Factors associated with postoperative shivering after total knee arthroplasty and development and validation of a predictive model

Abstract

Postoperative shivering is a common complication following anesthesia, which can increase oxygen consumption, prolong recovery, and affect patient comfort and safety. Understanding its risk factors is important for improving postoperative outcomes and guiding preventive strategies. To investigate the associated risk factors for postoperative shivering after total knee arthroplasty (TKA) and to develop and validate a predictive model. A retrospective review of medical records of patients who underwent TKA at our hospital from January 2023 to December 2024 was conducted. Single-factor and multi-factor logistic regression analysis was used to identify independent risk factors for postoperative shivering, and a nomogram was created to visualize the model. The discriminatory ability of the model was evaluated using receiver operating characteristic curves and the area under the curve, while the goodness-of-fit was assessed using the Hosmer–Lemeshow test. To enhance the robustness of the validation results, internal assessment was conducted using the Bootstrap method combined with 10-fold cross-validation, and calibration plots and decision curves were used to analyze the clinical application value of the model. A total of 685 patients who underwent TKA were included in the study, and 143 patients developed postoperative shivering, with an incidence rate of 20.88%. Through logistic regression analysis, 5 independent risk factors for postoperative shivering were identified: age over 65 years (OR: 1.784, 95% CI: 1.234–2.654), operating room temperature not exceeding 21°C (OR: 3.024, 95% CI: 2.083–6.174), intraoperative fluid administration exceeding 1500 mL (OR: 1.970, 95% CI: 1.288–3.194), use of a pain pump (OR: 1.573, 95% CI: 1.116–2.309). Anesthesia duration exceeding 150 minutes (OR: 2.549, 95% CI: 1.607–4.621). Based on the results of receiver operating characteristic curves and the Hosmer–Lemeshow test, combined with bootstrap and cross-validation, the model demonstrates good discriminative ability and adaptability, strong stability, and high clinical reference value. Postoperative shivering after TKA are influenced by multiple factors, and the nomogram model established in this study has good predictive performance, providing a scientific basis for clinical identification of high-risk patients and early intervention.

1. Introduction

With the acceleration of population aging, the incidence of knee joint diseases has been increasing year by year, severely affecting patients’ quality of life. Total knee arthroplasty (TKA) is the primary surgical treatment for end-stage knee osteoarthritis and has been widely applied in clinical practice.^[1] TKA not only significantly alleviates pain and restores knee joint function but also improves patients’ quality of life and mobility. However, as a relatively invasive orthopedic surgery, TKA is associated with various postoperative complications, including infection, thrombosis, nausea and vomiting, and postoperative shivering. Among these, postoperative shivering is a relatively common short-term complication. Previous studies have reported an incidence ranging from 5% to 65%, depending on the type of anesthesia, surgical procedure, and perioperative care strategies.^[2] Although often transient, it can significantly impair recovery, increase oxygen consumption, and cause hemodynamic instability.

Postoperative shivering typically manifests as involuntary muscle tremors and is one of the common stress responses during the perioperative period.^[3] Its underlying mechanisms remain unclear and may be associated with factors such as the effects of anesthetic drugs, impaired intraoperative temperature regulation, intraoperative fluid administration, intraoperative environmental temperature, and individual variability. Shivering not only causes significant discomfort and anxiety in patients but also increases cardiac and pulmonary workload, oxygen consumption, and metabolic rate. It may also lead to physiological disturbances such as electrocardiogram changes, blood pressure elevation, and decreased oxygen saturation, posing particular risks for elderly patients or those with underlying cardiac or pulmonary conditions.^[4,5] Therefore, identifying the relevant risk factors for postoperative shivering after TKA and conducting effective prediction is of great significance for clinicians to take preventive measures in advance and improve patient outcomes.

Currently, there are many studies on postoperative shivering after general anesthesia or spinal anesthesia, primarily in fields such as gynecology, general surgery, or urology. However, research specifically focused on postoperative shivering in orthopedic surgery, particularly in TKA patients, remains limited. The available literature is scarce and mostly confined to descriptive studies, lacking systematic risk assessment tools and predictive models.^[6,7] Additionally, existing studies have not combined large-scale data with scientific modeling methods, resulting in incomplete identification of influencing factors. The practicality and generalizability of predictive tools require further improvement. Therefore, it is necessary to develop a scientific and accurate predictive model in the context of orthopedic surgery to assess the risk of postoperative shivering and guide clinical individualized management.

