Early versus delayed necrosectomy in pancreatic necrosis: A population-based cohort study on readmission, healthcare utilization, and in-hospital mortality

Abstract

BACKGROUND

Acute necrotizing pancreatitis is a severe and life-threatening condition. It poses a considerable challenge for clinicians due to its complex nature and the high risk of complications. Several minimally invasive and open necrosectomy procedures have been developed. Despite advancements in treatment modalities, the optimal timing to perform necrosectomy lacks consensus.

AIM

To evaluate the impact of necrosectomy timing on patients with pancreatic necrosis in the United States.

METHODS

A national retrospective cohort study was conducted using the 2016-2019 Nationwide Readmissions Database. Patients with non-elective admissions for pancreatic necrosis were identified. The participants were divided into two groups based on the necrosectomy timing: The early group received intervention within 48 hours, whereas the delayed group underwent the procedure after 48 hours. The various intervention techniques included endoscopic, percutaneous, or surgical necrosectomy. The major outcomes of interest were 30-day readmission rates, healthcare utilization, and inpatient mortality.

RESULTS

A total of 1309 patients with pancreatic necrosis were included. After propensity score matching, 349 cases treated with early necrosectomy were matched to 375 controls who received delayed intervention. The early cohort had a 30-day readmission rate of 8.6% compared to 4.8% in the delayed cohort (P = 0.040). Early necrosectomy had lower rates of mechanical ventilation (2.9% vs 10.9%, P < 0.001), septic shock (8% vs 19.5%, P < 0.001), and in-hospital mortality (1.1% vs 4.3%, P = 0.01). Patients in the early intervention group incurred lower healthcare costs, with median total charges of $52202 compared to $147418 in the delayed group. Participants in the early cohort also had a relatively shorter median length of stay (6 vs 16 days, P < 0.001). The timing of necrosectomy did not significantly influence the risk of 30-day readmission, with a hazard ratio of 0.56 (95% confidence interval: 0.31-1.02, P = 0.06).

CONCLUSION

Our findings show that early necrosectomy is associated with better clinical outcomes and lower healthcare costs. Delayed intervention does not significantly alter the risk of 30-day readmission.

Key Words: Acute necrotizing pancreatitis; Pancreatic necrosis; Early necrosectomy; Delayed necrosectomy; Readmission, Healthcare costs; Mortality

Core Tip: Clinical evidence regarding the impact of necrosectomy timing on patient outcomes and healthcare costs remains limited. Utilizing propensity-matched cohorts, this nationwide study evaluates the clinical and economic implications of early versus delayed necrosectomy in patients with pancreatic necrosis. Our findings show that early intervention within 48 hours is associated with lower rates of mechanical ventilation, septic shock, and in-hospital mortality. Early necrosectomy also results in substantial cost savings and shorter hospital stays. Intriguingly, the timing of the procedure does not significantly influence the 30-day readmission hazard ratio. These results contribute to the ongoing debate on the optimal timing of necrosectomy, offering evidence-based insights that could improve patient outcomes.

INTRODUCTION

Acute pancreatitis is a complex and potentially lethal disease. It is a leading cause of gastrointestinal-related hospitalization burden in the United States[1]. Necrosis of the pancreas or peripancreatic tissue can occur in about 20% of patients with acute pancreatitis[2]. Mortality rates can be as high as 20%-30% for patients with infected pancreatic necrotic collections[3,4]. Patients with necrotizing pancreatitis have historically benefited from surgically removing the necrotic tissue, even in the early phase of the illness[5,6]. It has been suggested that delaying surgery leads to the immune system encapsulating the necrotic pancreatic tissue, making necrosectomy technically easier and possibly decreasing mortality[7,8]. This hypothesis was validated by a randomized controlled trial from 1997 that showed a delay in surgical intervention beyond the first 12 days lowered mortality, as opposed to intervention within the first 72 hours of admission[9]. The mortality rate for those who underwent the procedure after 12 days dropped from 56% to 27%[9]. Since then, there has been a paradigm shift from surgical procedures to less-invasive endoscopic or percutaneous treatment approaches[10]. The debate over when to intervene has resurfaced as mortality and morbidity have declined due to the recent multidisciplinary nonsurgical management of necrotizing pancreatitis[11-13]. The development of lumen-apposing or large-bore metal stents for endoscopic ultrasound-guided drainage shortened procedure time and resulted in fewer adverse events[14,15]. The available data shows that the optimal timing to perform necrosectomy for patients with pancreatic necrosis is still evolving[16-20].

