1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54
PAN383_proof ■ 6 April 2015 ■ 1/6
Pancreatology xxx (2015) 1e6
Contents lists available at ScienceDirect
Pancreatology journal homepage: www.elsevier.com/locate/pan
Original article Q8
Severity and natural history of acute pancreatitis in diabetic patients*
Q7
Haq Nawaz, Michael O'Connell, Georgios I. Papachristou, Dhiraj Yadav* University of Pittsburgh Medical Center, Pittsburgh, PA, USA
Q2
a r t i c l e i n f o
a b s t r a c t
Article history: Available online xxx
Background: There is limited data on the prevalence of coexistent diabetes in acute pancreatitis and subsequent natural history in these patients. Methods: Using Pennsylvania Health Care Cost Containment Council data set, we identified 7399 unique White and Black Allegheny County, PA residents with first hospitalization for acute pancreatitis from 1996 to 2005. We evaluated the prevalence of coexistent diabetes, demographic and etiologic profile, severity of index hospitalization, and risk of readmission for acute or chronic pancreatitis during follow up (median 39 months) in this cohort. Results: The prevalence of coexisting diabetes was 18% (1349/7399). When compared with non-diabetics, diabetics were significantly more likely to be older (63 vs. 56 yrs), male (OR 1.4, 95% CI 1.2e1.6), black (OR 2.4, 95% CI 2.1e2.7) and have non-alcoholic etiologies (biliary, OR 1.5, 95% CI 1.2e1.9; idiopathic, OR 2.0, 95% CI 1.7e2.5; metabolic, OR 5.2, 95% CI 4.0e6.7). While diabetic patients had a significantly longer length of stay (median 5 vs. 4 days, p < 0.05), their severity of acute pancreatitis (multivariable OR 1.18, 95% CI 0.94e1.48) or in-hospital mortality (1.9% each, p ¼ 0.98) did not differ than non-diabetics. The overall risk of pancreatitis-related readmissions in diabetics was similar (33% each, p ¼ 0.99), but their risk of subsequent admission for chronic pancreatitis (multivariable HR 0.65, 95% CI 0.44e0.97) was lower than non-diabetics. Conclusion: About 20% patients with acute pancreatitis have coexistent diabetes. Demographic and etiologic profile of diabetic patients with acute pancreatitis differs from non-diabetics. Diabetic status has limited effect on the severity of and natural history following acute pancreatitis. Copyright © 2015, IAP and EPC. Published by Elsevier India, a division of Reed Elsevier India Pvt. Ltd. All rights reserved.
Keywords: Diabetes Pancreatitis Hospitalization
Introduction Acute pancreatitis (AP) is an inflammatory condition resulting from pancreatic injury from various causes [1]. Clinically, patients with AP present with sudden onset of usually severe upper abdominal pain sometimes radiating to the back, elevations in serum pancreatic enzymes and morphologic changes in the pancreas or peripancreatic area [2]. Almost all patients with a clinically significant episode of AP require hospitalization [3]. The
Q1
* An abstract based on this data will be submitted for the Digestive Disorders Meeting 2015. * Corresponding author. Division of Gastroenterology & Hepatology, University of Pittsburgh Medical Center, 200 Lothrop Street, M2, C-wing, Pittsburgh, 15213, PA, USA. Tel.: þ1 412 648 9825; fax: þ1 412 383 8992. E-mail address: [email protected] (D. Yadav).
clinical course of AP can vary from mild to severe and several host factors may affect its severity [4]. The relationship between AP and diabetes has generated interest in recent years. In population-based studies, diabetes is noted to increase the risk of AP by 1.5e3 folds [5e8]. This excess risk can be reduced with better blood sugar control [6,8]. Patients with AP can also develop diabetes. In a recent systematic review of 24 studies, the pooled prevalence of prediabetes and diabetes after a sentinel episode of AP was 16% and 23% respectively, and 15% patients needed insulin treatment [9]. While parenchymal destruction from necrosis may explain the development of diabetes in severe AP, diabetes can also develop in AP patients with mild disease by yet to be defined mechanisms. There is limited data on the prevalence of coexistent diabetes and the subsequent natural history in these patients. Diabetic patients tend to have a higher number of co-morbidities, some of which may be associated with AP etiology (e.g. hypertriglyceridemia
http://dx.doi.org/10.1016/j.pan.2015.03.013 1424-3903/Copyright © 2015, IAP and EPC. Published by Elsevier India, a division of Reed Elsevier India Pvt. Ltd. All rights reserved.