In recent years, nomograms have gained widespread application in clinical risk prediction due to their intuitive and practical characteristics.^[8] As a visualization tool based on a multi-factor regression model, nomograms can convert complex statistical results into simple and intuitive charts, helping clinicians quickly assess the probability of specific events occurring in individual patients.^[9] Its core advantages lie not only in integrating multiple clinical variables to enhance predictive accuracy but also in its excellent interpretability and practicality, making it easy to implement in clinical practice. Compared with traditional scoring systems, nomograms offer significant advantages in terms of precision and flexibility, enabling personalized risk assessment and stratified management, particularly in scenarios requiring rapid decision-making before or during surgery.^[10] Therefore, under the backdrop of the continuous advancement of precision medicine, nomograms have become an important tool for constructing clinical prediction models and have shown broad application prospects in multiple fields such as orthopedics, oncology, cardiovascular medicine, and critical care.^[11–14]

Based on this, this study aims to retrospectively analyze the clinical data of TKA patients admitted to our hospital from January 2023 to December 2024, systematically analyze the relevant factors affecting postoperative shivering, and construct a visualizable nomogram prediction model to achieve early identification and individualized intervention of postoperative shivering. Through a rigorous modeling and validation process, we aim to enhance the stability and reliability of the model, providing a practical risk prediction tool for clinical practice. This study will also offer theoretical foundations and data support for the prevention and control of postoperative shivering, thereby optimizing perioperative management and improving the overall treatment outcomes and postoperative satisfaction of TKA patients.

2. Materials and methods

2.1. Data sources and data collection

This study was based on a retrospective design, systematically organizing and analyzing the clinical data and perioperative information of inpatients who underwent TKA at our hospital from January 2023 to December 2024. All data were obtained from actual inpatient medical records. The study process strictly adhered to medical ethics norms and complied with the review standards and relevant ethical requirements of our hospital’s ethics committee. The aim of this study is to identify potential influencing factors based on real clinical data, providing theoretical basis and data support for the construction of subsequent risk prediction models.

2.2. Inclusion and exclusion criteria

Inclusion criteria: to ensure the homogeneity of the study population and the reliability of the data, the following inclusion criteria were established: adult patients aged 18 years or older; patients scheduled to undergo unilateral primary TKA; preoperative American Society of Anesthesiologists physical status classification between I and IV; complete preoperative assessment records, anesthesia process records, postoperative recovery observation records, and temperature monitoring data.

Exclusion criteria: to avoid potential confounding factors, the following exclusion criteria were established: patients whose surgery was interrupted or required reintervention postoperatively; patients with fever (body temperature > 37.5°C) or confirmed infectious diseases preoperatively; patients with significant organ dysfunction, such as NYHA class III or higher heart failure, or severe pulmonary, hepatic, renal insufficiency; patients who received medications that may affect shivering assessment during surgery or postoperatively, such as neuromuscular blockers or sedatives; patients with incomplete medical records or missing critical clinical information.

Finally, based on the above inclusion and exclusion criteria, a number of TKA patients who met the criteria were selected and included in this study for relevant data analysis and model construction.

2.3. Collection of relevant variables

This study systematically collected multiple clinical and perioperative data from patients who underwent TKA, covering basic demographic characteristics and anesthesia-related indicators. These included: patient gender and age, body mass index, history of smoking and alcohol consumption, common chronic conditions such as hypertension and diabetes, preoperative fasting and fluid restriction duration, American Society of Anesthesiologists physical status classification, type of anesthesia administered, duration of anesthesia, total fluid infusion during surgery, use of a pain pump postoperatively, and intraoperative operating room temperature. The above variables were included in the analysis as potential factors influencing the occurrence of postoperative shivering, to support the subsequent identification of risk factors and the construction of a predictive model.