Despite the recognized clinical importance of pancreatic necrosis, there is a lack of large-scale, data-driven epidemiological studies evaluating the effects of the timing of interventions on clinical outcomes and healthcare expenditures. In this study, we aim to compare the impact of early (within 48 hours) versus delayed (after 48 hours) necrosectomy on 30-day readmission rates, healthcare utilization, and in-hospital mortality in patients with pancreatic necrosis using a large national database from the United States.

MATERIALS AND METHODS

Design and data source

We carried out a retrospective cohort study using the 2016-2019 Nationwide Readmissions Database (NRD) of the Healthcare Cost and Utilization Project[21]. The NRD comprises inpatient admissions and readmissions, accounting for around 60% of all-payer hospitalizations across the United States[21]. Diagnoses and procedures were identified using the International Classification of Diseases, Tenth Revision (ICD-10) codes. The study included patients aged 18 years or older with a non-elective admission for pancreatic necrosis (ICD-10 code K85.x). It was further classified based on patients with necrosectomy within 48 hours (cases) or after 48 hours (controls). The timing of necrosectomy was from admission, and all patients included in the study had a primary diagnosis code for necrotizing pancreatitis. Necrosectomy procedures were endoscopic (ICD-10-PCS 0F9G8), percutaneous (ICD-10-PCS 0F9G3), or surgical (ICD-10-PCS 0F9G0). Exclusions were made for patients admitted in December in order to monitor 30-day readmission rates; there were no necrosectomy or readmissions associated with traumatic injuries. The authors did not use NRD databases before 2016, as the data is coded using ICD-9 codes, which could lead to the misclassification of certain variables. This study report was prepared and revised according to the Strengthening the Reporting of Observational Studies in Epidemiology recommendations[22].

Baseline characteristics

Baseline patient information included demographic variables (age and gender), index admission length of hospital stay and charges, median household income category, primary insurance, discharge outcome, and 30-day readmission status. As in prior studies, the Elixhauser Comorbidity Index was computed to account for multiple comorbidities[23,24].

Outcome measures

The outcomes of interest were early readmission rates (within 30 days of index admission), length of stay, hospital costs, and clinical outcomes, including mechanical ventilation, septic shock, portal venous thrombosis, intensive care unit (ICU) admission, acute kidney injury, new renal replacement therapy during admission, and all-cause inpatient mortality.

Statistical analysis

Little's test was applied to establish if data were missing completely at random (MCAR) with a significance threshold of P < 0.05. The variables with over 2% missing data that failed Little's MCAR evaluation underwent multiple imputations (25 datasets) for sensitivity analysis. Data were analyzed employing descriptive statistics for nonparametric databases. Categorical values were reported as percentages, and continuous variables as medians and interquartile ranges. Pearson's chi-square test was utilized to compare categorical variables and the Mann-Whitney test for continuous variables. The impact of the timing of each procedure on outcome variables was evaluated using multivariable regression analysis. The risk of readmission and mortality was ascertained by applying Cox proportional hazard regression analysis. Kaplan-Meier estimates were used to demonstrate differences in 30-day readmission among procedure timings. All statistical analyses were executed using the Statistical Software for Data Science (StataCorp LLC, College Station, TX, United States), version 16.1.

Ethical considerations

The data were acquired from the NRD, a de-identified, publicly accessible registry. This database protects the privacy of patients, physicians, and hospitals. Therefore, the informed consent was waived as the patient identifiers were removed from the hospitalization data. Institutional review board approval was also not required for this study. According to the Healthcare Cost and Utilization Project Data Use Agreement, any individual table cell counts of ≤ 10 have been masked to ensure privacy and compliance. In such instances, data are designated as < 10.

RESULTS

Patient demographics and comorbidities

In the unmatched cohort, 1309 participants met the selection criteria for the study period. Of these, 69 (5.27%) patients were readmitted within 30 days (Table 1). After propensity score matching, 349 cases (patients who underwent necrosectomy within 48 hours) were matched to 375 controls (patients who received necrosectomy after 48 hours). The median age at the time of hospital admission was 55.0 years in both cohorts. With regard to age distribution, the early necrosectomy cohort had more patients aged ≥ 65 years compared to the delayed intervention cohort (29.5% vs 21.3%). The distribution of comorbidities among unmatched and matched patients who underwent pancreatic necrosectomy has been outlined (Table 2). There was no significant difference in cardiovascular diseases, pulmonary disorders, diabetes mellitus, renal failure, liver disease, coagulopathy, or obesity in the matched cohorts.

