Accepted Manuscript
Risk of and Factors Associated with Readmission After a Sentinel Attack of AcutePancreatitis
Kishore Vipperla, Georgios I. Papachristou, Jeffrey Easler, Venkata Muddana,Adam Slivka, David C. Whitcomb, Dhiraj Yadav
PII: S1542-3565(14)00670-3DOI: 10.1016/j.cgh.2014.04.035Reference: YJCGH 53816
To appear in: Clinical Gastroenterology and HepatologyAccepted Date: 29 April 2014
Please cite this article as: Vipperla K, Papachristou GI, Easler J, Muddana V, Slivka A, Whitcomb DC,Yadav D, Risk of and Factors Associated with Readmission After a Sentinel Attack of Acute Pancreatitis,Clinical Gastroenterology and Hepatology (2014), doi: 10.1016/j.cgh.2014.04.035.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service toour customers we are providing this early version of the manuscript. The manuscript will undergocopyediting, typesetting, and review of the resulting proof before it is published in its final form. Pleasenote that during the production process errors may be discovered which could affect the content, and alllegal disclaimers that apply to the journal pertain.
All studies published in Clinical Gastroenterology and Hepatology are embargoed until 3PM ET ofthe day they are published as corrected proofs on-line. Studies cannot be publicized as acceptedmanuscripts or uncorrected proofs.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
CGH-D-14-00062 - Revision 2 Title: Risk of and Factors Associated with Readmission After a Sentinel Attack of Acute Pancreatitis Short Title: Readmissions after sentinel acute pancreatitis 1Kishore Vipperla, 1Georgios I. Papachristou, 1Jeffrey Easler, 1Venkata Muddana, 1Adam Slivka, 1David C. Whitcomb, 1Dhiraj Yadav Drs. Yadav and Papachristou co-directed this project 1Division of Gastroenterology & Hepatology, University of Pittsburgh Medical Center, Pittsburgh, PA Word Count: Abstract with keywords– 327 (approved by Dr Novak); Text (with table and figure legends including footnotes and references) - 3990 Conflict of Interest: The authors report no conflicts relevant to this manuscript. Funding: None. Acknowledgement: Presented in part at the Digestive Diseases Week 2013 and published in abstract form in Gastroenterology 2013;144(5):Suppl 1, Page S-272-273. Address for Correspondence: Dhiraj Yadav MD MPH Associate Professor Division of Gastroenterology & Hepatology University of Pittsburgh Medical Center 200 Lothrop Street, M2, C-wing Pittsburgh, PA 15213 Tel: 412 648 9825 Fax: 412 383 8992 e-mail: [email protected] Authorship criteria and contributions: Kishore Vipperla: Study design, data collection, data interpretation, revising the article, final approval of the version to be published. Jeffrey Easler, Venkata Muddana: Data collection, data interpretation, revising the article, final approval of the version to be published. Adam Slivka, David C. Whitcomb: Data interpretation, revising the article, final approval of the version to be published. Georgios I. Papachristou, MD: Study design and supervision, data interpretation, revising the article, final approval of the version to be published. Dhiraj Yadav: Study design and supervision, data analysis and interpretation, drafting and revising the article, final approval of the version to be published.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
Abstract:
Background & Aims: Few data are available on how many patients are readmitted to the
hospital after attacks of acute pancreatitis. We aimed to determine the risk and factors that
determine early (within 30 days) and late (after 30 days) readmission of patients with acute
pancreatitis.
Methods: In a retrospectively study, we collected and analyzed data on 127 surviving patients
(median age, 53 years; 52% male; 83% white) hospitalized at the University of Pittsburgh
Medical Center for a sentinel attack of acute pancreatitis, enrolled in the Severe Acute
Pancreatitis Study from June 2003 through April 2010, and had follow up data. Information was
collected on demographics, clinical profile, risk score at discharge (based on a recently
developed scoring system), and details of readmissions during follow up (median 36 months).
