Computed tomography-derived nasojejunal tube tip-to-gastric antrum path length and short-term outcomes in severe acute pancreatitis

Abstract

BACKGROUND

Nasojejunal (NJ) feeding is commonly used in severe acute pancreatitis (SAP), but whether deeper tube advancement improves clinical outcomes is uncertain. Computed tomography (CT)-based three-dimensional reconstruction can quantify the in-situ tube trajectory.

AIM

To determine whether CT-derived NJ tube path length is associated with short-term outcomes in patients with confirmed SAP.

METHODS

This prospective single-center observational cohort included 63 patients with confirmed SAP. CT-based three-dimensional reconstruction measured the curved path length from the gastric antrum-pylorus junction to the tube tip. The primary outcome was intensive care unit (ICU) length of stay. Associations were assessed using Spearman correlation and Acute Physiology and Chronic Health Evaluation II-adjusted median regression. Firth penalized logistic regression evaluated the exploratory composite outcome. Anatomical tube-tip position was independently classified on multiplanar CT.

RESULTS

Median path length was 17.55 cm (interquartile range, 14.35-22.27 cm). Path length was not associated with ICU stay by Spearman analysis (rho = 0.055, P = 0.671) or adjusted median regression (difference per 5-cm increase = 0.192-day, 95% confidence interval: -0.922 to 1.305, P = 0.732). Post-CT ICU stay showed a similar null association. Anatomical position was duodenal in 28 patients, at the duodenojejunal flexure in 18, and jejunal in 17; path length correlated strongly with independently measured flexure-relative distance (rho = 0.997, P < 0.001). The exploratory composite outcome occurred in 17 patients, and path length remained non-significant in Firth regression.

CONCLUSION

CT-derived NJ tube path length was not associated with ICU outcomes but closely reflected distal anatomical tube position. These findings do not establish an optimal advancement target.

Key Words: Severe acute pancreatitis; Nasojejunal tube; Gastric antrum; Enteral nutrition; Computed tomography; Prognosis

Core Tip: This prospective single-center observational cohort used computed tomography-based three-dimensional reconstruction to quantify the curved path length from the gastric antrum-pylorus junction to the nasojejunal tube tip. Path length strongly corresponded to independently assessed distal tube-tip anatomy but was not associated with intensive care unit length of stay in unadjusted, Acute Physiology and Chronic Health Evaluation II-adjusted, or post-computed tomography analyses. These findings support the metric as an imaging-derived surrogate of tube advancement but do not establish a prognostic or therapeutic depth target.

INTRODUCTION

Acute pancreatitis (AP) is one of the most common acute gastrointestinal diseases requiring hospitalization. Its reported incidence varies substantially across regions and has increased over time, ranging from 4.9 to 73.4 per 100000 people, and has exhibited a gradual upward trend in recent years[1]. Approximately 20% of patients develop moderately severe or severe disease, and severe AP (SAP), characterized by persistent organ failure, carries substantial morbidity and mortality[2]. Nutritional support is a central component of management. Current guidelines recommend early oral or enteral nutrition when feasible and reserve parenteral nutrition for patients in whom oral or enteral feeding is not possible or tolerated[3-5].

For critically ill patients with SAP who cannot maintain adequate oral intake, enteral nutrition may be delivered through a nasogastric or nasojejunal (NJ) route. NJ feeding bypasses the pylorus and is often selected when gastric emptying is impaired, aspiration risk is high, or gastric feeding is poorly tolerated. Historically, more distal jejunal feeding was also proposed to reduce pancreatic exocrine stimulation and promote pancreatic “rest”. However, the clinical importance of progressively deeper jejunal tube advancement remains uncertain, and the position of the NJ tube tip varies considerably among patients.

The present study used computed tomography (CT)-based three-dimensional reconstruction to quantify the in-situ curved catheter path length from a predefined gastric antrum-pylorus junction to the radiopaque NJ tube tip. This metric represents the reconstructed catheter trajectory and was considered an imaging-derived surrogate of tube advancement rather than a direct anatomical distance from the ligament of Treitz to the tube tip. To examine its anatomical meaning, actual tube-tip location and a separate signed distance relative to the duodenojejunal flexure were independently assessed on multiplanar CT. We prospectively evaluated whether this CT-derived path length was associated with intensive care unit (ICU) length of stay and other short-term outcomes in patients with confirmed SAP.

