Abstract
Objectives:
There is a paucity of studies evaluating predictors of new-onset diabetes mellitus (DM) following acute pancreatitis (AP-related DM). We utilized a population-based database to evaluate predictors of AP-related DM.
Methods:
The Nationwide Readmissions Database (2010–2014) was utilized to identify all non-diabetic adults with an index primary diagnosis of acute pancreatitis (AP). Multiple exclusions were applied to identify cohorts with and without AP-related DM. A case-control study was conducted to identify risk factors for developing AP-related DM within the calendar year.
Results:
We identified 2510 subjects with AP-related DM and 40,308 controls with AP who did not develop DM. Multivariable analysis revealed that increasing age (50–64 years, adjusted Odds Ratio [aOR], 1.35, 95% confidence interval [CI], 1.14–1.60), male sex (aOR, 1.2, 95% CI, 1.03–1.40), lowest income quartile (aOR, 1.48, 95% CI, 1.18–1.84), Elixhauser comorbidity index ≥3 (aOR, 1.47, 95% CI, 1.23–1.75), components of metabolic syndrome (aOR, 2.12, 95% CI, 1.21–3.70), severe AP (aOR, 1.60, 95% CI, 1.34–1.90), and recurrent AP (aOR, 1.46, 95% CI, 1.24–1.72) were independently associated with increased risk of AP-related DM.
Conclusion:
These population-level variables predictive of developing AP-related DM can potentially identify patients who may benefit from closer follow-up, intensive education, and implementation of preventative strategies.
Keywords: diabetes, acute pancreatitis, population database, metabolic syndrome
Introduction
Acute pancreatitis (AP) is exceedingly common in the United States, accounting for nearly 250,000 admissions per year.1 Not only does AP result in significant morbidity and mortality in the acute phase, but it can also result in long-term metabolic sequelae. The process of AP is driven primarily by autodigestion of pancreatic parenchyma through inappropriate activation of digestive enzymes, which disrupts the global architecture of the pancreas and results in acinar cell death, pancreatic atrophy and loss of both endocrine and exocrine function.1–3 While pancreatic exocrine function tends to recover over time following AP, endocrine function frequently continues to deteriorate.4, 5
One of the many intrinsic functions of the pancreas is its role in glucose metabolism. The current classification of diabetes mellitus (DM) includes different subtypes.6 Diabetes secondary to a disease of the exocrine pancreas is referred to as pancreatogenic DM (or type 3c diabetes), and often results from the destruction of β-islet cells secondary to various insults including acute and chronic pancreatitis, pancreatic cancer, and surgical pancreatectomy.6–10 Until recently, most cases of pancreatogenic DM have been misdiagnosed as either type 1 or type 2 DM, due in part to a lack of universally accepted diagnostic criteria.11 While these misclassifications distort the true prevalence of pancreatogenic DM, current estimates range between 1% and 9% of all patients with DM.9, 11 Historical studies suggest pancreatic endocrine function achieves full recovery after AP, but multiple studies have disproven this early theory.12, 13 In fact, a growing pool of literature supports the evidence that even a single episode of AP can potentially precipitate sustained pancreatic insufficiencies.14–17
Subsequent studies have convincingly shown that a single episode of AP raises the risk of new-onset DM two- to three-fold.15, 18–20 A recent systematic review and meta-analysis reported development of impaired glucose tolerance in nearly 40% of patients after index AP, with an estimated pooled prevalence of post-AP new onset diabetes (AP-related DM) of 23%, nearly 70% of which require insulin.14 Despite the increasingly robust correlation between AP and development of DM, there remains a paucity of high-quality literature describing risk factors for AP-related DM. Therefore, we sought to conduct a case-control study using a nationally representative population database to identify predictors of incident DM in patients discharged after an index episode of AP.
Materials and Methods
Study design and data source
We utilized the Nationwide Readmissions Database (NRD) from 2010 to 2014 to conduct a retrospective case-control study. The NRD is developed by the Agency for Healthcare Research and Quality (AHRQ) as part of the Healthcare Cost and Utilization Project (HCUP) for our study. The NRD is the largest publicly available database of all-payer hospital inpatient stays and contains data from approximately 18 million discharges each year from 28 geographically distributed states, representing 60.0% of the U.S. population and 58.2% percent of all hospitalizations. The database is specifically designed to support national readmission analyses of healthcare services. Unique, de-identified patient-linkage numbers are used to follow individual patients within the database across institutions, allowing for the analysis of all discharge records from included hospitals continuously for up to one calendar year.21 Since the NRD is a publicly available population-based dataset, the study was exempt from approval by The Ohio State University Institutional Review Board.
The NRD was queried using validated International Classification of Diseases, Ninth Edition Clinical Modification (ICD-9-CM) codes to define variables of interest (Supplemental Table 1). Adult subjects (≥18 years) with a primary discharge diagnosis of AP were identified (Figure 1). Stringent exclusion criteria were then applied to identify a cohort without prior DM (i.e., prevalent DM) who were discharged with an index episode of AP. Subjects with pregnancy, chronic pancreatitis, pancreatic neoplasm, prior DM in the current admission or any previous calendar year admission, lack of data due to the absence of any preceding calendar year admission, mortality during index admission, and any December discharge (to allow for sufficient follow up time) were excluded. From the remaining group, we further excluded patients without same calendar year readmission data. Our final study population included cases and controls defined as subjects with an index episode of AP with and without new onset DM within one calendar year, respectively. For subgroup analysis, cases were further restricted to those with early onset DM within 30 days from the initial hospital discharge.
FIGURE 1.
Study Flowchart
* Excluded: (i) age <18 year, (ii) pregnancy, (iii) chronic pancreatitis, (iv) pancreatic neoplasm, (v) prior DM in the current admission or any previous calendar year admission, (vi) patients without prior calendar year admission, (vii) mortality during index admission, (viii) any December discharge (to allow for sufficient follow-up time), (ix) patients without same calendar-year readmissions.
