Search

Purpose: The necessity of routine drain placement in pancreatic resections is controversial. Some randomized controlled trials have shown that the omission of drainage is safe for some patients, whereas reintervention rates and mortality rates are substantial for others. The present study aimed to assess fistula-associated outcomes in the setting of routine drain placement and drain irrigation on demand., Methods: Between 01/2017 and 12/2022, perioperative and outcome data from patients who underwent consecutive pancreatoduodenectomies (PD, n = 253) or distal pancreatectomies (DP, n = 72) were prospectively collected in the electronic StuDoQ database and analysed. All patients underwent intraoperative drain placement. Drains were removed starting at postoperative day 2 in PD or at day 5 in DP after testing for amylase concentration. In case of high amylase levels or macroscopically suspicious pancreatic fistulas, drain irrigation was started. Nondrained fluid collections underwent percutaneous radiologic or transluminal endoscopic evacuation., Results: Clinically relevant pancreatic fistulas were detected in 53 of 325 patients (POPF grade B 16.3%, grade C 1.2%). 43.3% of those had drain irrigation. Additional interventional or endoscopic drainage was necessary in 14 and 5 patients, respectively (overall 5.8%), and were observed in 4.0% of patients with PD and in 12.5% with DP (p = 0.009). Delayed fistula-associated postpancreatectomy haemorrhage (PPH) was present in 1.2% (4/325) of patients. The fistula- and delayed PPH-associated reoperation rate was 1.5% (5/325). The 30-day and in-hospital mortality rates were both 1.5% (5/325), and the rate of fistula-associated mortality was 0.6% (2/325). The overall 90-day mortality rate was 4.5%., Conclusions: In pancreatectomies, a standardized drainage protocol including on-demand drain irrigation results in very low fistula-associated morbidity and mortality and an infrequent need for interventional or surgical reintervention as compared to previously published drainage studies., (© 2024. The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.)

Objective: To assess the perioperative and long-term outcome following pulmonary resection in patients with metachronous metastasis of pancreatic ductal adenocarcinoma (PDAC)., Background: Most patients with PDAC relapse or develop tumor spread to secondary organs. Currently, it remains unclear how to proceed with pulmonary metastasis in the metachronous setting. In particular, the role of surgery remains controversial., Methods: Data of patients with pulmonary metachronous metastasis after PDAC collected from 2003 to 2015 in databases of two high-volume pancreatic cancer centers were retrospectively analyzed. Clinical and pathological aspects of primary PDAC as well as the perioperative and long-term outcome following pulmonary metastasectomy (PM) was evaluated, respectively. Patients with synchronous liver metastasis or metastasis to other secondary organs were excluded. Univariate survival analysis was performed., Results: We identified 15 patients undergoing pulmonary resection for suspected metastasis after primary pancreatic resection. Operative and histopathologic evaluation revealed resectable pancreatic pulmonary metastasis in 11 patients (73.3%). The median disease-free survival (DFS) and overall survival (OS) after PM diagnosis was 18 months and 26 months, respectively. The median time to metachronous metastasis (TMM) was 17 months [3-64 months]. Perioperative morbidity was low with only one readmission (8.3%). There was no perioperative mortality. Patients who developed pulmonary metastasis later than 17 months after primary surgery showed better OS compared to those who did earlier (32.2 vs. 14.75 months, p = 0.025). In addition, patients with high-grade tumors had worse survival (12.4 vs. 31 months, p = 0.02). Elevated serum CEA levels or CA 19-9 levels were also not associated with shortened OS., Conclusions: This study suggests that pulmonary metastasectomy after PDAC is safe and effective. Patients with extended DFS after primary pancreatic surgery as well as favorable tumor grading seem to particularly benefit from pulmonary surgery., (Copyright © 2019. Published by Elsevier Ltd.)

Background: Recent literature suggests that chyle leak may complicate up to 10% of pancreatic resections. Treatment depends on its severity, which may include chylous ascites. No international consensus definition or grading system of chyle leak currently is available., Methods: The International Study Group on Pancreatic Surgery, an international panel of pancreatic surgeons working in well-known, high-volume centers, reviewed the literature and worked together to establish a consensus on the definition and classification of chyle leak after pancreatic operation., Results: Chyle leak was defined as output of milky-colored fluid from a drain, drain site, or wound on or after postoperative day 3, with a triglyceride content ≥110 mg/dL (≥1.2 mmol/L). Three different grades of severity were defined according to the management needed: grade A, no specific intervention other than oral dietary restrictions; grade B, prolongation of hospital stay, nasoenteral nutrition with dietary restriction, total parenteral nutrition, octreotide, maintenance of surgical drains, or placement of new percutaneous drains; and grade C, need for other more invasive in-hospital treatment, intensive care unit admission, or mortality., Conclusion: This classification and grading system for chyle leak after pancreatic resection allows for comparison of outcomes between series. As with the other the International Study Group on Pancreatic Surgery consensus statements, this classification should facilitate communication and evaluation of different approaches to the prevention and treatment of this complication., (Copyright © 2016 Elsevier Inc. All rights reserved.)

