Development and validation of an intraoperative hypothermia nomograph model for patients undergoing video-assisted thoracoscopic lobectomy: a retrospective study

This study aimed to develop and internally validate a nomogram model for assessing the risk of intraoperative hypothermia in patients undergoing video-assisted thoracoscopic (VATS) lobectomy. This study is a retrospective study. A total of 530 patients who undergoing VATS lobectomy from January 2022 to December 2023 in a tertiary hospital in Wuhan were selected. Patients were divided into hypothermia group (n = 346) and non-hypothermia group (n = 184) according to whether hypothermia occurred during the operation. Lasso regression was used to screen the independent variables. Logistic regression was used to analyze the risk factors of hypothermia during operation, and a nomogram model was established. Bootstrap method was used to internally verify the nomogram model. Receiver operating characteristic (ROC) curve was used to evaluate the discrimination of the model. Calibration curve and Hosmer Lemeshow test were used to evaluate the accuracy of the model. Decision curve analysis (DCA) was used to evaluate the clinical utility of the model. Intraoperative hypothermia occurred in 346 of 530 patients undergoing VATS lobectomy (65.28%). Logistic regression analysis showed that age, serum total bilirubin, inhaled desflurane, anesthesia duration, intraoperative infusion volume, intraoperative blood loss and body mass index were risk factors for intraoperative hypothermia in patients undergoing VATS lobectomy (P < 0.05). The area under ROC curve was 0.757, 95% CI (0.714–0.799). The optimal cutoff value was 0.635, the sensitivity was 0.717, and the specificity was 0.658. These results suggested that the model was well discriminated. Calibration curve has shown that the actual values are generally in agreement with the predicted values. Hosmer–Lemeshow test showed that χ2 = 5.588, P = 0.693, indicating that the model has a good accuracy. The DCA results confirmed that the model had high clinical utility. The nomogram model constructed in this study showed good discrimination, accuracy and clinical utility in predicting patients with intraoperative hypothermia, which can provide reference for medical staff to screen high-risk of intraoperative hypothermia in patients undergoing VATS lobectomy.


Data analysis
All data analysis was performed using SPSS version 26.0 and R version 4.3.2.Categorical variables were described by frequency and percentage, and comparison among groups was used by chi-square test (χ 2 ).Continuous variables were described by mean and standard deviation (SD), and comparison among groups was used by T-test.Kolmogorov-Smirnov test was used for normality test.Continuous variables of skewed distribution were described by M (P25, P75), and comparison among groups was used by Mann-Whitney U test.Lasso regression was used to screen the independent variables.Logistic regression was used to construct the prediction model.R version 4.3.2 was used to construct the nomogram and visualize the prediction model.Bootstrap method was used to repeat 1000 times for internal validation of the nomogram model.Receiver operating characteristic (ROC) curve was used to evaluate the discrimination of the model.Calibration curve and Hosmer Lemeshow test were used to evaluate the accuracy of the model.Decision curve analysis (DCA) was used to evaluate the clinical utility of the model.All statistical tests were conducted by two-sided tests, and P values of < 0.05 indicated statistical significance.

Sample characteristics
A total of 530 patients who underwent VATS lobectomy were included in this study.Of the 530 participants, 290 were male, accounting for 54.72%.The median age was 63.00 years.The superior lobe of the left lung was surgically removed in 142 cases (26.79%), the inferior lobe of the left lung in 105 cases (19.81%), the superior lobe of the right lung in 164 cases (30.94%), the middle lobe of the right lung in 22 cases (4.15%), and the inferior lobe of the right lung in 97 cases.Intraoperative hypothermia occurred in 346 of 530 patients undergoing VATS lobectomy (65.28%).More sample characteristics are shown in Table 1.

Univariate analysis of intraoperative hypothermia
Patients were divided into hypothermia group (n = 346) and non-hypothermia group (n = 184) according to whether hypothermia occurred during the operation.The results showed that TBIL, age, albumin, BMI, combined hypertension, operation history, inhaled desflurane, anesthesia duration, intraoperative infusion volume  www.nature.com/scientificreports/and intraoperative blood loss, differences were statistically significant (P < 0.05).However, there was no significant difference in red blood cells, ASA and surgical spot (P > 0.05).More sample characteristics are shown in Table 1.

