World J Surg (2014) 38:1461–1467 DOI 10.1007/s00268-013-2439-1
Limitations of ACS-NSQIP in Reporting Complications for Patients Undergoing Pancreatectomy: Underscoring the Need for a Pancreas-Specific Module Irene Epelboym • Irmina Gawlas • James A. Lee • Beth Schrope • John A. Chabot • John D. Allendorf
Published online: 10 January 2014 Ó Socie´te´ Internationale de Chirurgie 2014
Abstract Background Large centralized databases are used with increasing frequency for reporting hospital-specific and nationwide trends and outcomes after various surgical procedures in order to improve quality of surgical care. American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) is a risk-adjusted, case-weighted complication tracking initiative that reports 30-day outcomes from more than 400 academic and community institutions in the US. However, the accuracy of event reporting specific to pancreatic surgery has never been examined in depth. Methods We retrospectively reviewed medical records of patients, the information on whose postoperative course was originally reported through ACS-NSQIP between 2006 and 2010. Preoperative characteristics, operative data, and postoperative events were recorded after review of electronic medical records including physician and nursing notes, operative room records and anesthesiologist reports. Fidelity of reported clinical events was assessed. Accuracy,
This work was described in a quick-shot presentation at the 54th annual meeting of the Society for Surgery of the Alimentary Tract. It was also given as a long oral presentation at the 47th annual meeting of the Pancreas Club, Orlando, FL, May 2013. I. Epelboym I. Gawlas J. A. Lee B. Schrope J. A. Chabot J. D. Allendorf Department of Surgery, College of Physicians and Surgeons, Columbia University, New York, NY, USA Present Address: J. D. Allendorf (&) 61 Fort Washington Avenue, Suite 820, New York, NY 10032-3784, USA e-mail:
[email protected]
sensitivity, and specificity were calculated for each variable of interest. Results Two hundred and forty-nine pancreatectomies were reviewed, including 145 (58.2 %) Whipple procedures, 19 (7.6 %) total pancreatectomies, 65 (26.1 %) distal pancreatectomies, and 15 (6.0 %) central or partial resections. Median age was 65.7, males comprised 41.5 % of the group, and 74.3 % of patients were Caucasian. The overall rate of complications reported by NSQIP was 44.0 %, compared with 45.0 % in our review, however discordance was observed in 27.3 % of the time, including 34 cases of reporting a complication where there was not one, and 34 cases of missed complication. The most frequently reported event was postoperative bleeding requiring transfusion, however this was also the event most commonly misclassified. Additionally, three procedures unrelated to the index operation were recorded as reoperation events. While a pancreasspecific module does not yet exist, ACS-NSQIP reports a 7.6 % rate of organ-space surgical site infections; when compared with our institutional rate of Grades B and C postoperative fistula (10.4 %), we observed discordance 4.4 % of the time. Delayed gastric emptying, a common post-pancreatectomy morbidity, was not captured at all. Additionally, there were significant inaccuracies in reporting urinary tract infections, postoperative pneumonia, wound complications, and postoperative sepsis, with discordance rates of 4.4, 3.2, 3.6, and 6.8 %, respectively. Conclusions ACS-NSQIP data are an important and valuable tool for evaluating quality of surgical care, however pancreatectomy-specific postoperative events are often misclassified, underscoring the need for a hepatopancreatobiliary-specific module to better capture key outcomes in this complex and unique patient population.