Background: In the recent International Study Group of Pancreatic Surgery (ISGPS) consensus on extended pancreatectomy, several issues on perioperative outcome and long-term survival remained unclear. Robust data on outcomes are sparse. The present study aimed to assess the outcome of extended pancreatectomy for borderline resectable and locally advanced pancreatic cancer., Methods: A consecutive series of patients with primary pancreatic adenocarcinoma undergoing extended pancreatectomies, as defined by the new ISGPS consensus, were compared with patients who had a standard pancreatectomy. Univariable and multivariable analysis was performed to identify risk factors for perioperative mortality and characteristics associated with survival. Long-term outcome was assessed by means of Kaplan-Meier analysis., Results: The 611 patients who had an extended pancreatectomy had significantly greater surgical morbidity than the 1217 patients who underwent a standard resection (42·7 versus 34·2 per cent respectively), and higher 30-day mortality (4·3 versus 1·8 per cent) and in-hospital mortality (7·5 versus 3·6 per cent) rates. Operating time of 300 min or more, extended total pancreatectomy, and ASA fitness grade of III or IV were associated with increased in-hospital mortality in multivariable analysis, whereas resections involving the colon, portal vein or arteries were not. Median survival and 5-year overall survival rate were reduced in patients having extended pancreatectomy compared with those undergoing a standard resection (16·1 versus 23·6 months, and 11·3 versus 20·6 per cent, respectively). Older age, G3/4 tumours, two or more positive lymph nodes, macroscopic positive resection margins, duration of surgery of 420 min or above, and blood loss of 1000 ml or more were independently associated with decreased overall survival., Conclusion: Extended resections are associated with increased perioperative morbidity and mortality, particularly when extended total pancreatectomy is performed. Favourable long-term outcome is achieved in some patients., (© 2016 BJS Society Ltd Published by John Wiley & Sons Ltd.)

Background: Postoperative pancreatic fistula (POPF) causes significant morbidity and mortality after distal pancreatectomy. Patch coverage of the pancreatic stump is often used with the intention to prevent POPF. Despite numerous investigations, the effects of patch coverage remain unclear. The present meta-analysis aims to clarify the effects of patch coverage in distal pancreatectomy on the incidence of POPF., Methods: A systematic search of MEDLINE/PubMed and the Cochrane Database according to the PRISMA Statement was performed. Subsequently a meta-analysis on rates and overall incidence of POPF and length of hospital stay was carried out. By applying the inverse variance weighting method, the combined effect size and 95 % confidence interval were calculated. Heterogeneity was assessed using I 2 statistics., Results: Five randomized controlled trials and six observational clinical studies were included for final analysis. A cumulative incidence of 43 % of POPF grades A-C was identified. Patch coverage in distal pancreatectomy is significantly associated with a decreased rate of POPF grade C (p = 0.006). Patches of autologous vascularized tissue significantly reduce the overall incidence of POPF (p = 0.04) and clinically relevant POPF grade B and C (p = 0.002). Fibrin sealant patches do not influence rates of POPF after distal pancreatectomy. None of the outcomes evaluated showed adverse results for the patch group., Conclusions: Patch coverage after distal pancreatectomy can reduce the rate of POPF. Patch coverage with autologous vascularized tissue but not fibrin sealant patches may be used to reduce clinically relevant POPF and postoperative morbidity in distal pancreatectomy.

Background: Despite all improvements in tumor diagnostics and treatment, pancreatic cancer is still the fourth leading cause of cancer-related death in the Western world. It is mostly diagnosed at a locally advanced or metastasized stage because of the lack of early symptoms. A radical margin-free surgical resection offers the only potential cure for locoregional disease. Over the last decades, several surgical strategies and techniques have evolved to optimize oncologic radical resections and thus to improve long-term outcome of patients., Purpose: The purpose of this review was to describe the various surgical strategies and techniques for locally advanced pancreatic cancer and to evaluate their influence on long-term outcome., Conclusions: Locally advanced pancreatic cancer should not generally be deemed unresectable. Various surgical techniques offer a good chance of margin-free tumor resection, even if surrounding organs or vessels are involved. Because of potentially higher peri- and postoperative morbidity rates, patients should be selected properly and are best treated in specialized high-volume centers.