MATERIALS AND METHODS

Research design

This prospective single-center observational cohort study was conducted in the ICU of Nanping First Hospital Affiliated to Fujian Medical University between December 2021 and December 2024. Consecutive eligible patients were prospectively screened during the prespecified recruitment period, and the study size was determined by the number of eligible patients enrolled during this period. The study protocol was approved by the Ethics Committee of Nanping First Hospital Affiliated to Fujian Medical University, approval No. NPSY202101017. Written informed consent was obtained from all patients or their legal guardians. Because this was a non-interventional observational cohort with no investigator-assigned treatment, interventional clinical-trial registration was not applicable.

Clinical variables were extracted from the electronic health record and nursing documentation, whereas imaging variables were obtained from the picture archiving and communication system. Two investigators independently cross-checked the analytic dataset against source records. Apparent outliers and inconsistencies were reverified in the original clinical or imaging records, and range and logical consistency checks were completed before statistical analysis.

Participants

Patients with predicted SAP were prospectively screened at ICU admission. The final analytic cohort included only patients who subsequently fulfilled the 2012 Revised Atlanta Classification criteria for SAP, defined as persistent organ failure lasting more than 48 hours[6], consistent with the 2021 Chinese guidelines[3].

The inclusion criteria were: (1) Age ≥ 18 years; (2) ICU hospitalization > 24 hours; (3) Provision of written informed consent; and (4) Acceptance of NJ tube placement for enteral nutrition. The diagnostic criteria and treatment protocols for SAP were based on the 2021 Chinese Guidelines for the Diagnosis and Treatment of AP[3].

The exclusion criteria were: (1) Pyloric obstruction, intestinal obstruction, severe Gastrointestinal bleeding, or another condition precluding enteral nutrition or NJ tube placement; and (2) CT imaging unsuitable for the predefined tube-path measurement. Patients transferred before completion of the prespecified exposure and outcome assessments were treated as attrition. The participant screening and inclusion process is shown in Figure 1.

Figure 1 Flow diagram of participant screening, enrollment, and analytic classification. Seventy-eight patients with predicted severe acute pancreatitis were screened. ICU: Intensive care unit.

Outcome measures

Baseline clinical variables included age, sex, pancreatitis etiology, Acute Physiology and Chronic Health Evaluation II (APACHE II) score at ICU admission, and Ranson pancreatitis score. The Ranson score was calculated using five parameters obtained at admission and six parameters obtained 48 h after admission.

The primary exposure was the CT-derived NJ tube tip-to-gastric antrum curved path length, and the primary clinical outcome was ICU length of stay from ICU admission to ICU recovery/discharge or death. Secondary outcomes included time to initiation of enteral nutrition, time to achievement of the enteral nutrition target, requirement for mechanical ventilation, and duration of mechanical ventilation. Post-CT ICU length of stay, defined as the remaining ICU stay after the index CT examination, was evaluated as a temporal sensitivity outcome.

Treatment failure was operationally defined as the occurrence during ICU treatment of any of the following: (1) ICU death; (2) New or worsening persistent organ failure, defined according to the Modified Marshall scoring system as a score ≥ 2 in a previously non-failing respiratory, cardiovascular, or renal system, or an increase in the number of failing organ systems persisting for > 48 hours[6]; (3) Infected pancreatic necrosis requiring unplanned surgical or interventional treatment; or (4) Unplanned escalation of organ support, including new initiation of vasopressor therapy, invasive mechanical ventilation, or renal replacement therapy. For the exploratory short-term composite outcome, an ICU stay of ≤ 10 days without treatment failure was classified as favorable, whereas an ICU stay of > 10 days or treatment failure was classified as poor. The 10-day threshold reflected the average ICU length of stay in our department.

Tube placement and enteral nutrition protocol

A single-lumen 14-Fr, 110-cm disposable nasogastrointestinal feeding tube (Jiangsu Kangbainian Medical Technology Co., Ltd., China) was used. All tubes were placed at the bedside using electromagnetic guidance, and initial clinical tube position was confirmed by abdominal radiography. Repositioning or replacement was performed for migration back into the stomach, mechanical dysfunction, or a clinical indication such as feeding intolerance. Tube position was not altered for study purposes, and no predefined distal jejunal depth was mandated.