123
1462
Introduction In 1994, the Department of Veteran’s Affairs began collecting perioperative outcomes after surgical procedures in a systematic way through what eventually became known as the National Surgical Quality Improvement Program, NSQIP [1]. In 2004, this initiative was extended to academic, teaching, and community hospitals nation-wide, and today it serves as a risk-adjusted, case-weighted complication tracking database that reports 30-day outcomes from more than 400 academic and community institutions in the US [2, 3]. Several studies have used data collected by this program to demonstrate reduced morbidity, mortality, and cost of general surgical procedures at university and community hospitals after implementation of regular reporting at participating institutions [4–6]. Subsequently, similar analyses were performed to evaluate outcomes and improve quality in patients undergoing cardiac, vascular, bariatric and colorectal surgery [5, 7–12]. In recent years, this aggregate data have also been used to formally assess outcomes after pancreatic surgery, specifically related to effects and consequences of blood loss [13], complications related to classic compared with pylorus-preserving pancreaticoduodenectomies [14], and overall risk of morbidity and mortality following pancreatic surgery [15]. Even though the primary aim of American College of Surgeons (ACS)-NSQIP was to assess the delivery of care in the postoperative period, events tracked by this initiative have also been used as a quality and performance rubric. However, although ACS-NSQIP has already demonstrated a great potential to improve patient outcomes, it is important to assess its limitations, especially as the data it contains are increasingly being used to inform patient care and health policy. ACS-NSQIP includes a process to validate interreviewer reliability and consistency [16], but no group to date has formally assessed the accuracy of the collected variables. The aim of our study was therefore to cross validate outcomes specific to patients undergoing pancreatic surgery using data collected by ACS-NSQIP compared with our institutional database, in an effort to evaluate the accuracy and depth of ACS-NSQIP in patients undergoing pancreatic surgery.
Methods After obtaining IRB approval and in compliance with the Health Insurance Portability and Accountability Act, we reviewed in detail the medical records of patients whose postoperative course was reported through ACS-NSQIP by our institution between 2006 and 2010.
123
World J Surg (2014) 38:1461–1467
We queried the electronic medical record to record preoperative variables such as age, gender, race, presenting symptoms of pancreatic disease, family history of pancreatic disease and pancreatic cancer, use of tobacco or alcohol, and presence of major comorbidity (chronic kidney disease, chronic obstructive pulmonary disease, coronary artery disease, or diabetes mellitus). Intraoperative variables were obtained from nurse, anesthesiologist, and surgeon reports. Operating room (OR) time was defined as time between patient entry into and exit from the OR. Pathologic diagnosis was determined from final pathology reports. Perioperative complications were gathered from daily progress notes and discharge summaries, and graded using the system proposed by DeOliveira et al. [17]. Overall morbidity was defined as any complication, and major morbidity was defined as complication of grade III or greater (requiring surgical, radiologic, or endoscopic intervention). Pancreatic fistula was assessed and graded according to International Study Group on Pancreatic Fistula [18]. Length of stay was calculated from date of operation to date of discharge. Readmission rate was defined as readmission within 30 days of hospital discharge, but 90-day readmission was tracked as well. Perioperative mortality was defined as death within 30 days or within the same hospital admission as index operation. Relevant variables contained in the ACS-NSQIP database included demographic data, preoperative comorbidities, operative time, and postoperative complications. Among these were intraoperative death or cardiac arrest requiring CPR, myocardial infarction, postoperative requirement for ventilator support beyond 48 h or an unplanned intubation, deep vein thrombosis, pulmonary embolism, wound complications including superficial and deep incisional surgical site infections as well as wound disruptions, postoperative organ space infection, pneumonia, urinary tract infection, progressive renal insufficiency, acute renal failure, coma, bleeding event requiring transfusion, sepsis, and septic shock. Length of stay was also recorded. We compared overall 30-day complication rates reported through ACS-NSQIP with events tracked in our database, and examined each in detail. Chart review was considered the gold standard. Patient-level data were entered by members of the Pancreas Center Data Core, and each patient’s record was then additionally independently reviewed in detail by two clinicians. Data submitted to ACS-NSQIP were collected and reported by a designated nurse practitioner affiliated with the program who received special training and certification by the ACS. For both our local database as well as the NSQIP database, primary source was the electronic medical record. We stratified events by concordant (those present in both databases), events reported in error, and events not
