Pancreatectomy risk calculator: an ACS-NSQIP resource

ORIGINAL ARTICLE

Pancreatectomy risk calculator: an ACS-NSQIP resourcePurvi Parikh1, Mira Shiloach2, Mark E. Cohen2, Karl Y. Bilimoria3, Clifford Y. Ko4, Bruce L. Hall5 & Henry A. Pitt1

1Department of Surgery, Indiana University, Indianapolis, 2American College of Surgeons, 3Department of Surgery, Northwestern University, Chicago, IL,4Department of Surgery, University of California Los Angeles, Los Angeles, CA, and 5Department of Surgery, Washington University, St. Louis, MI, USA

Abstracthpb_216 488..497

Background: The morbidity of pancreatoduodenectomy remains high and the mortality may be signifi-

cantly increased in high-risk patients. However, a method to predict post-operative adverse outcomes

based on readily available clinical data has not been available. Therefore, the objective was to create a

‘Pancreatectomy Risk Calculator’ using the American College of Surgeons-National Surgical Quality

Improvement Program (ACS-NSQIP) database.

Methods: The 2005–2008 ACS-NSQIP data on 7571 patients undergoing proximal (n = 4621), distal

(n = 2552) or total pancreatectomy (n = 177) as well as enucleation (n = 221) were analysed. Pre-operative

variables (n = 31) were assessed for prediction of post-operative mortality, serious morbidity and overall

morbidity using a logistic regression model. Statistically significant variables were ranked and weighted

to create a common set of predictors for risk models for all three outcomes.

Results: Twenty pre-operative variables were statistically significant predictors of post-operative mor-

tality (2.5%), serious morbidity (21%) or overall morbidity (32%). Ten out of 20 significant pre-operative

variables were employed to produce the three mortality and morbidity risk models. The risk factors

included age, gender, obesity, sepsis, functional status, American Society of Anesthesiologists (ASA)

class, coronary heart disease, dyspnoea, bleeding disorder and extent of surgery.

Conclusion: The ACS-NSQIP ‘Pancreatectomy Risk Calculator’ employs 10 easily assessable clinical

parameters to assist patients and surgeons in making an informed decision regarding the risks and

benefits of undergoing pancreatic resection. A risk calculator based on this prototype will become

available in the future as on online ACS-NSQIP resource.

KeywordsACS-NSQIP, pancreatectomy, risk calculator, pancreatic resections

Received 12 April 2010; accepted 29 June 2010

CorrespondenceHenry A. Pitt, 535 Barnhill Drive RT 130D, Indianapolis, IN 46202, USA. Tel: 317 274 2304; Fax: +317 274

4554; E-mail: [email protected]

Introduction

During the past decade, the number of pancreatic resections beingperformed at high volumes centres has progressively increased.1

This change is multifactorial and includes increased detection ofmalignant and premalignant pancreatic lesions, improved out-comes at regional referral institutions and surgeons’ willingness tooperate on older and higher-risk patients. The annual number ofpancreatic resections has increased by 15% in the past 20 years,

and resection for benign pancreatic disease has increased by 27%.2

However, the morbidity of pancreatic surgery remains high, andthe mortality may be significantly increased in high-risk patients.With a narrow therapeutic margin, careful patient selection isimperative to minimize post-operative complications and opera-tive mortality.

Single-institution studies have reported a low peri-operativemortality rate of 1–2% for these procedures, but these results arenot always reproducible at other institutions.3–5 In contrast,population-based studies have reported a higher peri-operativemortality rate ranging from 4.6% to 7.8%.6,7 Morbidity after pan-creatic resections still remains high with complication rates

This paper was presented at the International Hepato-Pancreato-Biliary

Association Meeting, 18–22 April 2010, Buenos Aires, Argentina.

DOI:10.1111/j.1477-2574.2010.00216.x HPB

HPB 2010, 12, 488–497 © 2010 International Hepato-Pancreato-Biliary Association

varying from 20% to 50%.8,9 Recently, The Nationwide InpatientSample (NIS) has been used to develop both a nonogram and arisk score to predict in-hospital mortality for cancer patientsundergoing pancreatectomy.10,11 These NIS-based analyses givegeneralized estimates of national inpatient mortality rates forpatients after pancreatic resections for cancer. However, they donot include estimates for other pancreatic diseases nor do theyprovide information about the risk for developing post-operativecomplications.

The American College of Surgeons-National Surgery QualityImprovement Program (ACS-NSQIP) currently collects data onpre-operative risk factors as well as post-operative morbidity andmortality to assess surgical quality at more than 200 hospitals. TheNSQIP was first developed in Veterans Affairs hospitals in the1990s12,13 and was then piloted in selected university medicalcentres in the early 2000s.14 As the ACS-NSQIP has evolved, thepotential to provide robust outcomes on patients undergoingpancreatic surgery and to develop risk calculators has become areality.15,16 Therefore, the objective of this analysis was to use theACS-NSQIP database to develop a pancreatectomy risk calculatorto predict post-operative adverse outcomes based on readily avail-able clinical data.

