cluster research and strong performance in dimensionality reduction for generating composite cluster scores. Factors were retained according to the following criteria: eigenvalue ≥ 1, factor loadings ≥ 0.5, and a minimum of two items per factor. Tucker’s congruence coefficient (φ) was used to assess factorial similarity across time points. Its formula is \documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document}$$\rm \varphi = (\sum {f_1}_i{f_{2i}})^2{\mathrm{ }}/{\mathrm{ }}(\sum {f_{1i}}^2{\mathrm{ }}\cdot{\mathrm{ }}\sum {f_{2i}}^2)$$\end{document}, where f1i and f2i represent the factor loadings of the overlapping symptoms at the two time points, respectively. According to the criteria proposed by Lorenzo-Seva and ten Berge (2006): φ ≥ 0.95 indicates high stability; 0.85 ≤ φ < 0.95 indicates fair stability; and φ < 0.85 suggests structural change.29 Bootstrap analysis was used to assess the internal stability of symptom clusters. Tucker’s φ calculation and bootstrap stability analysis were implemented in R 4.3 (psych package). For each symptom cluster identified by EFA, latent class mixed models (LCMM) were fitted to identify heterogeneous trajectory subtypes using data from all three perioperative time points (T1–T3). Nonlinear trajectories were modeled by including a quadratic time term (time + time2). A single-class model was first fitted, and the number of classes was then increased sequentially. Each multi-class model was initialized using parameter estimates from the single-class model. Unequal variances between classes were permitted (nwg = TRUE). Model fit was assessed using the Akaike information criterion (AIC) and the Bayesian information criterion (BIC), with lower values indicating better fit; BIC served as the primary decision criterion. Classification accuracy was evaluated via entropy (range 0–1), where higher values reflect greater accuracy. To guard against spurious classes, each class was required to contain more than 5% of the total sample. Classification quality was further examined using the average posterior probability (AvePP) for each class: AvePP ≥ 0.80 indicated good classification, 0.70–0.80 acceptable, and < 0.70 poor. LCMM analyses were implemented with the lcmm package in R 4.3. Generalized estimating equations (GEE) were used to examine longitudinal trends in QOL, with assessment time included as a fixed factor. Monthly per capita household income and postoperative tumor stage were adjusted for as covariates. An unstructured working correlation matrix was adopted, and a gamma distribution with a log link function was specified. Robust standard errors were used for statistical inference. Pairwise comparisons were adjusted with the Bonferroni correction. Spearman correlation analysis was used to assess unadjusted bivariate associations between symptom clusters and QOL. All tests were two-sided, with statistical significance set at p < 0.05. Results Demographic and Clinical Characteristics A total of 183 participants were included in the study. Among them, 47 (25.7%) were aged 18–44 years, 115 (62.8%) were aged 45–65 years, and 21 (11.5%) were aged >65 years, as shown in Table 1.Table 1Demographic and Clinical Characteristics of Patients with Cervical Cancer (N = 183)VariableValuesAge (years) 18–4447 (25.7) 45–65115 (62.8) >6521 (11.5)Marital status Unmarried3 (1.6) Married157 (85.8) Divorced/Widowed23 (12.6)Ethnicity Han91 (49.7) Other92 (50.3)Residence Urban99 (54.1) Suburban23 (12.6) Rural61 (33.3)Educational level Primary school and below60 (32.7) Junior high school61 (33.3) High school/Technical school24 (13.2) Associate degree and above38 (20.8)Employment status Employed60 (32.8) Retired24 (13.1) Other99 (54.1)Monthly per capita household income (CNY) <20004 (2.2) 2000–400087 (47.5) >400092 (50.3)Health insurance Insured180 (98.4) Self-funded3 (1.6)Histologic type Squamous cell carcinoma140 (76.5) Adenocarcinoma25 (13.7) Other18 (9.8)Postoperative tumor stage I112 (61.2) II54 (29.5) III17 (9.3) IV0 (0)Surgical approach Laparoscopic120 (65.6) Laparotomy63 (34.4)Notes: Data are presented as n (%). Table 1Demographic and Clinical Characteristics of Patients with Cervical Cancer (N = 183)VariableValuesAge (years) 18–4447 (25.7) 45–65115 (62.8) >6521 (11.5)Marital status Unmarried3 (1.6) Married157 (85.8) Divorced/Widowed23 (12.6)Ethnicity Han91 (49.7) Other92 (50.3)Residence Urban99 (54.1) Suburban23 (12.6) Rural61 (33.3)Educational level Primary school and below60 (32.7) Junior high school61 (33.3) High school/Technical school24 (13.2) Associate degree and above38 (20.8)Employment status Employed60 (32.8) Retired24 (13.1) Other99 (54.1)Monthly per capita household income (CNY) <20004 (2.2) 2000–400087 (47.5) >400092 (50.3)Health insurance Insured180 (98.4) Self-funded3 (1.6)Histologic type Squamous cell carcinoma140 (76.5) Adenocarcinoma25 (13.7) Other18 (9.8)Postoperative tumor stage I112 (61.2) II54 (29.5) III17 (9.3) IV0 (0)Surgical approach Laparoscopic120 (65.6) Laparotomy63 (34.4)Notes: Data are presented as n (%). Demographic and Clinical Characteristics of Patients with Cervical Cancer (N = 183) Notes: Data are presented as n (%). Incidence and Severity Scores of Perioperative Symptoms At T1, the three most prevalent symptoms among patients with cervical cancer were distress, sleep disturbance, and fatigue, with sleep disturbance being the most severe. At T2, pain, fatigue, and sleep disturbance were the most common symptoms, and fatigue was the most severe. At T3, pain, fatigue, and sleep disturbance remained the most frequently reported symptoms, with fatigue showing the highest severity (Table 2).Table 2Perioperative Symptom Incidence and Severity in Patients with Cervical Cancer (N = 183)SymptomT1T2T3IncidenceSeverityIncidenceSeverityIncidenceSeverityPain49 (26.8)0 (0, 1)177 (96.7)4 (3, 5)132 (72.1)2 (0, 2)Fatigue103 (56.3)2 (0, 3)171 (93.4)5 (4, 6)151 (82.5)3 (2, 4)Nausea20 (10.9)0 (0, 0)74 (40.4)0 (0, 2)23 (12.6)0 (0, 0)Sleep disturbance128 (69.9)4 (0, 4)159 (86.9)4 (3, 5)109 (59.6)2 (0, 3)Distress121 (66.1)3 (0, 4)135 (73.8)3 (3, 5)104 (56.8)2 (0, 5)Palpitations/shortness of breath23 (12.6)0 (0, 0)32 (17.5)0 (0, 0)19 (10.4)0 (0, 0)Forgetfulness57 (31.1)0 (0, 1)33 (18.0)0 (0, 0)38 (20.8)0 (0, 0)Decreased appetite69 (37.7)0 (0, 2)113 (61.7)2 (0, 3)77 (42.1)0 (0, 2)Drowsiness23 (12.6)0 (0, 0)73 (39.1)0 (0, 1)21 (11.5)0 (0, 0)Dry mouth33 (18.0)0 (0, 0)67 (36.6)0 (0, 1)31 (16.9)0 (0, 0)Sadness99 (54.1)2 (0, 4)106 (57.9)2 (0, 4)74 (40.4)0 (0, 3)Vomiting6 (3.3)0 (0, 0)21 (11.5)0 (0, 0)13 (7.1)0 (0, 0)Numbness29 (15.8)0 (0, 0)69 (37.7)0 (0, 2)41 (22.4)0 (0, 0)Abdominal bloating32 (17.5)0 (0, 0)70 (38.3)0 (0, 2)47 (25.7)0 (0, 1)Abdominal cramps15 (8.2)0 (0, 0)64 (35.0)0 (0, 2)19 (10.4)0 (0, 0)Constipation19 (10.4)0 (0, 0)21 (11.5)0 (0, 0)24 (13.1)0 (0, 0)Diarrhea7 (3.8)0 (0, 0)17 (9.3)0 (0, 0)6 (3.3)0 (0, 0)Dizziness16 (8.7)0 (0, 0)61 (33.3)0 (0, 1)20 (10.9)0 (0, 0)Grogginess/Confusion6 (3.3)0 (0, 0)36 (19.7)0 (0, 0)10 (5.5)0 (0, 0)Urinary urgency8 (4.4)0 (0, 0)5 (2.7)0 (0, 0)7 (3.8)0 (0, 0)Dysuria4 (2.2)0 (0, 0)10 (5.5)0 (0, 0)9 (4.9)0 (0, 0)Hot flashes46 (25.1)0 (0, 1)86 (47.0)0 (0, 4)50 (27.3)0 (0, 1)Notes: Data are presented as n (%) or median (P25, P75). T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. Table 2Perioperative Symptom Incidence and Severity in Patients with Cervical Cancer (N = 183)SymptomT1T2T3IncidenceSeverityIncidenceSeverityIncidenceSeverityPain49 (26.8)0 (0, 1)177 (96.7)4 (3, 5)132 (72.1)2 (0, 2)Fatigue103 (56.3)2 (0, 3)171 (93.4)5 (4, 6)151 (82.5)3 (2, 4)Nausea20 (10.9)0 (0, 0)74 (40.4)0 (0, 2)23 (12.6)0 (0, 0)Sleep disturbance128 (69.9)4 (0, 4)159 (86.9)4 (3, 5)109 (59.6)2 (0, 3)Distress121 (66.1)3 (0, 4)135 (73.8)3 (3, 5)104 (56.8)2 (0, 5)Palpitations/shortness of breath23 (12.6)0 (0, 0)32 (17.5)0 (0, 0)19 (10.4)0 (0, 0)Forgetfulness57 (31.1)0 (0, 1)33 (18.0)0 (0, 0)38 (20.8)0 (0, 0)Decreased appetite69 (37.7)0 (0, 2)113 (61.7)2 (0, 3)77 (42.1)0 (0, 2)Drowsiness23 (12.6)0 (0, 0)73 (39.1)0 (0, 1)21 (11.5)0 (0, 0)Dry mouth33 (18.0)0 (0, 0)67 (36.6)0 (0, 1)31 (16.9)0 (0, 0)Sadness99 (54.1)2 (0, 4)106 (57.9)2 (0, 4)74 (40.4)0 (0, 3)Vomiting6 (3.3)0 (0, 0)21 (11.5)0 (0, 0)13 (7.1)0 (0, 0)Numbness29 (15.8)0 (0, 0)69 (37.7)0 (0, 2)41 (22.4)0 (0, 0)Abdominal bloating32 (17.5)0 (0, 0)70 (38.3)0 (0, 2)47 (25.7)0 (0, 1)Abdominal cramps15 (8.2)0 (0, 0)64 (35.0)0 (0, 2)19 (10.4)0 (0, 0)Constipation19 (10.4)0 (0, 0)21 (11.5)0 (0, 0)24 (13.1)0 (0, 0)Diarrhea7 (3.8)0 (0, 0)17 (9.3)0 (0, 0)6 (3.3)0 (0, 0)Dizziness16 (8.7)0 (0, 0)61 (33.3)0 (0, 1)20 (10.9)0 (0, 0)Grogginess/Confusion6 (3.3)0 (0, 0)36 (19.7)0 (0, 0)10 (5.5)0 (0, 0)Urinary urgency8 (4.4)0 (0, 0)5 (2.7)0 (0, 0)7 (3.8)0 (0, 0)Dysuria4 (2.2)0 (0, 0)10 (5.5)0 (0, 0)9 (4.9)0 (0, 0)Hot flashes46 (25.1)0 (0, 1)86 (47.0)0 (0, 4)50 (27.3)0 (0, 1)Notes: Data are presented as n (%) or median (P25, P75). T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. Perioperative Symptom Incidence and Severity in Patients with Cervical Cancer (N = 183) Notes: Data are presented as n (%) or median (P25, P75). T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. Symptom Clusters at Different Time Points Symptoms with an incidence ≥ 20% were included in EFA to extract symptom clusters. Items with factor loadings < 0.5 were excluded. A total of 8, 14, and 9 symptoms were retained at T1, T2, and T3, respectively. The KMO values at the three time points were 0.669, 0.692, and 0.663; Bartlett’s test of sphericity was significant (p < 0.001) at all time points, confirming the suitability of the data for factor analysis. The cumulative variance explained at T1, T2, and T3 was 70.034%, 64.055%, and 64.540%, respectively (Table 3).Table 3Symptom Count and Distribution of Each Symptom