nature publishing group ORIGINAL CONTRIBUTIONS 435

see related editorial on page 443

CME

PANCREAS
Comparison of BISAP, Ranson’s, APACHE-II, and CTSI
Scores in Predicting Organ Failure, Complications, and
Mortality in Acute Pancreatitis
Georgios I. Papachristou, MD1,2, Venkata Muddana, MD1, Dhiraj Yadav, MD1, Michael O’Connell, PhD1, Michael K. Sanders, MD1,
Adam Slivka, MD, PhD1 and David C. Whitcomb, MD, PhD1,3,4

OBJECTIVES: Identification of patients at risk for severe disease early in the course of acute pancreatitis (AP)
is an important step to guiding management and improving outcomes. A new prognostic scoring
system, the bedside index for severity in AP (BISAP), has been proposed as an accurate method
for early identification of patients at risk for in-hospital mortality. The aim of this study was to
compare BISAP (blood urea nitrogen >25 mg/dl, impaired mental status, systemic inflammatory
response syndrome (SIRS), age > 60 years, and pleural effusions) with the “traditional”
multifactorial scoring systems: Ranson’s, Acute Physiology and Chronic Health Examination
(APACHE)-II, and computed tomography severity index (CTSI) in predicting severity, pancreatic
necrosis (PNec), and mortality in a prospective cohort of patients with AP.

METHODS: Extensive demographic, radiographic, and laboratory data from consecutive patients with AP
admitted or transferred to our institution was collected between June 2003 and September
2007. The BISAP and APACHE-II scores were calculated using data from the first 24 h from
admission. Predictive accuracy of the scoring systems was measured by the area under the
receiver-operating curve (AUC).

RESULTS: There were 185 patients with AP (mean age 51.7, 51% males), of which 73% underwent contrast-
enhanced CT scan. Forty patients developed organ failure and were classified as severe AP (SAP;
22%). Thirty-six developed PNec (19%), and 7 died (mortality 3.8%). The number of patients with a
BISAP score of ≥3 was 26; Ranson’s ≥3 was 47, APACHE-II ≥8 was 66, and CTSI ≥3 was 59. Of the
seven patients that died, one had a BISAP score of 1, two had a score of 2, and four had a score of
3. AUCs for BISAP, Ranson’s, APACHE-II, and CTSI in predicting SAP are 0.81 (confidence interval
(CI) 0.74–0.87), 0.94 (CI 0.89–0.97), 0.78 (CI 0.71–0.84), and 0.84 (CI 0.76–0.89), respectively.

CONCLUSIONS: We confirmed that the BISAP score is an accurate means for risk stratification in patients with
AP. Its components are clinically relevant and easy to obtain. The prognostic accuracy of BISAP is
similar to those of the other scoring systems. We conclude that simple scoring systems may have
reached their maximal utility and novel models are needed to further improve predictive accuracy.

Am J Gastroenterol 2010; 105:435–441; doi:10.1038/ajg.2009.622; published online 27 October 2009

INTRODUCTION hospital admissions per annum in the United States (1–4). It is a

Acute pancreatitis (AP) is a common disorder with substan- complex process in which pancreatic enzyme activation causes
tial burden on the health-care system. AP accounts for 210,000 local pancreatic damage, resulting in an acute inflammatory

1
Division of Gastroenterology, Hepatology and Nutrition, Department of Medicine, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania, USA;
2
Division of Gastroenterology, Department of Medicine, Veterans Affairs Pittsburgh Health System, Pittsburgh, Pennsylvania, USA; 3Department of Cell Biology
& Physiology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA; 4Department of Human Genetics, University of Pittsburgh School
of Medicine, Pittsburgh, Pennsylvania, USA. Correspondence: Georgios I. Papachristou, MD, Department of Medicine, UPMC Presbyterian, M2 C Wing, 200
Lothrop Street, Pittsburgh, Pennsylvania 15213, USA. E-mail: papachri@pitt.edu
Received 21 February 2009; accepted 24 June 2009

© 2010 by the American College of Gastroenterology The American Journal of GASTROENTEROLOGY