MethodsACS-NSQIPThe American College of Surgeons-National Surgical QualityImprovement Program (ACS-NSQIP) is a prospective, multicen-tre clinical registry that was created to provide feedback on risk-adjusted outcomes to hospitals for quality-improvementpurposes. The sampling strategy, data abstraction procedures,variables collected and structure have already been published.17–20

The programme collects detailed information on patient demo-graphics, pre-operative risk factors and laboratory values, opera-tive variables and post-operative events using standardizeddefinitions.12 From the ACS NSQIP database for 1 January 2005 to31 December 2008, patients >16 years of age who underwent amajor pancreatic resection were identified using Current Proce-dural Terminology (CPT) codes. These data were used for thedevelopment of a pancreatectomy risk calculator.

Pre-operative variablesA set of potential predictive variables was constructed from ACS-NSQIP data fields. The patient demographic variables of age(<65, 65–74, 75–84 and >85 years) and gender, and the lifestylefactor of smoking status (within 1 year of operation) and alcoholstatus were considered. The pre-operative factors considered wereAmerican Society of Anesthesiologists (ASA) classification (I/II,normal healthy or mild systemic disease; III, severe systemicdisease; IV/V, severe systemic disease that is a constant threat tolife or moribund), pre-operative functional status (independentvs. partially or totally dependent), dyspnoea (none, moderateexertion, at rest) and body mass index (BMI) (normal, under-weight, overweight, three levels of obesity as classified by the

World Health Organization). Comorbidities considered were ven-tilator dependence, sepsis, a history of chronic obstructive pulmo-nary disease (COPD), hypertension requiring medication, currentpneumonia, ascites, congestive heart failure (within 30 days priorto the procedure), coronary heart disease (includes angina, myo-cardial infarction within 30 days before the operation, percutane-ous cardiac intervention, or coronary artery bypass surgery),peripheral vascular disease (includes revascularization for periph-eral vascular disease, claudication, rest pain, amputation, or gan-grene) and a neurological event or disease (includes stroke with orwithout residual deficit, transient ischaemic attack, haemiplegia,paraplaegia, quadriplaegia, or impaired sensation), diabetes (oralmedication or insulin dependent), dialysis, acute renal failure,weight loss (>10% in past 6 months), bleeding disorders, currentchemotherapy or recent radiotherapy, oesophageal varices,chronic steroid use and red blood cell transfusion prior to theprocedure. All of these comorbidities are rigorously definedwithin the ACS-NSQIP. A total of 31 pre-operative variables wereassessed for prediction of post-operative mortality, serious mor-bidity, or overall morbidity.

OutcomesOutcomes were assessed at 30 days regardless of whether thepatient was discharged, remained hospitalized or was admitted toa different institution. Outcomes included mortality (all-causedeath within 30 days after the operation), serious morbidity and30-day overall morbidity. Overall morbidity includes superficialsurgical site infection (without pre-operative wound infection),deep incisional surgical site infection (without pre-operativewound infection), pneumonia (without pre-operative pneumo-nia), unplanned intubation (without pre-operative ventilatordependence), progressive renal insufficiency (without pre-operative renal failure or dialysis), urinary tract infection anddeep venous thrombosis. Variables that were assigned to seriousmorbidity were organ space surgical site infection (without pre-operative wound infection), wound disruption, cerebrovascularaccident or stroke, myocardial infarction, cardiac arrest requiringcardiopulmonary resuscitation, pulmonary embolism, ventilatordependence longer than 48 h (without pre-operative ventilatordependence), acute renal failure (without pre-operative renalfailure or dialysis), bleeding complications defined by transfusionsin excess of four units and sepsis or septic shock (if pre-operativesepsis exists, it must worsen post-operatively).

Risk calculator developmentThe risk calculator was developed after all pre-operative variableswere made categorical and entered with CPT codes and ICD-9(International Classification of Disease Codes, 9th edition) codes.CPT codes were used to categorize surgical procedure by extent ortype (proximal pancreatectomy, distal pancreatectomy, total pan-createctomy or enucleation). The CPT codes used were 48140,48145, 48146, 48150, 48152, 48153, 48154, 48155 and 48120. Thediagnoses were categorized according to the indication using

HPB 489


ICD-9 codes: acute pancreatitis, chronic pancreatitis, benign neo-plasm, malignant neoplasm or other. Operations were also classi-fied by emergent/non-emergent procedures and wound class(class 1/2: clean, clean/contaminated vs. class 3/4: contaminated ordirty/infected).

Pre-operative laboratory variables examined included haemat-ocrit, white blood count, platelet count, sodium, blood urea nitro-gen, creatinine, albumin, partial thromboplastin time andprothrombin. Values were categorized using ACS-NSQIP defini-tions of normal and abnormal, and missing data constituted athird categorical level, an indicator variable. Of note, all of thelaboratory values were forced into the model and were found notto have a substantial impact on the overall determination of pre-dictors for the risk calculator. Thus, they were not included in thefinal risk calculator model.