Objective: To assess the long-term survival and quality of life in total pancreatectomies and to identify risk factors for perioperative morbidity and mortality., Background: Total pancreatectomy may be required in locally advanced or centrally located pancreatic neoplasms to achieve complete tumor clearance, but available data on short- and long-term results are limited., Methods: A total of 434 consecutive total pancreatectomies for primary pancreatic or periampullary tumors were performed between October 2001 and September 2012 at the authors' institution and were prospectively documented and analyzed. Long-term outcome was assessed using Kaplan-Meier and quality of life analysis (EORTC-QLQ-C30 and PAN26). Uni- and multivariate analysis was performed to identify perioperative risk factors and predictors for long-term survival., Results: Extended total pancreatectomies were performed in 54% of cases, with arterial and portal vein resections in 15% and 32%, respectively. Overall 30-day and in-hospital mortality rates were 3.7% and 7.8%, respectively. High blood loss, long operative time, and arterial resections were independently associated with increased perioperative mortality (P ≤ 0.018). In malignant disease, median and 5-year survival were good for standard total pancreatectomies (28.6 months and 24.3%, respectively) and were significantly impaired after vascular resections (P < 0.001). Poor tumor grading, high American Joint Commission on Cancer tumor stage, age more than 70 years, and an R1 resection were independent prognostic parameters. Long-term global quality of life was comparable with a matched healthy control group., Conclusions: Standard total pancreatectomy, if needed, is associated with good long-term outcome in pancreatic cancer. Marked surgical morbidity and impaired survival associated with vascular resections reflect the invasiveness of extended total pancreatectomies and the underlying advanced malignant disease.

Background: The lymph node (Ln) status of patients with resectable pancreatic ductal adenocarcinoma is an important predictor of survival. The survival benefit of extended lymphadenectomy during pancreatectomy is, however, disputed, and there is no true definition of the optimal extent of the lymphadenectomy. The aim of this study was to formulate a definition for standard lymphadenectomy during pancreatectomy., Methods: During a consensus meeting of the International Study Group on Pancreatic Surgery, pancreatic surgeons formulated a consensus statement based on available literature and their experience., Results: The nomenclature of the Japanese Pancreas Society was accepted by all participants. Extended lymphadenectomy during pancreatoduodenectomy with resection of Ln's along the left side of the superior mesenteric artery (SMA) and around the celiac trunk, splenic artery, or left gastric artery showed no survival benefit compared with a standard lymphadenectomy. No level I evidence was available on prognostic impact of positive para-aortic Ln's. Consensus was reached on selectively removing suspected Ln's outside the resection area for frozen section. No consensus was reached on continuing or terminating resection in cases where these nodes were positive., Conclusion: Extended lymphadenectomy cannot be recommended. Standard lymphadenectomy for pancreatoduodenectomy should strive to resect Ln stations no. 5, 6, 8a, 12b1, 12b2, 12c, 13a, 13b, 14a, 14b, 17a, and 17b. For cancers of the body and tail of the pancreas, removal of stations 10, 11, and 18 is standard. Furthermore, lymphadenectomy is important for adequate nodal staging. Both pancreatic resection in relatively fit patients or nonresectional palliative treatment were accepted as acceptable treatment in cases of positive Ln's outside the resection plane. This consensus statement could serve as a guide for surgeons and researchers in future directives and new clinical studies., (Copyright © 2014 Mosby, Inc. All rights reserved.)

