Prediction model establishment and validation for enteral nutrition aspiration during hospitalization in patients with acute pancreatitis

Abstract

BACKGROUND

Acute pancreatitis (AP) is a disease caused by abnormal activation of pancreatic enzymes and can lead to self-digestion of pancreatic tissues and dysfunction of other organs. Enteral nutrition plays a vital role in the treatment of AP because it can meet the nutritional needs of patients, promote the recovery of intestinal function, and maintain the barrier and immune functions of the intestine. However, the risk of aspiration during enteral nutrition is high; once aspiration occurs, it may cause serious complications, such as aspiration pneumonia, and suffocation, posing a threat to the patient’s life. This study aims to establish and validate a prediction model for enteral nutrition aspiration during hospitalization in patients with AP.

AIM

To establish and validate a predictive model for enteral nutrition aspiration during hospitalization in patients with AP.

METHODS

A retrospective review was conducted on 200 patients with AP admitted to Chengdu Shangjin Nanfu Hospital, West China Hospital of Sichuan University from January 2020 to February 2024. Clinical data were collected from the electronic medical record system. Patients were randomly divided into a validation group (n = 40) and a modeling group (n = 160) in a 1:4 ratio, matched with 200 patients from the same time period. The modeling group was further categorized into an aspiration group (n = 25) and a non-aspiration group (n = 175) based on the occurrence of enteral nutrition aspiration during hospitalization. Univariate and multivariate logistic regression analyses were performed to identify factors influencing enteral nutrition aspiration in patients with AP during hospitalization. A prediction model for enteral nutrition aspiration during hospitalization was constructed, and calibration curves were used for validation. Receiver operating characteristic curve analysis was conducted to evaluate the predictive value of the model.

RESULTS

There was no statistically significant difference in general data between the validation and modeling groups (P > 0.05). The comparison of age, gender, body mass index, smoking history, hypertension history, and diabetes history showed no statistically significant difference between the two groups (P > 0.05). However, patient position, consciousness status, nutritional risk, Acute Physiology and Chronic Health Evaluation (APACHE-II) score, and length of nasogastric tube placement showed statistically significant differences (P < 0.05) between the two groups. Multivariate logistic regression analysis showed that patient position, consciousness status, nutritional risk, APACHE-II score, and length of nasogastric tube placement were independent factors influencing enteral nutrition aspiration in patients with AP during hospitalization (P < 0.05). These factors were incorporated into the prediction model, which showed good consistency between the predicted and actual risks, as indicated by calibration curves with slopes close to 1 in the training and validation sets. Receiver operating characteristic analysis revealed an area under the curve (AUC) of 0.926 (95%CI: 0.8889-0.9675) in the training set. The optimal cutoff value is 0.73, with a sensitivity of 88.4 and specificity of 85.2. In the validation set, the AUC of the model for predicting enteral nutrition aspiration in patients with AP patients during hospitalization was 0.902, with a standard error of 0.040 (95%CI: 0.8284-0.9858), and the best cutoff value was 0.73, with a sensitivity of 91.9 and specificity of 81.8.

CONCLUSION

A prediction model for enteral nutrition aspiration during hospitalization in patients with AP was established and demonstrated high predictive value. Further clinical application of the model is warranted.

Key Words: Acute pancreatitis; Hospitalization; Enteral nutrition; Predictive model; Aspiration

Core Tip: The study aims to establish and verify a predictive model for the risk of enteral nutrition aspiration during hospitalization in patients with acute pancreatitis. By comprehensively analyzing patient clinical data, independent risk factors are identified to provide early risk warning, guide clinical nursing intervention, and reduce the incidence of complications. Rate, improve patient prognosis.