 436 Papachristou et al.
response. The clinical course of AP is usually mild and often
resolves without sequelae. Nonetheless, between 10 and 20%
of patients experience a severe AP (SAP) attack, resulting in
PANCREAS
an intense inflammatory response, a variety of local and sys-
temic complications, a prolonged hospital course, significant
morbidity and mortality (5–8). In such patients, the acute
inflammatory response may progress to systemic inflamma-
tory response syndrome (SIRS), multiorgan failure, and/or
pancreatic necrosis (PNec). However, the individual patient
response to pancreatic injury is highly variable and often
unpredictable. Clinical biomarkers that predict the develop-
ment of these life-threatening complications are important
to help guide patient triage and management. However, no
simple test or group of tests have proven to perform signifi-
cantly better in clinical settings than good clinical judgment
(1), and the available scoring systems are used primarily for
comparing clinical research studies.
The Ranson’s (9) score represented a major advance in evalu-
ating the severity of AP and has been used for over three dec-
ades to assess AP severity. The Ranson’s score is moderately
accurate in classifying patients in terms of severity, but has the
disadvantage of requiring a full 48 h to be completed, missing a
potentially valuable early therapeutic window (10). In addition,
it contains data not routinely ordered or collected during hos-
pitalization at the current time (eg, lactate dehydrogenase, fluid
sequestration, and base deficit). In the United States, the most
commonly used prediction scoring system for clinical research
studies in AP is the Acute Physiology and Chronic Health
Examination (APACHE)-II (11,12). The APACHE-II score is
as accurate as Ranson’s score and can be administered on any
day. However, as a tool that was originally designed to predict
intensive care unit survival, APACHE-II requires the collection
of a large number of parameters, some of which may not be
relevant to AP prognosis, whereas other important measures,
such as pancreatic injury and significant regional complications
are missed. Balthazar et al. (13) developed a grading system
based on contrast-enhanced CT scan (CECT) findings. This
prognostic score uses CT scan features of AP and PNec, allow-
ing the calculation of a CT severity index (CTSI). The CTSI,
however, is based on local complications and has the drawback
of not reflecting the systemic inflammatory response. Other
scoring systems have also been evaluated, but most of them are
modifications of the three systems outlined above.
Recently, a new prognostic scoring system, the bedside index
for severity in AP (BISAP), has been proposed as an accurate
method for early identification of patients at risk for in-hospital
mortality. The BISAP uses five points: urea nitrogen (BUN)
>25 mg/dl, impaired mental status by evidence of disorienta-
tion or disturbance in mental status, presence of the SIRS, age
>60 years, and pleural effusions (14,15). SIRS is defined by the
presence of ≥2 of the following criteria: pulse >90 beats/min,
respirations >20 per min, or PaCO2 < 32 mm Hg, temperature
>100.4 °F or < 96.8 °F and white blood cell count >12,000
or < 4,000 cells per mm3 or >10% immature neutrophils
(bands) (16,17).
The American Journal of GASTROENTEROLOGY
The BISAP score was retrospectively derived and validated
based on a large population data set (Cardinal Health Clini-
cal Outcomes Research Database, Cardinal Health, Marl-
borough, MA, USA) (14). A prospective validation of BISAP
as a prognostic scoring system in AP was recently published
(15), concluding that it is indeed a reliable and accurate means
of stratifying patients with AP for clinical care and research.
The aim of this study is (1) to evaluate the BISAP score in an
independent prospectively enrolled cohort of patients from
a different geographic area, and (2) to compare its accuracy
to the “traditional” multifactorial scoring systems: Ranson’s,
APACHE-II, and CTSI in predicting disease severity, PNec,
and mortality.
METHODS
The Severity of Acute Pancreatitis Study protocol was approved
by the Institutional Review Board of the University of Pitts-
burgh. Enrollment was conducted at two tertiary care hospitals
of the University of Pittsburgh Medical Center. Informed con-
sent was obtained from all patients or appropriate surrogates
before study enrollment. The diagnosis of AP was based on the
presence of two of the following three features: (i) abdominal
pain characteristic of AP, (ii) serum amylase and/or lipase ≥3
times the upper limit of normal, and (iii) characteristic find-
ings of AP on abdominal CT scan.
Extensive demographic, radiographic, and laboratory data
from consecutive patients with AP admitted or transferred
to our institution was collected. On rare occasions not all
the laboratory values or vital signs were available in patients
included in the study who were transferred to our institution
from outside hospitals. No points were added to the scoring
systems for missing values. The BISAP and APACHE-II scores
were calculated using data from the first 24 h from admission
and the Ranson’s score using data from the first 48 h. CTSI was
calculated in patients who underwent CECT within 48 h from
admission.
Patients were classified as mild AP or severe AP, based on
the presence of organ failure for more than 48 h. Organ failure
included shock (systolic blood pressure < 90 mm Hg), pulmo-
nary insufficiency (arterial PO2 < 60 mm Hg at room air or the
need for mechanical ventilation), or renal failure (serum cre-
atinine level >2 mg/dl after rehydration or hemodialysis). PNec
was assessed by CECT. Evidence of PNec on CT was defined as
lack of enhancement of pancreatic parenchyma with contrast
(18). All CT scans were reviewed by radiologists dedicated to
abdominal imaging, who were blinded to laboratory data and
clinical course.
Statistics
Descriptive data are presented as median and interquartile
range for continuous variables. Categorical data are pres-
ented as proportions. Bivariate relationship for categorical
variables was assessed using odds ratios (OR) calculated
based on Pearson’s χ2-test. The distribution of severity, PNec,
VOLUME 105 | FEBRUARY 2010 www.amjgastro.com