Cluster at Three Time Points (Based on Severity) (N = 183)Symptom ClusterSymptomFactor LoadingT1T2T3Pain-FatiguePain0.6180.8740.769Fatigue0.8230.8410.702Sleep disturbance—0.8710.763Decreased appetite0.782——Number of symptoms in this cluster3/43/43/4PsychologicalDistress0.9010.8610.757Sadness0.8490.8280.814Sleep disturbance0.742——Number of symptoms in this cluster3/32/32/3Hot flashes–neurologicalHot flashes0.8710.7300.883Forgetfulness0.867—0.707Numbness—0.6860.813Number of symptoms in this cluster2/32/33/3Digestive tractAbdominal bloating—0.7200.714Decreased appetite—0.7040.840Abdominal cramps—0.680—Nausea—0.630—Number of symptoms in this cluster4/42/4Energy-DeficitDrowsiness—0.720—Dry mouth—0.709—Dizziness—0.664—Number of symptoms in this cluster3/3Cumulative variance explained (%)70.03464.05564.540Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. — indicates the symptom was not included in the symptom cluster at the given time point. Table 3Symptom Count and Distribution of Each Symptom Cluster at Three Time Points (Based on Severity) (N = 183)Symptom ClusterSymptomFactor LoadingT1T2T3Pain-FatiguePain0.6180.8740.769Fatigue0.8230.8410.702Sleep disturbance—0.8710.763Decreased appetite0.782——Number of symptoms in this cluster3/43/43/4PsychologicalDistress0.9010.8610.757Sadness0.8490.8280.814Sleep disturbance0.742——Number of symptoms in this cluster3/32/32/3Hot flashes–neurologicalHot flashes0.8710.7300.883Forgetfulness0.867—0.707Numbness—0.6860.813Number of symptoms in this cluster2/32/33/3Digestive tractAbdominal bloating—0.7200.714Decreased appetite—0.7040.840Abdominal cramps—0.680—Nausea—0.630—Number of symptoms in this cluster4/42/4Energy-DeficitDrowsiness—0.720—Dry mouth—0.709—Dizziness—0.664—Number of symptoms in this cluster3/3Cumulative variance explained (%)70.03464.05564.540Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. — indicates the symptom was not included in the symptom cluster at the given time point. Symptom Count and Distribution of Each Symptom Cluster at Three Time Points (Based on Severity) (N = 183) Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. — indicates the symptom was not included in the symptom cluster at the given time point. At T1, three symptom clusters were identified: psychological symptom cluster (distress, sadness, sleep disturbance), pain-fatigue symptom cluster (pain, fatigue, decreased appetite), and hot flashes–neurological symptom cluster (hot flashes, forgetfulness). Detailed factor loadings are presented in Table S1. At T2, five symptom clusters were identified: pain-fatigue symptom cluster (pain, fatigue, sleep disturbance), psychological symptom cluster (distress, sadness), hot flashes–neurological symptom cluster (hot flashes, numbness), digestive tract symptom cluster (abdominal bloating, decreased appetite, abdominal cramps, nausea), and energy-deficit symptom cluster (drowsiness, dry mouth, dizziness). Detailed factor loadings are presented in Table S2. At T3, four symptom clusters were identified: pain-fatigue symptom cluster (pain, fatigue, sleep disturbance), psychological symptom cluster (distress, sadness), hot flashes–neurological symptom cluster (hot flashes, numbness, forgetfulness), and digestive tract symptom cluster (abdominal bloating, decreased appetite). Detailed factor loadings are presented in Table S3. Stability of Each Symptom Cluster Across Time Points To evaluate the structural stability of each symptom cluster across the three perioperative time points, Tucker’s congruence coefficient (φ) was calculated for pairwise comparisons between time points. Internal stability at each time point was assessed via bootstrap analysis with 1000 resamples. The results indicated that the psychological (mean φ = 0.895), hot flashes–neurological (mean φ = 0.941), and pain-fatigue (mean φ = 0.854) symptom clusters exhibited fair stability over time. The digestive tract symptom cluster (mean φ = 0.967) emerged postoperatively and showed a stable structure. The energy-deficit symptom cluster also emerged after surgery but could not be evaluated across time points (Table S4). Bootstrap analysis demonstrated that the mean consistency values at all three time points were above 0.85 (Table S5). Trajectory Classification of Symptom Clusters Across the Perioperative Period in Patients with Cervical Cancer The digestive tract and energy-deficit symptom clusters emerged only at T2 and were therefore excluded from this analysis. LCMM was used to fit 1- to 3-class models for each remaining symptom cluster (Table 4). For the pain-fatigue symptom cluster, the 3-class model contained a subgroup with a proportion < 5% of the total sample and was thus discarded. Consequently, a 2-class solution was retained (BIC = 1734.5). For the psychological symptom cluster, no meaningful reduction in BIC was observed across multiclass models, so a single-class