Results: Of the 127 patients, 52% were transfers from another care center and 32% required
admission to the intensive care unit. Etiologies for pancreatitis were biliary (47%), idiopathic
(13%), alcohol associated (12%), and others (28%). Pancreatic necrosis (28%), persistent organ
failure (27%), and peripancreatic fluid collections (19%) were common. The median length of
stay was 9 days. A total of 108 readmissions occurred, for 43 patients (34%). Early readmissions
(n=21) occurred more frequently for patients with smoldering (ongoing) symptoms or local
complications than for those without. Late readmissions (n=22) occurred more frequently for
patients with recurrent pancreatitis than for those without. Male sex, alcohol-associated disease,
and severe disease increased the risks of readmission and recurrence. Risk for readmission was
lower among non-transferred patients (23%) and patients without necrosis or organ failure
(16%). Risk for readmission increased with the number of points on the weighted scoring
system.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
Conclusions: About one third of patients hospitalized for acute pancreatitis are readmitted,
usually due to smoldering symptoms, local complications, or recurrent attacks. Studies are
needed to determine whether individualized discharge planning, with consideration of the
etiology of acute pancreatitis, can reduce the risk for readmission.
KEY WORDS: pancreas, inflammation, prognostic factor, outcome
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
Background:
With implementation of the Affordable Care Act, significant changes are expected to
occur over the next few years in the way healthcare is delivered in the United States. An
important emphasis of the new health care law is to provide high quality care at the lowest
possible cost. Inpatient care accounts for the bulk of health care costs, and readmissions after
hospitalization are common. Therefore, identification of the risk and reasons for readmissions,
and addressing them can potentially reduce hospitalizations and healthcare costs.
Acute Pancreatitis (AP) is the most common pancreas disorder, requiring hospitalization
in almost all patients1. In fact, AP is now the leading cause of gastrointestinal tract related
hospital admissions in the United States2. However, there is paucity of data on the risk of and
reasons for readmissions after an episode of AP. Focused studies have evaluated the functional
consequences after severe AP3-7. Two single center studies evaluated the risk of readmissions in
all AP patients, and developed a weighted scoring system to identify patients at high-risk of
readmission8, 9.
We recently analyzed population-based data from Allegheny County, Pennsylvania in
patients hospitalized for first-attack of AP from 1996-200510, 11. We found the overall risk of
readmissions to be 22%. Readmissions were common in younger patients and those who were
male, black and had alcohol etiology. Due to limitations of administrative data, we were unable
to know the precise reasons for readmissions. Since we determined the risk of admissions only
on a yearly basis, we were unable to know the risk and reasons of early and late readmissions.
The aim of our current study was to evaluate the risk and determinants of readmission
after a sentinel attack of AP in a well-established, prospectively ascertained cohort of AP
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
patients. We also evaluated the performance of the weighted scoring system recently proposed to
determine the risk of readmissions in AP patients8, 9.
Methods
Severe Acute Pancreatitis Study (SAPS)
SAPS is an ongoing prospective observational study at the University of Pittsburgh
Medical Center (UPMC) designed to study the role of patient, diseaseand genetic factors that
affect the severity of AP. Data from this study on the role of clinical scoring systems in disease
severity has been published12, 13. The study protocol is approved by the Institutional Review
Board of the University of Pittsburgh and a written informed consent was obtained from all
patients prior to study enrollment.
Patient cohort
Between June 2003-April 2010, in three one-year periods, SAPS enrolled 256 AP
patients of varying severity from two tertiary care hospitals of the UPMC system (UPMC-
Presbyterian, UPMC-Shadyside). All patients were enrolled within 24 hours of admission or
transfer from another institution(s) after obtaining an informed consent from patients or their
designated surrogates. AP diagnosis was based on presence of two of the following three
features: abdominal pain characteristic of AP, serum pancreatic enzyme elevation to ≥3 times the
upper limit of normal, and characteristic findings of AP on abdominal computed tomography
(CT) scan.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
Since the present study focuses on readmissions, we limited the study cohort to patients
who were admitted with their first AP attack, had no prior history of CP, survived the index
admission and had data available on follow up.
Index admission
Data during index admission was collected prospectively and included detailed
information on demographic, clinical, laboratory and radiographic parameters. For patients
transferred to our institution, pertinent information from outside records generated prior to
admission was obtained by contacting the transferring center. Clinical and laboratory parameters
were used to determine a variety of severity scores (Ranson, Acute Physiology and Chronic
Health Evaluation II [APACHE-II], Bedside index of severity in AP [BISAP]). Organ failure
was defined by the presence of shock (systolic blood pressure <90mmHg), pulmonary
insufficiency (arterial PO2 <60mmHg at room air or the need for mechanical ventilation), or
renal failure (serum creatinine level >2 mg/dl after rehydration or hemodialysis). Data on
treatment and interventions for either pancreatic necrosis or peripancreatic fluid collections were
noted.