Variables that entered the model for mortality included: agegroup, systemic sepsis, functional health status, ASA classification,history of congestive heart failure, dyspnoea, previous or concur-rent chemotherapy, oesophageal varices and type of surgery. Themortality model had four forced variables including gender, BMIclassification, coronary heart disease and bleeding disorder. Vari-ables that entered the predictive model for morbidity included:age group, gender, BMI classification, systemic sepsis, functionalstatus, ASA classification, surgical extent, coronary heart disease,history of severe COPD, smoking status, dyspnoea, bleeding dis-orders and weight loss greater than 10%.

Statistical analysisAll variables were converted to a categorical format. The statisticalmodel was constructed in two stages. First, forward stepwise logis-tic regression models for mortality, serious morbidity and overallmorbidity were run which included all the independent variablesmentioned above.21 This modelling method adds variables to themodel when they provide significant, independent contributions.A second step, the Firth penalized likelihood approach, wasneeded in order to account for some predictors that have emptycells as a result of a low occurrence rate or small sample sizes.22 Forexample, mortality is an infrequent outcome and, when crossedwith certain predictors, may result in a cell that has no mortalityoutcome under at least one level of the predictor. The presence ofthese empty cells can compromise the validity of the ordinarylogistic model fitting algorithm. Firth’s penalized likelihoodapproach was therefore used to achieve model consequence underthese empty cell conditions.22 However, as Firth’s method pre-cludes step-wise methods, a two-step process was used wherestep-wise selected variables were forced into Firth-adjustedmodels for final parameter specifications.

Model quality was evaluated using Hosmer–Lemeshowgoodness-of-fit tests for calibration (correspondence in predic-tions and observations across the range of predictions) andc-statistics for discrimination.23,24 The c-statistic was consideredthe most relevant measure of model success and refers to theability of the risk estimate to discriminate cases from non-cases. If

discrimination is no better than chance, the c-statistic will equal0.50. All analysis and data manipulation were done using SAS 9.2(SAS Institute, Inc., Cary, NC, USA).

Patient examplesTo test the model and determine if it gave results similar to nation-ally published data, we developed three example patients who fellinto low-, intermediate- and high-risk areas. The low-risk patientwas a 45-year-old female with a mucinous cyst neoplasm (MCN)in the neck of the pancreas. This theoretical patient had an ASAclass of II, a BMI of 25 and no pre-operative sepsis. She is nowfully functional with no coronary heart disease, no dyspnoea orbleeding disorders. The pre-operative plan is enucleation of theMCN. The intermediate risk patient was a 75-year-old male withan intraductal papillary mucinous neoplasm (IPMN) in the tail ofthe pancreas. He has a BMI of 35, an ASA class of III, no pre-operative sepsis and is fully functional. He had a prior coronaryartery bypass graft but did not have dyspnea or bleeding disorders.He is scheduled for a laparoscopic distal pancreatectomy. Thehigh-risk patient was a 55-year-old male with chronic pancreatitisas a result of excess alcohol intake. He has chronic pain requiringnarcotics and a distal bile duct stricture which was stented endo-scopically. This patient has a BMI of 30, sepsis because of cholan-gitis, is dependent in an intensive care unit, has an ASA class of IV,had a prior coronary angioplasty, dyspnea with moderate exertionand has a bleeding disorder. His surgeon is contemplating a pan-creatoduodenectomy for relief of pain and biliary obstruction.

ResultsPatient characteristicsThe 2005 to 2008 ACS-NSQIP dataset yielded 7571 pancreaticprocedures at 193 hospitals (Table 1). The average patient age was61.9 years with 47.7% being male. ASA Class III was mostcommon (60.8%). The most frequent procedure was a proximalpancreatectomy at 61.0%, and the most common indication wasneoplasm at 66.5%. Mortality was 2.5%, serious morbidity was21.2% and overall morbidity was 31.8%.

The individual mortality rates for the procedures were 2.9% forproximal pancreatectomy, 1.7% for distal pancreatectomy, 0.4%for pancreatic enucleation and 4.8% for total pancreatectomy.

Risk calculatorApplication of the variable selection process for the datasetyielded six variables that appeared in models for all three out-comes and five variables that appeared in two outcomes (Table 2).One variable, coronary heart disease, was only involved in onemodel but had significant impact and early selection and thereforewas considered to be important enough to be entered into thedataset. Based on entry into individual models, the 10 variableswhich were most highly weighted and found to have the highestrank in all three models, were chosen as the universal data set todevelop the pancreatic risk calculator. Odds ratios for the variablesselected for the universal model showed findings consistent with

490 HPB


clinical expectations (Table 3). The relative odds ratios for age andsurgical extent are illustrated in Fig. 1a,b, respectively. Modelsusing the 10-variable universal dataset had acceptable discrimina-tion and calibration for each outcome (Table 4). The c-statisticsfor mortality, serious morbidity and overall morbidity were 0.74,0.61 and 0.61, respectively, and the Hosmer–Lemoshow fit statis-tics were not significant, indicating that the model had adequatefit for the variables entered.