and mortality by the BISAP point score was assessed using
the Cochran–Armitage trend test. Sensitivity, specificity,
positive predictive value (PPV), and negative predictive
value (NPV) were calculated for individual scoring systems.
Receiver-operating characteristic curves for SAP, PNec, and
mortality were calculated for BISAP, Ranson’s, APACHE-
II, and CTSI scores using cutoff values, and the predictive
accuracy of each scoring system was measured by the area
under the receiver-operating curve (AUC). Pairwise AUC
comparisons were also performed between two scoring sys-
tems at a time using the nonparametric approach developed
by DeLong et al. (19).
RESULTS
Patient characteristics
A total of 185 patients with AP were prospectively enrolled
between June 2003 and September 2007. All patients were
recruited within 24 h from the time of admission or transfer
(18). Of these, 83 (44.8%) patients were initially admitted at
outside hospitals and subsequently transferred to our institu-
tion. Data within 24 h from the initial admission were collected
in all transferred patients and used for the calculation of the
BISAP and APACHE-II scores. Only 3% of the laboratory val-
ues or vital signs required were not available in these patients.
The median age was 52 years (interquartile range 37–66;
range 15–90), with 51% men and 87% white. The etiolo-
gies of AP included biliary (36%), idiopathic (27%), alcohol
(14%), post-endoscopic retrograde cholangiopancreatogra-
phy (14%), hypertriglyceridemia (4%), and others (5%). One
hundred-thirty five (73%) patients underwent CECT early in
their hospitalization. Forty patients (22%) developed persist-
ent organ failure for ≥48 h and were classified as SAP (multi-
organ failure 48%). Thirty-six patients (19%) had evidence of
PNec on CECT. The median length of stay was 7 days (inter-
quartile range 4–16; range 1–105). Seven patients died during
hospitalization (mortality 3.8%).
BISAP score
The proportion of subjects with severe disease, PNec, and mor-
tality stratified by the BISAP point score is presented in Table 1.
Table 1. Number of patients and their proportion with SAP, PNec, and mortality stratified by the BISAP point score
BISAP score Number of patients
0 64
1 59
2 36
3 18
4 8
5 0
BISAP, bedside index for severity in acute pancreatitis; PNec, pancreatic necrosis; SAP, severe AP.
Cochran–Armitage trend test P values for disease severity, PNec, and mortality were < 0.001, < 0.001, and < 0.01, respectively.
© 2010 by the American College of Gastroenterology
Comparison of BISAP, Ranson’s, APACHE, and CTSI Scores 437
A significant trend for disease severity, PNec, and mortality
was seen with increasing BISAP score.
PANCREAS
Comparison of scoring systems in predicting SAP, PNec, and
mortality
Receiver-operating characteristic curves yielded an AUC of
0.81 (95% confidence intervals (CI) 0.74–0.87) for BISAP
in predicting disease severity, 0.78 (CI 0.69–0.85) in pre-
dicting the development of PNec and 0.82 (CI 0.67–0.91)
in predicting mortality. AUCs for each scoring system in
predicting SAP, PNec, and mortality are shown in Table 2.
Ranson’s score showed a slightly higher accuracy for pre-
dicting SAP 0.94 (CI 0.89–0.97) and mortality 0.95 (CI
0.90–0.98). CTSI, as expected, was the most accurate in
predicting PNec, as evidence of PNec is based on CT scan
readings, with an AUC of 0.98 (CI 0.94–1.00). Pairwise
comparison revealed an AUC of 0.83 (CI 0.76–0.88) for
BISAP score and 0.93 (CI 0.88–0.97) for Ranson’s score in
predicting severe disease with a small overlap between the
95% CI (Figure 1a). Pairwise comparison of BISAP with
APACHE-II score revealed similar accuracy in predicting
SAP (AUC 0.83 (CI 0.76–0.88) and 0.78 (CI 0.68–0.85),
respectively; Figure 1b).
On the basis of highest sensitivity and specificity values
generated from the receiver-operating characteristic curves,
the following cutoffs were selected for further analysis:
BISAP score ≥3, Ranson’s ≥3, APACHE-II ≥8, and CTSI ≥3.
The observed incidence of severe disease, PNec, and mortal-
ity stratified by the above BISAP, Ranson’s, APACHE-II, and
CTSI cutoffs with the corresponding OR are seen in Table 3.
The number of patients with a BISAP score of ≥3 was 26,
Ranson’s ≥3 was 47, APACHE-II ≥8 was 66, and CTSI ≥3 was
59. On χ2-test, patients with a BISAP score ≥3 had a 7-fold
higher likelihood of developing SAP (OR 7.3), five times for
PNec (OR 4.8) and 10-fold higher likelihood of death (OR
9.5). In regards to mortality, all seven patients who died had
a Ranson’s score ≥3 and APACHE-II score ≥8. On the other
hand, one patient who died had a BISAP score of 1, two
patients had a score of 2, four patients had a BISAP score of
3, and no such patients had a score of 4 or 5. Using the above
cutoffs, the sensitivity, specificity, PPV, and NPV of different
SAP PNec Mortality
1 (1.6%) 3 (4.7%) 0 (0%)
10 (16.9%) 6 (10.2%) 1 (1.7%)
14 (38.9%) 15 (41.7%) 2 (5.6%)
11 (61.1%) 7 (38.9%) 4 (22.2%)
4 (50%) 5 (62.5%) 0 (0%)
NA NA NA
The American Journal of GASTROENTEROLOGY
 438 Papachristou et al.