Table 1 Clinical characteristics of all patients (index hospitalizations) who underwent pancreatic necrosectomy with a primary diagnosis of pancreatic necrosis, n (%).

Factor	Unmatched patients			Matched patients
	Within 48 hours (cases)	After 48 hours (controls)	P value	Within 48 hours (cases)	After 48 hours (controls)	P value
Total patients	420	889		349	375
Length of stay, median (IQR)	6.0 (3.0, 10.0)	23.0 (12.0, 42.0)	< 0.001	6.0 (4.0, 11.0)	16.0 (9.0, 31.0)	< 0.001
Total hospital charges in USD, median (IQR)	51773.0 (28907.0, 92300.0)	196571.0 (89244.0, 408072.0)	< 0.001	52202.0 (29417.0, 101413.0)	147418.0 (73243.5, 316993.0)	< 0.001
30-day readmission	35 (8.3)	34 (3.8)	< 0.001	30 (8.6)	18 (4.8)	0.040
Age in years at admission, median (IQR)	55.0 (41.0, 66.0)	52.0 (39.0, 64.0)	0.080	55.0 (42.0, 66.0)	55.0 (39.0, 64.0)	0.13
Age groups (years)			0.068			0.061
18-34	51 (12.1)	144 (16.2)		38 (10.9)	57 (15.2)
35-49	113 (26.9)	241 (27.1)		91 (26.1)	100 (26.7)
50-64	134 (31.9)	301 (33.9)		117 (33.5)	138 (36.8)
≥ 65	122 (29.0)	203 (22.8)		103 (29.5)	80 (21.3)
Elixhauser comorbidity index score			< 0.001			0.32
0	32 (7.6)	18 (2.0)		20 (5.7)	18 (4.8)
1	37 (8.8)	46 (5.2)		24 (6.9)	40 (10.7)
2	71 (16.9)	91 (10.2)		65 (18.6)	65 (17.3)
≥ 3	280 (66.7)	734 (82.6)		240 (68.8)	252 (67.2)
Primary payer			0.65			0.37
Medicare	129 (33.9)	260 (30.3)		112 (35.8)	109 (30.4)
Medicaid	73 (19.2)	166 (19.4)		57 (18.2)	70 (19.6)
Private	158 (41.5)	381 (44.5)		127 (40.6)	151 (42.2)
Other	21 (5.5)	50 (5.8)		17 (5.4)	28 (7.8)
Median household income national quartile for patient ZIP Code			0.28			0.53
1st (0-25th)	105 (25.5)	197 (22.3)		88 (25.7)	88 (23.6)
2nd (26th-50th)	114 (27.7)	256 (28.9)		94 (27.4)	107 (28.7)
3rd (51st-75th)	109 (26.5)	271 (30.6)		87 (25.4)	109 (29.2)
4th (76th-100th)	84 (20.4)	161 (18.2)		74 (21.6)	69 (18.5)
Disposition of patient (uniform)			< 0.001			0.12
Routine	264 (62.9)	396 (44.6)		211 (60.5)	214 (57.2)
Transfer to SNF, STH, ICF, and another facility, or AMA	57 (13.6)	182 (20.5)		47 (13.5)	60 (16.0)
HHC	94 (22.4)	259 (29.2)		87 (24.9)	84 (22.5)

STH: short-term hospital; SNF: Skilled nursing facility; ICF: Intermediate care facility; HHC: Home health care; AMA: Against medical advice.

Table 2 Distribution of Elixhauser comorbidities among all patients (index hospitalizations) who underwent pancreatic necrosectomy with a primary diagnosis of pancreatic necrosis, n (%).