Patient examplesThe patient examples that were entered into the model showedlevels of mortality, serious morbidity and overall morbidity com-parable to most literature on this topic (Table 5, Fig. 2). The low-risk patient who had no abnormal variables and received apancreatic enucleation for a benign neoplasm had a risk of mor-tality, serious morbidity and overall morbidity of 0.08%, 12.6%and 18.6%, respectively. The intermediate-risk patient who had

four abnormal variables and received a distal pancreatectomy foran IPMN had a 2.1% risk of mortality. The risk for serious mor-bidity and overall morbidity were higher at 23% and 32.6%. Thehigh-risk patient who had eight abnormal variables and was beingconsidered for a proximal pancreatectomy had a mortality risk of33.6%, with a risk of serious morbidity at 77.2% and overallmorbidity at 87.6%.

Discussion

The ACS-NSQIP dataset from 2005–08 had 7571 patients whohad undergone a pancreatic resection. Thirty-one preoperativevariables were analysed for prediction of post-operative mortality,serious morbidity and overall morbidity using logistic regressionmodels. Twenty pre-operative variables were found to be statisti-cally significant. Ten out of the 20 risk factors were employed toproduce mortality and morbidity risk models. The risk factorsincluded age >74 years, male gender, BMI higher than 40, pre-operative sepsis, dependent functional status, ASA class more thanII, history of coronary heart disease, dyspnoea on moderate exer-tion, a bleeding disorder and the contemplated procedure(Table 6). All of these variables can be easily assessed at the time ofinitial presentation and entered into the model so that a surgeoncan provide an accurate assessment of operative risk, and a patientcan receive individualized estimates of the risk of mortality,serious morbidity and overall morbidity.

Patients who present with pancreatic pathology are oftenelderly and have multiple medical comorbidities. During initialevaluation, while performing a patient’s history and physical, thecomorbidities in the Pancreatectomy Risk Calculator can be iden-tified and can be used to determine if the patient is an appropriatecandidate for surgery. During this pre-operative counselling, therisk and benefits of the procedure are explained, and consent isobtained. Currently, the pre-operative counselling needed toobtain informed consent includes knowledge of published peri-operative morbidity and mortality rates. However, the majority ofthis information comes from referral centres and is not individu-alized for a particular patient. The Pancreatectomy Risk Calcula-tor that was developed from the ACS-NSQIP database usesvariables that can easily be discovered by a careful history andphysical examination and will be available online as an ACS-NSQIP resource. This information also may be employed to planresources for the patient and might be used to encourage riskstratification.

Predictive models that calculate the risk of post-operative mor-tality after pancreatectomy for cancer have been developed usingthe Nationwide Inpatient Sample.10,11 The NIS is managed by theHealthcare Cost and Utilization Project and is the largest all-payerdatabase of hospital discharges, providing a 20% stratified sampleof all non-federal hospitals in a given year. The data for the hos-pital and patients are entered retrospectively after discharge of thepatient using ICD-9 codes. The integer-based risk score that wasdeveloped from the NIS uses age, gender, Charlson comorbidity

Table 1 Patient demographics in ACS-NSQIP used for pancreatec-tomy risk calculator

Variable 2005–2008

n 7571

Hospitals, n 193

Cases/hospital, n (range) 39.2 (1–519)

Age, years,a mean � SD 61.9 � 13.8

Gender, % male 47.7

ASA, %

I/II. Normal healthy/Mild systemic disease 34.7

III. Severe systemic disease 60.8

IV/V. Severe systemic disease/Moribund 4.5

Surgical extent, %

Proximal pancreatectomy 61.0

Distal pancreatectomy 33.7

Enucleation 2.9

Total pancreatectomy 2.3

Indication for surgery, %b

Malignant neoplasm 66.5

Benign neoplasm 18.5

Chronic pancreatitis 5.0

Acute pancreatitis 1.3

Other 8.7

Outcomes

Mortality 2.5

Serious morbidity 21.2

Overall morbidity 31.8

aAge recorded as 90+ had been recorded to 90.bTaken from the reported ICD-9 code for post-operative diagnosis in ACSNSQIP.ACS NSQIP, American College of Surgeons National Quality Improve-ment Program; ASA, American Society of Anesthesiologists PhysicalStatus Classification.

HPB 491


score, pancreatectomy type and hospital volume as its predictorsfor in-patient mortality for patients with pancreatic adenocarci-noma.10 This model may be able to predict in-patient mortality;however, it does not have the ability to predict serious morbidityand overall morbidity. With published reports showing that mor-bidity for pancreatic resections varies from 20% to 50%, thechance of developing a post-operative complication is importantfor patients to appreciate prior to pancreatic resection.25

Nonograms are graphical devices or models that use algorithmsor mathematical formulae to estimate the probability of anoutcome for each individualized patient. The benefit of post-operative nonograms in predicting long-term survival has beenproven in patients with cancers of various organ systems.Recently, a nonogram has been developed to preoperativelypredict in-patient mortality for patients after pancreatic resectionusing the Nationwide Inpatient Sample (NIS).10 This nonogramhas similar limitations as the integer-based risk score. Importantfactors that might contribute to peri-operative risk such as ASAclass, functional status, weight loss, coronary heart disease andserum albumin are not available in the NIS. Furthermore, thenonogram, while web accessible, has numerous questions thatneed to be answered, making it lengthy and cumbersome for use.In comparing the ACS-NSQIP Pancreatectomy Risk Calculator tothe NIS Risk Score and Nonogram, the Risk Calculator employed

more patients (7571 vs. 5481 or 5715), over a shorter time (4 vs. 6or 9 years), with more diagnoses (all vs. cancer) and procedures(all vs. major).