Table 2. AUC of different scoring systems in predicting SAP, PNec, and mortality

AUC (95% CI) Severity PNec Mortality

PANCREAS

BISAP 0.81 (0.74 – 0.87) 0.78 (0.69 – 0.85) 0.82 (0.67 – 0.91)
Ranson’s 0.94 (0.89 – 0.97) 0.85 (0.79 – 0.90) 0.95 (0.90 – 0.98)
APACHE-II 0.78 (0.71 – 0.84) 0.72 (0.64 – 0.78) 0.94 (0.89 – 0.97)
CTSI 0.84 (0.76 – 0.89) 0.98 (0.94 – 1.00) 0.83 (0.75 – 0.89)
APACHE-II, Acute Physiology and Chronic Health Examination-II; AUC, area under the receiver-operating curve; BISAP, bedside index for severity in acute pancreatitis;
CTSI, computed tomography severity index; PNec, pancreatic necrosis; SAP, severe AP.

a ROC curve of severity b ROC curve of severity

1.00 1.00

Criteria Criteria
0.75 BISAP 0.75 BISAP
Ranson’s APACHE-II

Sensitivity
Sensitivity

0.50 0.50

0.25 0.25

0.00 0.00
0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00
1–Specificity 1–Specificity

Figure 1. Pairwise AUC comparison of BISAP with Ranson’s and APACHE-II scores in predicting severe disease. (a) Pairwise area under the receiver-
operating curve (AUC) comparison of bedside index for severity in acute pancreatitis (BISAP) and Ranson’s scores in predicting severe AP (SAP).
(b) Pairwise AUC comparison of BISAP and Acute Physiology and Chronic Health Examination (APACHE)-II scores in predicting SAP.