AP: acute pancreatitis, DM: diabetes mellitus
Demographic and Clinical Data
Subject characteristics included age, sex, income quartile and type of insurance. Comorbidities were quantified using the Elixhauser comorbidity index and stratified into <3 or ≥3 coexisting comorbid conditions.22 Length of stay was stratified into <7 and ≥7 days in accordance with the timeline for early- versus late-phase AP, the latter by definition classified as moderately severe or severe AP.23 Hospital characteristics included teaching status (urban non-teaching, urban teaching, rural) and size (small, medium, large) as defined by the HCUP algorithm.24 Recurrent AP was defined as a second admission with a principal discharge diagnosis of AP after the index admission.
To prevent redundancy in etiological association with AP, we used a hierarchical algorithm using related diagnosis codes to assign patients into different groups: gallstone-related, alcohol-related, metabolic syndrome associated, and other (not classified into the preceding groups).25, 26 Metabolic syndrome was defined as the presence of at least three of the following conditions: diabetes, hypertension, hypertriglyceridemia, low high density lipoprotein (HDL) cholesterol, and abdominal obesity.27 Since the NRD is based on ICD-9-CM codes and thus lacks specific coding for HDL cholesterol, and by excluding the outcome measure of diabetes, for the purpose of this study we defined metabolic syndrome as the presence of all three remaining criteria (obesity, high triglycerides, high blood pressure). Additional variables included related in-hospital procedures (endoscopic retrograde cholangiopancreatography [ERCP], cholecystectomy, percutaneous biliary procedures, open biliary procedures, any surgical pancreatic procedure), complications of pancreatitis (sepsis, intra-abdominal infections, cyst and pseudocyst of the pancreas) and presence of severe AP. Severe AP was defined per the revised Atlanta classification as the presence of acute respiratory failure, acute renal failure, or intravenous vasopressor requirement (as a surrogate marker for hypotension/cardiac failure).23, 25
The primary outcome of interest was to identify independent predictors of DM after an index admission for AP during the same calendar year (which occurred at a variable timeframe and is not necessarily equivalent to 12 months). Acute pancreatitis-related DM was considered present if subsequent admissions after AP included a diagnostic code for DM. The presence of AP-related DM was also evaluated within 30 days of the index AP event.
Statistical analysis
Differences in subjects with and without AP-related DM were evaluated with the use of chi-squared tests and t-tests, as appropriate. Multivariable logistic regression models were fit to evaluate predictors of AP-related DM both within 30 days and the same calendar year. Stepwise selection was used to determine the terms included in the models where all subject demographics, hospital characteristics, complications of pancreatitis, severe AP, and procedures were eligible for inclusion. Results of the models were summarized with adjusted odds ratios (aOR) and 95% confidence intervals (CI). There was very minimal missing data among our population with less than 2% missing income or insurance type (Supplemental Table 2). Participants missing either of these variables were excluded from the multivariable models when necessary. SAS 9.4 (SAS Institute, Cary, NC) was used for all analyses incorporating survey weights and accounting for the complex survey design. All results are weighted and represent national estimates.
Results
Baseline characteristics
Following application of stringent exclusion criteria, 42,818 subjects with an index admission for AP, without any prior diagnosis of DM, and with readmission data for analysis of AP-related DM were included in the study (Figure 1). Within the same calendar year and among those with readmission data, 2510 ‘cases’ who developed AP-related DM were identified; among these, 25.5% (641/2510) of subjects developed AP-related DM within 30 days after hospital discharge.
Univariate Comparisons
Comparison of baseline demographic characteristics, comorbidities, and features of the index AP episode (etiology, complications, related procedures) for subjects are shown in Tables 1 and 2. Similarly, Table 3 and Supplemental Table 3 compare characteristics for patients who developed AP-related DM within 30 days of discharge and those who did not. Those who developed AP-related DM were more likely to have severe AP, recurrent AP, were slightly older, more commonly fell into the lower income quartiles, had government-funded insurance, had a higher number of comorbidities, and had a longer duration of hospitalization (≥7 days). Over half of the patients who developed AP-related DM had at least one comorbidity characteristic of non-diabetic metabolic syndrome.
TABLE 1.