In recent years, considerable attention has been given to usingcomputer-based methods to classify medical data to help predictoutcomes. The general approach has been to develop computeralgorithms that learn decision characteristics for data classifica-tion and then use them to classify future patients with unknowndisease states or therapy outcomes. Several quantitative modelsranging from simple linear analysis to more complex logisticregression and artificial neural networks (ANN) have been pro-posed. ANN is based on finding an optimal path from the samplespace to the decision space. This process involves feeding uniqueinput samples (features) and the matching responses (outcomes)to let the network learn from the examples and compose a mapthat inter-relates inputs to outputs through a complex set of inter-connecting pathways or operations.26 Unlike logistic regression,which fits the data to a descriptive function, ANN transforms thedata on each layer, changing its dimensional space to define therule to get to the decision region. Thus, the two approaches areinherently different, raising the question if one approach hasbetter diagnostic performance than the other. A meta-analysiscomparing ANN with regression models in 28 studies foundthat both modes have similar performance.27 In another study,

Table 2 Variables selected in construction of the pancreatectomy risk calculator

ACS NSQIP variablesa Mortalityb Seriousmorbidityb

Overallmorbidityb

Models, n Included inuniversal model

ASA classification 2 2 1 3 Yes

Functional health status 1 3 4 3 Yes

Sepsis 6 1 2 3 Yes

Surgical extent 5 6 3 3 Yes

Age group 3 8 8 3 Yes

Dyspnoea 4 5 11 3 Yes

Body mass index 7 7 2 Yes

Coronary heart disease 5 1 Yes

Gender 4 12 2 Yes

Bleeding disorder 11 6 2 Yes

Oesophageal varices 8 13 2 No

COPD 12 9 2 No

Congestive heart failure 7 1 No

Chemotherapy 9 1 No

Wound class 9 1 No

Peripheral vascular disease 10 1 No

Smoking status 10 1 No

Weight loss 13 1 No

Ascites 14 1 No

Neurological disease 14 1 No

aSome variables have been restructured. bSelection order.ACS NSQIP, American College of Surgeons National Quality Improvement Program; ASA, American Society of Anesthesiologists Physical StatusClassification; COPD, chronic obstructive pulmonary disease.

492 HPB


Dreiseitl surveyed 72 articles comparing ANN with logistic regres-sion and found there was no difference in models for predictingoutcomes.28 Currently, only one paper uses ANN for predictingclinical outcomes in patients with acute biliary pancreatitis andnone for pancreatic surgery.29

A recent analysis of complication rates after pancreatectomyagain employing the Nationwide Inpatient Sample also has beenpublished.30 This analysis reported a 22.7% complication rate afterpancreatectomy with no change over the 9 years of the analysis(1998–2006). Independent predictors of complications includeage >74 years, total pancreatectomy, and low hospital resectionvolume. In comparison, the ACS-NSQIP Pancreatectomy risk Cal-culator provides robust information for prediction of both seriousand overall morbidity with more recent clinical rather thanadministrative data.

The potential benefit of undergoing resection at high-volumescentres has led to regionalization of care for patients with pancre-atic malignancies.1,31–33 However, contradictory data exist as towhat defines high volume and whether volume should be definedby surgeon volume or hospital volume.34,35 A study using theNationwide Inpatient Sample showed that volume aloneaccounted for less than 2% of the variance in peri-operative mor-tality after pancreatic resection.35 Furthermore, the NationwideInpatient Sample only provides information on hospital volumebut is not able to provide surgeon specific data. The ACS-NSQIPhas the potential to adjust outcomes by hospital volume andindividual surgeon. Analyses of these options are underwaybut are not represented in the present. In this study of the 193hospitals, 60% were academic/teaching institutions which couldbe described as high-volume hospitals for pancreatic resections.