World J Surg (2014) 38:1461–1467
1463
Table 1 Demographic characteristics for 249 patients
Table 2 Operative characteristics in detail for 249 patients
Characteristics
Data
Characteristics
Age (median, IQR)
65.7 [48.5, 82.9]
Operation
Male sex
103 (41.5 %)
Pancreaticoduodenectomy
White race
185 (74.3 %)
Distal pancreatectomy
Co-morbidities
Number
145 (58.2 %) 65 (26.1 %)
Total pancreatectomy
19 (7.6 %)
Connective tissue disease
14 (5.7 %)
Central/partial pancreatectomy
15 (6.0 %)
Peptic ulcer disease
66 (27.1 %)
Othera
Diabetes (with end-organ damage)
21 (8.6 %)
Hypothyroidism
26 (10.7 %)
Venous repair
Arrhythmia
23 (9.4 %)
Venous resection, primary repair
15 (6.0 %)
History of preoperative MI
9 (3.7 %)
Venous resection, interposition graft
23 (9.2 %)
Coronary artery disease Peripheral vascular disease
24 (9.4 %) 4 (1.64 %)
Arterial resection EBL (median, IQR)
CTwo co-morbidities
Vascular component
5 (2.0 %) 44 (17.7 %) 3 (1.2 %)
3 (1.2 %) 700 (850)
52 (20.9 %)
Final pathology
Preoperative biliary drainagea
63 (25.3 %)
Benign (n, %)
129 (51.8 %)
Neoadjuvant therapy
38 (15.3 %)
Cystic disease
70 (28.1 %)
Unless stated otherwise, the results are given as the number and percent
PDAC
98 (39.4 %)
Periampullary adenoma
3 (1.2 %)
MI myocardial infarction
Periampullary carcinoma
16 (6.4 %)
a
Pancreatitis
13 (5.2 %)
PNET
39 (15.7 %)
Defined as internal or external stenting
captured. Events described as organ space infection in the ACS-NSQIP database were presumed to reflect Grades B or C postoperative fistula. Superficial and deep surgical site infections as well as wound disruptions were grouped into a single variable. Sepsis and septic shock were analyzed together as cases of infection. Fidelity of reported clinical events was assessed. Accuracy, sensitivity, and specificity were calculated for each individual variable of interest. Statistical analyses were conducted using STATA software (version 12, College Station, TX).
Results Two hundred and forty-nine pancreatectomies were reviewed in detail, among them 145 (58.2 %) Whipples, 19 (7.6 %) total pancreatectomies, 65 (26.1 %) distal pancreatectomies, and 15 (6.0 %) central or partial pancreatic resections. Median age was 65.7, males comprised 41.5 % of the group, and 74.3 % of patients were Caucasian. A fifth of all patients (52, 20.9 %) had greater than two preoperative comorbidities, and the most commonly seen condition was peptic ulcer disease. Demographic characteristics of this cohort are presented in detail in Table 1. One hundred and thirteen operations were performed for benign disease, including 69 (61.1 %) cystic lesions, 25 (22.1 %) pancreatic neuroendocrine tumors (PNETs), 13 (11.5 %) resections for pancreatitis, and 3 (2.7 %) for
Metastatic lesion
6 (2.4 %)
Otherb
4 (1.6 %)
EBL estimated blood loss, PDAC pancreatic ductal adenocarcinoma, PNET pancreatic neuroendocrine tumor a
Lateral pancreaticojejunostomy, enucleation, dorsal pancreatectomy, and pancreatic debridement
b
Solid pseudopapillary tumor, benign pancreatic tissue, and pancreatic intraepithelial neoplasia
ampullary adenomas. There were 130 resections for malignant lesions, the majority of which (98, 75.4 %) were pancreatic ductal adenocarcinoma; 16 (12.3 %) ampullary cancers and 14 (10.8 %) malignant neuroendocrine tumors were resected as well. Six secondary pancreatic malignant lesions, all metastatic renal cell carcinoma, were also resected. Operative indications and final pathology are described in detail in Table 2. There were 9 (3.6 %) perioperative deaths observed in this group. Within 30 days of the index operation, the overall rate of any complication reported by ACS-NSQIP was 44.0 %, compared with 45.0 % in our review, however discordance was observed in 73 (29.3 %) cases, including 34 cases of falsely reporting a complication where there was not one, and 34 cases of missed complication. The overall sensitivity of ACS-NSQIP for capturing any complication was 69.6 %, and the specificity 75.2 %. The most frequently reported event was postoperative bleeding requiring transfusion, however this was also the event most commonly misclassified. There were 56 (22.7 %) instances of postoperative bleeding recorded in
123
1464
World J Surg (2014) 38:1461–1467
Fig. 1 Event-specific accuracy of the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP). Postoperative bleeding was defined as the need for transfusion or invasive intervention for bleeding [24 h after the index operation.