Summary of the Population and Comparison of Subjects With and Without AP-related DM Within the Same Calendar Year as Index Admission AP: Analysis of the Nationwide Readmission Database 2010–2014
| Overall (n = 42,818) |
AP without DM (n = 40,308) |
AP-related DM (n = 2510) |
||
|---|---|---|---|---|
| n (%) | n (%) | n (%) | P | |
| Patient Demographics and Hospital Characteristics | ||||
| Age (mean ± SE), y | 53.15 ± 0.17 | 53.08 ± 0.18 | 54.38 ± 0.62 | 0.043 |
| Age | <0.001 | |||
| 18 – 49 | 19,548 (45.65) | 18,562 (46.05) | 985 (39.25) | |
| 50 – 64 | 12,113 (28.29) | 11,251 (27.91) | 862 (34.33) | |
| ≥65 | 11,158 (26.06) | 10,495 (26.04) | 663 (26.42) | |
| Male Sex | 20,065 (46.86) | 18,804 (46.65) | 1261 (50.25) | 0.049 |
| Income (National Quartile) | <0.001 | |||
| First | 14,350 (34.11) | 13,310 (33.62) | 1039 (41.92) | |
| Second | 10,651 (25.32) | 10,071 (25.44) | 580 (23.39) | |
| Third | 9464 (22.50) | 8975 (22.67) | 490 (19.75) | |
| Fourth | 7603 (18.07) | 7232 (18.27) | 370 (14.94) | |
| Type of Insurance | 0.014 | |||
| Medicare | 17,263 (40.40) | 16,123 (40.09) | 1140 (45.46) | |
| Medicaid | 10,364 (24.25) | 9777 (24.31) | 587 (23.40) | |
| Private | 8910 (20.85) | 8431 (20.96) | 479 (19.12) | |
| Other | 6192 (14.49) | 5891 (14.65) | 301 (12.02) | |
| AHRQ-Elixhauser Index | <0.001 | |||
| < 3 | 16,663 (38.92) | 15,913 (39.48) | 750 (29.86) | |
| ≥ 3 | 26,155 (61.08) | 24,395 (60.52) | 1760 (70.14) | |
| Etiology and Associations of AP | 0.001 | |||
| Gallstone Related | 7181 (16.77) | 6802 (16.87) | 379 (15.10) | |
| Alcohol Related | 12,977 (30.31) | 12,296 (30.50) | 681 (27.15) | |
| Components of Metabolic Syndrome* | 376 (0.88) | 329 (0.82) | 48 (1.90) | |
| Other | 22,285 (52.04) | 20,883 (51.81) | 1402 (55.86) | |
| Severe AP | 7252 (16.94) | 6590 (16.35) | 662 (26.38) | <0.001 |
| Recurrent AP | 12,257 (28.63) | 11,394 (28.27) | 863 (34.38) | <0.001 |
| Type of Hospital | 0.431 | |||
| Urban Non-teaching | 17,968 (41.96) | 16,859 (41.82) | 1109 (44.18) | |
| Urban Teaching | 19,555 (45.67) | 18,458 (45.79) | 1098 (43.74) | |
| Rural | 5295 (12.37) | 4992 (12.38) | 303 (12.08) | |
| Hospital Size | 0.671 | |||
| Small | 6219 (14.52) | 5871 (14.57) | 347 (13.83) | |
| Medium | 10,380 (24.24) | 9793 (24.29) | 587 (23.41) | |
| Large | 26,220 (61.23) | 24,645 (61.14) | 1575 (62.76) | |
| LOS (mean ± SE), d | 5.28 ± 0.05 | 5.21 ± 0.05 | 6.36 ± 0.42 | 0.007 |
| LOS | 0.001 | |||
| <7 days | 33,161 (77.45) | 31,337 (77.74) | 1824 (72.66) | |
| ≥7 days | 9657 (22.55) | 8971 (22.26) | 686 (27.34) | |
| Complications of Pancreatitis | ||||
| Sepsis | 1099 (2.57) | 1015 (2.52) | 84 (3.34) | 0.111 |
| Intra-abdominal infections | 203 (0.47) | 186 (0.46) | 17 (0.67) | 0.316 |
| Cyst and pseudocyst of pancreas | 1776 (4.15) | 1637 (4.06) | 140 (5.56) | 0.043 |
| Procedures | ||||
| Any ERCP | 2718 (6.35) | 2575 (6.39) | 143 (5.70) | 0.473 |
| Cholecystectomy | 2190 (5.11) | 2064 (5.12) | 126 (5.01) | 0.895 |
| Percutaneous biliary procedures | 183 (0.43) | 177 (0.44) | ≤10 (≤0.40) | 0.317 |
| Open biliary procedures | 95 (0.22) | 88 (0.22) | ≤10 (≤0.40) | 0.682 |
| Any surgical pancreatic procedure | 1034 (2.41) | 954 (2.37) | 80 (3.20) | 0.165 |
Non-diabetic components of metabolic syndrome (obesity, high triglycerides, high blood pressure)
AP: acute pancreatitis; DM: diabetes mellitus; SE: standard error; AHRQ: Agency for Healthcare Research and Quality; LOS: length of stay; ERCP: endoscopic retrograde cholangiopancreatography
In cooperation with the AHRQ Heathcare Cost and Utilization Project (HCUP) data use agreement, observation values 1–10 are not reported to avoid risk of unintentional identification of individual subjects.
TABLE 2.
Summary of Comorbidities and Comparison of Subjects With and Without AP-related DM Within the Same Calendar Year as Discharge for Index Admission AP: Analysis of the Nationwide Readmission Database 2010–2014
| Overall (n = 42,818) |
AP without DM (n = 40,308) |
AP-related DM (n = 2510) |
||
|---|---|---|---|---|
| n (%) | n (%) | n (%) | P | |
| Comorbidities | ||||
| Hyperlipidemia* | 8487 (19.82) | 7865 (19.51) | 621 (24.76) | <0.001 |
| Obesity* | 3105 (7.25) | 2800 (6.95) | 305 (12.16) | <0.001 |
| Morbid obesity | 1326 (3.10) | 1182 (2.93) | 144 (5.74) | <0.001 |
| Hypertension† | 21,805 (50.92) | 20,190 (50.09) | 1615 (64.35) | <0.001 |
| Tobacco use | 15,068 (35.19) | 14,254 (35.36) | 814 (32.43) | 0.095 |
| Cholangitis | 376 (0.88) | 358 (0.89) | 18 (0.71) | 0.528 |
| Acute cholecystitis | 153 (0.36) | 151 (0.37) | ≤10 (≤0.40) | 0.180 |
| NAFLD | 1311 (3.06) | 1242 (3.08) | 69 (2.75) | 0.602 |
| Pancreatic anomalies | 93 (0.22) | 87 (0.22) | ≤10 (≤0.40) | 0.945 |
| Cardiovascular disease | 8993 (21.00) | 8309 (20.61) | 684 (27.26) | <0.001 |
| Chronic kidney disease | 6104 (14.25) | 5478 (13.59) | 625 (24.91) | <0.001 |
| Alcoholic liver disease | 4663 (10.89) | 4412 (10.95) | 251 (9.98) | 0.359 |
| Hepatitis C | 1455 (3.40) | 1349 (3.35) | 106 (4.21) | 0.150 |
| COPD | 8451 (19.74) | 7869 (19.52) | 583 (23.21) | 0.016 |
| Depression | 7195 (16.80) | 6834 (16.95) | 361 (14.39) | 0.054 |
Non-diabetic components of metabolic syndrome
AP: acute pancreatitis; DM: diabetes mellitus; NAFLD: non-alcoholic fatty liver disease; COPD: chronic obstructive pulmonary disease
In cooperation with the AHRQ Heathcare Cost and Utilization Project (HCUP) data use agreement, observation values 1–10 are not reported to avoid risk of unintentional identification of individual subjects.