solution was selected. For the hot flashes–neurological symptom cluster, a 2-class solution was optimal (BIC = 1432.6). Classification quality was good for all classes, with AvePP ≥ 0.80 (Table 5).Table 4LCMM Fit Indices for Symptom Cluster Trajectories in Patients with Cervical Cancer (N = 183)Symptom ClusterModelBICAICEntropyPain-fatigue11761.71748.9—21734.51705.70.63431738.41693.40.779Psychological12204.82192.0—22216.22187.30.69132235.42190.50.568Hot flashes–neurological11569.01556.2—21432.61403.70.86531442.01397.10.912Note: — indicates not applicable (entropy is undefined for a single class).Abbreviations: LCMM, latent class mixed models; AIC, Akaike Information Criterion; BIC, Bayesian Information Criterion. Table 5Average Posterior Probability (AvePP) of Each Latent ClassSymptom ClusterClassCountPercentageAvePPPain-fatigue16334.40.821212065.60.932Psychological11831001.000Hot flashes–neurological115283.10.96123116.90.963Notes: Data are presented as n (%).Abbreviation: AvePP, average posterior probability. Table 4LCMM Fit Indices for Symptom Cluster Trajectories in Patients with Cervical Cancer (N = 183)Symptom ClusterModelBICAICEntropyPain-fatigue11761.71748.9—21734.51705.70.63431738.41693.40.779Psychological12204.82192.0—22216.22187.30.69132235.42190.50.568Hot flashes–neurological11569.01556.2—21432.61403.70.86531442.01397.10.912Note: — indicates not applicable (entropy is undefined for a single class).Abbreviations: LCMM, latent class mixed models; AIC, Akaike Information Criterion; BIC, Bayesian Information Criterion. LCMM Fit Indices for Symptom Cluster Trajectories in Patients with Cervical Cancer (N = 183) Note: — indicates not applicable (entropy is undefined for a single class). Abbreviations: LCMM, latent class mixed models; AIC, Akaike Information Criterion; BIC, Bayesian Information Criterion. Table 5Average Posterior Probability (AvePP) of Each Latent ClassSymptom ClusterClassCountPercentageAvePPPain-fatigue16334.40.821212065.60.932Psychological11831001.000Hot flashes–neurological115283.10.96123116.90.963Notes: Data are presented as n (%).Abbreviation: AvePP, average posterior probability. Average Posterior Probability (AvePP) of Each Latent Class Notes: Data are presented as n (%). Abbreviation: AvePP, average posterior probability. Characteristics of Symptom Cluster Trajectories Across the Perioperative Period in Patients with Cervical Cancer For each symptom cluster, T1 was set as the reference point for defining the latent classes. Since all classes exhibited an increasing trend from T1 to T2, class labels were determined mainly based on trajectory changes from T2 to T3. In the pain-fatigue symptom cluster, Class 1 (34.4%) showed a gradual decline from T2 to T3 and was labeled as the “gradual rise–gradual decline” group. Class 2 (65.6%) showed a sharp decline and was labeled as the “rapid rise–rapid decline” group (Figure 1A). The psychological symptom cluster consisted of only one class, which was labeled the “gradual rise–gradual decline” group (Figure 1B). For the hot flashes–neurological symptom cluster, Class 1 (83.1%) showed a gradual decline from T2 to T3 and was labeled as the “gradual rise–gradual decline” group. Class 2 (16.9%) showed a sharp decline and was labeled as the “rapid rise–rapid decline” group (Figure 1C). Figure 1Trajectory of Symptom Clusters in Patients with Cervical Cancer During the Perioperative Period: (A) LCMM trajectory of the pain-fatigue symptom cluster; (B) LCMM trajectory of the psychological symptom cluster; (C) LCMM trajectory of the hot flashes–neurological symptom cluster.Image A: Line graph ′Pain-Fatigue Cluster′. X-axis: Time (T1, T2, T3). Y-axis: Mean Symptom Severity Score (1-5). Legend: Trajectory Class. Class 1: T1=0.97, T2=2.48, T3=1.93. Class 2: T1=0.82, T2=4.84, T3=2.14. Image B: Line graph ′Psychological Cluster′. X-axis: Time (T1, T2, T3). Y-axis: Mean Symptom Severity Score (1.9-2.6). Legend: Trajectory Class. Class 1: T1=2.37, T2=2.53, T3=1.92. Image C: Line graph ′Hot Flashes-Neurological Cluster′. X-axis: Time (T1, T2, T3). Y-axis: Mean Symptom Severity Score (0-4). Legend: Trajectory Class. Class 1: T1=0.46, T2=0.91, T3=0.37. Class 2: T1=0.44, T2=3.75, T3=0.34.A set of three line graphs showing symptom cluster trajectories over time.Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. Trajectory of Symptom Clusters in Patients with Cervical Cancer During the Perioperative Period: (A) LCMM trajectory of the pain-fatigue symptom cluster; (B) LCMM trajectory of the psychological symptom cluster; (C) LCMM trajectory of the hot flashes–neurological symptom cluster. Longitudinal Changes in Quality of Life in Patients with Cervical Cancer Across the Perioperative Period GEE analysis revealed a significant main effect of time on total QOL scores (Wald χ2 = 274.97, p < 0.001). The total score followed a V-shaped trajectory: it peaked at T1 (74.70 ± 1.176), declined to its lowest point at T2 (48.97 ± 0.967), and subsequently rose to 62.30 ± 0.782 at T3. All pairwise comparisons for the total score reached significance (all p < 0.001). The physical, social/family, and functional well-being domains showed a similar pattern (Wald χ2 = 631.30, 359.55, and 215.55, respectively; all p < 0.001). In contrast, the emotional well-being domain displayed a distinct trajectory (Wald χ2 = 18.46, p < 0.001). Scores were highest at T1 (14.59 ± 0.390) and dropped significantly at T2 (12.25 ± 0.357), with no statistically significant change from T2 to T3 (12.99 ± 0.294; p = 0.319). The estimated marginal means are presented in Table 6 and Table 7, and overall trajectories in Figure 2.Table 6Longitudinal Comparison of Domain Scores and Total Score of QOL Across Perioperative Time Points Among Patients with Cervical Cancer (N = 183)DomainTimeMeanSE95% Wald CIPhysical well-beingT124.890.14824.60–25.18T212.860.35212.19–13.57T319.850.24919.37–20.35Model effectWald χ2(2) = 631.30, p < 0.001Social/Family well-beingT119.930.35719.25–20.65T216.770.22416.34–17.22T317.970.22517.54–18.42Model effectWald χ2(2) = 359.55, p < 0.001Emotional well-beingT114.590.39013.85–15.37T212.250.35711.57–12.97T312.990.29412.42–13.58Model effectWald χ2(2) = 18.46, p < 0.001Functional well-beingT115.280.43014.46–16.15T27.080.2796.56–7.65T311.480.25510.99–11.99Model effectWald χ2(2) = 215.55, p < 0.001Total QOLT174.701.17672.43–77.04T248.970.96747.11–50.90T362.300.78260.78–63.85Model effectWald χ2(2) = 274.97, p < 0.001Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative.Abbreviations: CI, confidence interval; SE, standard error; QOL, quality of life. Table 7Pairwise Comparisons of QOL Across Perioperative Time Points (Rate Ratios with 95% Confidence Intervals) (N = 183)DomainComparisonRatio (exp(β))95% Wald CIp-value(Bonferroni)Physical well-beingT1 vs T21.941.83–2.04<0.001 ***T1 vs T31.251.22–1.29<0.001 ***T2 vs T30.650.61–0.68<0.001 ***Social/Family well-beingT1 vs T21.191.14–1.24<0.001 ***T1 vs T31.111.06–1.16<0.001 ***T2 vs T30.930.90–0.97<0.001 ***Emotional well-beingT1 vs T21.191.10–1.29<0.001 ***T1 vs T31.121.05–1.200.004 **T2 vs T30.940.88–1.010.319Functional well-beingT1 vs T22.161.96–2.37<0.001 ***T1 vs T31.331.24–1.43<0.001 ***T2 vs T30.620.56–0.67<0.001 ***Total QOLT1 vs T21.531.45–1.60<0.001 ***T1 vs T31.201.15–1.25<0.001 ***T2 vs T30.790.75–0.82<0.001 ***Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. ** indicates p < 0.01; *** indicates p < 0.001. The ratio (exp(β)) represents the multiplicative change in the predicted mean of the outcome (eg, a ratio of 1.94 means the mean at T1 is 1.94 times that at T2). p-values were adjusted using Bonferroni correction for multiple comparisons.Abbreviations: CI, confidence interval; QOL, quality of life. Figure 2Longitudinal Trends in Domain-Specific Quality of Life Among Patients with Cervical Cancer During the Perioperative Period.A line graph showing mean quality of life domain scores over time for four well-being domains.Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. Table 6Longitudinal Comparison of Domain Scores and Total Score of QOL Across Perioperative Time Points Among Patients with Cervical Cancer (N = 183)DomainTimeMeanSE95% Wald CIPhysical well-beingT124.890.14824.60–25.18T212.860.35212.19–13.57T319.850.24919.37–20.35Model effectWald χ2(2) = 631.30, p < 0.001Social/Family well-beingT119.930.35719.25–20.65T216.770.22416.34–17.22T317.970.22517.54–18.42Model effectWald χ2(2) = 359.55, p < 0.001Emotional well-beingT114.590.39013.85–15.37T212.250.35711.57–12.97T312.990.29412.42–13.58Model effectWald χ2(2) = 18.46, p < 0.001Functional well-beingT115.280.43014.46–16.15T27.080.2796.56–7.65T311.480.25510.99–11.99Model effectWald χ2(2) = 215.55, p < 0.001Total QOLT174.701.17672.43–77.04T248.970.96747.11–50.90T362.300.78260.78–63.85Model effectWald χ2(2) = 274.97, p < 0.001Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative.Abbreviations: CI, confidence interval; SE, standard error; QOL, quality of life. Longitudinal Comparison of Domain Scores and Total Score of QOL Across Perioperative Time Points Among Patients with Cervical Cancer (N = 183) Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. Abbreviations: CI, confidence interval; SE, standard error; QOL, quality of life. Table 7Pairwise Comparisons of QOL Across Perioperative Time Points (Rate Ratios with 95% Confidence Intervals) (N = 183)DomainComparisonRatio (exp(β))95% Wald CIp-value(Bonferroni)Physical well-beingT1 vs T21.941.83–2.04<0.001 ***T1 vs T31.251.22–1.29<0.001 ***T2 vs T30.650.61–0.68<0.001 ***Social/Family well-beingT1 vs T21.191.14–1.24<0.001 ***T1 vs T31.111.06–1.16<0.001 ***T2 vs T30.930.90–0.97<0.001 ***Emotional well-beingT1 vs T21.191.10–1.29<0.001 ***T1 vs T31.121.05–1.200.004 **T2 vs T30.940.88–1.010.319Functional