2.4. Definition of outcomes

Postoperative shivering refers to involuntary muscle tremors that occur during the anesthesia recovery period after TKA, often manifested as involuntary shaking of the facial, upper limb, lower limb, or trunk muscles. In this study, postoperative shivering was observed from the end of surgery until the patient regained consciousness and was transferred out of the operating room. It was assessed and recorded by anesthesiologists or circulating nurses in the operating room based on the patient’s presentation. Following the Crossley and Mahajan scoring criteria,^[15] the occurrence of grade 2 or higher shivering (i.e., visible muscle tremors requiring consideration of intervention) was defined as “shivering occurrence.” Postoperative shivering was recorded as a binary variable: occurrence was coded as 1, and nonoccurrence as 0. Patients who received muscle relaxants, strong sedatives, or other drugs that could affect muscle activity during or after surgery were excluded from the outcome analysis.

2.5. Statistical analysis

To construct and validate a predictive model for the risk of shivering after TKA, this study adopted a stratified random sampling strategy to group the collected clinical data. Specifically, using R software (version 4.2.1, R Foundation for Statistical Computing, Vienna, Austria), all samples were randomly allocated to a modeling set (70% of the total sample) and a validation set (30% of the total sample) in a 7:3 ratio. The modeling set was mainly used for preliminary model construction and internal fitting evaluation, while the validation set was used to independently verify the robustness and extrapolation performance of the model.

First, SPSS 27.0 (IBM Corp., Armonk) statistical software was used to perform preliminary univariate analysis on the modeling group samples. For each clinical variable and its association with postoperative shivering, the chi-square test was used to perform statistical tests on categorical variables to explore differences in their distribution across different groups. The significance level was set at a 2-tailed P-value < .05 to identify potential factors significantly associated with postoperative shivering, which were then included as candidate variables for further analysis.

Subsequently, the statistically significant variables identified through univariate analysis were incorporated into a multivariate logistic regression model to control for confounding factors and determine the independent predictive value of each variable for the occurrence of shivering after adjustment. The same significance criterion of P < .05 was applied to ensure that variables included in the final model had high statistical and clinical significance. Based on the results of the multivariate regression analysis, a nomogram was constructed in the R platform to visually illustrate the specific contributions of each independent risk factor to the risk of postoperative shivering and provide a basis for individualized risk assessment.

In terms of model performance evaluation, receiver operating characteristic (ROC) curves were plotted on the training and validation sets, and the area under the curve (AUC) was calculated to quantify the model’s discriminative ability. The closer the AUC value is to 1, the stronger the model’s ability to distinguish between the occurrence of shivering. Additionally, to assess the consistency between the model’s predicted values and actual observed results, calibration curves were further plotted. If the prediction curve is close to the ideal diagonal line, it indicates that the model has good calibration.

To assess the applicability of this prediction tool under different clinical risk thresholds, this study also introduced decision curve analysis. By comparing the net benefit corresponding to different decision strategies, the feasibility and application prospects of the model in actual clinical environments are judged. If the model achieves higher net benefit values across a wide range of risk probability intervals, it indicates that it has good clinical guidance value.

To enhance the reliability and robustness of model validation, this study combined the Bootstrap resampling method (with 1000 repetitions) and a 10-fold cross-validation strategy for internal validation, ensuring the stability and generalization ability of the model results from multiple dimensions. Finally, based on the results of the ROC curves, calibration plots, and decision curves, a comprehensive and systematic evaluation of the discriminative performance, goodness-of-fit, and clinical application value of the constructed predictive model was conducted.

3. Results

3.1. General situation

A total of 685 patients who underwent TKA were included in this study, all of whom met the established inclusion and exclusion criteria, and their clinical data were complete and reliable. Postoperative observation revealed that 143 patients experienced shivering, with an overall incidence rate of 20.88%, indicating that this complication has a certain risk of occurrence in patients after TKA. To construct and validate a postoperative shivering prediction model, this study randomly divided all cases into a modeling group and a validation group in a 7:3 ratio. Ultimately, 480 patients were assigned to the modeling set for model development and internal evaluation, while the remaining 205 patients were assigned to the validation set for independence testing and external stability verification.