Factor	Unmatched patients			Matched patients
	Within 48 hours (cases)	After 48 hours (controls)	P value	Within 48 hours (cases)	After 48 hours (controls)	P value
Total patients	420	889		349	375
Congestive heart failure	35 (8.3)	81 (9.1)	0.64	30 (8.6)	29 (7.7)	0.67
Cardiac arrhythmias	55 (13.1)	287 (32.3)	< 0.001	48 (13.8)	58 (15.5)	0.51
Uncomplicated hypertension	230 (54.8)	458 (51.5)	0.27	194 (55.6)	200 (53.3)	0.54
Chronic pulmonary diseases	81 (19.3)	135 (15.2)	0.062	72 (20.6)	59 (15.7)	0.087
Uncomplicated diabetes	70 (16.7)	119 (13.4)	0.11	58 (16.6)	53 (14.1)	0.35
Complicated diabetes	66 (15.7)	170 (19.1)	0.13	55 (15.8)	64 (17.1)	0.64
Hypothyroidism	19 (4.5)	83 (9.3)	0.002	17 (4.9)	27 (7.2)	0.19
Renal failure	29 (6.9)	66 (7.4)	0.74	24 (6.9)	25 (6.7)	0.91
Liver disease	69 (16.4)	227 (25.5)	< 0.001	54 (15.5)	55 (14.7)	0.76
Coagulopathy	31 (7.4)	136 (15.3)	< 0.001	20 (5.7)	25 (6.7)	0.60
Obesity	60 (14.3)	204 (22.9)	< 0.001	43 (12.3)	43 (11.5)	0.72
Weight loss	118 (28.1)	444 (49.9)	< 0.001	108 (30.9)	124 (33.1)	0.54
Fluid and electrolyte disorder	227 (54.0)	634 (71.3)	<0.001	215 (61.6)	211 (56.3)	0.14
Iron-deficiency anemia	33 (7.9)	67 (7.5)	0.84	29 (8.3)	24 (6.4)	0.32
Alcohol abuse	110 (26.2)	271 (30.5)	0.11	97 (27.8)	108 (28.8)	0.76
Drug abuse	21 (5.0)	78 (8.8)	0.016	15 (4.3)	26 (6.9)	0.13
Depression	72 (17.1)	163 (18.3)	0.60	59 (16.9)	73 (19.5)	0.37
Complicated hypertension	32 (7.6)	86 (9.7)	0.23	26 (7.4)	27 (7.2)	0.90

Inpatient outcomes and causes of readmissions

Cases had a lower rate of mechanical ventilation (2.9% vs 10.9%, P < 0.001), septic shock (8% vs 19.5%, P < 0.001), ICU admission (0.6% vs 99%, P < 0.001), acute kidney injury (15.5% vs 30.4%, P < 0.001), and a lower all-cause inpatient mortality (1.1% vs 4.3%, P = 0.01) compared to controls (Table 3). The Elixhauser Comorbidity Index score of ≥ 3 was 68.8% in the early cohort compared to 67.2% in the delayed cohort. In the matched cohort, patients who underwent pancreatic necrosectomy within 48 hours had a significantly shorter median length of stay (6 vs 16 days, P < 0.001) and lower median total hospital charges ($52202 vs $147418, P < 0.001), compared to those who underwent necrosectomy after 48 hours. There was no increased risk of mortality among matched cohorts, with a hazard ratio (HR) of 0.46 [95% confidence interval (CI): 0.11-1.88, P = 0.28].

Table 3 Clinical outcomes of all patients (index hospitalizations) who underwent pancreatic necrosectomy with a primary diagnosis of pancreatic necrosis, n (%).

Factor	Unmatched patients			Matched patients
	Within 48 hours (cases)	After 48 hours (controls)	P value	Within 48 hours (cases)	After 48 hours (controls)	P value
Total patients	420	889		349	375
Mechanical ventilation	14 (3.3)	155 (17.4)	< 0.001	< 10	41 (10.9)	< 0.001
Septic shock	38 (9.0)	239 (26.9)	< 0.001	28 (8.0)	73 (19.5)	< 0.001
Portal venous thrombosis	35 (8.3)	71 (8.0)	0.83	30 (8.6)	27 (7.2)	0.49
ICU-level admission	< 10	145 (16.3)	< 0.001	< 10	37 (9.9)	< 0.001
Acute kidney injury	64 (15.2)	330 (37.1)	< 0.001	54 (15.5)	114 (30.4)	< 0.001
New RRT during admission	15 (3.6)	43 (4.8)	0.30	13 (3.7)	< 10	0.20
Died during hospitalization	< 10	50 (5.6)	< 0.001	< 10	16 (4.3)	0.010

ICU: Intensive care unit; RRT: Renal replacement therapy. NOTE: According to the Healthcare Cost and Utilization Project Data Use Agreement, any individual table cell counts of ≤ 10 have been masked to ensure privacy and compliance. In such instances, data are designated as < 10.