Time to initiation of enteral nutrition was defined as the interval from ICU admission to the first administration of enteral nutrition. Time to achievement of the enteral nutrition target was defined as the interval from initiation of enteral feeding to the first achievement of ≥ 20 kcal/kg/day. A standard polymeric formula was used as the first-line enteral formulation; a semi-elemental peptide-based formula [Peptison; Nutricia Pharmaceutical (Wuxi) Co., Ltd., China] was used when clinically indicated for suspected malabsorption or intolerance, and feeding was advanced according to clinical tolerance.

CT acquisition and measurement

All index examinations were routine clinically indicated contrast-enhanced abdominal CT studies obtained after tube placement and before initiation of enteral nutrition. Images were acquired using a SOMATOM go.Top scanner (Siemens Healthineers, Germany), with portal venous phase imaging approximately 70-80 seconds after intravenous contrast administration and reconstruction at 1.0-mm slice thickness. Source images were transferred to a Siemens post-processing workstation. CT-based three-dimensional reconstruction was used to measure the NJ tube tip-to-gastric antrum curved path length (Figure 2). A three-dimensional volume-rendered image of the tube was generated, the gastric antrum-pylorus junction was predefined as the measurement starting point, and the radiopaque NJ tube tip was used as the endpoint. The curved centerline of the tube was manually traced using the workstation distance-line measurement tool; straightening of the reconstructed trajectory was used for visualization. All CT images used for study analysis and figure preparation were de-identified.

Figure 2 Computed tomography-based measurement of the nasojejunal tube tip-to-gastric antrum curved path length. A: Three-dimensional volume-rendered reconstruction. The measurement began at the predefined gastric antrum-pylorus junction, followed the selected curved centerline (CL_3), and ended at the nasojejunal tube tip; B: Straightened centerline reconstruction showing the selected centerline and the 0-mm reference point with orthogonal cross-sectional images; C: Anatomical schematic showing the gastric antrum-pylorus junction, duodenum, duodenojejunal flexure/Ligament of Treitz region, and nasojejunal tube tip. The measured variable was the curved catheter path length from the antrum-pylorus junction to the tube tip; the duodenojejunal flexure was an anatomical reference. NJ: Nasojejunal.

The measured value represented the curved catheter path length from the gastric antrum-pylorus junction to the NJ tube tip rather than a straight anatomical distance or the distance from the ligament of Treitz to the tube tip. Two radiologists blinded to clinical outcomes independently performed the path-length measurements using the same predefined landmarks and workstation protocol, and the finalized study measurement was used for analysis. Anatomical tube-tip location was independently classified on multiplanar CT according to visible landmarks of the duodenojejunal flexure as duodenal, located in the duodenojejunal flexure region, or jejunal beyond the flexure. Classification was based on anatomical appearance rather than a predefined distance threshold. A signed distance relative to the duodenojejunal flexure was measured separately as a continuous anatomical variable; negative values indicated a proximal position and positive values indicated extension beyond the flexure. The signed distance was not used to determine the categorical anatomical location.

Statistical analysis

Descriptive and univariate analyses were performed using SPSS version 27.0, with additional continuous-outcome and bootstrap analyses performed in Python 3.13.5 using SciPy 1.17.0 and statsmodels 0.14.6. Normality of continuous variables was assessed using the Shapiro-Wilk test (Supplementary Table 1). Continuous variables showing departures from normality were summarized as median and interquartile range (IQR) and compared using the Mann-Whitney U test. Categorical variables were expressed as n (%) and compared using the χ² test or an exact test, as appropriate; the Fisher-Freeman-Halton exact test was used for three-category comparisons. Spearman rank correlation was used for associations between CT-derived path length and continuous outcomes, with 95% confidence intervals (CIs) estimated by 10000 bootstrap resamples. Complete data were available for all variables included in the primary, secondary, and adjusted analyses.