Background: Complete macroscopic tumor resection is one of the most relevant predictors of long-term survival in pancreatic ductal adenocarcinoma. Because locally advanced pancreatic tumors can involve adjacent organs, "extended" pancreatectomy that includes the resection of additional organs may be needed to achieve this goal. Our aim was to develop a common consistent terminology to be used in centers reporting results of pancreatic resections for cancer., Methods: An international panel of pancreatic surgeons working in well-known, high-volume centers reviewed the literature on extended pancreatectomies and worked together to establish a consensus on the definition and the role of extended pancreatectomy in pancreatic cancer., Results: Macroscopic (R1) and microscopic (R0) complete tumor resection can be achieved in patients with locally advanced disease by extended pancreatectomy. Operative time, blood loss, need for blood transfusions, duration of stay in the intensive care unit, and hospital morbidity, and possibly also perioperative mortality are increased with extended resections. Long-term survival is similar compared with standard resections but appears to be better compared with bypass surgery or nonsurgical palliative chemotherapy or chemoradiotherapy. It was not possible to identify any clear prognostic criteria based on the specific additional organ resected., Conclusion: Despite increased perioperative morbidity, extended pancreatectomy is warranted in locally advanced disease to achieve long-term survival in pancreatic ductal adenocarcinoma if macroscopic clearance can be achieved. Definitions of extended pancreatectomies for locally advanced disease (and not distant metastatic disease) are established that are crucial for comparison of results of future trials across different practices and countries, in particular for those using neoadjuvant therapy., (Copyright © 2014 Mosby, Inc. All rights reserved.)

Background: This position statement was developed to expedite a consensus on definition and treatment for borderline resectable pancreatic ductal adenocarcinoma (BRPC) that would have worldwide acceptability., Methods: An international panel of pancreatic surgeons from well-established, high-volume centers collaborated on a literature review and development of consensus on issues related to borderline resectable pancreatic cancer., Results: The International Study Group of Pancreatic Surgery (ISGPS) supports the National Comprehensive Cancer Network criteria for the definition of BRPC. Current evidence supports operative exploration and resection in the case of involvement of the mesentericoportal venous axis; in addition, a new classification of extrahepatic mesentericoportal venous resections is proposed by the ISGPS. Suspicion of arterial involvement should lead to exploration to confirm the imaging-based findings. Formal arterial resections are not recommended; however, in exceptional circumstances, individual therapeutic approaches may be evaluated under experimental protocols. The ISGPS endorses the recommendations for specimen examination and the definition of an R1 resection (tumor within 1 mm from the margin) used by the British Royal College of Pathologists. Standard preoperative diagnostics for BRPC may include: (1) serum levels of CA19-9, because CA19-9 levels predict survival in large retrospective series; and also (2) the modified Glasgow Prognostic Score and the neutrophil/lymphocyte ratio because of the prognostic relevance of the systemic inflammatory response. Various regimens of neoadjuvant therapy are recommended only in the setting of prospective trials at high-volume centers., Conclusion: Current evidence justifies portomesenteric venous resection in patients with BRPC. Basic definitions were identified, that are currently lacking but that are needed to obtain further evidence and improvement for this important patient subgroup. A consensus for each topic is given., (Copyright © 2014 Mosby, Inc. All rights reserved.)

Background: Pancreatic fistulas after distal pancreatectomy occur in up to 60% of patients with distal pancreatectomy. Several techniques for closure of the pancreatic stump have been advocated, but the best management of stump closure remains controversial. Our aim was to evaluate the clinical benefits of coverage of the pancreatic resection margin by autologous tissue., Methods: One hundred seventeen consecutive patients underwent distal pancreatectomy at the university hospital in Heidelberg between May 2009 and September 2010. A coverage procedure was performed in 73 of these patients. All patients were recorded prospectively, and the clinical course was evaluated focusing on the occurrence of pancreatic fistula as defined by the International Study Group on Pancreatic Fistula. A treatment cost analysis was performed., Results: The rate of clinically relevant pancreatic fistulas (types B and C) was decreased in patients with coverage compared to the standard controls (type B, 7% vs 9%; type C, 7% vs 25%; P < .002). Patients with a coverage procedure had a shorter duration of stay in the hospital (P < .02), and treatment costs were lower (P < .001) compared to patients without coverage., Conclusion: Coverage of the pancreatic remnant after distal pancreatectomy decreases the rate of clinically relevant pancreatic fistulas, duration of stay, and treatment costs. A randomized trial is needed to verify these results., (Copyright © 2012 Mosby, Inc. All rights reserved.)