The 30-day readmission rate was higher for patients who underwent necrosectomy within 48 hours (8.6% vs 4.8%, P = 0.040) compared to controls. The top five causes of readmission are delineated (Figure 1). The most common cause of readmission in patients who underwent pancreatic necrosectomy within 48 hours was acute pancreatitis with uninfected necrosis (12.5%). In those who underwent pancreatic necrosectomy after 48 hours, it was sepsis due to infection with an unspecified organism (7.1%).

Figure 1 Absolute rates of cause-specific 30-day readmission stratified by pancreatic necrosectomy timing on index admission in the matched cohort. A: Before 48 hours; B: After 48 hours.

Comparative analysis of necrosectomy timing

Pancreatic necrosectomy timing did not significantly affect the risk of early readmission [HR 0.56 (95%CI: 0.31-1.02), P = 0.06]. However, pancreatic necrosectomy after 48 hours had a reduced probability of early readmission over necrosectomy within 48 hours with a log rank of 0.07 (Figure 2).

Figure 2 The 30-day readmission risk based on pancreatic necrosectomy timing on index acute pancreatitis admission in the matched cohort (log rank = 0.07).

DISCUSSION

This is the first population-based study to evaluate readmission, healthcare utilization, and in-hospital mortality for early (< 48 hours) versus delayed necrosectomy (> 48 hours) using a national multicenter database. Our findings indicate that early necrosectomy significantly reduces rates of mechanical ventilation, septic shock, and in-hospital mortality. There is a reduced likelihood of readmission after delayed necrosectomy compared to early intervention. However, patients undergoing early intervention have shorter hospital stays and lower inpatient costs.

Hospital readmission constitutes a significant problem in the context of health care policy and reform[25,26]. A number of organizations view readmission rates as a barometer for the quality of healthcare facilities. The body of research on necrotizing pancreatitis-related readmissions has expanded dramatically in recent years. However, there is currently a paucity of clinical evidence comparing the effects of early versus delayed necrosectomy treatments on readmission. In our study, we specifically compared readmission as one of the parameters between early and delayed necrosectomy cohorts. The timing of pancreatic necrosectomy did not significantly affect the risk of early readmission (HR 0.56, P = 0.06). However, necrosectomy after 48 hours showed a reduced probability of 30-day readmission over necrosectomy within 48 hours. One possible reason could be the differences in patient characteristics between the matched cohorts. The early necrosectomy cohort had a relatively higher frequency of patients aged 65 years or older compared to the delayed intervention cohort (29.5% vs 21.3%). A retrospective analysis of 623 patients who underwent pancreatectomy showed that the patient age of ≥ 65 years independently predicted 30-day unplanned readmissions[27]. Similarly, the Elixhauser Comorbidity Index score of ≥ 3 has been designated as an important variable for readmission prediction in acute pancreatitis[28,29]. In our analysis, the Elixhauser Comorbidity Index score of ≥ 3 was also higher in the early compared to the delayed necrosectomy cohort (68.8% vs 67.2%).

The optimal timing to perform necrosectomy for patients with pancreatic necrosis is still evolving[30]. According to the 2020 clinical practice update from the American Gastroenterological Association, the optimal timing for pancreatic debridement should be around four weeks[31]. Early debridement (< 2 weeks after onset) correlates with increased morbidity and mortality[31]. The International Association of Pancreatology/American Pancreatic Association guidelines also suggest that endoscopic treatment for walled-off necrosis be delayed for at least four weeks after the onset of pancreatitis to allow for the encapsulation of necrotic tissue[32]. The delay leads peripancreatic collections to encapsulate, reducing the risk of procedural complications such as bleeding and perforation[33,34]. However, experts recommend early percutaneous drainage for infected or symptomatic necrotic collections[35-39]. Nonetheless, there has been limited investigation into the potential advantages and drawbacks of initiating drainage procedures before the 4-week mark.