3.2. Independent risk factors for the occurrence of shivering after TKA

In the modeling group, 14 clinical variables were subjected to univariate logistic regression analysis to identify potential risk factors associated with the occurrence of postoperative shivering after TKA. The results showed that 7 variables demonstrated statistically significant associations, including patient age, duration of preoperative fasting and fluid restriction, type of anesthesia, duration of anesthesia, intraoperative fluid administration, operating room temperature, and use of a pain pump (see Table 1). Multivariate logistic regression analysis identified age > 65 years (OR: 1.784; 95% CI: 1.234–2.654), operating room temperature ≤ 21°C (OR: 3.024; 95% CI: 2.083–6.174), intraoperative fluid administration > 1500 mL (OR: 1.970; 95% CI: 1.288–3.194), and use of a pain pump (OR: 1.573; 95% CI: 1.116–2.309) anesthesia duration > 150 minutes (OR: 2.549; 95% CI: 1.607–4.621) were identified as independent risk factors for the occurrence of shivering after TKA (Table 2). The identification of these independent risk factors provides important evidence for the establishment of subsequent predictive models and helps clinicians develop individualized prevention and management strategies for high-risk patients.

Table 1.

Univariate analysis of the occurrence of postoperative shivering after total knee arthroplasty.

Risk factors	Postoperative shivering group (N = 100)	Control group (N = 380)	P
Gender			.143
Male	32	152
Female	68	228
Age			.002
≤65 yr	34	195
>65 yr	66	185
BMI			.441
≤24 kg/m²	42	176
>24 kg/m²	58	204
Smoking			.312
Yes	38	124
No	62	256
Drinking			.416
Yes	41	139
No	59	241
Hypertension			.067
Yes	35	98
No	65	282
Diabetes			.094
Yes	31	87
No	69	293
Fasting time			.004
≤12 hr	26	159
>12 hr	74	221
ASA classification			.075
I–II	45	209
>II	55	171
Anesthesia method			.009
General anesthesia	42	108
General anesthesia + nerve block anesthesia	58	272
Anesthesia duration			.001
≤150 min	19	136
>150 min	81	244
Intraoperative fluid administration			.013
≤1500 mL	39	201
>1500 mL	61	179
Operating room temperature			.007
≤21°C	54	148
>21°C	46	232
Use of a pain pump			.010
Yes	66	196
No	34	184

Table 2.

Multivariate analysis of the occurrence of postoperative shivering after total knee arthroplasty.

Factors	β	SE	Wald χ2	P	OR	95% CI
Age > 65 yr	0.579	0.194	8.923	.003	1.784	1.234–2.654
Operating room temperature ≤ 21°C	1.107	0.305	13.174	.000	3.024	2.083–6.174
Intraoperative fluid administration > 1500 mL	0.678	0.237	8.182	.004	1.970	1.288–3.194
Use of a pain pump	0.453	0.181	6.273	.012	1.573	1.116–2.309
Anesthesia duration > 150 min	0.936	0.292	10.273	.001	2.549	1.607–4.621

3.3. Nomogram development and validation

Based on the independent risk factors identified through multivariate logistic regression analysis, a nomogram model was developed to predict the risk of postoperative shivering after TKA (Fig. 1). The model assigns scores to each predictor, which are summed to estimate an individual’s risk of postoperative shivering. To evaluate the model’s performance, ROC curves were plotted. The AUC reached 0.823 in the modeling group and 0.822 in the validation group (Fig. 2A and B), indicating strong discriminative ability. The Hosmer–Lemeshow goodness-of-fit test yielded χ² = 7.965 (P = .415) in the modeling group and χ² = 7.247 (P = .488) in the validation group, suggesting good calibration. Internal validation using the Bootstrap method and 10-fold cross-validation showed consistent performance, with AUC = 0.822, sensitivity = 82.4%, and specificity = 81.9%. Calibration curves (Fig. 2C and D) demonstrated good agreement between predicted and observed outcomes. Decision curve analysis (Fig. 2E and F) showed that the model provides a positive net benefit across a range of threshold probabilities.

Figure 1.

Nominal graph prediction model for the occurrence of chills after total knee arthroplasty.

Figure 2.