INTRODUCTION

Acute pancreatitis (AP) is a disease caused by abnormal activation of pancreatic enzymes and can lead to self-digestion of pancreatic tissues and dysfunction of other organs. AP usually occurs in adults, with an annual incidence rate of 5/100000 to 30/100000, and the incidence rate is increasing every year[1]. The main inducing factors of AP include cholelithiasis, alcohol intake, and hyperlipidemia, which account for more than 70% of the causes of pancreatitis in China[2]. In addition, excessive drinking and overeating are common inducing factors. AP can be classified into mild, moderate, and severe according to its severity as well as into interstitial edematous type and necrotizing type according to its pathology. The clinical manifestations of AP are diverse, but the most typical symptom is sudden and sustained upper abdominal pain[3,4]. Pain is often severe and difficult to bear, and the location may vary, radiating from the left upper abdomen to the waist and back. Patients may also experience nausea, vomiting, abdominal distension, and fever[5]. These symptoms not only cause great pain to patients but also may affect their daily life and work. Enteral nutrition plays a vital role in the treatment of AP because it can meet the nutritional needs of patients, promote the recovery of intestinal function, and maintain the barrier and immune functions of the intestine[6]. However, the risk of aspiration during enteral nutrition is high; once aspiration occurs, it may cause serious complications, such as aspiration pneumonia, and suffocation, posing a threat to the patient’s life[7]. Therefore, a prediction model for enteral nutrition aspiration during hospitalization should be established. This evaluated the data of 200 patients with AP in our hospital, established a prediction model, and validated it to provide a reference for clinical practice.

MATERIALS AND METHODS

Subject selection

A retrospective review was conducted on 200 patients diagnosed with AP and admitted to Chengdu Shangjin Nanfu Hospital, West China Hospital of Sichuan University from January 2020 to February 2024. The inclusion criteria were as follows: (1) All patients who met the clinical diagnosis of AP; (2) Patients who received enteral nutrition for more than one week; and (3) Patients with complete clinical data. The exclusion criteria included the following: (1) Patients with other intestinal diseases; (2) Patients with mental disorders; and (3) Patients with contraindications to nasogastric feeding. This study was approved by the Medical Ethics Committee of West China Hospital, Sichuan University (No. 2023-1702), and patient informed consent was exempted.

Methods

Clinical data were collected from the electronic medical record system, and patients were randomly divided into a validation group (n = 40) and a modeling group (n = 160) in a 1:4 ratio, matched with 200 patients from the same time period. The modeling group was further categorized into an aspiration group (n = 25) and a non-aspiration group (n = 135) based on the occurrence of enteral nutrition aspiration during hospitalization. Patients were observed for one month after treatment and hospitalization, and the criteria for aspiration judgment were as follows: (1) Typical clinical manifestations, such as coughing, rapid breathing, and increased heart rate during enteral nutrition; (2) Residual enteral nutrition fluid in the patient’s oral and nasal cavity or enteral nutrition fluid found in sputum after suctioning; (3) Potential of hydrogen (pH) value monitoring of respiratory secretions, with pH less than 7; and (4) Bronchoscopy examination showing the presence of gastric contents in the patient’s respiratory tract. Any of the above criteria was considered as aspiration.

Research process

(1) The general information of patients in the validation and modeling groups was compared; (2) Univariate analysis of factors affecting enteral nutrition aspiration in patients with AP during hospitalization was conducted in the modeling group; (3) Multivariate logistic regression analysis was performed for factors affecting enteral nutrition aspiration in patients with AP during hospitalization; (4) A prediction model was established for enteral nutrition aspiration in patients with AP during hospitalization and validated with a calibration curve; and (5) The predictive value of the model was analyzed using the receiver operating characteristic (ROC) curve, and area under the curve (AUC), sensitivity, and specificity were obtained.

Statistical analysis

Experimental data collected were analyzed using SPSS 27.0. Normally distributed metric data in the experimental data were represented as “mean ± SD” and independent sample one-tailed t-test was used for comparison. Count data were represented as the number of cases or rates, and χ² test or Fisher’s exact test was used for comparison. Univariate and binary logistic regression analyses were conducted to evaluate factors affecting enteral nutrition aspiration in patients with AP during hospitalization. ROC curve was used to evaluate the predictive value of the model for enteral nutrition aspiration in patients with AP during hospitalization, and a difference with P < 0.05 was considered statistically significant.