Background: The majority of pancreatic cancers are diagnosed at an advanced stage. As surgical resection remains the only hope for cure, more aggressive surgical approaches have been advocated to increase resection rates. Institutions have begun to release data on their experience with pancreatectomy and simultaneous arterial resection (AR), which has traditionally been considered a general contraindication to resection. The aim of the present meta-analysis was to evaluate the perioperative and long-term outcomes of patients with AR during pancreatectomy for pancreatic cancer., Methods: The Medline, Embase, and Cochrane Library and J-East databases were systematically searched to identify studies reporting outcome of patients who underwent pancreatectomy with AR for pancreatic cancer. Studies that reported perioperative and/or long-term results after pancreatectomy with AR were eligible for inclusion. Meta-analyses included comparative studies providing data on patients with and without AR and were performed using a random effects model., Results: The literature search identified 26 studies including 366 and 2243 patients who underwent pancreatectomy with and without AR. All studies were retrospective cohort studies and the methodological quality was moderate to low. Meta-analyses revealed AR to be associated with a significantly increased risk for perioperative mortality [Odds ratio (OR) = 5.04; 95% confidence interval (CI), 2.69-9.45; P < 0.0001; I² = 24%], poor survival at 1 year (OR = 0.49; 95% CI, 0.31-0.78; P = 0.002; I² = 35%) and 3 years (OR = 0.39; 95% CI, 0.17-0.86; P = 0.02; I² = 49%) compared with patients without AR. The increased perioperative mortality (OR = 8.87; 95% CI, 3.40-23.13; P < 0.0001; I² = 5%) and lower survival rate at 1 year (OR = 0.50; 95% CI, 0.31-0.82; P = 0.006; I² = 40%) was confirmed in the comparison to patients undergoing venous resection. Despite substantial perioperative mortality, pancreatectomy with AR was associated with more favorable survival compared with patients who did not undergo resection for locally advanced disease., Conclusions: AR in patients undergoing pancreatectomy for pancreatic cancer is associated with a poor short and long-term outcome. Pancreatectomy with AR may, however, be justified in highly selected patients owing to the potential survival benefit compared with patients without resection. These patients should be treated within the bounds of clinical trials to assess outcomes after AR in the era of modern pancreatic surgery and multimodal therapy.

Purpose: Pancreatic surgery is a technically challenging intervention with high demands for preoperative diagnostics and perioperative management. A perioperative mortality rate below 5% is achieved in high-volume centers due to the high level of standardization in surgical procedures and perioperative care. Besides standard resections, certain indications may require individualized surgical concepts such as tumor enucleations. The aim of the study was to evaluate the indications, technique, and outcome of this limited local approach compared to major resections., Materials and Methods: Data from patients undergoing pancreatic surgery were prospectively recorded. All patients with tumor enucleations were compared with classical resections (pancreaticoduodenectomy or left resection) in a matched-pair analysis (1:2). Tumor type, localization, operative parameters, complications, and outcome were evaluated., Results: Fifty-three patients underwent pancreatic tumor enucleation between October 2001 and December 2009. Indications included cystic lesions, IPMNs, and neuroendocrine pancreatic tumors. Enucleations were associated with shorter operation time, less blood loss as well as shorter ICU and hospital stay compared to pancreaticoduodenectomy and left resections. The overall surgical morbidity of enucleations was 28.3% without major complications. Leading clinical problems were ISGPF type A fistulas (20.8%) requiring prolonged primary drainage. No surgical revisions were necessary, and no deaths occurred., Conclusions: Pancreatic tumor enucleations can be carried out with good results and no mortality. Decisions regarding enucleations are highly individual compared to standard resections, underlining the importance of treatment in experienced high-volume institutions. Enucleations should be carried out whenever possible and oncologically feasible to prevent the typical complications of major pancreatic resection.

Background: Pancreatic fistula is an unresolved problem after distal pancreatectomy. The current study investigated the safety of LigaSure for distal pancreatic resection in a porcine model., Methods: A distal pancreatectomy was performed in 22 pigs. Animals were randomly assigned to undergo conventional scalpel transection with hand-sewn closure of the pancreatic remnant or pancreatic transection and sealing by LigaSure. Closed-suction drainage was collected daily. Animals were sacrificed on postoperative day (POD) 7 and the pancreatic remnant was sampled for histology., Results: Two grade A postoperative temporary pancreatic fistulas (ISGPF definition) developed after hand suturing but none after LigaSure sealing. Amylase and lipase levels in drainage fluid were higher in the suture group during the first postoperative days but showed no differences after the fourth day. All but two animals that had to be euthanized because of small bowel invagination survived until POD 7. No significant differences were found in macroscopic changes between groups at reexploration. Histology demonstrated focal, chronic granulating inflammation with minor necrosis in all animals., Conclusions: The LigaSure sealing device provides a safe alternative to conventional hand-sewn closure of the pancreatic stump in distal pancreatectomy. This is the first study that investigated this effective and highly applicable technique for pancreatic transection.