Receiver operating characteristic curves for the prediction of chills after total knee arthroplasty in the training set (A) and validation set (B). Calibration curves for the prediction of chills after total knee arthroplasty in the training set (C) and validation set (D). Decision curves for the nomination diagram predicting the occurrence of postoperative shivering after total knee arthroplasty in the training set (E) and validation set (F).

4. Discussion

This study conducted a retrospective analysis of clinical data from 685 patients who underwent TKA and identified several independent risk factors significantly associated with postoperative shivering, including age > 65 years, operating room temperature ≤ 21°C, intraoperative fluid infusion > 1500 mL, use of a pain pump, and anesthesia duration > 150 minutes. Based on these findings, a visualizable nomogram prediction model was constructed and systematically evaluated using receiver operating characteristic curves, the Hosmer–Lemeshow goodness-of-fit test, bootstrap sampling, and 10-fold cross-validation. The results demonstrated that the model exhibits high predictive accuracy, good fitting performance, and stable internal consistency.

Advanced age (>65 years) was found to significantly increase the risk of postoperative shivering following TKA. Elderly patients have significantly impaired temperature regulation. With advancing age, the central nervous system’s ability to perceive and regulate temperature decreases, leading to delayed or abnormal responses to temperature changes induced by surgery, thereby increasing the likelihood of shivering.^[16] Additionally, elderly patients have a reduced basal metabolic rate, insufficient heat production, and thinner fat layers, leading to increased heat loss and greater susceptibility to temperature imbalance during or after surgery.^[17] From an immunological perspective, elderly patients exhibit immune aging, with elevated levels of pro-inflammatory factors such as IL-6 and TNF-α, which intensify and prolong inflammatory responses. Inflammation-mediated thermal regulation abnormalities further promote the occurrence of shivering.^[18] During surgery, the release of a large amount of inflammatory mediators activates the hypothalamic thermoregulatory center, triggering shivering as a compensatory response to inflammation and body temperature decline. Additionally, elderly patients often have multiple chronic conditions (such as diabetes, cardiovascular, or pulmonary diseases), which may impair microcirculation and tissue perfusion, exacerbating postoperative insufficient heat production and impaired function of heat-producing organs (such as skeletal muscle), thereby reducing cold tolerance.^[19] Psychological and neurological factors should also not be overlooked. In the elderly, slowed neural conduction speeds and abnormal transmission of cold stimuli signals result in more pronounced and prolonged shivering responses. Clinical studies support these mechanisms. A study^[20] found that the incidence of shivering in patients aged 65 years and older was approximately twice that of younger patients (OR = 2.1, 95% CI: 1.4–3.2). Other analyses have shown that levels of pro-inflammatory factors in the body are significantly elevated in elderly patients postoperatively, and shivering symptoms are positively correlated with inflammatory markers (R = 0.56, P < .01), suggesting a key role of inflammatory responses in the development of shivering.^[21]

An operating room temperature at or below 21°C was independently associated with a higher incidence of postoperative shivering. During surgery, excessive heat loss by the patient and an imbalance between heat production and heat loss lead to a decrease in core body temperature, triggering central nervous system-mediated involuntary muscle tremors, that is, the shivering response. Intraoperative exposure, anesthesia-induced vasodilation, and evaporative heat loss further exacerbate hypothermia.^[22] When operating room temperatures are low, patients experience significantly increased heat loss through radiation, conduction, convection, and evaporation, especially under epidural or general anesthesia, where thermoregulatory function is suppressed, making it more difficult for the body to maintain core temperature through autonomic mechanisms. From a neurophysiological perspective, a hypothermic environment stimulates skin and deep receptors, activating the cold-sensitive areas in the anterior hypothalamus via the spinothalamic pathway, leading to enhanced sympathetic nervous activity and skeletal muscle contraction, which manifest as postoperative shivering.^[23] Additionally, a study^[24] noted that even with the use of warmed fluids or warming measures during surgery, low operating room temperatures can still lead to delayed shivering within 1 to 2 hours postoperatively. Clinical studies have also confirmed these mechanisms. A study^[25] found that the incidence of postoperative shivering was 39% in the group with an operating room temperature set at 20°C, significantly higher than the 16% in the 23°C group (P < .001). Another study^[26] noted that for every 1°C decrease in intraoperative environmental temperature, the risk of postoperative shivering increased by approximately 11% (OR = 1.11, 95% CI: 1.05–1.17).