Pancreatic fistula includes Grades B and C pancreatic fistula incidence. UTI urinary tract infection, PNA pneumonia, Infection, other any septic complication not a direct consequence of UTI, PNA, or wound infection
ACS-NSQIP, but true incidence was actually 19.0 %, with ACS-NSQIP missing 27 (57.5 %) and incorrectly reporting 36 (64.3 %) events. ACS-NSQIP records unplanned returns to the OR as one of the quality measures for postoperative recovery. In general, we understand these to represent instances where a reoperation is necessary as a direct consequence of an intraabdominal catastrophe following the initial surgery, such as hemorrhage or peritonitis. However, of the 16 procedures recorded as reoperation events, 3 (18.8 %) were unrelated to the index operation. These included a postoperative bronchoscopy, a harvest of internal jugular vein that occurred simultaneously with the pancreatectomy but was performed by a different surgeon and dictated as a separate operative note, and an instance where the original operative note had been appended 2 days after the original operation. ACS-NSQIP began collecting readmission data from select institutions starting in 2011, so this data was unavailable for our sample. In our own pancreas center database, however, we do track readmissions and have found the 30-day readmission rate to be 8.8 %. Additional 17 (6.8 %) patients were readmitted between postoperative days 31 and 90. While a pancreas-specific module does not yet exist, ACS-NSQIP reports a 7.6 % rate of organ-space surgical site infections; when compared with our institutional rate of Grades B and C postoperative fistula (10.4 %), we observed discordance 4.4 % of the time. Nine of 26 (34.6 %) clinically significant pancreatic fistulas were not captured, and two were reported in error. Delayed gastric
emptying, a common post-pancreatectomy morbidity that we observed in 61 patients in this group (24.5 %), was not captured at all. Additionally, there were significant inaccuracies in reporting urinary tract infections, postoperative pneumonia, wound complications, and postoperative sepsis, with discordance rates of 4.4, 3.2, 3.6, and 6.8 %, respectively. Summary data representing sensitivity and specificity of ACS-NSQIP for individual postoperative complications, as well as event concordance are presented in Fig. 1 and Table 3.
123
Discussion We evaluated the accuracy of ACS-NSQIP data pertaining to outcomes after pancreatic surgery at our institution using comprehensive independent chart review. We found that overall and pancreatectomy-specific postoperative events are often misclassified, suggesting that better, standardized, and protocolized reporting measures are necessary in order to draw reliable conclusions about outcomes in this complex and unique patient population. In this cohort, we found the overall 30-day morbidity rate to be 45 %. In their study on morbidity and mortality associated with hepatopancreatobiliary surgery, Kneuertz et al. [15] found a complication rate after pancreatectomy approximating 30 % using ACS-NSQIP data. Our higher observed morbidity, however, is actually more consistent with reports from individual high-volume centers [19–22]. This suggests potential underreporting of overall
World J Surg (2014) 38:1461–1467
1465
Table 3 Event concordance between ACS-NSQIP and pancreas center database Events
a
Event concordance
Events reported in error
Events not captured
Total events
Any complication
88 (64.2 %)
24 (21.4 %)
49 (35.8 %)
137 (55.0 %)
Postoperative bleedingb
20 (42.6 %)
36 (64.3 %)
27 (57.5 %)
47 (19.0 %)
Pancreatic fistulac
17 (65.4 %)
2 (10.5 %)
9 (34.6 %)
26 (10.4 %)
UTI
11 (61.1 %)
4 (26.7 %)
7 (38.9 %)
18 (7.2 %)
PNA
9 (60.0 %)
2 (18.2 %)
6 (40.0 %)
15 (6.0 %)
Wound infection
16 (88.9 %)
7 (34.8 %)
2 (11.1 %)
18 (7.2 %)
Infection, otherd
11 (47.8 %)
5 (31.3 %)
12 (52.2 %)
23 (9.2 %)
UTI urinary tract infection, PNA pneumonia a
Compared with events captured by our institutional database
b
Defined as the need for transfusion or invasive intervention for bleeding [24 h after the index operation
c
Represents Grades B and C pancreatic fistula incidence
d
Any septic complication not a direct consequence of UTI, PNA, or wound infection