Table 3 Percentages, odds ratios and confidence intervals for variables in pancreatectomy risk calculator

ACS NSQIP variablesa % of patients Mortality Serious morbidity Overall morbidity

Age (<65 years) 53.3

65–74 years 27.1 1.70 (1.17–2.47)b 1.10 (0.96–1.26)b 1.13 (1.00–1.27)b

75–84 years 17.6 2.28 (1.54–3.38) 1.30 (1.11–1.51) 1.26 (1.10–1.45)

�85 years 2.0 3.54 (1.77–7.05) 1.57 (1.08–2.27) 1.78 (1.27–2.49)

Gender Male 47.7 1.16 (0.86–1.57) 1.22 (1.09–1.37)b 1.13 (1.02–1.25)c

Body mass index (normal) 36.4

Underweight 3.4 0.96 (0.40–2.33) 0.94 (0.67–1.31)b 0.95 (0.71–1.26)b

Overweight 34.8 1.29 (0.91–1.83) 1.15 (1.00–1.31) 1.16 (1.02–1.30)

Class 1 obesity 15.7 1.28 (0.82–2.02) 1.34 (1.13–1.58) 1.34 (1.16–1.56)

Class 2 obesity 6.1 0.91 (0.44–1.91) 1.22 (0.96–1.56) 1.24 (1.00–1.53)

Class 3 obesity 3.7 2.32 (1.17–4.63) 1.83 (1.38–2.44) 1.60 (1.23–2.09)

Sepsis 3.0 2.62 (1.30–3.95)b 2.26 (1.70–3.01)b 2.10 (1.60–2.79)b

Functional health status(dependent)

3.3 3.27 (2.05–5.21)b 1.73 (1.30–2.29)b 1.75 (1.33–2.30)b

ASA classification I/II (no/milddisturbance)

34.7

Class III (severedisturbance)

60.8 2.33 (1.49–3.64)b 1.18 (1.03–1.34)c 1.20 (1.06–1.33)b

Life-threatening/moribund 4.5 3.20 (1.70–6.03) 1.33 (1.01–1.75) 1.40 (1.09–1.80)

Coronary heart disease 10.8 1.18 (0.80–1.73) 1.20 (1.01–1.43)c 1.26 (1.08–1.48)b

Dyspnoea 9.2

Moderate exertion 8.4 1.72 (1.15–2.58)b 1.38 (1.14–1.66)b 1.36 (1.14–1.61)b

At rest 0.8 4.70 (2.15–2.58) 1.44 (0.82–2.53) 1.08 (0.63–1.86)

Bleeding disorder 2.9 1.15 (0.58–2.29) 1.43 (1.06–1.94)c 1.68 (1.27–2.23)b

Surgical extent (proximal) 61.0

Distal pancreatectomy 33.7 0.62 (0.43–0.88)b 0.77 (0.68–0.88)b 0.73 (0.66–0.82)b

Enucleation 2.9 0.10 (0.01–1.52) 0.74 (0.52–1.06) 0.63 (0.46–0.87)

Total pancreatectomy 2.3 1.86 (0.91–3.79) 1.07 (0.74–1.53) 0.91 (0.66–1.27)

aSome variables have been restructured.bP-values < 0.01.cP-values < 0.001 for the variable.ACS NSQIP, American College of Surgeons National Quality Improvement Program; ASA, American Society of Anesthesiologists Physical StatusClassification.

HPB 493


(a)

(b)

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

Od

ds

rati

oO

dd

s ra

tio

P < 0.01 vs < 65

< 65 65–74 75–84 > 85

Age

2.0

1.8

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

Enucleation Distal pancreatectomy Proximal pancreatectomy Total pancreatectomy

Figure 1 (a) Odds ratios for increasing patient age in predicting mortality. (b) Odds ratios for the four pancreatectomy procedures

Table 4 Model performance for pancreatectomy risk calculator

Model performance Mortality Serious morbidity Overall morbidity

Rate (n, %) 186 (2.5) 1605 (21.2) 2411 (31.8)

C-statistic 0.74 0.61 0.61

Hosmer–Lemeshow 0.28 0.61 0.79

494 HPB


Furthermore, the average number of cases per hospital was 39with the range from 1 case to 519 cases over that time period.

This study has several limitations. The ACS-NSQIP databaseonly includes information from 2005–08 and from fewer than 200hospitals nationwide. However, these institutions perform themajority of pancreatic surgery in the United States. In the 4-yearperiod, 7571 pancreatic resections were performed and includedin the development of the pancreatic risk calculator. Because of

the limited number of pancreatic resections, the pancreatic riskmodel has not been validated with any subsequent data. Thepresent plan is to carry out analyses of 2009 and future data forvalidation of this model. Being able to further evaluate bothhospital- and surgeon-specific data would help to further deter-mine patient outcome. Another limitation of the present study isthe inability to define the frequency and morbidity related tooperation-specific complications. For example, a major contribu-tor to surgical site infection after pancreatectomy is pancreaticleak. A pancreatic fistula occurs in approximately 10% to 15%of pancreatoduodenectomy36 and 30% of distal pancreatectomypatients.37,38 In this current dataset, pancreatic fistula is groupedwith other organ space infections. However, in an upcomingupdate to the ACS-NSQIP data structure pancreatectomy-specificpre-operative risk factors and post-operative outcomes will becollected. Having data on pancreatectomy-specific outcomes suchas pancreatic fistula will be very helpful in determining the causeand improving outcomes.