Intraoperative fluid administration exceeding 1500 mL was identified as a significant contributor to postoperative shivering risk. Intraoperative massive fluid infusion (especially unwarmed fluids) can significantly increase the body’s cold load, leading to a rapid decrease in core body temperature, even when crystalloid fluid temperature is only 21 to 22°C. It is estimated that each liter of fluid infused at 20°C results in a decrease in body temperature of approximately 0.25°C.^[27] Therefore, when the infusion volume exceeds 1500 mL, the cumulative effect of negative energy balance becomes more pronounced, activating the hypothalamic cold center, leading to sympathetic nervous system excitation and involuntary muscle tremors, manifested as shivering. Additionally, massive fluid infusion may cause dilutional hypovolemia, reducing tissue perfusion and disrupting peripheral blood flow distribution, thereby exacerbating heat loss from the body surface.^[7] Furthermore, the vasoconstrictive effects of anesthesia limit heat production mechanisms and increase heat loss, further reducing cold tolerance. Additionally, rapid and large-volume fluid administration can trigger fluid shifts and disturbances in the internal environment, indirectly activating the release of inflammatory factors, which may also serve as one of the triggers for the shivering reflex. Clinical study results support the findings of this study. A study^[28] demonstrated a positive correlation between intraoperative fluid input and postoperative shivering, particularly when fluid input exceeded 1500 mL, with the incidence of shivering increasing from 12% to 37% (P < .01).

Use of a postoperative pain pump was shown to independently increase the likelihood of shivering in TKA patients. Pain pumps typically administer opioid medications (such as fentanyl, morphine, or sufentanil) in combination with local anesthetics (such as ropivacaine or bupivacaine) in a continuous or intermittent manner. While effective in alleviating postoperative pain, they may also interfere with normal temperature regulation mechanisms. Studies have shown that opioid drugs can activate μ-opioid receptors, thereby inhibiting the hypothalamic temperature regulation center’s sensitivity to hypothermia and delaying the reactive thermogenesis triggered by cold stimuli, such as shivering.^[29] Additionally, pain pumps may suppress postoperative metabolic heat production and stress-induced sympathetic excitation caused by pain, leading to impaired autonomic regulatory capacity in response to cold stimuli. Systemic absorption of local anesthetics can also increase heat loss through peripheral vasodilation. Notably, certain adjuvants in some analgesic pump formulations (e.g., unwarmed diluents) may accumulate latent cold loads during continuous infusion, exacerbating the risk of core temperature decline. This association has been confirmed by existing studies. A study^[30] reported that the incidence of postoperative shivering in patients using intravenous patient-controlled opioid analgesia was as high as 38%, significantly higher than the 16% in those not using pain pumps (P < .01). Another study^[31] also found that the use of pain pumps was independently associated with postoperative shivering (OR = 2.12, 95% CI: 1.34–3.31), suggesting that opioid analgesia strategies, while ensuring adequate pain relief, may increase the risk of shivering complications.

Anesthesia lasting more than 150 minutes was significantly associated with a greater risk of postoperative shivering. Anesthetic drugs (including inhaled anesthetics and intravenous anesthetics) exert dose- and time-dependent inhibitory effects on the hypothalamic temperature regulation center. As anesthesia duration prolongs, core temperature regulation mechanisms gradually weaken, leading to reduced heat production, especially when maintaining low environmental temperatures (e.g., ≤21°C) during surgery, resulting in significant heat loss and inducing reflexive shivering.^[32] Second, prolonged anesthesia often indicates significant surgical trauma or complex procedures, placing the body in a state of prolonged high stress. This leads to prolonged suppression of sympathetic nerve activity, causing peripheral vasodilation and heat transfer from the core to the periphery, exacerbating core temperature decline. Additionally, continuous fluid infusion is often required during anesthesia to maintain circulatory stability. If fluids are not adequately warmed, this can further increase the cold load. Studies have shown that for every 30-minute increase in anesthesia duration, the risk of postoperative shivering significantly increases.^[33] A study^[34] reported that the incidence of shivering in patients with anesthesia duration exceeding 2 hours was 35.8%, compared to 15.4% in those with anesthesia duration <2 hours (P < .001). Another study^[35] found that anesthesia duration exceeding 150 minutes was an independent predictor of shivering (OR = 2.48, 95% CI: 1.62–3.78, P < .01), suggesting that the duration of anesthesia management during surgery has a significant impact on temperature regulation.