We retrospectively noted information at the time of discharge from index admission for:
symptoms (nausea, vomiting or diarrhea), diet (solids, less than solids), pain (none, narcotics),
surgical drains (yes/no) and antibiotics (yes/no). We noted presence of abnormal vital signs
within 24 hours prior to discharge and defined these by the worst readings on any of the
following: fever ≥102°F, tachycardia ≥100 beats/min, tachypnea ≥20 breaths/min, hypertension
or hypotension – systolic pressure ≥140 mm Hg or <100 mm Hg, or diastolic pressure ≥90 mm
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
Hg or <60 mm Hg, hypoxia - oxygen saturation <95% or need for supplemental oxygen to
maintain saturation ≥95%).
Readmissions
Information on emergency room visits with/without inpatient admissions at our
institution and treatment received during subsequent care was collected retrospectively. For
patients who had an inpatient admission, we determined the timing, reason and the number of
readmissions (see below). We noted whether and how many additional attacks of AP (recurrent
AP, RAP) a patient had during the follow up period.
Early vs. late readmission
For the purposes of this study, we classified readmissions as early (i.e. occurring within
30 days of discharge) and late (i.e. occurring after 30 days of discharge). The reason for
admission was classified as related to - smoldering (ongoing) symptoms of AP (pain, inability to
tolerate oral diet), complications of AP (i.e. pancreatic necrosis, peripancreatic fluid collections),
surgical complications (drain issues, wound infection etc.), nasojejunal feeding tube issues
(clogging, dislodgement) and/or RAP.
We used the scoring system proposed by Whitlock et al to determine the risk of early
readmission9. This scoring system stratifies patients into three risk categories based on the total
number of points present: low (0-1 points), moderate (2-3 points) or high (≥4 points). The five
risk factors included in the scoring system include - discharge on less than a solid diet (3 points),
gastrointestinal symptoms including nausea, vomiting, or diarrhea at discharge (3 points),
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
pancreatic necrosis (2 points), use of antibiotics at discharge (2 points) and pain at discharge (1
point).
Data Analysis
Descriptive information is presented as proportions for categorical data and mean ±
standard deviation (SD) or median and interquartile range (IQR) for continuous data as
applicable. Information on readmissions was used to determine absolute risk. Bivariate
comparisons for categorical data were performed using chi-squared test or proportional trend
test, and for continuous data using Mann-Whitney U test. All p-values <0.05 were considered
significant. Data were analyzed using SPSSv19, SPSS Inc. Chicago, Ill, USA.
Results
Study cohort and baseline characteristics
Of the 256 AP patients in the SAPS study, 162 were admitted for first-attack of AP. Of
these, 10 patients died during the index admission, and 25 had no follow up information
available. The final sample size for the current study was therefore 127 patients who survived the
index admission and had data available for follow up.
Table 1 shows data on demographics, etiology, severity parameters and local
complications in patients who formed the final study cohort. The median age was 54 years, 52%
were male, 83% were white, and 52% were transfers from other institutions. The most common
etiologies were biliary in 59 (47%), idiopathic in 17 (13%), post-ERCP in 17 (13%) and alcohol
in 15 (12%) patients. Other etiologies included medications in 6 patients, hypertriglyceridemia in
5, medications in 6, hereditary in 2, malignancy/neoplasm in 5, hereditary in 2, and
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
hypercalcemia in 1patient. Overall, ICU care was needed in 33% and organ failure developed in
27% patients (single 15%; multiple 12%). During the course of AP, pancreatic necrosis
with/without peripancreatic fluid collections was observed in 28% and pancreatic/peripancreatic
fluid collections without necrosis in 19% patients. The median length of stay was 9 days.
Among patients with biliary pancreatitis, 30/59 (50.8%) patients underwent an
endoscopic retrograde cholangiopancreatography. Of these, 28 had biliary sphincterotomy, 22
had stone extraction, and 6 had stent placement (in two without sphincterotomy or stone
removal).
Risk and burden of readmission
During a median follow up of 36 months (IQR - 9.5, 65), 43/127 (34%) patients had at
least one inpatient admission at our institution (Figure 1). The median time to first readmission
was 1.3 months (IQR 0.5, 4). The risk of readmission was determined by demographic factors,
etiology and disease severity (Table 1). Younger patients, males and those with certain etiologies
(alcohol, idiopathic, hypertriglyceridemia) were at a significantly higher risk of readmission.