TABLE 3.
Comparison of Subjects With and Without AP-related DM Within 30 Days of Discharge for Index Admission AP: Analysis of the Nationwide Readmission Database 2010–2014
| AP without DM (n=42,177) |
AP-related DM (n=641) |
||
|---|---|---|---|
| n (%) | n (%) | P | |
| Patient Demographics and Hospital Characteristics | |||
| Age (mean ± SE), y | 53.10 ± 0.18 | 56.65 ± 1.11 | 0.002 |
| Age | 0.001 | ||
| 18 – 49 | 19,329 (45.83) | 218 (34.03) | |
| 50 – 64 | 11,912 (28.24) | 200 (31.21) | |
| ≥65 | 10,936(25.93) | 223 (34.75) | |
| Male Sex | 19,761 (46.85) | 304 (47.37) | 0.879 |
| Income (National Quartile) | 0.005 | ||
| First | 14,094 (34.02) | 256 (40.38) | |
| Second | 10,478 (25.29) | 173 (27.21) | |
| Third | 9319 (22.49) | 145 (22.89) | |
| Fourth | 7542 (18.20) | 60 (9.53) | |
| Type of Insurance | 0.037 | ||
| Medicare | 16,948 (40.27) | 315 (49.13) | |
| Medicaid | 10,233 (24.31) | 130 (20.35) | |
| Private | 8782 (20.87) | 128 (20.02) | |
| Other | 6125 (14.55) | 67 (10.49) | |
| AHRQ-Elixhauser Index | 0.004 | ||
| < 3 | 16,474 (39.06) | 189 (29.50) | |
| ≥ 3 | 25,704 (60.94) | 452 (70.50) | |
| Etiology and Associations of AP | 0.103 | ||
| Gallstone Related | 7073 (16.77) | 107 (16.75) | |
| Alcohol Related | 12,817 (30.39) | 160 (24.95) | |
| Components of Metabolic Syndrome* | 364 (0.86) | 12 (1.84) | |
| Other | 21,923 (51.98) | 362 (56.47) | |
| Severe AP | 7079 (16.78) | 173 (27.01) | <0.001 |
| Recurrent AP | 12,025 (28.51) | 232 (36.14) | 0.013 |
| Type of Hospital | 0.083 | ||
| Urban Non-teaching | 17,659 (41.87) | 308 (48.13) | |
| Urban Teaching | 19,313 (45.79) | 242 (37.80) | |
| Rural | 5205 (12.34) | 90 (14.07) | |
| Hospital Size | 0.954 | ||
| Small | 6126 (14.52) | 93 (14.47) | |
| Medium | 10,219 (24.23) | 161 (25.15) | |
| Large | 25,833 (61.25) | 387 (60.37) | |
| LOS (mean ± SE), d | 5.24 ± 0.05 | 7.54 ± 0.67 | 0.001 |
| LOS | 0.001 | ||
| <7 days | 32,729 (77.60) | 432 (67.41) | |
| ≥7 days | 9449 (22.40) | 209 (32.59) | |
| Complications of Pancreatitis | |||
| Sepsis | 1072 (2.54) | 27 (4.21) | 0.128 |
| Intra-abdominal infections | 196 (0.46) | ≤10 (≤1.56) | 0.095 |
| Cyst and pseudocyst of pancreas | 1735 (4.11) | 41 (6.38) | 0.132 |
| Procedures | |||
| Any ERCP | 2673 (6.34) | 45 (6.97) | 0.724 |
| Cholecystectomy | 2146 (5.09) | 43 (6.75) | 0.223 |
| Percutaneous biliary procedures | 180 (0.43) | ≤10 (≤1.56) | 0.773 |
| Open biliary procedures | 92 (0.22) | ≤10 (≤1.56) | 0.200 |
| Any surgical pancreatic procedure | 1020 (2.42) | 14 (2.20) | 0.805 |
Non-diabetic components of metabolic syndrome (obesity, high triglycerides, high blood pressure)
AP: acute pancreatitis; DM: diabetes mellitus; SE: standard error; AHRQ: Agency for Healthcare Research and Quality; LOS: length of stay; ERCP: endoscopic retrograde cholangiopancreatography
In cooperation with the AHRQ Heathcare Cost and Utilization Project (HCUP) data use agreement, observation values 1–10 are not reported to avoid risk of unintentional identification of individual subjects.
Independent Predictors of AP-related DM within one calendar year
A multivariable logistic regression analysis for developing AP-related DM within the same calendar year as the index AP event (Table 4; full model in Supplemental Table 4) demonstrated that age was independently associated with AP-related DM, with subjects age 50–64 having higher odds (aOR, 1.35, 95% CI, 1.14–1.60) than those under 50 years of age. Lower income was also a predictor of developing AP-related DM, with those subjects in the lowest national quartile (aOR, 1.48, 95% CI, 1.18–1.84) at greater odds compared to the highest quartile. Severe AP was another significant predictor of AP-related DM within one calendar year (aOR, 1.60, 95% CI, 1.34–1.90), as was recurrent AP (aOR, 1.46, 95% CI, 1.24–1.72). Finally, men (aOR, 1.20, 95% CI, 1.03–1.40), those with Elixhauser comorbidity index ≥3 (aOR, 1.47, 95% CI, 1.23–1.75), and those with associated non-diabetic metabolic syndrome (aOR, 2.12, 95% CI, 1.21–3.70) were more likely to experience AP-related DM.