Please cite this article in press as: Nawaz H, et al., Severity and natural history of acute pancreatitis in diabetic patients, Pancreatology (2015), http://dx.doi.org/10.1016/j.pan.2015.03.013
55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
PAN383_proof ■ 6 April 2015 ■ 2/6
2
H. Nawaz et al. / Pancreatology xxx (2015) 1e6
[10]) or severity (e.g. obesity [11]). It is unclear if the etiologic profile of AP in patients with coexisting diabetes is similar to non-diabetic patients. Mortality in patients with AP increases with the number of co-morbidities [12], but published data on the effect of coexistent diabetes on the severity of or mortality from AP is limited. Shen et al. reported an increased risk of severe AP (intensive care unit admission or local complications defined by drainage for pancreatic abscess or cysts), similar risk of organ failure and a lower mortality in patients with coexistent diabetes when compared with nondiabetics [13]. Frey et al. found that presence of diabetes increased the risk of organ failure but did not affect mortality in the first two weeks in AP patients [13]. Other studies not limited to AP have observed that diabetic status does not affect disease course in patients with sepsis [14,15] and diabetic patients admitted to the ICU have a lower risk of death when compared with non-diabetics [16]. Lastly, whether diabetic status affects the risk of recurrent acute (RAP) or chronic (CP) pancreatitis is unclear. We have previously reported on the natural history [17], risk and patterns of readmissions after sentinel AP [18] and hospitalizations for CP [19] in Allegheny County, PA population. In this cohort, we were able to evaluate the prevalence of coexistent diabetes and the natural history after sentinel AP based on diabetic status with the specific aims to determine the e prevalence of coexistent diabetes in patients with first attack of AP, demographic and etiologic profile and severity of AP in patients with coexistent diabetes, and natural history after AP in patients with coexistent diabetes. Methods This study was approved by the Institutional Review Board of the University of Pittsburgh Medical Center.
analyses. We combined patients with rheumatic diseases, postERCP and post-surgical etiologies into “other etiology” group. Comparison groups for etiology therefore consisted alcohol, biliary, alcohol plus biliary, idiopathic, metabolic and others. Due to small sample size and uncertainty about the definite cause, the alcohol plus biliary group were not included in analyses focusing on etiologies. AP severity We used associated diagnoses and procedure codes to determine organ failure, length of hospital stay and in-hospital mortality as markers of disease severity. Organ failure was specified as acute renal failure (ICD-9 diagnosis code 584 or 586 or procedure code 29.95 without concurrent diagnosis code 585), pulmonary (ICD-9 diagnosis code 518.81 or 518.82 or 518.85 or 786.09 or 799.1 or procedure codes 96.70e96.72), cardiovascular (ICD-9 diagnosis code 458.0 or 458.8 or 458.9 or 785.5 or 785.51 or 785.59 or 796.3). While we were able to identify the presence of organ failure, the duration of organ failure could not be determined. We also used the MediQual Atlas Admission Severity, available in PHC4 dataset for a defined number of diagnosis related groups or principal diagnoses, including AP to define severity. It is calculated by MediQual using information on demographic and clinical variables collected by individual facility and predicts in-hospital mortality at the time of admission. It has been used to assess the clinical severity of illness when comparing outcomes at different hospitals [20,21]. The probability of in-patient mortality increases with group assignment based on clinical instability (group 0 [None]: 0e0.1%; 1 [Minimal]: 0.2e1.1%; 2 [Moderate]: 1.2e5.7%; 3 [Severe]: 5.8e49.4%; 4 [Maximal]: 50e100%). Risk of readmissions for RAP or CP during follow up period
Data source Data source, ascertainment of cohort and methodology for the current study is similar to our prior published studies [17e19]. Briefly, data for this study was obtained from the Pennsylvania Health Care Cost Containment Council (PHC4) (www.phc4.org). PHC4 is an independent state agency which collects detailed information on all inpatient hospitalizations (from 1990 onwards), and select ambulatory procedures at freestanding clinics and hospital outpatient departments (from 1996 onwards) in Pennsylvania (>4 million records/year). PHC4 de-identifies the data and makes it available for purchase to interested parties as standardized datasets for each year. Linking data files over time permits longitudinal evaluation at the level of unique subjects and data can be customized to study diseases