The primary adjusted estimand was the difference in median ICU length of stay associated with a 5-cm increase in CT-derived path length after adjustment for APACHE II score, estimated using median regression. APACHE II was selected as the principal baseline severity covariate; Ranson score was not entered simultaneously to avoid redundant adjustment for disease severity. The same adjusted model was applied to post-CT ICU length of stay. Anatomical validation included Spearman correlation between path length and the independently measured signed flexure-relative distance, Kruskal-Wallis comparisons across anatomical tip-location categories, and an exact comparison of the exploratory composite outcome across categories. Potential nonlinearity was assessed using a centered quadratic path-length term, and a sensitivity analysis excluded the single ICU death.

Firth penalized logistic regression was retained for the exploratory binary outcome because of the limited number of events. Model 1 included CT-derived NJ path length, expressed per 5-cm increase, and APACHE II score, expressed per 1-point increase. Model 2 additionally included time to initiation of enteral nutrition, expressed per 24-hour delay; because enteral nutrition initiation occurred after the index CT exposure, model 2 was considered exploratory. Wald 95%CIs were reported. Sensitivity analyses used 7-, 10-, and 14-day thresholds for prolonged ICU stay, both as threshold-only outcomes and as composite outcomes retaining treatment failure. Adjusted threshold analyses included NJ path length and APACHE II score. All tests were two-sided, and P < 0.05 was considered statistically significant.

RESULTS

Participant characteristics and short-term outcomes

During the study period, 78 patients with predicted SAP were screened. Eleven were excluded before enrollment: 5 had a contraindication to enteral nutrition or tube placement, 3 declined written informed consent, 2 had CT images unsuitable for the predefined path measurement, and 1 was younger than 18 years. Sixty-seven patients were enrolled; 4 were transferred to a higher-level hospital before completion of the prespecified assessments, leaving 63 patients with confirmed SAP in the final analytic cohort (Figure 1). The final cohort comprised 36 men and 27 women, with a median age of 46.0 years (IQR, 37.0-62.0 years). The etiology was hyperlipidemic in 39 patients (61.9%), biliary in 7 (11.1%), and other or unidentified in 17 (27.0%).

According to the exploratory short-term composite outcome, 46 patients were classified as favorable and 17 as poor. Fifteen patients had an ICU stay > 10 days, and 4 patients (6.3%) experienced treatment failure, with 2 patients meeting both components. Treatment failure comprised 1 ICU death (1.6%), 2 cases of infected pancreatic necrosis requiring surgical or interventional treatment (3.2%), and 1 new initiation of renal replacement therapy (1.6%). Compared with the favorable-outcome group, the poor-outcome group had higher APACHE II and Ranson scores, later enteral nutrition initiation, and more frequent mechanical ventilation (all P < 0.05), whereas age, sex, etiology, time to enteral nutrition target, and CT-derived path length did not differ significantly (Table 1).

Table 1 Baseline and clinical characteristics according to exploratory short-term outcome, n (%).

Characteristic	Favorable outcome (n = 46)	Poor outcome (n = 17)	P value
Sex	-	-	0.461
Male	25 (54.3)	11 (64.7)	-
Female	21 (45.7)	6 (35.3)	-
Age, years	45.5 (38.3-56.5)	55.0 (35.0-83.0)	0.230
Etiology	-	-	0.058
Hyperlipidemic	32 (69.6)	7 (41.2)	-
Biliary	3 (6.5)	4 (23.5)	-
Other	11 (23.9)	6 (35.3)	-
APACHE II score	12.0 (10.0-17.0)	19.0 (16.0-22.0)	< 0.001
Ranson pancreatitis score	3.0 (2.0-4.0)	4.0 (3.0-5.0)	0.024
Time to initiation of enteral nutrition, hour	41.0 (35.0-55.8)	68.0 (43.0-93.0)	0.004
Time from EN initiation to enteral nutrition target, hour	94.0 (48.8-142.8)	56.0 (44.0-100.0)	0.127
CT-derived curved antrum-to-NJ-tip path length, cm	17.2 (14.0-22.0)	19.3 (15.2-22.7)	0.382
Mechanical ventilation	-	-	0.005
Yes	12 (26.1)	11 (64.7)	-
No	34 (73.9)	6 (35.3)	-

Continuous variables are presented as median (interquartile range) and were compared using the Mann-Whitney U test. Etiology was compared using the Fisher-Freeman-Halton exact test. Favorable outcome: Intensive care unit length of stay ≤10 days without treatment failure (n = 46). Poor outcome: Intensive care unit length of stay >10 days or treatment failure (n = 17). APACHE II: Acute Physiology and Chronic Health Evaluation II; NJ: Nasojejunal; EN: Enteral nutrition; CT: Computed tomography.