TABLE 4.
Multivariable Logistic Regression Model for AP-related DM Within the Same Calendar Year as Discharge for Index Admission AP: Analysis of the Nationwide Readmission Database 2010–2014
| aOR (95% CI) | P | |
|---|---|---|
| Age | 0.003 | |
| 18 – 49 | Reference | |
| 50 – 64 | 1.35 (1.14–1.60) | |
| ≥65 | 1.13 (0.91–1.39) | |
| Sex | 0.021 | |
| Male | 1.20 (1.03–1.40) | |
| Female | Reference | |
| Income (National Quartile) | <0.001 | |
| First | 1.48 (1.18–1.84) | |
| Second | 1.10 (0.87–1.39) | |
| Third | 1.06 (0.83–1.36) | |
| Fourth | Reference | |
| AHRQ-Elixhauser Index | <0.001 | |
| < 3 | Reference | |
| ≥ 3 | 1.47 (1.23–1.75) | |
| Etiology and Associations of AP | <0.001 | |
| Gallstone Related | Reference | |
| Alcohol Related | 0.85 (0.66–1.10) | |
| Non-diabetic Metabolic Syndrome* | 2.12 (1.21–3.70) | |
| Other | 1.21 (0.96–1.52) | |
| Severe AP | 1.60 (1.34–1.90) | <0.001 |
| Recurrent AP | 1.46 (1.24–1.72) | <0.001 |
Non-diabetic components of metabolic syndrome (obesity, high triglycerides, high blood pressure)
AP: acute pancreatitis; DM: diabetes mellitus; aOR: adjusted odds ratio; CI: confidence interval; AHRQ: Agency for Healthcare Research and Quality
Independent Predictors of AP-related DM within 30 days
A multivariable logistic regression model for developing AP-related DM within 30 days (Table 5; full model in Supplemental Table 5) revealed similar predictors. Older patients were similarly at higher risk, with subjects 65 years or older (aOR, 1.77, 95% CI, 1.24–2.51) conferring higher risk as compared to those younger than 50 years. As before, lower income was a predictor of developing AP-related DM at 30 days, with the risk highest for those patients in the lowest national quartile (aOR, 2.33, 95% CI, 1.53–3.54) as compared to the highest quartile. Severe AP (aOR, 1.70, 95% CI, 1.26–2.29) and recurrent AP (aOR, 1.65, 95% CI, 1.23–2.20) were again independent risk factors as well.
TABLE 5.
Multivariable Logistic Regression Model for AP-related DM Within 30 Days of Discharge for Index Admission AP: Analysis of the Nationwide Readmission Database 2010–2014
| aOR (95% CI) | P | |
|---|---|---|
| Age | ||
| 18 – 49 | Reference | 0.001 |
| 50 – 64 | 1.47 (1.07–2.02) | |
| ≥65 | 1.97 (1.37–2.83) | |
| Income (National Quartile) | 0.001 | |
| First | 2.33 (1.53–3.54) | |
| Second | 2.09 (1.35–3.24) | |
| Third | 2.00 (1.31–3.07) | |
| Fourth | Reference | |
| Severe AP | 1.74 (1.29–2.35) | <0.001 |
| Recurrent AP | 1.65 (1.23–2.20) | 0.001 |
AP: acute pancreatitis; DM: diabetes mellitus; aOR: adjusted odds ratio; CI: confidence interval
Discussion
In this case-control study using a large population database, the development of AP-related DM within the same calendar year was highest in those with increased AP severity and a recurrent episode of AP. Also, AP-related DM more commonly occurred in older age groups, male subjects, subjects with higher comorbidities, and those with non-diabetic components of metabolic syndrome (hypertension, obesity, and hyperlipidemia). While prevention of disease progression to severe AP can be facilitated by practicing adherence to clinical guidelines, post-discharge outpatient follow-up may also play a role in reducing the risk for recurrent episodes of AP. This study identifies a high-risk profile of patients for AP-related DM in whom close clinical follow-up, intensive education, and preventative strategies may be useful for improving both short- and long-term outcomes.
Recurrence of AP is perhaps predictably associated with AP-related DM. Acute pancreatitis and chronic pancreatitis have long been identified as two distinct diseases, though risk of progression from AP to chronic pancreatitis was originally thought to be minor.28 Over the past few decades, however, there has been increasing evidence of a spectrum of disease from AP to recurrent AP and chronic pancreatitis, with the theory that the first attack of AP sensitizes the pancreas to inflammation and fibrosis by subsequent mechanisms.29 A recent, large systematic review of high-quality cohort studies reports a 21% (95% CI, 17%−26%) risk of recurrence after first-episode AP in adults, with 36% (95% CI, 20%−53%) of those patients further progressing to CP.30 Prior studies have demonstrated this link between recurrent AP and AP-related DM, with reported odds ratios as high as 8.2 compared to a single episode alone.17 Another large, Taiwanese database study also found ≥2 admissions for AP to be a significant predictor of AP-related DM (OR, 1.94, 95% CI, 1.28–2.39).31 Prevention of recurrence of AP relies largely on lifestyle modifications and risk factor reduction. Smoking cessation, alcohol cessation after alcoholic AP, cholecystectomy during index admission for mild biliary AP or sphincterotomy after biliary AP in poor operative candidates, and reduction of triglyceride levels after hyperlipidemic AP reduce the risk of recurrence.32–36 In addition, one might expect that adherence and early outpatient patient follow-up after hospital discharge can potentially offset recurrence.