or events of interest. Study cohort, diabetic status and etiology of AP In the PHC4 dataset, we identified all unique White or Black Allegheny County residents who received first-time primary inpatient diagnosis of AP (ICD-9 CM code 577.0) from years 1996e2005. Patients with any prior diagnosis of AP or CP (ICD-9 577.1) were excluded. We used associated diagnoses codes for diabetes or diabetes related complications (ICD-9 250.0e250.9) to identify patients with coexisting diabetes. Associated diagnosis codes were also used in a hierarchical algorithm to assign patients into one of nine etiology groups: pancreatic cancer, rheumatic diseases (e.g. lupus), post-ERCP, both biliary and alcohol, biliary, alcohol, metabolic (hyperlipidemia, hyperparathyroidism), post-surgical and idiopathic [17]. Since patients with pancreatic cancer have a high prevalence of diabetes, often of short duration, and can be related to pancreatic cancer itself, we excluded these patients from
Follow up period was defined as the time between discharge from index hospitalization and final PHC4 data entry consisting of hospitalization for any reason or an ambulatory procedure reportable to PHC4 until the second quarter of 2007. We identified if a patient had subsequent hospitalization for primary or any (primary or secondary) diagnosis of AP (RAP) or CP during follow up and the timing after index hospitalization that such admission occurred. This information was used to determine and compare the risk of readmissions for RAP and CP during follow up based on coexisting diabetic status. 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. Bivariate comparison for categorical data was performed using chisquared test or the proportional trend test, and for continuous data using ManneWhitney U test. Multivariable logistic regression analysis was used to determine independent predictors for coexistent diabetes, performance of cholecystectomy and organ failure using age, race, sex, etiology, tobacco abuse and diabetic status as covariates. We did not use MediQual Severity group assignment as a covariate since it is calculated using a combination of demographic and clinic variables and would result in collinearity. Multivariable cox regression analysis was used to determine predictors of RAP and CP during follow up using age, sex, race, etiology, tobacco abuse, diabetic status and RAP (for prediction of CP) as covariates. A p-value of
PAN383_proof ■ 6 April 2015 ■ 1/6
Pancreatology xxx (2015) 1e6
Contents lists available at ScienceDirect
Pancreatology journal homepage: www.elsevier.com/locate/pan
Original article Q8
Severity and natural history of acute pancreatitis in diabetic patients*
Q7
Haq Nawaz, Michael O'Connell, Georgios I. Papachristou, Dhiraj Yadav* University of Pittsburgh Medical Center, Pittsburgh, PA, USA
Q2
a r t i c l e i n f o
a b s t r a c t
Article history: Available online xxx
Background: There is limited data on the prevalence of coexistent diabetes in acute pancreatitis and subsequent natural history in these patients. Methods: Using Pennsylvania Health Care Cost Containment Council data set, we identified 7399 unique White and Black Allegheny County, PA residents with first hospitalization for acute pancreatitis from 1996 to 2005. We evaluated the prevalence of coexistent diabetes, demographic and etiologic profile, severity of index hospitalization, and risk of readmission for acute or chronic pancreatitis during follow up (median 39 months) in this cohort. Results: The prevalence of coexisting diabetes was 18% (1349/7399). When compared with non-diabetics, diabetics were significantly more likely to be older (63 vs. 56 yrs), male (OR 1.4, 95% CI 1.2e1.6), black (OR 2.4, 95% CI 2.1e2.7) and have non-alcoholic etiologies (biliary, OR 1.5, 95% CI 1.2e1.9; idiopathic, OR 2.0, 95% CI 1.7e2.5; metabolic, OR 5.2, 95% CI 4.0e6.7). While diabetic patients had a significantly longer length of stay (median 5 vs. 4 days, p < 0.05), their severity of acute pancreatitis (multivariable OR 1.18, 95% CI 0.94e1.48) or in-hospital mortality (1.9% each, p ¼ 0.98) did not differ than non-diabetics. The overall risk of pancreatitis-related readmissions in diabetics was similar (33% each, p ¼ 0.99), but their risk of subsequent admission for chronic pancreatitis (multivariable HR 0.65, 95% CI 0.44e0.97) was lower than non-diabetics. Conclusion: About 20% patients with acute pancreatitis have coexistent diabetes. Demographic and etiologic profile of diabetic patients with acute pancreatitis differs from non-diabetics. Diabetic status has limited effect on the severity of and natural history following acute pancreatitis. Copyright © 2015, IAP and EPC. Published by Elsevier India, a division of Reed Elsevier India Pvt. Ltd. All rights reserved.