The effects of necrosectomy timing on healthcare resource utilization have not been well characterized. Our study revealed that patients in the delayed necrosectomy cohort had greater healthcare utilization and costs. It could be attributed to higher requirements for mechanical ventilation, a greater incidence of septic shock, ICU admission, and acute kidney injury associated with delayed necrosectomy (> 48 hours). Systemic inflammatory response syndrome, commonly seen in severe acute pancreatitis, can lead to multiorgan failure[40,41]. Therefore, delayed intervention may increase the risk of clinical deterioration due to infections and acute inflammatory responses, which can ultimately result in organ failure. It could lead to longer hospitalizations and higher inpatient costs in the delayed intervention group. Contrarily, a recent meta-analysis concluded that patients with early intervention were more likely to have mortality, organ failure, larger fluid collections, and less encapsulation[42]. However, it is possible that the retrospective nature of the studies included in the meta-analysis could have caused a between-group imbalance in patient profiles[42]. Consequently, the increased occurrence of organ failure in the early intervention group might have adversely influenced clinical outcomes[42]. In our cohorts, we could not stratify the individuals according to the specific etiologies of pancreatitis due to inconsistent data availability in the NRD database. Certain etiologies, like post-endoscopic retrograde cholangiopancreatography and autoimmune pancreatitis, may lead to more severe hospital courses. This observation indicates that such etiologies may have an impact on clinical outcomes such as length of stay and mortality. Therefore, future studies should aim to include etiological factors to better understand their influence on the disease course and outcomes.

Our findings show a mortality benefit associated with early compared to delayed necrosectomy (1.1% vs 4.3%, P = 0.01). It could be related to the relatively quick resolution of pancreatic necrosis, which led to a lower rate of complications. An Indian randomized control trial showed that patients with infected necrotizing pancreatitis had a significantly higher and faster resolution of organ failure with proactive percutaneous catheter drainage[43]. A meta-analysis of nine studies based on pooled data from 870 patients showed that there was no significant difference in mortality and complications between early and delayed minimally invasive intervention groups, denoting early intervention as safe for infected pancreatic necrosis[44]. A meta-analysis of four studies analyzing the data of 427 patients who underwent endoscopic treatment also showed no significant difference in rates of mortality and adverse events in early versus late groups[45]. Similarly, a meta-analysis of seven studies also showed no significant mortality or new-onset organ failure difference between early and delayed groups[46]. Therefore, early minimally invasive procedures do not have a negative impact on patient outcomes but may possibly lead to longer hospital and ICU stays[46]. A meta-analysis of six studies based on 630 patients revealed no statistically significant differences in overall adverse events or mortality, but early drainage may prolong the length of stay compared to standard endoscopic ultrasound-guided drainage[47]. Our cohort study revealed a mortality benefit, a shorter hospital stay, and a reduced need for ICU admission in the early group compared to the delayed group.

Sepsis was the most common cause of 30-day readmission in patients who underwent delayed necrosectomy. It indicates that more patients in our delayed necrosectomy cohort might have developed infected necrosis. It could also expose these patients to a number of inpatient complications. A retrospective study revealed that postponing necrosectomy until 30 days after index hospitalization had some mortality benefit, but it was associated with longer use of antibiotics and an increased occurrence of infections with Candida species and drug-resistant bacteria[48]. In patients with infected necrosis, Gram-negative infections are frequent, but Gram-positive enterococci and fungi have also been reported[49]. Nonetheless, several randomized controlled trials have revealed that empiric broad-spectrum antibiotics do not influence the likelihood of developing infected necrosis, multiorgan complications, mortality, or surgery in patients with severe acute necrotizing pancreatitis[50-52]. A randomized controlled trial from the United Kingdom advocated for procalcitonin-directed care to lower the administration of antibiotics in patients with acute pancreatitis[53]. Similarly, a retrospective study from China underscored the value of early prognostication of acute pancreatitis based on etiology and disease severity for antibiotic use[54]. Currently, there is insufficient evidence for routine prophylactic use of antifungals in these patients. The existing clinical data also show marked heterogeneity in terms of non-interventional supportive care[55]. Therefore, supportive treatment during the periprocedural period for pancreatic necrosis merits further research[55].