Table 3. Incidence of SAP, PNec, and mortality stratified by BISAP score, Ranson’s, APACHE-II, and CTSI with corresponding OR

Number of patients SAP PNec Mortality

BISAP
≤2 159 25 (15.7%) 24 (15.1%) 3 (1.9%)
>3 26 15 (57.7%) 12 (46.2%) 4 (15.4%)
OR (CI) 7.3 (2.7–19.6) 4.8 (1.8 –12.7) 9.5 (1.5 – 67.4)
Ranson’s
≤2 131 6 (4.6%) 7 (5.3%) 0 (0%)
>3 47 32 (68.0%) 24 (51.1%) 7 (14.9%)
OR (CI) 44.4 (14.7–146.5) 18.5 (6.6 – 55.7) 49.2 (4.0 – 244.7)
APACHE-II
<7 112 11 (9.8%) 11 (9.8%) 0 (0%)
>8 66 26 (39.4%) 19 (28.8%) 7 (10.6%)
OR (CI) 6.0 (2.5 –14.6) 3.7 (1.5 – 9.3) 28.6 (2.3 –141.8)
CTSI
<2 76 5 (6.6%) 1 (1.3%) 0 (0%)
>3 59 30 (50.8%) 35 (59.3%) 5 (8.5%)
OR (CI) 14.7 (4.9 – 52.1) 72.9 (13.4 – 397.7) 15.4 (1.2 –73.4)
APACHE-II, Acute Physiology and Chronic Health Examination-II; BISAP, bedside index for severity in acute pancreatitis; CI, confidence interval; CTSI, computed
tomography severity index; OR, odds ratio; PNec, pancreatic necrosis; SAP, severe AP.

The American Journal of GASTROENTEROLOGY VOLUME 105 | FEBRUARY 2010 www.amjgastro.com

 Comparison of BISAP, Ranson’s, APACHE, and CTSI Scores 439

Table 4. Sensitivity, specificity, PPV, and NPV of different scoring systems in predicting SAP, PNec, and mortality

% Sensitivity (95% CI) Specificity (95% CI) PPV (95% CI) NPV (95% CI)

PANCREAS
Severity
BISAP 37.5 (24.2 – 53.0) 92.4 (86.9 – 95.7) 57.7 (38.9 –74.5) 84.3 (77.8 – 89.1)
Ranson’s 84.2 (69.6 – 92.6) 89.8 (83.6 – 93.8) 69.6 (55.2 – 80.9) 95.3 (90.2– 97.9)
APACHE-II 70.3 (54.2 – 82.5) 71.9 (64.0 –78.7) 40.0 (29.0 – 52.1) 90.1 (83.1– 94.4)
CTSI 85.7 (70.6 – 93.7) 71.0 (61.5 –79.0) 50.8 (34.4 – 63.2) 93.4 (85.5 – 97.2)
PNec
BISAP 33.3 (20.2 – 49.7) 90.6 (84.8 – 94.3) 46.2 (28.8 – 64.5) 84.9 (78.5 – 89.6)
Ranson’s 77.4 (60.2 – 88.6) 88.4 (82.0 – 92.7) 52.2 (38.1– 65.9) 94.6 (89.2– 97.3)
APACHE-II 63.3 (45.5 –78.1) 68.5 (60.6 –75.5) 29.2 (19.6 – 41.2) 90.1 (83.1– 94.4)
CTSI 97.2 (85.8 – 99.5) 75.8 (66.5 – 83.1) 59.3 (46.6 –70.9) 98.7 (92.9 – 99.8)
Mortality
BISAP 57.1 (25.0 – 84.2) 87.6 (82.0 – 91.7) 15.4 (6.2– 33.5) 98.1 (94.6 – 99.4)
Ranson’s 100 (64.6 –100) 76.8 (69.8 – 82.5) 15.2 (7.6 – 23.2) 100 (97.1–100)
APACHE-II 100 (64.6 –100) 65.7 (58.2 – 72.4) 10.8 (5.3 – 20.6) 100 (96.7–100)
CTSI 100 (56.6 –100) 58.5 (49.9 – 66.6) 8.5 (3.7–18.4) 100 (95.2 –100)
APACHE-II, Acute Physiology and Chronic Health Examination-II; BISAP, bedside index for severity in acute pancreatitis; CI, confidence interval; CTSI, computed
tomography severity index; NPV, negative predictive value; PNec, pancreatic necrosis; PPV, positive predictive value; SAP, severe AP.