well-beingT1 vs T22.161.96–2.37<0.001 ***T1 vs T31.331.24–1.43<0.001 ***T2 vs T30.620.56–0.67<0.001 ***Total QOLT1 vs T21.531.45–1.60<0.001 ***T1 vs T31.201.15–1.25<0.001 ***T2 vs T30.790.75–0.82<0.001 ***Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. ** indicates p < 0.01; *** indicates p < 0.001. The ratio (exp(β)) represents the multiplicative change in the predicted mean of the outcome (eg, a ratio of 1.94 means the mean at T1 is 1.94 times that at T2). p-values were adjusted using Bonferroni correction for multiple comparisons.Abbreviations: CI, confidence interval; QOL, quality of life. Pairwise Comparisons of QOL Across Perioperative Time Points (Rate Ratios with 95% Confidence Intervals) (N = 183) Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. ** indicates p < 0.01; *** indicates p < 0.001. The ratio (exp(β)) represents the multiplicative change in the predicted mean of the outcome (eg, a ratio of 1.94 means the mean at T1 is 1.94 times that at T2). p-values were adjusted using Bonferroni correction for multiple comparisons. Abbreviations: CI, confidence interval; QOL, quality of life. Longitudinal Trends in Domain-Specific Quality of Life Among Patients with Cervical Cancer During the Perioperative Period. Correlation Between Perioperative Symptom Clusters and Quality of Life in Patients with Cervical Cancer Spearman correlation analysis revealed that severity scores of each symptom cluster at all three perioperative time points were significantly negatively correlated with total QOL scores (T1: r = −0.800 to −0.305; T2: r = −0.661 to −0.357; T3: r = −0.663 to −0.324; all p < 0.01). The digestive tract symptom cluster, which emerged at T2, maintained a moderate negative correlation afterward. It was significantly associated with both total QOL (r = −0.661 at T2 and r = −0.419 at T3) and all QOL dimensions. The energy-deficit symptom cluster was present only at T2 and showed a correlation coefficient of −0.401 with total QOL (Table 8).Table 8Correlation Coefficients Between Symptom Clusters and QOL Across Three Perioperative Time Points Among Patients with Cervical Cancer (N = 183)TimeSymptomClusterPhysicalWell-BeingSocial/FamilyWell-BeingEmotionalWell-BeingFunctionalWell-BeingTotalQOL ScoreT1Pain-fatigue−0.693**−0.462**−0.427**−0.554**−0.560**Psychological−0.531**−0.649**−0.818**−0.759**−0.800**Hot flashes–neurological−0.259**−0.284**−0.232**−0.251**−0.305**T2Pain-fatigue−0.698**−0.103−0.475**−0.715**−0.652**Psychological−0.197**−0.004−0.611**−0.324**−0.382**Hot flashes–neurological−0.363**−0.113−0.338**−0.252**−0.357**Digestive tract−0.627**−0.257**−0.605**−0.528**−0.661**Energy-deficit−0.422**−0.119−0.377**−0.336**−0.401**T3Pain-fatigue−0.706**−0.256**−0.267**−0.705**−0.633**Psychological−0.464**−0.235**−0.695**−0.584**−0.663**Hot flashes–neurological−0.311**−0.131−0.238**−0.280**−0.324**Digestive tract−0.445**−0.146*−0.237**−0.459**−0.419**Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. * indicates p < 0.05; ** indicates p < 0.01.Abbreviation: QOL, quality of life. Table 8Correlation Coefficients Between Symptom Clusters and QOL Across Three Perioperative Time Points Among Patients with Cervical Cancer (N = 183)TimeSymptomClusterPhysicalWell-BeingSocial/FamilyWell-BeingEmotionalWell-BeingFunctionalWell-BeingTotalQOL ScoreT1Pain-fatigue−0.693**−0.462**−0.427**−0.554**−0.560**Psychological−0.531**−0.649**−0.818**−0.759**−0.800**Hot flashes–neurological−0.259**−0.284**−0.232**−0.251**−0.305**T2Pain-fatigue−0.698**−0.103−0.475**−0.715**−0.652**Psychological−0.197**−0.004−0.611**−0.324**−0.382**Hot flashes–neurological−0.363**−0.113−0.338**−0.252**−0.357**Digestive tract−0.627**−0.257**−0.605**−0.528**−0.661**Energy-deficit−0.422**−0.119−0.377**−0.336**−0.401**T3Pain-fatigue−0.706**−0.256**−0.267**−0.705**−0.633**Psychological−0.464**−0.235**−0.695**−0.584**−0.663**Hot flashes–neurological−0.311**−0.131−0.238**−0.280**−0.324**Digestive tract−0.445**−0.146*−0.237**−0.459**−0.419**Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. * indicates p < 0.05; ** indicates p < 0.01.Abbreviation: QOL, quality of life. Correlation Coefficients Between Symptom Clusters and QOL Across Three Perioperative Time Points Among Patients with Cervical Cancer (N = 183) Notes: T1, preoperative baseline (the first day of admission); T2, 1–3 days postoperative; T3, 7–10 days postoperative. * indicates p < 0.05; ** indicates p < 0.01. Abbreviation: QOL, quality of life. Discussion This longitudinal observational study aimed to explore the dynamic trajectories of perioperative symptom clusters among cervical cancer patients and to analyze their associations with QOL. Our findings revealed that these symptom clusters evolved dynamically over time. Notably, the pain-fatigue and hot flashes–neurological clusters each exhibited two distinct recovery trajectories, whereas the psychological