RESULTS

Comparison of general information between validation and modeling groups

No statistically significant difference was found in general information between the two groups of patients (P > 0.05, Table 1).

Table 1 Factors affecting enteral nutrition aspiration in patients with acute pancreatitis.

		Validation (n = 40)	Modeling (n = 160)	t/χ2	P value
Age (years)		46.22 ± 5.24	46.18 ± 5.27	0.043	0.966
Gender	Male	24	88	0.325	0.850
	Female	16	72
BMI (kg/m2)		21.17 ± 2.36	21.21 ± 2.13	0.104	0.917
Position	Active	20	84	0.080	0.961
	Passive	20	76
Consciousness status	Alert	33	130	0.149	0.985
	Disturbance	6	24
	Coma	1	6
Nutritional risk	Low	33	127	0.195	0.907
	High	7	33
AP ACHE-II score		15.19 ± 3.16	15.37 ± 2.45	0.391	0.696
Length of nasogastric tube placement (cm)		47.22 ± 2.05	47.56 ± 2.18	0.892	0.373
Smoking history	Yes	20	81	0.005	0.998
	No	20	79
Hypertension	Yes	18	73	0.005	0.998
	No	22	87
Diabetes	Yes	8	28	0.136	0.935
	No	32	132

BMI: Body mass index; AP: acute pancreatitis; ACHE-II: Acute Physiology and Chronic Health Evaluation.

Univariate analysis of factors affecting enteral nutrition aspiration in patients with ap during hospitalization

Age, gender, body mass index (BMI), smoking history, hypertension history, and diabetes history showed no statistically significant difference between the two groups of patients (P > 0.05). However, body position, consciousness status, nutritional risk, Acute Physiology and Chronic Health Evaluation (APACHE-II) score, and nasogastric tube insertion length were statistically significantly different (P < 0.05, Table 2).

Table 2 Univariate analysis of factors affecting enteral nutrition aspiration in patients with acute pancreatitis during hospitalization.

		Aspiration group (n = 25)	Non-aspiration group (n = 135)	t/χ2	P value
Age (years)		46.16 ± 5.13	46.27 ± 5.50	0.093	0.926
Gender	Male	13	75	0.108	0.948
	Female	12	60
BMI (kg/m2)		21.16 ± 2.23	21.22 ± 1.61	0.160	0.873
Position	Active	19	65	6.562	0.038
	Passive	6	70
Consciousness status	Alert	8	122	56.281	< 0.001
	Disturbance	16	8
	Coma	1	5
Nutritional risk	Low	5	122	63.810	< 0.001
	High	20	13
AP ACHE-II score		17.46 ± 2.25	13.54 ± 2.16	8.282	< 0.001
Length of nasogastric tube placement (cm)		52.16 ± 2.21	45.16 ± 2.23	14.436	< 0.001
Smoking history	Yes	12	69	0.082	0.960
	No	13	66
Hypertension	Yes	10	63	0.378	0.828
	No	15	72
Diabetes	Yes	6	22	0.867	0.648
	No	19	113

BMI: Body mass index; AP: acute pancreatitis; ACHE-II: Acute Physiology and Chronic Health Evaluation.

Multivariate logistic regression analysis of factors affecting enteral nutrition aspiration in patients with ap during hospitalization

Body position, consciousness status, nutritional risk, APACHE-II score, and nasogastric tube insertion length were assigned values (actual values were substituted) as independent variables, and enteral nutrition aspiration was used as the dependent variable (aspiration = 1, non-aspiration = 0) for analysis. The results of multivariate logistic regression analysis showed that body position, consciousness status, nutritional risk, APACHE-II score, and nasogastric tube insertion length were independent factors affecting enteral nutrition aspiration in AP patients during hospitalization (P < 0.05), as shown in Table 3.