Patients with local complications or severe disease were more likely to be readmitted as reflected
by an overrepresentation of patients who were transferred, had an ICU admission, organ failure,
pancreatic necrosis and higher Ranson's score. The length of stay during index admission was
significantly higher in patients who were subsequently readmitted.
The total number of readmissions during the follow up period in these 43 patients was
108: 24 (56%) had one, 6 (14%) had two; 8 (19%) had three, and 5 (12%) had more than 5
admissions (one had 6; two had 7; one had 13, one had 15). The reasons for readmissions are
shown in Figure 1, Tables 2 and 3.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
More than one readmission was seen in 19 patients (15% overall, 44% of patients with
readmissions), and they accounted for 78% (84/108) of all readmissions. The breakdown of these
readmissions were – 38 RAP, 20 smoldering symptoms, 18 complications of AP, 6 surgical
complications, 1 nosocomial/central line associated blood stream infection, and 1 NJ tube
clogging (Figure 1, Table 2).
Demographic factors and etiology determined the risk of multiple readmissions. For e.g.
among 30 males who had readmissions, 16 (53%) were readmitted more than once (vs. 3/13,
23% in females). Similarly, multiple readmissions were more common with alcohol etiology (7/9
[78%] vs. 5/10 [50%] with idiopathic, 43% with other, 4/17 [24%] with biliary). Other
demographic factors and markers of severity did not differ in patients with one or more
readmissions (data not shown). All five patients with more than 3 readmissions were male; two
had alcohol etiology and one each had idiopathic, hereditary and biliary etiology (Table 3).
Twenty-two (17.3%) patients were discharged to a skilled nursing facility or long term
care. As expected, these patients were sicker as reflected by a higher prevalence of ICU
admission (19/22, 90.5%), organ failure (17/22, 77.3%) and pancreatic necrosis (15/22, 68%).
Their readmission rate was significantly higher than patients who were discharged to home
(13/22, 59% vs. 30/105, 28.6%, p=0.012).
Early vs. Late first readmission
Of the 43 patients who were readmitted, about half (21/43, 49%) had their first
readmission within 30 days after discharge from the index visit. The reason for readmission
within 30 days was more likely to be due to smoldering symptoms or AP related local
complications, while late readmissions were more likely due to a new attack of AP.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
Patients with early or late readmission were equally likely to be men, needed ICU admission, and
have multiorgan failure or pancreatic necrosis. When compared with patients with no
readmission, those with early readmission had a higher prevalence of idiopathic etiology, while
those with late readmission were more likely to have alcohol or other etiologies (data not
shown).
RAP
Readmission for at least one attack of RAP during the follow up was seen in 22 patients
(17% overall, 51% among patients with readmissions). The risk of RAP was greater in males
(17/66, 26% vs. 5/61, 8% in females, p=0.01), with alcohol etiology (7/15, 47% vs. with biliary
etiology 7/59, 12%, p=0.006). Patient with pancreatic necrosis had a higher risk of readmission
when compared with patients without pancreatic necrosis (9/36, 25% vs. no necrosis 13/91,
14%), but this did not reach statistical significance (p=0.19).
Cholecystectomy was performed in 47/59 (80%) patients with biliary etiology. RAP
during follow up occurred in 7 of these 47 (15%) patients. Specifically, RAP was seen in 1/7
patients who had a history of prior cholecystectomy (14%), in 1/14 (7%) who underwent
cholecystectomy during the index admission, and in 5/15 (33%) who underwent cholecystectomy
more than four weeks after discharge from index admission. Among 12 patients with biliary
etiology who did not have a cholecystectomy, interestingly, none developed RAP. Of these 12
patients, the status of eventual cholecystectomy was unknown in 4 patients due to short follow
up (7, 7, 12 and 16 weeks), 4 were poor surgical candidates, in 2 the diagnosis of biliary AP was
presumed but not certain, 1 also had a biliary stricture, and 1 patient who developed severe
necrotizing pancreatitis requiring multiple surgical interventions died 14 months before a
cholecystectomy could be performed.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
Relationship of symptom control at discharge with readmissions
Table 4 shows data on the risk of readmission (early or late) based on vital signs,
symptoms, diet, pain control, presence of surgical drains and use of antibiotics. The risk of
readmissions was significantly higher in patients who had unstable vital signs on discharge, were
on less than solid diet, had poor control of symptoms or pain and had a percutaneous drainage
catheter placed.