The median CT-derived NJ tube tip-to-gastric antrum path length was 17.55 cm (IQR, 14.35-22.27 cm; range, 9.55-39.74 cm). Median ICU length of stay was 7.0 days (IQR, 5.0-10.0 days; range, 2-31 days). Mechanical ventilation was required in 23 patients (36.5%), with a median ventilation duration of 105.0 hours (IQR, 82.0-201.0 hours). NJ tube placement occurred a median of 7.0 hours (IQR, 5.0-10.0 hours) after ICU admission, and CT was performed a median of 5.0 h (IQR, 4.0-7.0 hours) after tube placement and 12.0 h (IQR, 9.0-17.0 hours) after ICU admission. Enteral nutrition was initiated a median of 46.0 hours (IQR, 36.5-68.0 hours) after ICU admission, and the index CT preceded enteral nutrition initiation in all 63 patients (Table 2).

Table 2 Overall intensive care unit nutritional, and temporal characteristics, n (%).

Outcome/timing variable	Overall cohort (n = 63)
ICU length of stay, day	7.0 (5.0-10.0); range 20-31.0
Post-CT ICU length of stay, day	7.0 (5.0-10.0); range 20-31.0
Mechanical ventilation	23 (36.5)
Mechanical ventilation duration among ventilated patients, hour	105.0 (82.0-201.0)
NJ placement after ICU admission, hour	7.0 (5.0-10.0)
CT after NJ placement, hour	5.0 (4.0-7.0)
CT after ICU admission, hour	12.0 (9.0-17.0)
EN initiation after ICU admission, hour	46.0 (36.5-68.0)
Time from EN initiation to target ≥ 20 kcal/kg/day, hour	86.0 (48.0-135.0)
CT performed before EN initiation	63 (100.0)
Treatment failure	4 (6.3)
ICU death	1 (1.6)
Infected pancreatic necrosis requiring surgery/intervention	2 (3.2)
New renal replacement therapy	1 (1.6)
Exploratory poor composite outcome	17 (27.0)
ICU LOS > 10 days	15 (23.8)

Continuous variables are median (interquartile range), with range shown for intensive care unit length of stay. Treatment-failure components are mutually summarized by the primary qualifying event. ICU: Intensive care unit; CT: Computed tomography; NJ: Nasojejunal; EN: Enteral nutrition; CRRT: Continuous renal replacement therapy; LOS: Length of stay.

Association of NJ tube path length with ICU outcomes

In the overall cohort, CT-derived path length was not significantly correlated with total ICU length of stay (Spearman rho = 0.055, 95%CI: -0.216 to 0.317, P = 0.671; n = 63) (Figure 3A). In APACHE II-adjusted median regression, each 5-cm increase in path length was associated with a 0.192-day difference in median ICU length of stay (95%CI: -0.922 to 1.305, P = 0.732), whereas APACHE II score was associated with longer ICU stay (0.275 day per point, 95%CI: 0.082 to 0.468, P = 0.006) (Table 3). The post-CT analysis yielded a similar null association for path length (Spearman rho = 0.053, 95%CI: -0.218 to 0.315, P = 0.682; adjusted median difference per 5 cm = 0.192-day, 95%CI: -0.914 to 1.298, P = 0.730) (Figure 3B). Among the 23 mechanically ventilated patients, path length was not correlated with ventilation duration (Spearman rho = 0.064, 95%CI: -0.399 to 0.509, P = 0.771) (Figure 3C).

Figure 3 Associations of the computed tomography-derived nasojejunal tube tip-to-gastric antrum path length with intensive care unit outcomes. A: Total intensive care unit length of stay in all 63 patients (Spearman rho = 0.055, P = 0.671); B: Post-computed tomography intensive care unit length of stay in all 63 patients (Spearman rho = 0.053, P = 0.682); C: Mechanical ventilation duration among the 23 mechanically ventilated patients (Spearman rho = 0.064, P = 0.771). Trend lines are shown for visualization; statistical inference is based on Spearman rank correlation. CT: Computed tomography; ICU: Intensive care unit.