Table 3 Multivariate logistic regression analysis of factors affecting enteral nutrition aspiration in patients with acute pancreatitis during hospitalization.

Risk factors	β	SE	Ward	OR	95%CI	P value
Position	1.152	0.445	6.703	3.165	1.323-7.571	< 0.001
Consciousness status	0.903	0.322	7.871	2.468	1.313-4.639	< 0.001
Nutritional risk	1.243	0.484	6.593	3.465	1.342-8.947	< 0.001
APACHE-II score	1.072	0.362	8.768	2.921	1.437-5.938	< 0.001
Length of nasogastric tube placement (cm)	1.179	0.418	7.955	3.251	1.433-7.276	< 0.001

ACHE-II: Acute Physiology and Chronic Health Evaluation.

Prediction model for enteral nutrition aspiration in patients with ap during hospitalization

Based on the results of logistic regression analysis, body position, consciousness status, nutritional risk, APACHE-II score, and nasogastric tube insertion length were included in the prediction model for enteral nutrition aspiration in patients with AP during hospitalization (Figure 1). The calibration curve slope of this model was close to 1 in the training and validation sets, indicating good consistency between the predicted and actual risks of aspiration (Figure 2).

Figure 1 Prediction model for enteral nutrition aspiration in patients with acute pancreatitis during hospitalization.

Figure 2 Calibration curve. A: Training set; B: Validation set.

ROC analysis of the prediction model for enteral nutrition aspiration in patients with ap during hospitalization

The ROC analysis results showed that the AUC of the prediction model for enteral nutrition aspiration in patients with AP during hospitalization was 0.926 (standard error = 0.034, 95%CI: 0.8889-0.9675) in the training set, with a best cutoff value of 0.73, sensitivity of 88.4, and specificity of 85.2 (Figure 3A). In the validation set, the AUC of the prediction model was 0.902 (standard error = 0.040, 95%CI: 0.8284-0.9858), with a best cutoff value of 0.73, sensitivity of 91.9, and specificity of 81.8 (Figure 3B).

Figure 3 Receiver operating characteristic analysis of the training set, demonstrating the predictive value of the model for enteral nutrition aspiration in patients with acute pancreatitis during hospitalization. A: Training set; B: Validation set. AUC: Area under the curve.

DISCUSSION

Enteral nutrition aspiration during AP hospitalization is a common clinical problem that not only increases the difficulty of treatment but also prolongs hospital stay[8]. Aspiration of enteral nutrition has significant adverse effects on patients, including risk of serious complications, such as pulmonary infections and aspiration pneumonia, which can threaten the patient’s life. Aspiration can affect patient’s nutritional intake, leading to deterioration in their nutritional status and affecting the recovery process[9,10]. Furthermore, aspiration can cause psychological stress for patients, reducing their treatment confidence and quality of life. The present study investigated this issue and established a prediction model to provide clinical reference.