Keywords: Diabetes Pancreatitis Hospitalization
Introduction Acute pancreatitis (AP) is an inflammatory condition resulting from pancreatic injury from various causes [1]. Clinically, patients with AP present with sudden onset of usually severe upper abdominal pain sometimes radiating to the back, elevations in serum pancreatic enzymes and morphologic changes in the pancreas or peripancreatic area [2]. Almost all patients with a clinically significant episode of AP require hospitalization [3]. The
Q1
* An abstract based on this data will be submitted for the Digestive Disorders Meeting 2015. * Corresponding author. Division of Gastroenterology & Hepatology, University of Pittsburgh Medical Center, 200 Lothrop Street, M2, C-wing, Pittsburgh, 15213, PA, USA. Tel.: þ1 412 648 9825; fax: þ1 412 383 8992. E-mail address: [email protected] (D. Yadav).
clinical course of AP can vary from mild to severe and several host factors may affect its severity [4]. The relationship between AP and diabetes has generated interest in recent years. In population-based studies, diabetes is noted to increase the risk of AP by 1.5e3 folds [5e8]. This excess risk can be reduced with better blood sugar control [6,8]. Patients with AP can also develop diabetes. In a recent systematic review of 24 studies, the pooled prevalence of prediabetes and diabetes after a sentinel episode of AP was 16% and 23% respectively, and 15% patients needed insulin treatment [9]. While parenchymal destruction from necrosis may explain the development of diabetes in severe AP, diabetes can also develop in AP patients with mild disease by yet to be defined mechanisms. There is limited data on the prevalence of coexistent diabetes and the subsequent natural history in these patients. Diabetic patients tend to have a higher number of co-morbidities, some of which may be associated with AP etiology (e.g. hypertriglyceridemia
http://dx.doi.org/10.1016/j.pan.2015.03.013 1424-3903/Copyright © 2015, IAP and EPC. Published by Elsevier India, a division of Reed Elsevier India Pvt. Ltd. All rights reserved.
Please cite this article in press as: Nawaz H, et al., Severity and natural history of acute pancreatitis in diabetic patients, Pancreatology (2015), http://dx.doi.org/10.1016/j.pan.2015.03.013
55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
PAN383_proof ■ 6 April 2015 ■ 2/6
2
H. Nawaz et al. / Pancreatology xxx (2015) 1e6
[10]) or severity (e.g. obesity [11]). It is unclear if the etiologic profile of AP in patients with coexisting diabetes is similar to non-diabetic patients. Mortality in patients with AP increases with the number of co-morbidities [12], but published data on the effect of coexistent diabetes on the severity of or mortality from AP is limited. Shen et al. reported an increased risk of severe AP (intensive care unit admission or local complications defined by drainage for pancreatic abscess or cysts), similar risk of organ failure and a lower mortality in patients with coexistent diabetes when compared with nondiabetics [13]. Frey et al. found that presence of diabetes increased the risk of organ failure but did not affect mortality in the first two weeks in AP patients [13]. Other studies not limited to AP have observed that diabetic status does not affect disease course in patients with sepsis [14,15] and diabetic patients admitted to the ICU have a lower risk of death when compared with non-diabetics [16]. Lastly, whether diabetic status affects the risk of recurrent acute (RAP) or chronic (CP) pancreatitis is unclear. We have previously reported on the natural history [17], risk and patterns of readmissions after sentinel AP [18] and hospitalizations for CP [19] in Allegheny County, PA population. In this cohort, we were able to evaluate the prevalence of coexistent diabetes and the natural history after sentinel AP based on diabetic status with the specific aims to determine the e prevalence of coexistent diabetes in patients with first attack of AP, demographic and etiologic profile and severity of AP in patients with coexistent diabetes, and natural history after AP in patients with coexistent diabetes. Methods This study was approved by the Institutional Review Board of the University of Pittsburgh Medical Center.