Table 3 Association of computed tomography-derived path length with continuous intensive care unit outcomes.

Analysis	Outcome	Variable	Effect estimate	95%CI	P value
Unadjusted Spearman correlation	Total ICU length of stay, day	CT-derived curved path length	ρ = 0.055	-0.216 to 0.317	0.671
Unadjusted Spearman correlation	Post-CT ICU length of stay, day	CT-derived curved path length	ρ = 0.053	-0.218 to 0.315	0.682
Unadjusted Spearman correlation	Mechanical ventilation duration, hour	CT-derived curved path length	ρ = 0.064	-0.399 to 0.509	0.771
Adjusted median regression	Total ICU length of stay, day	CT-derived curved path length, per 5-cm increase	0.192 days	-0.922 to 1.305	0.732
Adjusted median regression	Total ICU length of stay, day	APACHE II, per 1-point increase	0.275 days	0.082 to 0.468	0.006
Adjusted median regression	Post-CT ICU length of stay, day	CT-derived curved path length, per 5-cm increase	0.192 days	-0.914 to 1.298	0.730
Adjusted median regression	Post-CT ICU length of stay, day	APACHE II, per 1-point increase	0.275 days	0.084 to 0.467	0.006

Effect estimates for Spearman-correlation rows are Spearman ρ and are dimensionless; effect estimates for median-regression rows are adjusted median differences in intensive care unit days. Spearman 95% confidence intervals were obtained by 10000 bootstrap resamples. Median regression included computed tomography-derived curved path length and Acute Physiology and Chronic Health Evaluation II. CT: Computed tomography; ICU: Intensive care unit; APACHE II: Acute Physiology and Chronic Health Evaluation II; CI: Confidence interval.

Sensitivity analyses using 7-, 10-, and 14-day thresholds for prolonged ICU stay showed no significant association between CT-derived path length and prolonged ICU stay after adjustment for APACHE II score. Consistent results were obtained when treatment failure was retained as a component of the corresponding exploratory composite outcomes (Supplementary Table 2).

Penalized logistic regression

In Model 1, CT-derived NJ tube path length was not significantly associated with the poor composite outcome after adjustment for APACHE II score [adjusted odds ratio (OR) per 5-cm increase = 1.343, 95%CI: 0.808-2.232, P = 0.255]. A higher APACHE II score was associated with higher odds of the poor composite outcome (adjusted OR per 1-point increase = 1.177, 95%CI: 1.053-1.315, P = 0.004).

In exploratory model 2, CT-derived NJ tube path length remained unassociated with the poor composite outcome (adjusted OR per 5-cm increase = 1.423, 95%CI: 0.826-2.451, P = 0.204). APACHE II score remained associated with higher odds of the poor composite outcome (adjusted OR per 1-point increase = 1.154, 95%CI: 1.026-1.297, P = 0.017), and later enteral nutrition initiation was associated with higher odds of the poor composite outcome (adjusted OR per 24-hour delay = 1.816, 95%CI: 1.036-3.182, P = 0.037) (Table 4).

Table 4 Firth penalized logistic regression for the exploratory short-term composite outcome.

Model	Variable	Adjusted OR	95%CI	P value	Events/predictor parameters
1	CT-derived curved path length, per 5-cm increase	1.343	0.808-2.232	0.255	17/2
1	APACHE II, per 1-point increase	1.177	1.053-1.315	0.004	17/2
2	CT-derived curved path length, per 5-cm increase	1.423	0.826-2.451	0.204	17/3
2	APACHE II, per 1-point increase	1.154	1.026-1.297	0.017	17/3
2	EN initiation, per 24-hour delay	1.816	1.036-3.182	0.037	17/3

Model 1 included nasojejunal tube path length and Acute Physiology and Chronic Health Evaluation II score. Model 2 additionally included time to initiation of enteral nutrition and was considered exploratory. The binary outcome comprised 17 poor composite events. Model 1 contained two predictor parameters (approximately 8.5 events per parameter), and model 2 contained three predictor parameters (approximately 5.7 events per parameter). Wald 95% confidence intervals are reported. OR: Odds ratio; CT: Computed tomography; APACHE II: Acute Physiology and Chronic Health Evaluation II; CI: Confidence interval; EN: Enteral nutrition.

