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: yadavd@upmc.edu 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