cluster followed a uniform trajectory. The digestive tract and energy-deficit symptom clusters emerged exclusively postoperatively. Overall QOL declined sharply in the early postoperative phase and then partially recovered. Nevertheless, all symptom clusters remained negatively correlated with QOL throughout the perioperative period. In this study, the pain-fatigue symptom cluster persisted throughout the perioperative period. Its core symptoms (pain and fatigue) are consistent with findings from previous longitudinal studies on gynecologic malignancies.30 Notably, the accompanying symptom changed from decreased appetite preoperatively to sleep disturbance postoperatively. This indicates that even with the relatively stable structure (mean Tucker’s φ = 0.854), symptom clusters may still evolve dynamically in response to surgical stress. This finding reinforces the dynamic nature of symptom clusters and highlights the necessity of repeated assessments across the perioperative period.13 Although the psychological symptom cluster gradually declined in severity after T2 and followed a homogeneous trajectory, emotional well-being scores did not improve significantly from the early postoperative phase (T2) to the recovery phase (T3) (p = 0.319). This finding differs from previous reports of heterogeneous psychological distress trajectories among patients with other gynecological cancers, a difference that may be attributable to our limited follow-up period.31 The rate ratio for emotional well-being was close to 1 (exp(β) = 0.94; 95% CI, 0.88–1.01), indicating negligible change from T2 to T3. By contrast, physical and functional well-being demonstrated marked postoperative recovery, with respective rate ratios of 0.65 and 0.62, both of which deviated substantially from 1. Notably, this disconnect between declining psychological symptom severity and persistently poor emotional well-being has rarely been highlighted in previous studies. We hypothesize that intensive inpatient support rapidly alleviated subjective distress, whereas existential concerns such as altered body image, loss of fertility, and fear of cancer recurrence contributed to sustained poor emotional QOL.32–34 Longitudinal qualitative research is needed to validate this hypothesis. In this study, the hot flashes–neurological symptom cluster demonstrated a highly stable structure (mean φ = 0.941). Its occurrence was presumably attributable to the perimenopausal status of 62.8% of patients (aged 45–65 years) and to surgery-induced pelvic nerve injury.35–37 LCMM identified a minority of patients (16.9%) who exhibited a “rapid rise–rapid decline” trajectory, characterized by a sharp postoperative increase in severity followed by a rapid decrease. In contrast, the majority of patients followed a “gradual rise–gradual decline” pattern. This heterogeneity suggests the existence of a subgroup that is highly susceptible to acute hormone withdrawal and nerve injury. These findings provide preliminary evidence for early postoperative risk stratification and individualized interventions. The digestive tract and energy-deficit symptom clusters emerged exclusively postoperatively. The digestive tract cluster exhibited high structural consistency between T2 and T3 (φ = 0.967). This finding aligns with the pathophysiological mechanisms of postoperative gastrointestinal dysfunction and residual sedative effects associated with Enhanced Recovery After Surgery (ERAS) protocols.38,39 Having examined the features of individual symptom clusters, we next assessed their collective associations with QOL. All symptom cluster severity measures were negatively associated with overall QOL, such that higher symptom burden corresponded to poorer quality of life. This result aligns with prior studies on cervical cancer patients undergoing chemoradiotherapy.19 Building on prior cross-sectional work, our longitudinal perioperative data demonstrated a V-shaped trajectory of overall QOL and distinct time-dependent correlations. Total QOL scores were 53% higher at T1 than at T2 (exp(β) = 1.53). Despite partial recovery at T3, scores remained approximately 17% below baseline (exp(β) = 1.20, T1 vs. T3). The psychological symptom cluster had the strongest impact on preoperative QOL (r = −0.800), whereas the pain-fatigue cluster predominated at early postoperative T2 (r = −0.652). Such time-varying patterns, including the magnitude of postoperative decline and the shifting dominance of symptom clusters across treatment phases, cannot be captured by cross-sectional designs. According to the Theory of Unpleasant Symptoms (TOUS),40 the symptom experience is shaped by three interrelated categories of factors: physiological (eg, physical function, disease and treatment status), psychological (eg, mental, affective and cognitive function), and situational (eg, social support and the