Based on the independent risk factors for the occurrence of shivering after TKA identified in this study, including age over 65 years, operating room temperature of 21°C or lower, intraoperative fluid administration exceeding 1500 mL, use of a pain pump, and anesthesia duration exceeding 150 minutes, clinical practice should develop scientific and effective intervention strategies targeting these high-risk factors. First, perioperative temperature management should be strengthened, and operating room temperature should be appropriately regulated, with a recommended maintenance of 22°C or above to reduce environment-induced hypothermia and the occurrence of shivering. Second, intraoperative fluid management should be optimized to avoid excessive fluid administration, and fluid infusion rate and total volume should be adjusted appropriately based on the patient’s specific condition to prevent fluid-related temperature dilution and increased heat loss. Additionally, anesthetic regimens should be selected with caution, particularly in high-risk patients, weighing the benefits and risks of pain pumps. Multimodal analgesia may be considered when necessary to mitigate the negative impact of single-modality analgesia on temperature regulation. Furthermore, efforts should be made to minimize anesthesia and surgical duration to reduce the inhibitory effects of anesthesia on the temperature regulation center, and intraoperative temperature should be monitored dynamically with proactive warming measures implemented promptly. For older patients, preoperative assessment and individualized management should be strengthened to prevent and promptly intervene in shivering, thereby improving postoperative recovery quality. In summary, through multidisciplinary collaboration and comprehensive management, the incidence of postoperative shivering after TKA can be effectively reduced, enhancing postoperative comfort and safety for patients.

Although this study systematically explored relevant risk factors for postoperative shivering after TKA and constructed a nomogram with strong predictive performance, several limitations should be acknowledged. First, the study was conducted using a single-center retrospective design with a relatively homogeneous sample population, which may introduce selection bias and limit the generalizability of the model. Second, as an observational study, the findings reflect associations rather than causal relationships; establishing causality would require prospective cohort studies or randomized controlled trials. Third, certain potentially important variables (such as perioperative basal metabolic rate, preoperative body temperature, individual physical and psychological characteristics) were not fully captured, which may have led to residual confounding. Fourth, the study lacks external validation using multicenter or diverse clinical settings; future validation in larger and more heterogeneous populations is essential to confirm the model’s stability and applicability. Additionally, surgical team experience, intraoperative management techniques, and procedure details were not analyzed, though these factors may also influence the occurrence of shivering. Finally, the diagnosis of postoperative shivering relied primarily on clinical observation without objective physiological markers, which may introduce subjectivity. Future research should employ multicenter, prospective designs incorporating a broader range of clinical and physiological indicators to further refine the model and enhance its clinical utility.

5. Conclusions

This study identified 5 independent risk factors for postoperative shivering after total knee arthroplasty: age > 65 years, operating room temperature ≤ 21°C, intraoperative fluid administration > 1500 mL, use of a pain pump, and anesthesia duration > 150 minutes. Based on these factors, a nomogram was developed and internally validated, showing good discrimination and calibration. This model offers a practical tool to help clinicians identify high-risk patients and implement targeted preventive measures. Further multicenter prospective studies are warranted to validate and optimize its clinical utility.

Acknowledgments

The author would like to thank Deng Guanghua for his guidance in analyzing the data and submitting the thesis.

Author contributions

Conceptualization: Tao He.

Data curation: Tao He.

Formal analysis: Zhi-Jun Qin.

Investigation: Zhi-Jun Qin.

Methodology: Zhi-Jun Qin.

Software: Zhi-Jun Qin.

Validation: Zhi-Jun Qin.

Visualization: Zhi-Jun Qin.

Writing – original draft: Tao He.

Writing – review & editing: Tao He.