Severity of AP is also of particular interest in regards to predicting AP-related DM, as it too is potentially modifiable. While several prior studies found severe AP to be an independent risk factor for impaired glucose tolerance or AP-related DM,18, 37–41 others have not.14, 15, 20, 42–44 Even meta-analyses and systemic reviews have divergent conclusions, albeit with alternative study designs; while one such study determined that severity of AP made minimal impact on development of AP-related DM,14 a more recent meta-analysis concluded that the incidence of AP-related DM was higher for severe compared to mild AP (39% vs 14%).37 In this large case-control study, we confirmed severe AP to be a significant predictor of AP-related DM at both the 30-day and longer term follow-up. This can logically be explained as increasingly severe AP results in greater pancreatic β-cell loss.1, 2 Currently, prevention of progression to severe AP relies largely on timely and effective fluid resuscitation, nutritional support and mitigation of complications. Aggressive, goal directed fluid administration is paramount to prevent hypoperfusion and organ failure, with some studies showing lactated Ringer’s solution to be superior to normal saline.35, 45–47
Etiology of AP as a predictor of AP-related DM has been another source of contention in prior studies. While some notable studies have linked alcoholic etiology to AP-related DM,31, 37 others found no such risk.14, 38, 42 Using this comprehensive U.S. database, we found no significant correlation between alcohol- and biliary-related etiology and risk of AP-related DM. Several studies recognize the components of metabolic syndrome as predictors of AP-related DM.20, 38, 42, 44 A recent, large population based cohort study including 1 million patients spanning several years reported hypertension, hypertriglyceridemia, and obesity to be independent risk factors for diabetes after AP.20 These results support our findings of metabolic syndrome as a predictor of AP-related DM, though more targeted studies characterizing the role of metabolic syndrome in AP are warranted.
Additive comorbidities as indicated by an Elixhauser index ≥3 also places patients at increased risk of AP-related DM at intermediate follow-up. While the majority of damage to pancreatic architecture is theoretically limited to the inciting AP episode, the development of AP-related DM is in part an insidious process and may develop over months to years. Previous studies have noted singular comorbidities to be predictive of AP-related DM, among them hypertriglyceridemia, obesity, hyperlipidemia, smoking and chronic liver disease.20, 38, 42, 44 We propose that an accumulation of comorbidities, more so than any single comorbidity, add to the propensity to develop delayed-onset AP-related DM (after 30 days), yet do not have significant effect in the acute period.
Similar to prior studies, we found increasing age to be another predictor of AP-related DM.20, 38, 42 One possible explanation is decreased pancreatic reserve in older patients leading to increased susceptibility to sustained damage and poorer functional recovery compared to younger age groups. Additionally, male sex is a predictor of AP-related DM within one calendar year, consistent with other large database studies.15, 20 A final non-modifiable risk factor for AP-related DM is lower income, both at the 30 day and one calendar year endpoints. While these non-modifiable risk factors cannot be targeted for intervention in the acute AP episode, they are useful in developing a profile for identification of at-risk patients.
Currently, the ADA recommends screening for DM every 3 years starting at age 45. A recent retrospective study of 2.3 million primary care records reports the incidence of AP-related DM to be even greater than that of type 1 DM.10 The same study reported a prevalence of insulin use of 20.9% (95% CI, 14.6–28.9%) following diagnosis of AP-related DM, more than 5-fold the requirement in type 2 DM after an equal time interval. Yet despite this, patients are not routinely screened for DM after AP. The current study highlights the growing need to establish thorough screening guidelines for AP-related DM. Patients of older age groups, male sex, lower socioeconomic strata, with a higher number of comorbidities, who have metabolic syndrome, and who developed severe and/or recurrent AP are at greatest risk of AP-related DM and should be screened earlier and more frequently.
There are some important considerations to avoid misinterpretation of our data. The NRD utilizes ICD-9-CM coding for diagnoses. Although this restricts which variables can be studied and relies on accuracy of diagnostic coding, the diagnostic codes for our primary disease (i.e., AP and DM) have been previously validated.1, 25, 29, 48 This study was specifically designed using a population of patients who were readmitted to the hospital and hence does not include any data on outpatients. While many patients who are admitted with an index case of AP can develop DM and are not hospitalized, we used cases and controls, both of whom had readmission data indicating the absence or presence of diabetes. Additionally, as this is a case-control study limited to inpatient data only, this dataset cannot be used to estimate the incidence or prevalence of AP-related DM, nor was this the intention of our study design. We applied strict exclusion criteria to limit our study population to an index episode of AP without prior evidence of DM. Although we sterilized the study population as much as possible within the confines of a diagnostic code-driven database, we cannot fully exclude the possibility of unintentional inclusion of undiagnosed DM or pre-diabetes. We do, however, feel this replicates the real-world burden of underdiagnosed disease. Additionally, the NRD does not provide information on race, ethnicity, medication use, or laboratory parameters, so potential confounding from one of these variables cannot be excluded. Lastly, analysis using the NRD is limited to a single calendar year, placing constraints on our ability to study AP-related DM onset greater than one year after index AP. Despite these considerations, in the absence of a prospective cohort study with longitudinal follow-up, these risk factors cannot otherwise be evaluated in a statistically valid manner.
In conclusion, DM following an episode of AP is gathering increased recognition as an important clinical sequela of AP. Acute pancreatitis-related DM is most strongly associated with increased AP severity and recurrence of AP. Therefore, prevention of AP-related DM can indirectly be pursued by aggressive goal-directed management to prevent severe AP and targeted risk factor modification to moderate AP recurrence. Patients with AP should be monitored for the development of DM, particularly when one or more predictors are identified. Prospective cohort studies are needed to definitively establish the incidence rate and to validate our predictors of AP-related DM. Future studies are also needed to understand the pathophysiology of AP-related DM, and to develop strategies for prevention following AP hospitalization.