The results showed no statistically significant difference in age, gender, BMI, smoking history, hypertension history, or diabetes history between the two groups of patients (P > 0.05). However, statistically significant differences were found in body position, consciousness status, nutritional risk, APACHE-II score, and nasogastric tube insertion length between the two groups (P < 0.05). Further analysis using multivariate logistic regression analysis showed that body position, consciousness status, nutritional risk, APACHE-II score, and nasogastric tube insertion length were independent risk factors for enteral nutrition aspiration in patients with AP during hospitalization (P < 0.05). This finding indicates the influence of body position, consciousness status, nutritional risk, APACHE-II score, and nasogastric tube insertion length on aspiration in patients with AP. During AP treatment, when patients are in an active position, they may change position or move their bodies frequently, which may cause the nasogastric tube to shift. If the nasogastric tube shifts or becomes loose, then nutritional fluid may reflux into the esophagus or respiratory tract, thereby increasing the risk of aspiration[11]. In addition, an active position may reduce the coordination between a patient’s breathing and swallowing actions and nasogastric tube operation[12]. For example, during a change in position, a patient may not be able to effectively control their breathing and swallowing, making it more likely for nutritional fluid or food to enter the respiratory tract instead of the digestive tract. Consciousness status is also an important factor. Patients with AP may experience consciousness disorders because of the severity of their condition. In this case, patient’s swallowing and cough reflexes may be weakened, reducing their defense against aspiration and making them more susceptible to aspiration[13]. Nutritional risk is also a key factor in aspiration. Patients with AP often have a risk of malnutrition because of the impact of the disease. Malnutrition not only affects patients’ recovery speed but also may reduce the body’s immune function, increasing the risk of complications, including aspiration. Marchetti et al[14] reported a significant positive correlation between high nutritional risk and risk of in-hospital death (OR = 2.34; 95%CI: 1.31-4.19; P = 0.004). Patients with high nutritional risk usually require a number of treatment measures, such as tracheal intubation, tracheotomy, and nasogastric tube insertion. The implementation of these treatments not only increases the patient’s treatment burden but also may increase the risk of aspiration, which may have an adverse effect on prognosis[15]. Therefore, for patients with high nutritional risk, special attention should be paid to their nutritional status and risk management during treatment to reduce the probability of adverse events, such as aspiration, and improve patient’s survival rate and quality of life[16]. The APACHE-II score is an indicator used to evaluate the severity of a patient’s illness. A high score indicates severe patient’s condition. In this case, the patient’s intestinal function may be severely damaged, reducing their tolerance to enteral nutrition and increasing the risk of aspiration[17]. Finally, the length of nasogastric tube insertion is an important factor that affects aspiration. The length of the nasogastric tube directly affects whether nutritional fluid can accurately enter the intestine. If the nasogastric tube is inserted too short, then the nutritional fluid may flow directly into the esophagus, causing aspiration. If it is inserted too long, then it may stimulate the intestine, causing discomfort and complications[18].

Based on the results of the logistic regression analysis, body position, consciousness status, nutritional risk, APACHE-II score, and nasogastric tube insertion length were determined to be the key factors for constructing a prediction model for enteral nutrition aspiration in patients with AP during hospitalization. These factors not only independently affect the risk of aspiration but also provide abundant information for the construction of the model regarding their interactions. After incorporating these factors into the model, this study established a tool that can accurately predict the risk of aspiration. The analysis of the calibration curve showed that that the slope of the curve was close to 1, which means that the predicted risk of aspiration by the model is highly consistent with the observed risk. This result fully demonstrates the effectiveness and reliability of the model. Further ROC analysis showed that the AUC of the model for predicting enteral nutrition aspiration in patients with AP during hospitalization was 0.926 in the training set, with a standard error of 0.034 (95%CI: 0.8889-0.9675); the best cut-off value was 0.73, with a sensitivity of 88.4 and a specificity of 85.2. In the validation set, the AUC of the model for predicting enteral nutrition aspiration in patients with AP during hospitalization was 0.902, with a standard error of 0.040 (95%CI: 0.8284-0.9858); the best cutoff value was 0.73, with a sensitivity of 91.9 and a specificity of 81.8.

Although the model established in this study has good predictive value, it still has limitations. The construction of the model is mainly based on the currently available clinical data and variables, which may not fully represent all the factors that affect the risk of aspiration. Other potential influencing factors may not have been included in the model, so it may have a certain impact on the predictive results. Therefore, the model needs further improvement and refinement. By continuously collecting data, optimizing the model algorithm, and adjusting it in combination with clinical practice, a more accurate, stable, and practical prediction model can be developed in the future, providing more powerful support for the treatment and recovery of patients with AP.

CONCLUSION

The prediction model established for enteral nutrition aspiration in patients with AP during hospitalization has high predictive value and can provide effective reference for clinical doctors. It can help identify high-risk patients in advance, develop personalized enteral nutrition plans, reduce the incidence of aspiration, and improve patient’s treatment effect and quality of life. Therefore, this model should be used and promoted in clinical practice.