Information was available to calculate the risk score in 123/127 (96.9%) patients. Of
these 37 (30.1%) were in the low-risk category, 23 (18.7%) in the moderate-risk category and 63
(51.2%) were in the high-risk category. The risk of early readmission increased with the
discharge score (0% [0/37], 13% [3/23] and 29% [18/63] in patients with low-, moderate- and
high-risk discharge scores) (Figure 2). Data was similar when stratified by presence of necrosis
or organ failure and in non-transferred patients (data not shown).
Discussion
Our study provides a descriptive assessment of the risk and patterns of readmission after
AP at a major US tertiary care center, and how they relate to demographic factors, etiology and
disease severity. We found that readmissions after a sentinel attack of AP were common. About
half of the readmissions were related to smoldering (ongoing) symptoms or local complications
of AP, while the remaining were due to recurrent attacks of AP. Readmissions from ongoing
symptoms or local complications usually occur within a few days to weeks from discharge.
Recurrent attacks of AP usually occurred after a month of discharge and were was influenced by
sex and etiology of AP.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
A better understanding of the reasons and timing of readmissions can help to address the
question of whether and which readmissions are potentially preventable and what strategies may
be helpful to reduce readmission risk. We note that the risk of readmissions is higher in patients
with local complications and severe AP. Our data suggests that discharge planning in patients
with AP should be individualized and based on presence of symptoms and disease severity.
Patients with mild AP14 who have ongoing symptoms at the time of discharge, and all patients
with moderate or severe AP14 should have a close follow up with their primary care physician
and/or gastroenterologist, and other specialists depending on individual needs (surgery, pain
management, nutrition). This can help to assess for and address issues related to symptom
control (pain management, nausea, vomiting, bowel disturbances) and to identify local
complications. Patients with severe AP, local or systemic complications (necrosis, peripancreatic
fluid collections, organ failure), and those needing enteral feeding via nasojejunal feeding tubes
should be evaluated for placement in a short term rehabilitation or transitional facility where
appropriate care can be provided until their clinical condition allows for discharge to home.
Over half of all readmissions were due to RAP. The risk of RAP can be reduced by
addressing the etiology of AP, i.e. performing early cholecystectomy in biliary AP15,
triglyceride control16, discontinuing offending medication, and behavior modification17-19.
Patients with pancreatic necrosis tended to have a higher risk of RAP, although this was not
statistically significant, likely due to small sample size. Alcohol increases the risk of pancreatic
necrosis20, and continuing alcohol consumption can be a significant contributor to recurrences in
these patients (this was noted in 5/8 patients with pancreatic necrosis who had RAP and alcohol
etiology in our study). Patients with necrotizing pancreatitis are also at increased risk of
obstructive pancreatitis from development of pancreatic ductal strictures or disconnected duct
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
syndrome due to ductal disruption in the area of necrosis21. Identification of ductal dilatation and
inflammatory changes limited to the disconnected segment can provide a clue to this possibility.
Since CP patients may have frequent exacerbations needing hospital admissions, we only
included patients with sentinel AP in our study to avoid overestimation of RAP.
The overall risk of readmissions in our study was higher than a recent population study
using administrative data10 and in another large study based on chart review9. This may be
explained by a higher proportion of transferred patients with severe AP in our cohort. We
performed a sensitivity analysis after excluding transfers and patients with necrosis or organ
failure. The risk of readmission in patients who were not transferred (23%) and those who did
not have organ failure or necrosis (16%) was much lower and provides an insight into the
readmission risk in a community setting.
We validate the results and scoring system proposed by Whitlock et al9. However, there
were differences in the two cohorts which needs discussion. The proportion of patients in the
high risk category (51 vs. 19%) was higher in our cohort which is explained by a higher
prevalence of severe AP. Secondly, the risk of readmissions in the high risk category in our
cohort was lower (29 vs. 77%) than the Whitlock study. A limitation of both studies is that data
were collected retrospectively by review of patient records. Since two of the five criteria in the
scoring system are subjective (pain, gastrointestinal symptoms), differences may have resulted in
the way individual findings were interpreted and/or documented in clinical records. Finally,
unlike Whitlock et al who defined readmissions as visit to the emergency room on inpatient
admission, we have presented data only on inpatients admissions. Even if we include patients
seen in the emergency room but not admitted (n=2), our results would not be much different.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
Future studies in different populations should evaluate readmission risk and the performance of
this scoring system.