Objective: To evaluate the safety and outcome of multivisceral pancreatic resections for primary pancreatic malignancies., Background: Curative resection is the only potential cure for patients with pancreatic cancer, but some patients present with advanced tumors that are not resectable by a standard pancreatic resection. Data on risk and survival analysis of extended pancreatic resections is limited., Methods: One hundred one patients who had a multivisceral pancreatic resection between 10/2001 and 12/2007 were identified from a prospective database, and perioperative and long-term results were compared with those of 202 matched patients with a standard pancreatic resection. Uni- and multivariate regression analysis were performed to identify parameters that are associated with perioperative morbidity. Long-term survival was evaluated., Results: Colon, stomach, adrenal gland, liver, hepatic or celiac artery, kidney, or small intestine were resected in 37.6%, 33.7%, 27.7%, 18.8%, 16.8%, 11.9%, and 6.9% of the 101 patients with multivisceral resection, respectively. Additional portal vein resection was performed in 20.8% of patients. Overall and surgical morbidity but not mortality was significantly increased compared with standard pancreatic resections (55.5% vs. 42.8%, 37.6 vs. 25.3%, and 3.0% vs. 1.5%, respectively). Uni- and multivariate analysis identified a long operative time and the extended multivisceral resection of 2 or more additional organs as independent risk factors for intraabdominal complications or need for relaparotomy. Median survival was comparable to that of standard pancreatic resections., Conclusions: Multivisceral resections can be performed with increased morbidity but comparable mortality and long-term prognosis as compared with standard pancreatic resections at high volume centers. Increased morbidity is related to extended multivisceral resections with a long operative time.

Management of acute necrotizing pancreatitis has changed significantly over the past years. Early management is non-surgically and solely supportive. Today, more patients survive the early phase of severe pancreatitis due to improvements of intensive-care-medicine. Pancreatic infection is the major risk factor with regard to morbidity and mortality in the late phase of severe acute pancreatitis. Whereas early surgery and surgery for sterile necrosis can only be recommended in selected cases, pancreatic infection is a well accepted indication for surgical treatment. Surgery should ideally be postponed until four weeks after the onset of symptoms as necrosis is well demarcated at that time. Four surgical techniques can be performed with comparable results regarding mortality: necrosectomy combined with (1) open packing, (2) planned staged relaparotomies with repeated lavage, (3) closed continuous lavage of the retroperitoneum, and (4) closed packing. However, closed continuous lavage of the retroperitoneum, and closed packing seem to be associated with a lower morbidity compared to the other two approaches. Advances in radiologic imaging, new developments of interventional radiology and other minimal access interventions have revolutionized the management of many surgical conditions over the past decades. However, minimal invasive surgery and interventional therapy for infected necrosis should be limited to specific indications in patients who are critically ill and otherwise unfit for conventional surgery. Open surgical debridement is the "gold standard" for treatment of infected pancreatic and peripancreatic necrosis.

The indications for surgery in acute pancreatitis have changed significantly in the past two decades. Medical charts of patients with acute pancreatitis treated at our institution were analyzed to assess the effects of changes in surgical treatment on patient outcomes. A total of 136 patients with radiologically defined severe pancreatitis were primarily treated or referred to our institution between 1980 and 1997. Severity of the disease (Ranson score), indications for surgical intervention, timing of surgery, and mortality rates were compared during three study periods: 1980 to 1985 (period I), 1986 to 1990 (period II), and 1991 to 1997 (period III). In period I patients underwent exploratory laparotomy if their clinical status did not improve markedly within 72 hours of admission to the hospital, whereas during period II surgery was reserved for patients who had secondary organ failure together with pancreatic necrosis seen on CT scan. During period III the aim was to operate as late as possible in the presence of pancreatic necrosis or when infected necrosis was suspected. The policy of limiting the indications for surgery resulted in a decrease in surgically treated patients from 68% to 33% (P < 0.001). Likewise, surgical intervention was performed later. In period I, 73% of operations were performed within 72 hours of admission, compared to 32% in period III (P = 0.008). The mortality rate for patients who underwent early surgery (within 72 hours) was higher than for those who underwent late surgical exploration of the abdomen (P = 0.02). Overall, the mortality rate for patients with severe pancreatitis was reduced from 39% to 12% (P = 0.003). Mortality among patients treated nonoperatively did not change significantly. The present study supports the policy of delayed surgery in severe acute pancreatitis. Early surgical intervention often results in unnecessary procedures with an increase in the number of deaths. Whenever possible, prolonged observation allows selection of patients who are likely to benefit from delayed surgery or nonoperative treatment.