Despite these limitations, the purpose of the present study wasto develop a Pancreatectomy Risk Calculator using variables thatare easily obtainable by history and physical examination. ThePancreatectomy Risk Calculator is not intended to substitute forsurgeon judgment or experience but should be used as an addi-tional resource in counselling patients who are being considered

Table 5 Patients tested with pancreatectomy risk calculator

Patient performance Mortality Serious morbidity Overall morbidity

Low risk 0.08% 12.6% 18.6%

Intermediate risk 2.1% 23.0% 32.6%

High risk 33.6% 77.2% 87.6%

Risk

Per

cen

t

0

10

20

30

40

50

60

70

80

90

100

Mortality

Serious morbidity

Overall morbidity

Low Intermediate High

Figure 2 Outcomes in low-, intermediate- and high-risk patients

Table 6 Risk factors for increased mortality, serious morbidity andoverall morbidity

Age > 74

Gender Male

BMI > 40

Preoperative sepsis

Dependent functional health status

ASA classification > II

Coronary heart disease

Dyspnoea on moderate exertion

Bleeding disorder

Proximal or total pancreatectomy

BMI, body mass index.

HPB 495


for high-risk pancreatic surgery. In the future ACS-NSQIP datawill also provide increasingly specific data at the hospital-,surgeon- and procedure-specific levels. In summary, the ACS-NSQIP Pancreatectomy Risk Calculator employs 10 easily acces-sible clinical parameters to assist patients and surgeons inmaking an informed decision regarding risks and benefits ofpancreatic resection. This system also may be helpful in resourceplanning and risk modification. A risk calculator based on thisprototype will soon become available as an online ACS-NSQIPresource.

Disclosure of interest

None declared.

References

1. Zeigler KM, Nakeeb A, Pitt HA, Schmidt CM, Bishop SN, Moreno J et al.

(2010) Pancreatic surgery: evolution of a high volume center. Surgery (in

press).

2. Teh SH, Diggs BS, Deveney CW, Sheppard BC. (2009) Patient and

hospital characteristics on the variance of perioperative outcomes for

pancreatic resection in the United States. Arch Surg 144:713–717.

3. Winter JM, Cameron JL, Campbell KA, Arnold MA, Chang DC, Coleman

J et al. (2006) 1423 pancreaticoduodenectomies for pancreatic cancer: a

single institution experience. J Gastrointestinal Surg 10:1199–1210.

4. Cameron JL, Riall TS, Coleman J, Belcher KA. (2006) One thousand

consecutive pancreaticoduodenectomies. Ann Surg 244:10–15.

5. Vin Y, Sima CS, Getrajdman GI, Brown KT, Covey A, Brennan MF et al.

(2008) Management and outcomes of postpancreatectomy fistula, leak

and abscess: results of 908 patients resected at a single institution

between 2000 and 2005. J Am Coll Surg 207:490–498.

6. McPhee JT, Hill JS, Whalen GF, Zayaruzny M, Litwin DE, Sullivan ME

et al. (2007) Perioperative mortality for pancreatectomy: a national per-

spective. Ann Surg 246:246–253.

7. Ghaferi AA, Birkmeyer JD, Dimick JB. (2009) Variation in hospital

mortality associated with inpatient surgery. New Engl JMed 361:1368–

1375.

8. Yeo CJ, Cameron JL, Sohn TA, Lillemoe KD, Pitt HA, Talamini MA et al.

(1997) Six hundred fifty consecutive pancreaticoduodenectomies in the

1990s: pathology, complications, and outcomes. Ann Surg 226:248–257.

9. Buchler MW, Friess H, Muller MW, Wheatley AM, Beger HU. (1995)

Randomized trial of duodenum-preserving pancreatic head resection

versus pylorus-preserving Whipple in chronic pancreatitis. Am J Surg

169:65–70.

10. Are C, Afuh C, Ravipati L, Sasson A, Ullrich F, Smith L. (2009) Preopera-

tive nomogram to predict risk of perioperative mortality following pan-

creatic resections for malignancy. J Gastrointest Surg 13:2152–2162.

11. Hill JS, Zhou Z, Simons JP et al. (2010) A simple risk score to predict

in-hospital mortality after pancreatic resection for cancer. Ann Surg Onc

17:1802–1807.

12. Khuri SF, Henderson WG, Daley J, Jonasson O, Jones RS, Campbell DA

Jr et al. (2008) Successful implementation of the Department of Veterans

Affairs National Surgical Quality Improvement Program in the private

sector: the Patient Safety in Surgery Study. Ann Surg 248:329–336.

13. Khuri SF, Daley J, Henderson W, Hur K, Demakis J, Aust JB et al. (1998)

The Department of Veterans Affairs' NSQIP: the first national, validated,

outcome based, risk-adjusted, and peer-controlled program for the mea-

surement and enhancement of the quality of surgical care. National VA

Surgical Quality Improvement Program. Ann Surg 228:491–507.

14. Glasgow RE, Jackson HH, Neumayer L, Schifftner TL, Khuri SF,

Henderson WG et al. (2007) Pancreatic resection in Veterans Affairs and

selected university medical centers: results of the patient safety in

surgery study. J Am Coll Surg 204:1252–1260.

15. Pitt HA, Kilbane M, Strasberg SM, Pawlik TM, Dixon E, Zyromski NJ et al.

(2009) ACS-NSQIP has the potential to create an HPB-NSQIP option.