A major limitation of our study is its retrospective design. This could have led to a bias in
detection of readmissions only in patients who received care at our institution. Moreover, 25
patients had no follow-up data available. An ideal approach would be to contact all patients
and/or their physicians. However, given the time lag between enrollment and the conduct of the
current study, difficulty in contacting all patients (change of address, physicians, etc.) or recall
bias (timing, reason, number of admissions) may have resulted in differential ascertainment.
Finally, only a subset of AP patients treated at our institution enrolled in the SAPS study.
Therefore, our results may not be reflective of all AP patients treated at our institution.
When compared with patients in whom follow up was available, patients with no follow-
up data were less likely to be non-white (0/25, 0% vs. 21/105, p=0.025), more likely to be
transfers (19/25, 76% vs. 66/126, 52.4%, p=0.045), and less likely to have necrosis (1/25, 4% vs.
36/127, 28.3%, p=0.009). Extrapolating data from our results, it is likely that the readmission
rate in these patients would be ~15-20%. Having follow up information in these patients would
have resulted in a slight reduction in the overall readmission rate from 34 to 32%.
In summary, readmissions after the sentinel attack of AP are common and usually due to
smoldering (ongoing) symptoms, local complications and RAP. Readmissions from ongoing
symptoms or local complications of AP are common early during the follow up period, while
RAP risk increases with the duration of follow up and its risk is influenced by sex and etiology
of AP. Future studies should evaluate and quantify the benefits of individualized discharge
planning and addressing the etiology of AP on the risk of readmissions and reduction in health
care costs.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
Figure Legends:
Figure 1: Flowchart showing the study population, the timing and reasons for first readmission
Figure 2: Risk of early readmission in sentinel acute pancreatitis patients based on the weighted risk score at discharge proposed by Whitlock et al9
References: 1. Eland IA, Sturkenboom MC, van der Lei J, et al. Incidence of acute pancreatitis. Scand J
Gastroenterol 2002;37:124.
2. Peery AF, Dellon ES, Lund J, et al. Burden of gastrointestinal disease in the United States: 2012
update. Gastroenterology 2012;143:1179-87.
3. Angelini G, Cavallini G, Pederzoli P, et al. Long-term outcome of acute pancreatitis: a prospective
study with 118 patients. Digestion 1993;54:143-7.
4. Appelros S, Lindgren S, Borgstrom A. Short and long term outcome of severe acute pancreatitis.
Eur J Surg 2001;167:281-6.
5. Boreham B, Ammori BJ. A prospective evaluation of pancreatic exocrine function in patients
with acute pancreatitis: correlation with extent of necrosis and pancreatic endocrine
insufficiency. Pancreatology 2003;3:303-8.
6. Gupta R, Wig JD, Bhasin DK, et al. Severe acute pancreatitis: the life after. J Gastrointest Surg
2009;13:1328-36.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
7. Symersky T, van Hoorn B, Masclee AA. The outcome of a long-term follow-up of pancreatic
function after recovery from acute pancreatitis. JOP 2006;7:447-53.
8. Whitlock TL, Repas K, Tignor A, et al. Early readmission in acute pancreatitis: incidence and risk
factors. Am J Gastroenterol 2010;105:2492-7.
9. Whitlock TL, Tignor A, Webster EM, et al. A scoring system to predict readmission of patients
with acute pancreatitis to the hospital within thirty days of discharge. Clin Gastroenterol
Hepatol 2011;9:175-80.
10. Yadav D, Lee E, Papachristou GI, et al. A population based evaluation of readmissions after first
hospitalization for Acute Pancreatitis. Pancreas 2014 (In Press).
11. Yadav D, O'Connell M, Papachristou GI. Natural history following the first attack of acute
pancreatitis. Am J Gastroenterol 2012;107:1096-103.
12. Mounzer R, Langmead CJ, Wu BU, et al. Comparison of existing clinical scoring systems to
predict persistent organ failure in patients with acute pancreatitis. Gastroenterology
2012;142:1476-82.
13. Papachristou GI, Muddana V, Yadav D, et al. Comparison of BISAP, Ranson's, APACHE-II, and CTSI
scores in predicting organ failure, complications, and mortality in acute pancreatitis. Am J
Gastroenterol 2010;105:435-41.