Lasso and logistic regression analysis
Ten variables with statistically significant in the univariate analysis were included in Lasso regression for variable screening.tenfold cross-validation was used to select the optimal lambda parameter values.The lambda parameter value with the smallest cross-validation error was taken as the optimal value of the model, and the number of variables at this time was calculated (See Figs.2,3).Lasso regression has shown that Seven variables including TBIL, age, BMI, inhaled desflurane, anesthesia duration, intraoperative infusion volume and intraoperative blood loss were screened.The curves in Fig. 2 represent the change trajectory of independent variable coefficients.The lower x-axis represents Log Lambda value, the upper x-axis represents the number of nonzero coefficients, and the y-axis represents the penalty coefficient.Figure 3 was a tenfold cross-validation fit curve.The lower x-axis represents Log Lambda value, the upper x-axis represents the number of nonzero coefficients, and the y-axis represents binomial deviance.Further, the intraoperative hypothermia was taken as the dependent variable (non-hypothermia = 0, hypothermia = 1), and 7 variables selected by Lasso regression were incorporated into Logistic regression for analysis.Table 2 showed how the independent variables were assigned.Logistic regression analysis showed that age, TBIL, inhaled desflurane, anesthesia duration, intraoperative infusion volume, intraoperative blood loss and body mass index were risk factors for intraoperative hypothermia in patients undergoing VATS lobectomy (P < 0.05).More sample characteristics are shown in Table 3.

Evaluation of prediction efficiency of nomogram model
Bootstrap method was used to repeat 1000 times for internal validation of the nomogram model.ROC curve was used to evaluate the discrimination of the model.The area under ROC curve was 0.757, 95% CI (0.714-0.799).
The optimal cutoff value was 0.635, the sensitivity was 0.717, and the specificity was 0.658.These results suggested that the model was well discriminated.Calibration curve has shown that the actual values are generally in agreement with the predicted values.Hosmer-Lemeshow test showed that χ2 = 5.588, P = 0.693, indicating that the model has a good accuracy.The ROC curve is shown in Fig. 5.The calibration curve is shown in Fig. 6.

Evaluation of clinical utility of nomogram model
DCA was used to evaluate the clinical utility of the model.DCA is to determine the clinical utility of the prediction model by calculating the net benefit of each patient developing the risk threshold of intraoperative hypothermia 27 .Figure 7 shows the DCA.The x-axis represents threshold probability, and the y-axis represents net benefit.The results have shown that the model threshold probability is between 0 and 0.90, and the net benefit rate is greater than 0, which indicates that the model has a high clinical utility.