Anatomical validation and additional sensitivity analyses

On multiplanar CT, the tube tip was located in the duodenum in 28 patients (44.4%), in the duodenojejunal flexure region in 18 (28.6%), and in the jejunum in 17 (27.0%). CT-derived curved path length increased across these anatomical categories (median 13.60, 19.26, and 25.12 cm, respectively; P < 0.001) and correlated strongly with the independently measured signed distance relative to the duodenojejunal flexure (Spearman rho = 0.997, bootstrap 95%CI: 0.990-0.998, P < 0.001). ICU length of stay and the exploratory composite outcome did not differ significantly across anatomical tip-location categories (P = 0.869 and P = 0.716, respectively) (Supplementary Table 3; Supplementary Figure 1).

Baseline characteristics of the four enrolled patients transferred before completion of the prespecified assessments are summarized in Supplementary Table 4. Exclusion of the single ICU death did not materially change the association between path length and ICU length of stay (Spearman rho = 0.010, P = 0.936; adjusted median difference per 5 cm = 0.177 days, 95%CI: -1.014 to 1.368, P = 0.767). A quadratic path-length term was not significant in the adjusted median model (P = 0.358) (Supplementary Table 5). All four treatment-failure events occurred after the index CT examination; their components and timing are summarized in Supplementary Table 6.

DISCUSSION

Previous studies have compared nasogastric and NJ feeding in patients with predicted severe or severe AP. Petrov et al[7] reviewed 92 patients and found nasogastric feeding to be generally safe and well tolerated. Eatock et al[8] randomized 50 patients with severe AP to early nasogastric or NJ feeding and found no clear differences in APACHE II score, C-reactive protein response, pain, or analgesic use. Kumar et al[9] likewise found both routes feasible in a small randomized trial, and Singh et al[10] reported that early nasogastric feeding was not inferior to NJ feeding for infectious complications. A meta-analysis by Chang et al[11] found no significant differences in mortality, aspiration, diarrhea, pain aggravation, or energy balance between the two routes.

Zhu et al[12] subsequently reported no significant differences between nasogastric and NJ feeding in mortality, infectious complications, digestive complications, achievement of energy balance, or length of hospital stay. A Cochrane review by Dutta et al[13] similarly found insufficient evidence to establish superiority of either route. A 2024 meta-analysis by Wang et al[14] reported similar mortality between routes, although some analyses suggested differences in diarrhea, pain, and infection when feeding was initiated early. Current clinical guidance therefore supports enteral nutrition through either a gastric or post-pyloric route, with NJ feeding particularly relevant when gastric feeding is poorly tolerated, gastric emptying is delayed, or aspiration risk is high[15-18].

The question addressed in the present study differs from the comparison of nasogastric vs NJ feeding: All analyzed patients received NJ feeding, and we examined whether greater advancement of the tube was associated with better short-term outcomes once a post-pyloric route had been established. O’Keefe et al[19] reported that distal jejunal feeding may improve feeding delivery in selected patients with severe AP, but their trial focused primarily on feeding performance rather than demonstrating a general prognostic advantage of progressively greater tube depth. The traditional concept of complete pancreatic “rest” has also been questioned, and the benefits of enteral feeding may relate more to feeding tolerance and preservation of intestinal barrier function than to a simple linear relationship between tube depth and pancreatic exocrine suppression[20].

In the present cohort, the CT-derived NJ tube tip-to-gastric antrum path length was not significantly associated with the primary continuous ICU length-of-stay outcome in either unadjusted or APACHE II-adjusted analysis. The post-CT sensitivity analysis and the analysis of mechanical ventilation duration were similarly null, and the exploratory binary models produced concordant findings. At the same time, the curved path-length metric showed a very strong association with independently measured flexure-relative distance and increased stepwise from duodenal to flexure-region and jejunal tip locations. These findings support the construct validity of the CT-derived path length as an imaging-derived surrogate of distal tube advancement while confirming that it should not be interpreted as the direct anatomical distance from the ligament of Treitz to the tube tip.