14. Banks PA, Bollen TL, Dervenis C, et al. Classification of acute pancreatitis--2012: revision of the
Atlanta classification and definitions by international consensus. Gut 2013;62:102-11.
15. van Baal MC, Besselink MG, Bakker OJ, et al. Timing of cholecystectomy after mild biliary
pancreatitis: a systematic review. Ann Surg 2012;255:860-6.
16. Christian JB, Arondekar B, Buysman EK, et al. Clinical and economic benefits observed when
follow-up triglyceride levels are less than 500 mg/dL in patients with severe
hypertriglyceridemia. J Clin Lipidol 2012;6:450-61.
17. Nikkola J, Raty S, Laukkarinen J, et al. Abstinence after first acute alcohol-associated pancreatitis
protects against recurrent pancreatitis and minimizes the risk of pancreatic dysfunction. Alcohol
Alcohol 2013;48:483-6.
18. Nordback I, Pelli H, Lappalainen-Lehto R, et al. The recurrence of acute alcohol-associated
pancreatitis can be reduced: a randomized controlled trial. Gastroenterology 2009;136:848-55.
19. Sadr-Azodi O, Andren-Sandberg A, Orsini N, et al. Cigarette smoking, smoking cessation and
acute pancreatitis: a prospective population-based study. Gut 2012;61:262-7.
20. Papachristou GI, Papachristou DJ, Morinville VD, et al. Chronic alcohol consumption is a major
risk factor for pancreatic necrosis in acute pancreatitis. Am J Gastroenterol 2006;101:2605-10.
21. Pelaez-Luna M, Vege SS, Petersen BT, et al. Disconnected pancreatic duct syndrome in severe
acute pancreatitis: clinical and imaging characteristics and outcomes in a cohort of 31 cases.
Gastrointest Endosc 2008;68:91-7.
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
Table 1: Demographics, etiology and disease severity in acute pancreatitis patients, overall and based on readmission
Variable Groups All patients (n = 127)
Readmission p-value No
(n = 84) Yes
(n = 43) Age (years) – median (IQR) - 54 (38, 68) 57 (41, 72) 50 (36, 63) 0.04 Sex – n (%) Male
Female 66 (52) 61 (48)
36 (55) 48 (79)
30 (45) 13 (21)
0.005
Race – n (%)
White Other
105 (83) 22 (17)
68 (65) 16 (73)
37 (35) 6 (27)
0.62
BMI – median (IQR) - 29 (26, 33) 29 (25, 33) 30 (27, 33) 0.25 Etiology – n (%)
Biliary Alcohol
Idiopathic Others
59 (47) 15 (12) 17 (13) 36 (28)
42 (71) 6 (40) 7 (41) 29 (81)
17 (29) 9 (60) 10 (59) 7 (19)
0.004
Transfers – n (%)*
No Yes
60 (48) 66 (52)
46 (77) 38 (58)
14 (23) 28 (42)
0.025
ICU admission – n (%)*
No Yes
88 (67) 39 (33)
62 (70) 18 (46)
17 (30) 21 (54)
0.001
SIRS – n (%)* No Yes
72 (58) 51 (42)
53 (74) 30 (59)
19 (26) 21 (41)
0.12
BISAP (≥3) – n (%)
No Yes
114 (90) 13 (10)
77 (68) 7 (54)
37 (32) 6 (46)
0.36
APACHE II score (≥8) admission – n (%)*
No Yes
74 (62) 45 (38)
50 (68) 31 (69)
24 (32) 14 (31)
1.0
APACHE II score (≥8) 48 hours – n(%)*
No Yes
70 (62) 43 (38)
52 (74) 25 (58)
18 (26) 18 (41)
0.10
Ranson’s score (≥3) – n (%)*
No Yes
71 (61) 45 (39)
55 (77) 23 (51)
16 (23) 22 (49)
0.004
Persistent organ failure – n (%)
None Single
Multiorgan
93 (73) 19 (15) 15 (12)
69 (74) 10 (53) 5 (33)
24 (27) 9 (47) 10 (67)
0.003
Local complications – n (%) None PFCs
Necrosis +/- PFCs
57 (53) 24 (19) 36 (28)
56 (84) 19 (79) 9 (25)
11 (16) 5 (21) 27 (75)
<0.001
Length of stay – median (IQR)
- 9 (5, 19) 7 (5, 13) 20 (9, 36) <0.001
Percentages for all patients are column proportions and for readmission (yes/no) are row proportions; *missing data; IQR: interquartile range; SIRS defined as a score of ≥2. Readmissions for other etiologies: post-ERCP (0/17), hypertriglyceridemia (3/5); medications (2/6); hereditary (0/2), malignancy/neoplasm (0/5), hypercalcemia (0/1)