scoring systems in predicting SAP, PNec, and mortality are
seen in Table 4.
DISCUSSION
In this study, we compared the accuracy of three repre-
sentative prognostic multifactorial scoring systems and the
new BISAP system in a prospectively collected cohort of
patients with AP. We confirmed that the newly proposed
BISAP index is a reliable means of stratifying patients with
AP within 24 h from admission. In our cohort, the BISAP
score performed similar to the three “traditional” clinical
scoring systems.
This study was conducted at two tertiary care hospitals of the
University of Pittsburgh. The overall mortality in our cohort
was 3.8%, and 15.4% of the patients had a BISAP score ≥3. Our
cohort was appropriate for comparing BISAP point-score strat-
ification and the mortality rates with the tertiary care single-
center study performed at the Brigham and Woman’s Medical
Center in Boston (mortality 3.5% and BISAP ≥3: 14.4%) (15).
As expected, the proportion of patients with severe disease in
our cohort was higher than the initial population-based study
(mortality 1.2% and BISAP ≥3: 9.9%), which used data col-
lected from both community hospitals and tertiary care centers
(14). Because only a very small number of patients died in our
cohort (n = 7), all results related to mortality should be evalu-
ated with caution.
BISAP is a newly developed prognostic scoring system,
which was reported to perform well in preliminary studies,
especially in the early identification of patients with AP
at increased risk of in-hospital mortality (14). BISAP uses
findings of physical examination, vital signs, routine lab-
oratory data, and imaging findings to derive a five-point
score (15). It has been proposed that the primary advantage
of BISAP to the “traditional” scoring systems is simplicity.
However, our experience from this study is that the calcu-
lation of the BISAP score is more complicated than previ-
ously suggested. Even though, this is a five-point index,
SIRS calculation require the collection of multiple variables
(see Introduction); making it an eight-variable system to
calculate five points.
Ranson’s score is composed of 11 measures that are recorded
as binary values on admission and at 48 h, and its primary aim
was to evaluate the function of early operative intervention in
patients with AP (9). A composite score of 3 or more is com-
monly used to classify a patient as having severe disease. An
analysis of the components of Ranson’s score reveals that it is
weighted toward detecting multiorgan failure linked to sys-
temic inflammatory response and vascular leak syndromes.
Another commonly used severity index is the APACHE-II
index (11,12). This clinical tool measures the physiological
response to injury- and inflammation-driven stress and was
initially designed to predict prolonged intensive care unit
treatment and mortality.
BISAP has the advantage over Ranson’s score of being cal-
culated within 24 h of admission. BISAP appears to be more
heavily weighted toward the immune response to injury and
older age (>60 vs. >55 years with higher likelihood of elderly

© 2010 by the American College of Gastroenterology The American Journal of GASTROENTEROLOGY