Discussion
This study firstly developed a model for intraoperative hypothermia prediction in patients undergoing VATS lobectomy.It can provide reference for scientific temperature management in patients undergoing VATS lobectomy.In this study, the incidence of intraoperative hypothermia in patients undergoing VATS lobectomy was 65.28%, which was at a high level [11][12][13] .The reasons could be as follows: First, patients undergoing VATS lobectomy are routinely placed in a lateral position, exposing more body parts to the environment and losing more heat than supine position surgery.Second, the median age of the participants in this study was 63.00 years old, which indicates that the number of elderly patients is high.It has been reported that elderly patients are more likely to have intraoperative hypothermia than young patients 22 .Third, 80.57% (427 cases) of patients had anesthesia  for more than 3 h.It has been reported that the core body temperature of the patient will gradually decrease as the time of anesthesia is prolonged 28 .In addition, anesthetics can significantly impair normal autonomic thermoregulatory control, further affecting the patient's intraoperative body temperature 16 .
The results of this study showed that the older the age, the more likely to develop intraoperative hypothermia, which is consistent with a previous study 22 .As patients age, their basal metabolism decreases, blood circulation slows and sensitivity to temperature changes decreases 29 .In addition, elderly patients often suffer from basic diseases such as hypertension, diabetes and coronary heart disease, and their autoimmune function is poor, so they are more prone to intraoperative hypothermia.One study showed that patients with higher body fat percentage were less likely to develop intraoperative hypothermia, which is consistent with the results of this study (BMI > 23.9 kg/m 2 ) 30 .Body fat acts as insulation, helping the body retain heat 13 .In overweight patients (BMI > 23.9 kg/m 2 ), when the body temperature is reduced, patients with more body fat are more likely to stimulate vasoconstriction to reduce heat conduction to peripheral tissues, so as to maintain the body heat balance 31 .Therefore, these patients are not prone to intraoperative hypothermia.It is suggested that the medical staff should evaluate the BMI of the patients undergoing VATS lobectomy before surgery, and take stronger insulation measures for the patients with low body fat percentage, such as opening circulating hot water blanket and heating machine.
This study is the first to identify preoperative low serum TBIL levels and intraoperative inhaled desflurane as risk factors for intraoperative hypothermia in patients undergoing VATS lobectomy.This provides inspiration for future research on risk factors of intraoperative hypothermia.Serum TBIL, as an endogenous antioxidant in vivo, can produce anti-atherosclerosis properties through its antioxidant potential, and then provide certain protection against cardiovascular diseases 32 .The heart is the second largest energy user in the body, and half of the energy used is used to maintain body temperature.Therefore, higher serum TBIL levels can indirectly maintain body temperature.In addition, low total bilirubin is commonly seen in patients with malnutrition, anemia and other conditions.These diseases can indirectly lead to intraoperative hypothermia.Desflurane, as an inhaled anesthetic inducer, can enter the human body through the respiratory tract in the form of volatile gas, enter the blood through the alveolar artery, and with the blood circulation to the brain, block nerve transmission, and finally produce anesthesia 33 .It has been reported that desflurane inhibits the hypothalamic thermoregulatory center, reduces the temperature regulation ability of the autonomic nervous system, and slows down vasoconstriction, which increases heat loss and ultimately leads to lower body temperature in patients 34 .One study reported that desflurane inhibited vasoconstriction and chills to a greater extent when used in conventional concentrations than when propofol was administered intravenously 33 .Therefore, medical staff should pay attention to the anesthesia induction method of patients and take insulation measures in advance.
The odds ratio (OR) for the anesthesia duration was the largest among all risk factors (OR = 2.648), indicating the greatest advantage in predicting the effect of intraoperative hypothermia.When the anesthesia duration was greater than 180 min, the incidence of intraoperative hypothermia was higher in patients undergoing VATS lobectomy, which was consistent with the result of a previous study 28 .It has been reported that unheated anaesthetic can directly cause the core body temperature of patients to decrease by 1-2 °C16 .With the extension of anesthesia duration, the redistribution rate of heat from the core to the periphery will be accelerated, resulting in a significant increase in the incidence of hypothermia.This suggests that managers should improve the operating efficiency of the operating room, optimize the operation and anesthesia process, shorten the anesthesia duration of patients, and reduce the redistribution of heat.
The results of this study showed that intraoperative infusion volume greater than 1500 ml and intraoperative blood loss greater than 100 ml were risk factors for intraoperative hypothermia in patients undergoing VATS lobectomy, which was consistent with the results of previous studies 28,35 .It has been reported that a large amount of cold liquid into the body will absorb body heat, causing patients to have chills, thereby increasing the risk of intraoperative hypothermia 36 .Excessive blood loss during the operation will not only take away a lot of heat from the body, but also increase the amount of fluid replenishment during the operation and reduce the core body temperature.However, intraoperative hypothermia may also lead to potential intraoperative bleeding 37 .
Implications for clinical practice.The nomogram model constructed in this study for the first time has good discrimination, accuracy and clinical utility in predicting patients with intraoperative hypothermia.In clinical practice, nursing staff can evaluate and quantify the risk probability of intraoperative hypothermia in patients undergoing VATS lobectomy according to this nomogram model.The risk of intraoperative hypothermia was inferred based on the patient's age, BMI, serum TBIL levels, anesthesia duration, whether desflurane was inhaled, intraoperative infusion volume, and intraoperative blood loss, and individualized interventions were taken.For example, pay attention to the elderly and low BMI patients, and actively take measures to keep warm; the results of preoperative biochemical examination were monitored, and passive combined with active heat preservation measures were taken for patients with low serum TBIL levels 38 (Forced draft preheating, intraoperative fluid infusion and pleural flush were heated); before anesthesia induction, the patient was pre-insulated for at least 10 min and a closed respiratory circuit system was used.One study reported that 60 min after the start of surgery is the best intervention time window to prevent hypothermia in patients undergoing VATS surgery, and appropriate insulation measures should be selected within this time window 39 .We believe that the above preventive measures can reduce the incidence of intraoperative hypothermia in patients undergoing VATS lobectomy, reduce the intraoperative, postoperative and anesthesia complications, and improve the prognosis.
This study has several limitations.First, serum TBIL levels and inhaled desflurane as risk factors for intraoperative hypothermia have not been reported, and the results need to be further verified.But this result illustrates the innovation of the study.Second, this study is a single-center retrospective study, and a multi-center longitudinal study can be conducted in the future.Third, there may be some risk factors that are not included in the model for analysis, such as psychological factors of patients.

Conclusion
The nomogram model constructed in this study showed good discrimination, accuracy and clinical utility in predicting patients with intraoperative hypothermia, which can provide reference for medical staff to screen high-risk of intraoperative hypothermia in patients undergoing VATS lobectomy.This model is worthy of clinical promotion.

Figure 1 .
Figure 1.Flowchart of the study.

Figure 5 .
Figure 5. Receiver operating characteristic curve of prediction model.

Figure 7 .
Figure 7. Decision curve analysis of prediction model.

Table 1 .
Baseline characteristics of the patients.

Table 3 .
Logistic regression analysis of intraoperative hypothermia in patients undergoing VAST lobectomy.