HPB (Oxford) 11:405–413.

16. Cohen ME, Bilimora KY, Ko CY, Hall BL. (2009) Development of an

American College of Surgeons National Surgery Quality Improvement

Program: morbidity and mortality risk calculator for colorectal surgery.

J Am Coll Surg 208:1009–1016.

17. ACS-NSQIP participant use file user's guide. https://acsnsqip.org/

puf/docs/ACS_NSQIP_Participant_User_Data_File_User_Guide.pdf

[accessed on 12 April 2010].

18. ACS-NSQIP program specifics: surgical case inclusion/exclusion over-

view. http://acsnsqip.org/main/program_case_inclusion_exclusion.asp

[accessed on 12 April 2010].

19. ACS-NSQIP program specifics: surgical clinical nurse reviewer training.

http://acsnsqip.org/main/program_nurse_training.asp [accessed on 12

April 2010].

20. ACS-NSQIP program specfics: ACS NSQIP data; participant use

data file. http://acsnsqip.org/puf/PufRequestHomepage.aspx [accessed

on 12 April 2010].

21. Anderson RP, Jin R, Grunkemeier GL. (2003) Understanding logistic

regression analysis in clinical reports: an introduction. Ann Thorac Surg

75:753–757.

22. Firth penalized likelihood approach described at http://support.sas.com/

documentation/cdl/en/statug/63033/HTML/default/statug_logistic_

sect026.htm#statug.logistic.logisticdfirth [accessed on 12 April 2010].

23. Kramer AA, Zimmerman JE. (2007) Assessing the calibration of mortality

benchmarks in critical care: the Hosmer-Lemeshow test revisited. Crit

Care Med 35:2052–2056.

24. Hanley JA, McNeil BJ. (1982) The meaning and use of the area under a

receiver operating characteristic (ROC) curve. Radiology 143:29–36.

25. Simons JP, Shah SA, Ng SC, Whalen GF, Tseng JF. (2009) National

complication rates after pancreatectomy: beyond mere mortality. J Gas-

trointest Surg 13:1798–1805.

26. Song JH, Venkatesh SS, Conant EA, Arger PH, Sehgal CM. (2005) Com-

parative analysis of logistic regression and artificial neural network for

computer-aided diagnosis of breast masses. Acad Radiol 12:487–495.

27. Sargent DJ. (2001) Comparison of artificial neural networks with other

statistical approaches: results from medical data sets. Cancer 91:1636–

1642.

28. Dreiseitl S, Ohno-Machado L. (2002) Logistic regression and artificial

neural network classification models: a methodology review. J Biomed

Inform 35:352–359.

29. Yoldas O, Koc M, Karakose N, Kilic M, Tez M. (2008) Prediction of clinical

outcomes using artificial neural networks for patients with acute biliary

pancreatitis. Pancreas 36:90–92.

30. Lieberman MD, Kilburn H, Lindsey M, Brennan MF. (1995) Relation of

perioperative deaths to hospital volume among patients undergoing pan-

creatic resection for malignancy. Ann Surg 222:638–645.

31. Fong Y, Gonen M, Rubin D, Radzyner M, Brennan MF. (2005) Long term

survival is superior after resection for cancers in high-volume centers.

Ann Surg 242:540–544.

496 HPB


32. Birkmeyer JD, Siewers AE, Finlayson EV, Stukel TA, Lucas FL, Batista I

et al. (2002) Hospital volume and surgical mortality in the United States.

N Engl J Med 346:1128–1137.

33. Riall TS, Nealon WH, Goodwin JS, Townsend CM Jr, Freeman JL. (2008)

Outcomes following pancreatic resection: variability among high-volume

providers. Surgery 144:133–140.

34. Nathan H, Cameron JL, Choti MA, Schulick RD, Pawlik TM. (2009) The

volume-outcomes effect in hepato-pancreato-biliary surgery: hospital

versus surgeon contributions and specificity of the relationship. J Am Coll

Surg 208:528–538.

35. Meguid RA, Ahuja N, Chang DC. (2008) What constitutes a ‘high volume’

hospital for pancreatic resection? J Am Coll Surg 206:622–631.

36. Schmidt CM, Powell ES, Yiannoutsos CT, Howard TJ, Wiebke EA,

Wiesenauer CA et al. (2004) Pancreaticoduodenectomy: a 20-year

experience in 516 patients. Arch Surg 139:718–725; discussion 725–

727.

37. Rodríguez JR, Germes SS, Pandharipande PV, Gazelle GS, Thayer SP,

Warshaw AL et al. (2006) Implications and cost of pancreatic leak follow-

ing distal pancreatic resection. Arch Surg 141:361–365.

38. Fahy BN, Frey CF, Ho HS, Beckett L, Bold RJ. (2002) Morbidity,

mortality, and technical factors of distal pancreatectomy. Am J Surg

183:237–241.

HPB 497


Pancreatectomy risk calculator: an ACS-NSQIP resource

Documents

Transcript of Pancreatectomy risk calculator: an ACS-NSQIP resource