 440 Papachristou et al.
being confused or disoriented), whereas Ranson’s scoring
system seems to perform with higher accuracy in the predic-
tion of persistent organ dysfunction over 48 h. Thus, BISAP
PANCREAS
may be disadvantaged in that it cannot easily distinguish
transient organ dysfunction from persistent organ dysfunc-
tion at 24 h (sensitivity of 38%, specificity 92%, PPV 58%,
and NPV 84%), with the latter group suffering nearly all of
the morbidity and morality of AP (20). On the other hand,
BISAP score may be useful in the triage of patients to closer
observation and intensive care on initial evaluation rather
than being used to assess persistent organ failure and its
consequences.
In this study, the Ranson’s score performed well (sensitivity
of 84%, specificity 90%, PPV 70%, and NPV 95%), as we used
a definition of SAP based on presence of organ dysfunction for
at least 48 h. In contrast, Ranson’s score has performed with
moderate accuracy in previous studies, especially when sever-
ity was defined by the Atlanta Criteria, which also include local
pancreatic injury in the definition of severity. A meta-analysis
encompassing 1,300 patients reported that Ranson’s score has
an overall sensitivity of 75%, specificity 77%, PPV of 49%, and
NPV of 91% (3). However, the low PPV of Ranson’s score in
these studies may only reflect that half of the patients with a
score ≥3 did not meet the definition of severe disease that was
chosen for patient classification. On the other hand, inclusion
of similar measures for the calculation of Ranson’s score and
the definition of severity, as in our study, may bias the results
favorably compared to other scoring systems in terms of per-
formance characteristics. On stepwise comparison, APACHE-
II score performed similar to BISAP, as reported in previous
studies (14).
All the above multifactorial clinical scoring systems have
been very helpful in evaluating the severity of AP and start-
ing with Ranson’s score, and have been used for over three
decades to assess AP severity. However, the overall disadvan-
tage of such scoring systems, including the BISAP, is that they
are not designed to predict potentially preventable compli-
cations in AP, and are least useful in the middle prediction
range where the clinician needs most guidance. Therefore,
their use seems to be confined in medical decision-making
at the extreme of the prediction range, such as triaging inten-
sive care unit admission (20,21), and as enrollment criteria for
clinical trials.
AP is a dynamic and evolving process that involves multi-
ple systems and the risk for multiple organ complications.
Despite the complex and highly variable nature of AP, the
evaluation of the dynamic steps involved and the prediction
of future outcomes have followed the simplistic approach
of Ranson et al. (9,10) in dividing all patients with AP into
one of two categories—mild and severe. The Atlanta classifi-
cation has not significantly contributed to our understand-
ing of this dynamic process by classifying patients as mild or
severe, based on a wide variety of criteria and end points (22).
To improve the predictive accuracy of clinical tools, series of
The American Journal of GASTROENTEROLOGY
biomarkers, which measure each of the potentially pathologic
steps and are linked to specific end points and outcomes, may
be required. Additional structured analysis may be needed to
determine the effect of interacting pathways and the impact of
multiple factors in accurately predicting specific outcomes on
a patient-by-patient basis. Currently, all of the major scoring
systems, including BISAP, use threshold values to change con-
tinuous variables into binary values, and use the binary values
as equally weighted “points” to calculate a score. This second
derivative is then used to classify an episode of pancreatitis as
mild or severe, again based on selection of a cutoff value and
dependent on the definition of “severe”. The limitations of the
first step are shown by the use of artificial neural networks that
outperform Ranson’s score, APACHE-II, and similar systems
using the same data sets (23). The limitations of the second
step may be shown by the current analysis, in which it appears
that Ranson’s score outperforms BISAP based on our defini-
tion of severity.
In conclusion, BISAP is a reliable prognostic tool to classify
patients with AP into mild and severe groups, and its com-
ponents are clinically relevant and easy to obtain. However,
when used to assess persistent organ dysfunction, PNec, and
mortality, the BISAP index was not found to be either sim-
pler or more accurate than the existing multifactorial scoring
systems in our cohort. The limitation of all scoring systems
discussed here may be that they convert continuous variables
into binary values of equal weight, and fail to capture syn-
ergistic or multiplicative effects based on the interactions of
interdependent systems. Future research could focus on com-
prehensive reassessment of the pathologic mechanisms of AP
with attention to the effects of preexisting risk factors (eg, age,
obesity, genetic factors) and well-defined end points, identifi-
cation of accurate biomarkers to assess activity on these path-
ways, and mathematical models that have strong predictive
accuracy.
CONFLICT OF INTEREST
Guarantor of the article: Georgios I. Papachristou, MD.
Specific author contributions: Georgios I. Papachristou:
conception and design, acquisition of data, analysis
and interpretation of data, drafting of the paper, and obtain-
ing funding; Venkata Muddana: acquisition and analysis of
data and drafting of the paper; Dhiraj Yadav: analysis and
interpretation of data and drafting of the paper; Michael
O’Connell: statistical analysis; Michael K. Sanders: interpreta-
tion of data and drafting of the paper; Adam Slivka: critical
revision of the paper for important intellectual content and
supervision; David C. Whitcomb: critical revision of the
paper for important intellectual content, obtaining funding,
and supervision.
Financial support: This work was supported by the VA Stars
and Stripes Healthcare Network 2007 Competitive Pilot
Project Fund (GIP).
Potential competing interests: None.
VOLUME 105 | FEBRUARY 2010 www.amjgastro.com