complications in ACS-NSQIP. Interestingly, in our review, 12 patients experienced additional complications between postoperative days 31 and 90, underscoring that patients who have undergone pancreatic resections may present with complications in a delayed fashion, and therefore it is important to track outcomes in this population well beyond 30 days. In order to ensure high quality data, ACS-NSQIP conducts inter-rater reliability audits of the clinical reviewers, and gives an overall 2.5 % disagreement rate for all 136 variables collected [23]. For certain individual variables, however, we observed a much higher discrepancy between what is contained within ACS-NSQIP and chart review. Specifically and notably, there was a high rate of discordance with respect to transfusion/bleeding events. We found this to be due to the fact that events reported by ACS-NSQIP do not differentiate between transfusions given intraoperatively, and those administered after the operation. As the former may indicate increased complexity of the surgical procedure but not necessarily represent a postoperative complication, in our database we grade clinically significant bleeding events according to the guidelines proposed by the International Study Group on Pancreatic Surgery [24], as we feel it is a more clinically meaningful definition. It may be argued that since our definition for this particular event excludes intraoperative transfusion, the high rate of disagreement we found between the two databases is falsely inflated. However significant inaccuracy still remained even when we accounted for this difference in definition. In 2010, data in ACS-NSQIP underwent major definition change and currently intraoperative and postoperative transfusions are tracked separately, however we urge caution in using this variable in predictive modeling that relies on historically earlier data.
With regard to discrepancies in reporting of general post-surgical complications, the causal element is unclear. Both ACS-NSQIP and our institutional data are tracked by hand review of the chart and electronic medical record, where appropriate, therefore the argument of local versus administrative database does not apply. ACS-NSQIP reporters are specially trained by the program and must pass certifications in order to comply with the stringent requirements of the program, so theoretically data fidelity in what is reported by NSQIP and our internal database should be comparable. However, given the discrepancy we observe between the two, it may be worthwhile to consider enhancing the stringency of the training requirements for all data collectors, as well as using a dedicated specialist who is adept and versatile in identifying and correctly categorizing postoperative events in patients who undergo pancreatectomy, given that so many patients in this group experience protracted and complex set of events in the postoperative period. Readmission after pancreatectomy is increasingly being used as a quality metric for surgical care, however our prior work demonstrated this to be a multifactorial and often unavoidable event in this patient population [25]. We could not compare concordance between ACS-NSQIP and our own data because readmission tracking began after 2011, however for our entire population of patients undergoing pancreatic operations, the 30-day readmission rate was shown to be 11.6 % [25]. In the cohort we evaluated for the present study, 30 and 90 day readmission was 8.8 and 6.8 %, respectively. Given that nearly half of all readmissions (17 of 39, 43.6 %) occurred between postoperative days 31 and 90, we would urge that this particular outcome be tracked in a more longitudinal fashion as well. ACS-NSQIP data has been used to evaluate outcomes after pancreatic surgery [13–15], but many events specific
123
1466
to these operations are not captured by ACS-NSQIP. To that end, efforts are currently underway to create a procedure-specific version of ACS-NSQIP that would gather variables and outcomes pertinent to the particular ‘‘core’’ procedure. Indeed, in 2009, Pitt et al. [23] suggested that a hepato-pancreatico-biliary specific module be created in order to provide reliable, durable, and protocolized definitions of factors specific to and important in defining outcomes after complex and rare pancreatic operations. Among the elements members of this group propose including, and that we feel would be of greatest use to the scientific and surgical community with respect to understanding the perioperative experience of patients undergoing hepatopancreatobiliary operations are