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
Table 2: Reasons for readmission after sentinel attack of AP in 24 patients who had only one readmission
Reason for readmission Patients with only one readmission (n=24)
Smoldering symptoms 3 Local complications (pancreatic necrosis, peripancreatic fluid collections)
7
Surgical complications (drain, wound infection, etc.) 1
Nasojejunal tube problems 1 Hypoxia, pleural effusion 1 Recurrent AP 11
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
Table 3: Reasons for readmissions after sentinel attack of AP in 19 patients with multiple readmissions
Patient Sex Etiology Reason for readmission
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 M Hereditary RAP RAP RAP RAP RAP RAP RAP RAP RAP RAP RAP RAP RAP RA
P RAP
2 M Alcohol RAP RAP RAP RAP Sx Control
RAP RAP Sx Control
Sx Control
Sx Control
RAP RAP Sx Control
3 M Idiopathic Sx Control
Sx Control
Sx Control
Sx Control
Sx Control
Sx Control
Sx Control
4 M Alcohol AP comp
AP comp AP comp RAP Sx Control
Sx Control
RAP
5 M Biliary AP comp
AP comp AP comp Sx Control
RAP AP comp
6 M Alcohol AP comp
Surgical comp
AP comp
7 M Idiopathic AP comp
AP comp Surgical comp
8 M Alcohol Sx Control
AP comp Sx Control
9 M Idiopathic RAP RAP RAP 10 M Idiopathic Surgical
comp Surgical
comp Surgical
comp 11 F Biliary AP
comp AP comp Sx
Control 12 M HTG AP
comp Clogged NJ tube
AP comp
13 M Alcohol Surgical comp
Sx Control
RAP
14 F Alcohol RAP RAP 15 F Biliary RAP RAP 16 M Biliary RAP AP comp 17 M Idiopathic RAP RAP 18 M Drugs AP
comp Nosoco-mial inf
19 M Alcohol RAP Sx Control
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
AP Comp: complications of AP (Pseudocyst, necrosis, infection); HTG- Hypertriglyceridemia; Nosocomial Inf: nosocomial infection not related to pancreatitis; Sx Control: smoldering symptoms of AP for symptom control; Surgical comp: surgical complications (drain problems, wound infection, etc.)
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
Table 4: Risk of readmission based on discharge status, presence or necrosis or organ failure and transfer status
Discharge condition N (%)
All Patients (n=127)
No necrosis or organ failure (n=80)
No Transfer (n=60)
Readmission 43 (33.8)
p-value Readmission 13 (16.2)
p-value Readmission 14 (23.3)
p-value
Diet Solids Less than solid
4/41 (9.8)
39/82 (47.6)
<0.001
2/35 (5.7)
11/41 (26.8)
0.017
2/26 (7.7)
12/31 (38.7)
0.012
Symptoms None/Well controlled Poor controlled
28/95 (29.5) 15/29 (51.7)
0.044
10/64 (15.6) 3/13 (23.1)
0.684
10/50 (20) 4/7 (57.1)
0.054
Pain medications None/Good control Poor control/short-term narcotics
13/55 (23.6) 30/69 (43.5)
0.024
7/43 (16.3) 6/34 (17.6)
1.0
7/35 (20)
7/22 (31.8)
0.36
Antibiotics No Yes
36/111 (32.4)
7/13 (70)
0.14
13/74 (17.6)
0/3 (0)
1.0
14/55 (25.5)
0/2 (0)
1.0
Drain No Yes
36/114 (31.6)
7/10 (70)
0.032
13/77 (16.9)
-
-
14/57 (24.6)
-
-
Abnormal vital signs No Yes
19/72 (26.4) 18/31 (38.1)
0.003
7/52 (13.5) 2/7 (28.6)
0.29
7/39 (17.9)
3/5 (60)
0.069
*missing values for variables on discharge: diet (4); symptoms (3); pain (4); drain (3); antibiotics (3); vital signs (24)
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT
MANUSCRIP
T
ACCEPTED
ACCEPTED MANUSCRIPT