Study Highlights
WHAT IS CURRENT KNOWLEDGE
3 8. Banks PA, Freeman ML. Practice guidelines in acute pancreatitis. Am J
The individual patient response to pancreatic injury is
highly variable and often unpredictable.
3Traditional prognostic clinical scoring systems, such as
Ranson’s and Acute Physiology and Chronic Health
Examination (APACHE)-II, have been helpful in evaluating
the severity of acute pancreatitis (AP) and have been
used for three decades.
3A new prognostic scoring system, the bedside index
for severity in AP (BISAP), has been proposed as an
accurate method for early identification of patients at
risk for in-hospital mortality.
WHAT IS NEW HERE
3We confirmed that the BISAP score is an accurate
means for risk stratification in patients with AP. Its com-
ponents are clinically relevant and easy to obtain.
3The prognostic accuracy of BISAP is similar to those of
the other scoring systems.
3Clinical scoring systems may have reached their
maximal utility and novel models are needed to further
improve early predictive accuracy in patients with AP.
REFERENCES
1. Papachristou GI, Clermont G, Sharma A et al. Risk and markers of severe
acute pancreatitis. Gastroenterol Clin North Am 2007;36:277–96, viii.
2. Swaroop VS, Chari ST, Clain JE. Severe acute pancreatitis. JAMA
2004;291:2865–8.
3. Forsmark CE, Baillie J. AGA Institute technical review on acute pancreati-
tis. Gastroenterology 2007;132:2022–44.
4. Hughes SJ, Papachristou GI, Federle MP et al. Necrotizing pancreatitis.
Gastroenterol Clin North Am 2007;36:313–23, viii.
5. Isenmann R, Beger HG. Natural history of acute pancreatitis and the role of
infection. Baillieres Best Pract Res Clin Gastroenterol 1999;13:291–301.
© 2010 by the American College of Gastroenterology
Comparison of BISAP, Ranson’s, APACHE, and CTSI Scores 441
6. Whitcomb DC. Clinical practice. Acute pancreatitis. N Engl J Med
2006;354:2142–50.
7. Fagenholz PJ, Castillo CF, Harris NS et al. Increasing United States hospital
admissions for acute pancreatitis, 1988–2003. Ann Epidemiol 2007;17:491–7.
PANCREAS
Gastroenterol 2006;101:2379–400.
9. Ranson JH, Rifkind KM, Roses DF et al. Objective early identification of
severe acute pancreatitis. Am J Gastroenterol 1974;61:443–51.
10. Ranson JH, Pasternack BS. Statistical methods for quantifying the severity
of clinical acute pancreatitis. J Surg Res 1977;22:79–91.
11. Yeung YP, Lam BY, Yip AW. APACHE system is better than Ranson system
in the prediction of severity of acute pancreatitis. Hepatobiliary Pancreat
Dis Int 2006;5:294–9.
12. Larvin M, McMahon MJ. APACHE-II score for assessment and monitoring
of acute pancreatitis. Lancet 1989;2:201–5.
13. Balthazar EJ, Roinson DL, Megibow AJ et al. Acute pancreatitis: value of CT
in establishing prognosis. Radiology 1990;174:331–6.
14. Wu BU, Johannes RS, Sun X et al. The early prediction of mortality in acute
pancreatitis: a large population-based study. Gut 2008;57:1698–703.
15. Singh VK, Wu BU, Bollen TL et al. A prospective evaluation of the bedside
index for severity in acute pancreatitis score in assessing mortality and
intermediate markers of severity in acute pancreatitis. Am J Gastroenterol
2009;104:966–71.
16. Mofidi R, Duff MD, Wigmore SJ et al. Association between early systemic
inflammatory response, severity of multiorgan dysfunction and death in
acute pancreatitis. Br J Surg 2006;93:738–44.
17. Buter A, Imrie CW, Carter CR et al. Dynamic nature of early organ dysfunc-
tion determines outcome in acute pancreatitis. Br J Surg 2002;89:298–302.
18. Papachristou GI, Papachristou DJ, Avula H et al. Obesity increases the
severity of acute pancreatitis: performance of APACHE-O score and cor-
relation with the inflammatory response. Pancreatology 2006;6:279–85.
19. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under
two or more correlated receiver operating characteristic curves: a nonpara-
metric approach. Biometrics 1988;44:837–45.
20. Zimmerman JE, Rousseau DM, Duffy J et al. Intensive care at two
teaching hospitals: an organizational case study. Am J Crit Care
1994;3:129–38.
21. Papachristou GI. Prediction of severe acute pancreatitis: current knowledge
and novel insights. World J Gastroenterol 2008;14:6273–5.
22. Bradley EL III. A clinically based classification system for acute pancreatitis.
Arch Surg 1993;128:586–90.
23. Atkinson AJJ, Colburn WA, DeGruttola VG et al. Biomarkers and surrogate
endpoints: preferred definitions and conceptual framework. Clin Pharmacol
Ther 2001;69:89–95.
The American Journal of GASTROENTEROLOGY