surrounding environment). Pain, fatigue, and sleep disturbance share proinflammatory cytokine pathways, exacerbate one another, and substantially impair physical and functional status.41,42 The pain-fatigue cluster had the strongest correlations with postoperative physical and functional well-being (r = −0.698 and r = −0.715, respectively). This aligns with the finding that the physical domain showed the largest perioperative fluctuation on the ratio scale (exp(β) ranging from 0.65 to 1.94). At T2, the transient digestive tract cluster was moderately to strongly correlated with overall QOL (r = −0.661). This underscores the importance of proactive gastrointestinal symptom management in the early postoperative period, despite its transient nature. The hot flashes–neurological cluster affected all QOL domains even at mild severity, indicating that targeted interventions are essential for treatment-specific symptom clusters. Based on these findings, we propose a phased, symptom cluster–targeted management strategy. Preoperatively, psychological distress screening should be integrated into routine care. For patients with moderate-to-severe distress, cognitive-behavioral interventions or relaxation training are recommended, alongside prehabilitation to improve nutritional status and physical activity tolerance.43,44 In the early postoperative period, multimodal analgesia should be administered to relieve the acute exacerbation of the pain-fatigue symptom cluster and to reduce opioid reliance.38 For patients at high risk of postoperative nausea and vomiting, multimodal prophylactic antiemetics should be prescribed.39 Meanwhile, gastrointestinal function recovery can be promoted via gum chewing, early oral intake, and early mobilization.38 Based on trajectory classification results, patients in the “gradual rise–gradual decline” pain-fatigue subgroup (34.4%) and those in the “rapid rise–rapid decline” hot flashes–neurological subgroup (16.9%) should be referred to pain and fatigue management services or menopause clinics, respectively, to prevent persistent symptoms. In the recovery phase, to address persistently poor emotional well-being, a remote stepped psychological support system should be established after discharge. This program combines surgical menopause management with progressive functional rehabilitation.45,46 Limitations and Future Directions This study has several limitations. First, single-center convenience sampling may limit the generalizability of our findings. Second, follow-up was limited to 7–10 days postoperatively, so long-term postoperative complications, including lymphedema and sexual dysfunction, were not assessed. Third, although we evaluated group-level structural stability across time points using exploratory factor analysis and Tucker’s congruence coefficients, individual-level transitions in symptom cluster membership were not explored. Longitudinal confirmatory factor analysis and latent transition modeling were not performed due to the relatively small sample size. Fourth, the observational study design prohibits causal inference. Additionally, inflammatory and neuroendocrine biomarkers were not measured. Fifth, the MDASI-PeriOp-GYN and FACT-G have not been specifically validated in a perioperative cervical cancer population, although they showed acceptable reliability in our sample. Finally, our mechanistic explanations for the psychological–emotional disconnect and the high-risk hot flashes–neurological subgroup remain preliminary and require further validation via biomarker testing and qualitative research. Future studies should adopt multicenter, large-sample prospective designs with prolonged follow-up. Longitudinal confirmatory factor analysis and latent transition models are warranted to examine both group-level structural stability and individual-level transitions of symptom clusters. Incorporating biomarker detection and longitudinal qualitative interviews can further clarify the causal pathways between symptom clusters and quality of life, as well as the underlying mechanisms of distinct symptom trajectories, which will improve the robustness and generalizability of related conclusions. Conclusion This study found that patients with cervical cancer develop multiple symptom clusters throughout the perioperative period. These clusters change dynamically over time and exert adverse effects on patients’ quality of life. Notably, both the pain-fatigue and hot flashes–neurological symptom clusters presented heterogeneous recovery trajectories. To optimize symptom management and QOL, a phased, symptom cluster–targeted perioperative strategy is likely more effective than single-symptom interventions. A better understanding of symptom clusters can support refined symptom management and ultimately improve perioperative quality of life among patients with cervical cancer.