analyses. We combined patients with rheumatic diseases, postERCP and post-surgical etiologies into “other etiology” group. Comparison groups for etiology therefore consisted alcohol, biliary, alcohol plus biliary, idiopathic, metabolic and others. Due to small sample size and uncertainty about the definite cause, the alcohol plus biliary group were not included in analyses focusing on etiologies. AP severity We used associated diagnoses and procedure codes to determine organ failure, length of hospital stay and in-hospital mortality as markers of disease severity. Organ failure was specified as acute renal failure (ICD-9 diagnosis code 584 or 586 or procedure code 29.95 without concurrent diagnosis code 585), pulmonary (ICD-9 diagnosis code 518.81 or 518.82 or 518.85 or 786.09 or 799.1 or procedure codes 96.70e96.72), cardiovascular (ICD-9 diagnosis code 458.0 or 458.8 or 458.9 or 785.5 or 785.51 or 785.59 or 796.3). While we were able to identify the presence of organ failure, the duration of organ failure could not be determined. We also used the MediQual Atlas Admission Severity, available in PHC4 dataset for a defined number of diagnosis related groups or principal diagnoses, including AP to define severity. It is calculated by MediQual using information on demographic and clinical variables collected by individual facility and predicts in-hospital mortality at the time of admission. It has been used to assess the clinical severity of illness when comparing outcomes at different hospitals [20,21]. The probability of in-patient mortality increases with group assignment based on clinical instability (group 0 [None]: 0e0.1%; 1 [Minimal]: 0.2e1.1%; 2 [Moderate]: 1.2e5.7%; 3 [Severe]: 5.8e49.4%; 4 [Maximal]: 50e100%). Risk of readmissions for RAP or CP during follow up period
Data source Data source, ascertainment of cohort and methodology for the current study is similar to our prior published studies [17e19]. Briefly, data for this study was obtained from the Pennsylvania Health Care Cost Containment Council (PHC4) (www.phc4.org). PHC4 is an independent state agency which collects detailed information on all inpatient hospitalizations (from 1990 onwards), and select ambulatory procedures at freestanding clinics and hospital outpatient departments (from 1996 onwards) in Pennsylvania (>4 million records/year). PHC4 de-identifies the data and makes it available for purchase to interested parties as standardized datasets for each year. Linking data files over time permits longitudinal evaluation at the level of unique subjects and data can be customized to study diseases or events of interest. Study cohort, diabetic status and etiology of AP In the PHC4 dataset, we identified all unique White or Black Allegheny County residents who received first-time primary inpatient diagnosis of AP (ICD-9 CM code 577.0) from years 1996e2005. Patients with any prior diagnosis of AP or CP (ICD-9 577.1) were excluded. We used associated diagnoses codes for diabetes or diabetes related complications (ICD-9 250.0e250.9) to identify patients with coexisting diabetes. Associated diagnosis codes were also used in a hierarchical algorithm to assign patients into one of nine etiology groups: pancreatic cancer, rheumatic diseases (e.g. lupus), post-ERCP, both biliary and alcohol, biliary, alcohol, metabolic (hyperlipidemia, hyperparathyroidism), post-surgical and idiopathic [17]. Since patients with pancreatic cancer have a high prevalence of diabetes, often of short duration, and can be related to pancreatic cancer itself, we excluded these patients from
Follow up period was defined as the time between discharge from index hospitalization and final PHC4 data entry consisting of hospitalization for any reason or an ambulatory procedure reportable to PHC4 until the second quarter of 2007. We identified if a patient had subsequent hospitalization for primary or any (primary or secondary) diagnosis of AP (RAP) or CP during follow up and the timing after index hospitalization that such admission occurred. This information was used to determine and compare the risk of readmissions for RAP and CP during follow up based on coexisting diabetic status. 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. Bivariate comparison for categorical data was performed using chisquared test or the proportional trend test, and for continuous data using ManneWhitney U test. Multivariable logistic regression analysis was used to determine independent predictors for coexistent diabetes, performance of cholecystectomy and organ failure using age, race, sex, etiology, tobacco abuse and diabetic status as covariates. We did not use MediQual Severity group assignment as a covariate since it is calculated using a combination of demographic and clinic variables and would result in collinearity. Multivariable cox regression analysis was used to determine predictors of RAP and CP during follow up using age, sex, race, etiology, tobacco abuse, diabetic status and RAP (for prediction of CP) as covariates. A p-value of