preoperative factors specifically relevant to the type of surgery, such as presence of cirrhosis or biliary obstruction, concurrent local conditions (i.e. pancreatitis), antecedent related surgical or endoscopic procedures, and history (and type) of neoadjuvant therapy (if applicable). This module should also capture intraoperative and procedure-specific variables, such as whether vascular resection was performed, what was the conduit used, how were the enteric, pancreatic, and biliary reconstruction done, and what were the texture of the pancreas and sizes of the ducts. Additionally, postoperative data on outcomes, both short and long term, specific to pancreatic operations such as the development of fistula, delayed gastric emptying, readmission, pathology and staging, adjuvant therapy, and overall and diseasefree survival should be recorded. Realizing that a significant gap in knowledge exists on this subject, very recently, as part of the NSQIP Pancreas Demonstration Project, Parmar and colleagues [26] used institution-specific variables collected through prospective chart review among institutions who participate in NSQIP nationwide to assess the contribution of various intraoperative and postoperative factors to the development of delayed gastric emptying. This project required collaboration of 33 hospitals providing data points collected after reviewing the clinical course of 1,384 patients, an impressive undertaking and a key contribution to the body of literature on this topic. We feel this work only further underscores the need for a global NSQIP-wide initiative, the development of which we hope will be aided by the results of our study presented herein. The general shift from chart review to use of large scale national databases such as ACS-NSQIP reflects an important trend in surgical outcomes research as this paradigm improves efficiency, relies on consistent definitions that allow subsequent comparison of outcomes from different institutions, and allows for increased sample size, thereby enhancing statistical power in evaluating rare and complex conditions and operations. However, it is critical to take into consideration the potential tradeoff in accuracy when
123
World J Surg (2014) 38:1461–1467
drawing conclusions and building healthcare policy based on such data. We find there to be discrepancies in recording of general surgical and pancreatectomy-specific complications between ACS-NSQIP and our institutional database. Some of these can be attributed to inconsistent definitions and can therefore be easily rectified going forward, however many appear to be related to human error and methodology issues. In the current climate of healthcare change and reform, understanding the extent of this effect is imperative and to our knowledge has not been previously described with respect to the ACS-NSQIP database.
Conclusions ACS-NSQIP is an important initiative for standardizing and comparing outcomes after surgery. As mentioned previously, significant improvements in the delivery of care across many different types of institutions and for many different types of operations have already been demonstrated after analyzing post-implementation data. However, as this data is increasingly utilized to inform quality improvement policies, it is important to understand its strengths as well as its limitations. Our study demonstrates discordance between ACS-NSQIP data and our institutional database. As might be expected, this discordance was high for pancreas-specific complications, such as pancreatic fistula and delayed gastric emptying, but inaccuracies existed even in the reporting of general post-surgical problems. The accuracy of the former will likely be significantly improved with an HPB-specific module, but the latter suggests broader limitations of the ACS-NSQIP data. These limitations ought to be taken into consideration when using ACS-NSQIP data to inform quality improvement policies.
References 1. Khuri SF, Daley J, Henderson W et al (1998) The Department of Veterans Affairs’ NSQIP: the first national, validated, outcomebased, risk-adjusted, and peer-controlled program for the measurement and enhancement of the quality of surgical care: National VA Surgical Quality Improvement Program. Ann Surg 228:491–507 2. Ingraham AM, Richards KE, Hall BL et al (2010) Quality improvement in surgery: the American College of Surgeons National Surgical Quality Improvement Program approach. Adv Surg 44:251–267 3. American College of Surgeons (ACS) NSQIP. http://site.acsns qip.org/about-acs/. Accessed October 2013 4. Hall BL, Hamilton BH, Richards K et al (2009) Does surgical quality improve in the American College of Surgeons National Surgical Quality Improvement Program: an evaluation of all participating hospitals. Ann Surg 250:363–376