The participants in our study received endoscopic, percutaneous, or surgical necrosectomy procedures. A recent nationwide study of 4605 patients with infected necrosis showed that open debridement had a significantly higher risk of mortality, respiratory failure, mechanical ventilation, and acute kidney injury compared to minimally invasive procedures[56]. In our delayed cohort, patients experienced similar complications and had a higher mortality risk compared to the early cohort. We could not stratify patients based on the respective necrosectomy procedures they received due to NRD data limitations. However, it is plausible that a significant number of patients in our delayed cohort received surgical necrosectomy. A retrospective cohort study from Australia showed that delayed surgical intervention alone may have higher odds of additional complications such as pancreatic fistulae and new-onset diabetes mellitus compared to endoscopic and percutaneous approaches[57]. Furthermore, a retrospective study from the United States showed that endoscopic necrosectomy resulted in reduced morality risk, complications, hospital stay, and inpatient charges compared to percutaneous and surgical procedures[58]. A network meta-analysis of seven studies using pooled data from 400 patients designated the step-up approach with endoscopic debridement as the first choice for infected pancreatic necrosis[59]. Moreover, it was argued that surgical debridement (early and delayed) should be avoided[59]. A meta-analysis of 10 studies based on 570 patients revealed the delayed surgical step-up approach as the optimal choice for acute necrotizing pancreatitis and advocated avoiding drainage alone[60].

Current clinical evidence indicates a regimental shift towards a step-up approach in managing patients with pancreatic necrosis[61-63]. The step-up approach described in a trial from the Netherlands showed better outcomes regarding major complications and mortality than primary open necrosectomy[64]. Similarly, a trial conducted in the United States found no notable discrepancy in mortality rates but a significantly higher rate of complications in the surgical group (pancreatocutaneous fistula) compared to the endoscopic step-up approach[65]. A multicenter trial showed that upfront necrosectomy at the index intervention rather than as a step-up procedure may safely reduce the number of reinterventions in stable patients with fully encapsulated collections and infected necrotizing pancreatitis[66]. However, a cohort study from Germany found that an endoscopic step-up approach reduced peri-interventional morbidity and length of hospital stay[67]. Therefore, further research is warranted to evaluate the best therapeutic strategy utilizing novel technological advancements for patients with pancreatic necrosis[68,69].

This cohort study has several strengths. It has a large sample population, sourced from one of the largest all-payer data sets in the United States. This specific characteristic distinguishes our research from previous studies by providing a reasonable degree of generalizability about the outcomes of early versus delayed necrosectomy. It broadens the applicability of the results in clinical practice compared to single-center experiences with more restricted information on the subject. Using a robust analytical approach, we found that delayed necrosectomy (> 48 hours) is associated with significantly prolonged hospitalization and increased healthcare charges. It offers pertinent real-world insights and clinical evidence to gastroenterologists and gastrointestinal surgeons regarding necrosectomy timing. Therefore, it may aid in informed therapeutic decision-making and prognostic advice. Our results may serve to enable pancreatologists to conduct future studies expanding data on the risks and complications associated with early intervention compared to delayed strategies. It may help to refine patient selection criteria for early necrosectomy, potentially reducing postprocedure adverse clinical outcomes[70]. Further research is warranted to investigate the long-term impact of early versus delayed necrosectomy.

Limitations

There are certain limitations to our study. The retrospective cohort nature of our design renders it susceptible to the biases commonly associated with such studies. Furthermore, the NRD database lacks specific information on factors such as the severity of acute pancreatitis, the course of hospitalization, treatment modalities, and time intervals related to complications. The database specifies the interval between hospital admission and the necrosectomy procedure. However, it does not include patient data about the interval between hospital admission and the diagnosis of pancreatic necrosis. Our analysis did not account for the specific etiologies of pancreatitis, representing a potential limitation of our findings. We also could not stratify clinical outcomes by specific necrosectomy techniques or individual patients undergoing multiple procedures. Due to the lack of granular data in the database, this study could not specifically track the number or percentage of readmissions due to postprocedure complications. Finally, it is crucial to recognize that human coding errors may occur in the NRD because it is an administrative database reliant on ICD codes for data storage. Despite these constraints, this is the first study to compare patient outcomes between early and late necrosectomy procedures using a nationwide database. It will improve the paucity of data regarding the timing of intervention for pancreatic necrosis.

CONCLUSION

Our study showed that early necrosectomy was associated with improved clinical outcomes, including decreased risks of septic shock, mechanical ventilation, ICU admission, acute kidney injury, and lower all-cause inpatient mortality. Patients in the early cohort had relatively shorter hospital stays and less expensive medical care. The 30-day readmission rate was higher for patients who underwent early necrosectomy within 48 hours compared to those who received delayed intervention after 48 hours. As the management of necrotizing pancreatitis is continually evolving, our analysis shows that early necrosectomy may have certain clinical benefits over delayed intervention. Therefore, further research is required to stratify the long-term impact of various early interventions on patients with pancreatic necrosis.