These findings do not establish an optimal centimeter-based target for NJ tube advancement and are not sufficient to define a change in placement practice. The anatomical validation indicates that greater CT-derived path length generally corresponds to a more distal actual tube-tip position, but neither the continuous metric nor the categorical anatomical position showed a significant relationship with short-term ICU outcomes in this cohort. Tube-position decisions should therefore continue to be individualized according to established clinical considerations, including tube stability, gastric function, aspiration risk, feeding tolerance, anatomy, and procedural feasibility.

There is no universally accepted threshold for defining prolonged ICU stay in AP. Venkatesh et al[21], using a large critically ill AP cohort, evaluated a high-risk trajectory incorporating ICU mortality or an ICU stay longer than 7 days, whereas Lew et al[22] examined prolonged ICU admission longer than 14 days in SAP. In the present study, the 10-day threshold reflected the average ICU length of stay in our department and was used as a pragmatic clinical threshold. Sensitivity analyses at 7 days, 10 days, and 14 days consistently showed no significant association between NJ tube path length and outcome, indicating that the principal negative finding was not dependent on the selected cutoff.

APACHE II score was associated with both ICU length of stay in the adjusted median model and the exploratory short-term composite outcome, consistent with the relationship between systemic disease severity and adverse clinical trajectories in SAP. Later initiation of enteral nutrition was also associated with higher odds of a poor outcome in exploratory model 2. Because enteral nutrition initiation occurred after the index CT, this variable was evaluated as a subsequent clinical-process factor rather than a baseline confounder. Eckerwall et al[23] demonstrated that early nasogastric enteral nutrition was feasible in predicted SAP, and a meta-analysis by Liu et al[24] found associations between earlier enteral nutrition and improved clinical outcomes. Nevertheless, the present observational association does not establish that earlier feeding directly causes better outcomes. Greater disease severity, hemodynamic instability, impaired gastrointestinal function, and other clinical barriers may delay feeding and are themselves associated with prognosis. This interpretation is consistent with the randomized trial by Bakker et al[25], in which routine very early nasoenteric tube feeding did not improve major outcomes compared with an on-demand strategy.

The adjusted analyses were intentionally parsimonious. The primary continuous-outcome model estimated the APACHE II-adjusted median difference in ICU length of stay per 5-cm increase in CT-derived path length, while the exploratory binary models were based on 17 poor composite outcomes. Model 1 and model 2 contained two and three predictor parameters, respectively, corresponding to approximately 8.5 events and 5.7 events per parameter. Firth penalization was used to reduce small-sample bias in the binary models[26]. The CIs for path length remained compatible with effects in either direction, so the regression estimates should be interpreted as exploratory and in terms of their precision rather than statistical significance alone.

Several limitations should be considered. First, the single-center design and sample size limit precision and generalizability, particularly because hyperlipidemic pancreatitis accounted for most cases. Four enrolled patients were transferred before completion of the prespecified assessments, which may also affect cohort representativeness. Second, although independent anatomical assessment strongly supported the relationship between the CT-derived metric and distal tube position, the metric remains a curved in-situ catheter trajectory that can be influenced by individual anatomy, catheter redundancy, and manual centerline tracing. Third, the 10-day composite threshold was institution-specific, although analyses using 7-, 10-, and 14-day thresholds were concordant. Fourth, associations involving enteral nutrition timing remain susceptible to residual confounding, and the observational design does not establish causal effects of either tube depth or feeding timing. Multicenter prospective cohorts using standardized placement and imaging protocols should externally validate the imaging metric, its reproducibility, and its relationship with anatomical position, feeding tolerance, organ-failure trajectories, and clinically important outcomes, with participant flow reported according to observational-study standards[27].

CONCLUSION

The CT-derived NJ tube tip-to-gastric antrum curved path length was not significantly associated with ICU length of stay in unadjusted, APACHE II-adjusted, or post-CT analyses, and no significant association was observed with the exploratory short-term composite outcome. Its strong relationship with independently assessed distal tube-tip anatomy supports its use as an imaging-derived surrogate of tube advancement, but not as a direct anatomical depth measure. These findings do not establish an optimal tube-advancement target. APACHE II score and later enteral nutrition initiation were associated with the exploratory short-term outcome, while the observational design does not support causal interpretation of enteral nutrition timing. External prospective validation is warranted.