Supplementary Material
Source of funding:
Research reported in this publication was supported by the National Cancer Institute and National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) under award number U01DK108327 (PH, LL, ZC-M, DC, SK). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Abbreviations
- AP
acute pancreatitis
- DM
diabetes mellitus
- AP-related DM
Acute pancreatitis-related diabetes mellitus
- MRI
magnetic resonance imaging
- CT
computed tomography
- NRD
Nationwide Readmissions Database
- AHRQ
Agency for Healthcare Research and Quality
- HCUP
Healthcare Cost and Utilization Project
- ICD-9-CM
International Classification of Diseases, Ninth Edition Clinical Modification
- HDL
high density lipoprotein
- ERCP
endoscopic retrograde cholangiopancreatography
- LOS
length of stay
- aOR
adjusted odds ratio
- CI
confidence interval
- SE
standard error
- LDL
low-density lipoprotein
Footnotes
All authors have approved the final version of this article.
Conflict of interest: The authors have no conflict of interest.
References
- 1.Garg SK, Campbell JP, Anugwom C, et al. Incidence and Predictors of Readmissions in Acute Pancreatitis: A Nationwide Analysis. Pancreas. 2018;47:46–54. [DOI] [PubMed] [Google Scholar]
- 2.Saluja A, Dudeja V, Dawra R, et al. Early Intra-Acinar Events in Pathogenesis of Pancreatitis. Gastroenterology. 2019;156:1979–1993. [DOI] [PubMed] [Google Scholar]
- 3.Lankisch PG, Löhr-Happe A, Otto J, et al. Natural course in chronic pancreatitis. Pain, exocrine and endocrine pancreatic insufficiency and prognosis of the disease. Digestion. 1993;54:148–155. [DOI] [PubMed] [Google Scholar]
- 4.Sand J, Nordback I. Acute pancreatitis: risk of recurrence and late consequences of the disease. Nat Rev Gastroenterol Hepatol. 2009;6:470–477. [DOI] [PubMed] [Google Scholar]
- 5.Pelli H, Lappalainen-Lehto R, Piironen A, et al. Pancreatic damage after the first episode of acute alcoholic pancreatitis and its association with the later recurrence rate. Pancreatology. 2009;9:245–251. [DOI] [PubMed] [Google Scholar]
- 6.American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care. 2011;34:S62–S69 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Slezak LA, Andersen DK. Pancreatic resection: effects on glucose metabolism. World J Surg. 2001;25:452–60. [DOI] [PubMed] [Google Scholar]
- 8.Burkhart RA, Gerber SM, Tholey RM, et al. Incidence and severity of pancreatogenic diabetes after pancreatic resection. J Gastrointest Surg. 2015;19:217–225. [DOI] [PubMed] [Google Scholar]
- 9.Hart PA, Bellin MD, Andersen DK, et al. Type 3c (pancreatogenic) diabetes mellitus secondary to chronic pancreatitis and pancreatic cancer. Lancet Gastroenterol Hepatol. 2016;1:226–237. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Woodmansey C, McGovern AP, McCullough KA, et al. Incidence, Demographics, and Clinical Characteristics of Diabetes of the Exocrine Pancreas (Type 3c): A Retrospective Cohort Study. Diabetes Care. 2017;40:1486–1493. [DOI] [PubMed] [Google Scholar]
- 11.Ewald N, Kaufmann C, Raspe A, et al. Prevalence of diabetes mellitus secondary to pancreatic diseases (type 3c). Diabetes Metab Res Rev. 2012;28:338–342. [DOI] [PubMed] [Google Scholar]
- 12.Angelini G, Cavallini G, Pederzoli P, et al. Long-term outcome of acute pancreatitis: a prospective study with 118 patients. Digestion. 1993;54:143–147. [DOI] [PubMed] [Google Scholar]
- 13.Mitchell CJ, Playforth MJ, Kelleher J, et al. Functional recovery of the exocrine pancreas after acute pancreatitis. Scand J Gastroenterol. 1983;18:5–8. [DOI] [PubMed] [Google Scholar]
- 14.Das SL, Singh PP, Phillips AR, et al. Newly diagnosed diabetes mellitus after acute pancreatitis: a systematic review and meta-analysis. Gut. 2014;63:818–831. [DOI] [PubMed] [Google Scholar]
- 15.Shen HN, Yang CC, Chang YH, et al. Risk of Diabetes Mellitus after First-Attack Acute Pancreatitis: A National Population-Based Study. Am J Gastroenterol. 2015;110:1698–1706. [DOI] [PubMed] [Google Scholar]
- 16.Shen HN, Lu CL, Li CY. Epidemiology of first-attack acute pancreatitis in Taiwan from 2000 through 2009: a nationwide population-based study. Pancreas. 2012;41:696–702. [DOI] [PubMed] [Google Scholar]
- 17.Nikkola J, Laukkarinen J, Lahtela J, et al. The Long-term Prospective Follow-up of Pancreatic Function After the First Episode of Acute Alcoholic Pancreatitis: Recurrence Predisposes One to Pancreatic Dysfunction and Pancreatogenic Diabetes. J Clin Gastroenterol. 2017;51:183–190. [DOI] [PubMed] [Google Scholar]
- 18.Malecka-Panas E, Gasiorowska A, Kropiwnicka A, et al. Endocrine pancreatic function in patients after acute pancreatitis. Hepatogastroenterology. 2002;49:1707–1712. [PubMed] [Google Scholar]
- 19.Ibars EP, Sánchez de Rojas EA, Quereda LA, et al. Pancreatic function after acute biliary pancreatitis: does it change? World J Surg. 2002;26:479–486. [DOI] [PubMed] [Google Scholar]
- 20.Lee YK, Huang MY, Hsu CY, et al. Bidirectional Relationship Between Diabetes and Acute Pancreatitis: A Population-Based Cohort Study in Taiwan. Medicine (Baltimore). 2016;95:e2448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.NRD Overview: Healthcare Cost and Utilization Project (HCUP). December 1,2019. Available at: http://hcup-us.ahrq.gov/nrdoverview.jsp.AccessedMarch 31, 2020.