World J Surg (2014) 38:1461–1467 5. Vaid S, Bell T, Grim R et al (2012) Predicting risk of death in general surgery patients on the basis of preoperative variables using American College of Surgeons National Surgical Quality Improvement Program data. Perm J 16:10–17 6. Fink AS, Campbell DA Jr, Mentzer RM Jr et al (2002) The National Surgical Quality Improvement Program in non-veterans administration hospitals: initial demonstration of feasibility. Ann Surg 236:344–353 (discussion 353–344) 7. Kazaure HS, Roman SA, Rosenthal RA et al (2013) Cardiac arrest among surgical patients: an analysis of incidence, patient characteristics, and outcomes in ACS-NSQIP. JAMA Surg 148:14–21 8. Khan MA, Grinberg R, Johnson S et al (2013) Perioperative risk factors for 30-day mortality after bariatric surgery: is functional status important? Surg Endosc 27:1772–1777 9. Lee DY, Chung EL, Guend H et al (2013) Predictors of mortality after emergency colectomy for Clostridium difficile colitis: an analysis of ACS-NSQIP. Ann Surg 259:148–156 10. Madenci AL, Keith Ozaki C, Belkin M et al (2013) Carotid-subclavian bypass and subclavian-carotid transposition in the thoracic endovascular aortic repair era. J Vasc Surg 57:1275–1282 11. Cima R, Dankbar E, Lovely J et al (2013) Colorectal surgery surgical site infection reduction program: a national surgical quality improvement program-driven multidisciplinary singleinstitution experience. J Am Coll Surg 216:23–33 12. Nelson MT, Greenblatt DY, Soma G et al (2012) Preoperative factors predict mortality after major lower-extremity amputation. Surgery 152:685–694 (discussion 686–694) 13. Sun RC, Button AM, Smith BJ et al (2013) A comprehensive assessment of transfusion in elective pancreatectomy: risk factors and complications. J Gastrointest Surg 17:627–635 14. Leichtle SW, Kaoutzanis C, Mouawad NJ et al (2013) Classic Whipple versus pylorus-preserving pancreaticoduodenectomy in the ACS NSQIP. J Surg Res 183:170–176 15. Kneuertz PJ, Pitt HA, Bilimoria KY et al (2012) Risk of morbidity and mortality following hepato-pancreato-biliary surgery. J Gastrointest Surg 16:1727–1735
1467 16. Shiloach M, Frencher SK Jr, Steeger JE et al (2010) Toward robust information: data quality and inter-rater reliability in the American College of Surgeons National Surgical Quality Improvement Program. J Am Coll Surg 210:6–16 17. DeOliveira ML, Winter JM, Schafer M et al (2006) Assessment of complications after pancreatic surgery: a novel grading system applied to 633 patients undergoing pancreaticoduodenectomy. Ann Surg 244:931–937 (discussion 937–939) 18. Bassi C, Dervenis C, Butturini G et al (2005) Postoperative pancreatic fistula: an international study group (ISGPF) definition. Surgery 138:8–13 19. Hackert T, Werner J, Bu¨chler MW (2011) Postoperative pancreatic fistula. Surgeon 9:211–217 20. Gouma DJ, van Geenen RC, van Gulik TM et al (2000) Rates of complications and death after pancreaticoduodenectomy: risk factors and the impact of hospital volume. Ann Surg 232:786–795 21. Makni A, Bedioui H, Jouini M et al (2011) Pancreaticojejunostomy vs. pancreaticogastrostomy following pancreaticoduodenectomy: results of comparative study. Minerva Chir 66:295–302 22. Pecorelli N, Balzano G, Capretti G et al (2012) Effect of surgeon volume on outcome following pancreaticoduodenectomy in a high-volume hospital. J Gastrointest Surg 16:518–523 23. Pitt HA, Kilbane M, Strasberg SM et al (2009) ACS-NSQIP has the potential to create an HPB-NSQIP option. HPB 11:405–413 24. Grutzmann R, Ruckert F, Hippe-Davies N et al (2012) Evaluation of the International Study Group of Pancreatic Surgery definition of post-pancreatectomy hemorrhage in a high-volume center. Surgery 151:612–620 25. Gawlas I, Sethi M, Winner M et al (2012) Readmission after pancreatic resection is not an appropriate measure of quality. Ann Surg Oncol 20:1781–1787 26. Parmar AD, Sheffield KM, Vargas GM et al (2013) Factors associated with delayed gastric emptying after pancreaticoduodenectomy. HPB 15:763–772
123