- 22.Elixhauser A, Steiner C, Harris DR, et al. Comorbidity measures for use with administrative data. Med Care. 1998;36:8–27. [DOI] [PubMed] [Google Scholar]
- 23.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;62:102–111. [DOI] [PubMed] [Google Scholar]
- 24.Healthcare Cost and Utilization Project (HCUP). November, 2019. Avalable at: https://www.ahrq.gov/data/hcup/index.html.AccessedMarch 31, 2020.
- 25.Krishna SG, Kruger AJ, Patel N, et al. Cholecystectomy During Index Admission for Acute Biliary Pancreatitis Lowers 30-Day Readmission Rates. Pancreas. 2018;47:996–1002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Yadav D, Lee E, Papachristou GI, et al. A population-based evaluation of readmissions after first hospitalization for acute pancreatitis. Pancreas. 2014;43:630–637. [DOI] [PubMed] [Google Scholar]
- 27.Alberti KG, Eckel RH, Grundy SM, et al. Harmonizing the metabolic syndrome: a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation. 2009;120:1640–1645. [DOI] [PubMed] [Google Scholar]
- 28.Singer MV, Gyr K, Sarles H. Revised classification of pancreatitis. Report of the Second International Symposium on the Classification of Pancreatitis in Marseille, France, March 28–30, 1984. Gastroenterology. 1985;89:683–685. [PubMed] [Google Scholar]
- 29.Vipperla K, Papachristou GI, Easler J, et al. Risk of and factors associated with readmission after a sentinel attack of acute pancreatitis. Clin Gastroenterol Hepatol. 2014;12:1911–1919. [DOI] [PubMed] [Google Scholar]
- 30.Petrov MS, Yadav D. Global epidemiology and holistic prevention of pancreatitis. Nat Rev Gastroenterol Hepatol. 2019;16:175–184. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Ho TW, Wu JM, Kuo TC, et al. Change of Both Endocrine and Exocrine Insufficiencies After Acute Pancreatitis in Non-Diabetic Patients: A Nationwide Population-Based Study. Medicine (Baltimore). 2015;94:e1123. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Adiamah A, Psaltis E, Crook M, et al. A systematic review of the epidemiology, pathophysiology and current management of hyperlipidaemic pancreatitis. Clin Nutr. 122018;37:1810–1822. [DOI] [PubMed] [Google Scholar]
- 33.Magnusdottir BA, Baldursdottir MB, Kalaitzakis E, et al. Risk factors for chronic and recurrent pancreatitis after first attack of acute pancreatitis. Scand J Gastroenterol. 2019;54:87–94. [DOI] [PubMed] [Google Scholar]
- 34.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;136:848–855. [DOI] [PubMed] [Google Scholar]
- 35.Guidelines WGIAAP. IAP/APA evidence-based guidelines for the management of acute pancreatitis. Pancreatology. 2013;13:e1–15. [DOI] [PubMed] [Google Scholar]
- 36.Takeyama Y Long-term prognosis of acute pancreatitis in Japan. Clin Gastroenterol Hepatol. 2009;7:S15–S17. [DOI] [PubMed] [Google Scholar]
- 37.Zhi M, Zhu X, Lugea A, et al. Incidence of New Onset Diabetes Mellitus Secondary to Acute Pancreatitis: A Systematic Review and Meta-Analysis. Front Physiol. 2019;10:637. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Soo DHE, Pendharkar SA, Jivanji CJ, et al. Derivation and validation of the prediabetes self-assessment screening score after acute pancreatitis (PERSEUS). Dig Liver Dis. 2017;49:1146–1154. [DOI] [PubMed] [Google Scholar]
- 39.Andersson B, Pendse ML, Andersson R. Pancreatic function, quality of life and costs at long-term follow-up after acute pancreatitis. World J Gastroenterol. 2010;16:4944–4951. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Garip G, Sarandöl E, Kaya E. Effects of disease severity and necrosis on pancreatic dysfunction after acute pancreatitis. World J Gastroenterol. 2013;19:8065–8070. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Tu J, Yang Y, Zhang J, et al. Effect of the disease severity on the risk of developing new-onset diabetes after acute pancreatitis. Medicine (Baltimore). 2018;97:e10713. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Ma JH, Yuan YJ, Lin SH, et al. Nomogram for predicting diabetes mellitus after the first attack of acute pancreatitis. Eur J Gastroenterol Hepatol. 2019;31:323–328. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.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;7:447–453. [PubMed] [Google Scholar]
- 44.Wu D, Xu Y, Zeng Y, Wang X. Endocrine pancreatic function changes after acute pancreatitis. Pancreas. 2011;40:1006–1011. [DOI] [PubMed] [Google Scholar]
- 45.Crockett SD, Wani S, Gardner TB, et al. American Gastroenterological Association Institute Guideline on Initial Management of Acute Pancreatitis. Gastroenterology. 2018;154:1096–1101. [DOI] [PubMed] [Google Scholar]
- 46.Wu BU, Hwang JQ, Gardner TH, et al. Lactated Ringer’s solution reduces systemic inflammation compared with saline in patients with acute pancreatitis. Clin Gastroenterol Hepatol. 2011;9:710–717. [DOI] [PubMed] [Google Scholar]
- 47.Tenner S, Baillie J, DeWitt J, et al. American College of Gastroenterology guideline: management of acute pancreatitis. Am J Gastroenterol. 2013;108:1400–1415; 1416. [DOI] [PubMed] [Google Scholar]
- 48.Patel K, Li F, Luthra A, et al. Acute Biliary Pancreatitis is Associated With Adverse Outcomes in the Elderly: A Propensity Score-Matched Analysis. J Clin Gastroenterol. 2019;53:e291–e297. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.