Abstract
Background: Pancreatitis accounts for more than $2.5 billion of healthcare costs and remains the most common gastrointestinal (GI) admission. Few contemporary studies have assessed temporal trends of incidence, complications, management, and outcomes for acute pancreatitis in hospitalized patients at the national level.
Methods: We used data from one of the largest hospital-based databases available in the United States, the Healthcare Cost and Utilization Project's (HCUP) State Inpatient Database, from 10 states between 2008 and 2015. We included patients with a diagnosis of acute pancreatitis (ICD-9 CM 577.0). Patient- and hospital-level data were used to estimate incidence and inpatient mortality rates.
Results: From 80,736,256 hospitalizations, 929,914 (1.15%) cases of acute pancreatitis were identified, 186,226 (20.2%) of which were caused by gallbladder disease). The median age was 53 years (interquartile range [IQR], 41–67) and 50.8% were men. In-hospital mortality was 2.5% and crude mortality rates declined from 2.9% to 2.0% over the study period. Admission year remained significant after adjusting for patient demographics and comorbidities (odds ratio [OR], 0.90; 95% confidence interval [CI], 0.89–0.90; p < 0.001). Gallbladder disease was associated with decreased odds of mortality (OR, 0.60; 95% CI, 0.57–0.62). Median length of stay was four days (IQR, 2–7) and decreased over time. The rates of surgical and endoscopic interventions were highest in 2011 (peak incidence of 16.1% and 9.5%, respectively) and have been decreasing since. Surgical providers were, on average, more likely than medical providers to perform surgery in both those with and without gallbladder disease etiology (gallbladder disease OR, 7.11; 95% CI, 5.46–9.25; non-gallbladder disease OR, 20.50; 95% CI, 16.81–25.01), endoscopy (gallbladder disease OR, 1.22; 95% CI, 0.87–1.72; non-gallbladder disease OR, 1.60; 95% CI, 1.18–2.16), or both (gallbladder disease OR, 7.00; 95% CI, 5.22–9.37; non-gallbladder disease OR, 8.85; 95% CI, 5.61–13.96).
Conclusions: The incidence of pancreatitis, from 2008 to 2015, has increased whereas inpatient mortality (i.e., case fatality) has decreased. Understanding temporal trends in outcomes and management along with provider, hospital, and regional variation can better identify areas for future research and collaboration in managing these patients.
Keywords: pancreatic diseases; minimally invasive surgical procedures; surgical procedures, operative; pancreatitis; health personnel; mortality
Pancreatitis accounts for more than 200,000 hospitalizations in the United States each year [1]. An estimated 5 to 35 per 100,000 people are treated for pancreatitis annually and these estimates have increased over the last decade [2]. Pancreatitis is defined as autodigestion of the pancreas because of leaking pancreatic enzymes that damage the pancreas and surrounding tissue. It can result in pancreatic exocrine and endocrine dysfunction and causes a large spectrum of illness severity. Multiple etiologies and risk factors for pancreatitis exist, including alcohol abuse, tobacco smoking, gallstones, cannabis, inflammatory bowel disease, end-stage renal disease, anatomic and obstructive abnormalities, and genetic factors such as mutations in PRSS1, SPINK1, CTRC, CLDN2, or CFTR genes [3,4]. Although the majority of pancreatitis cases can be attributed to the factors listed above, 10%–30% of cases have no identifiable cause [4].
Hospitalizations because of acute pancreatitis were increasing in the United States during the first decade of the 21st century. Comparing the average incidence of hospitalizations from pancreatitis during 2002–2005 versus 2009–2012, a 13.3% increase was observed, making it the most common gastrointestinal illness requiring medical professional intervention during this time period [1,3]. Despite the increase in pancreatitis incidence, recent studies have demonstrated a decrease in case-fatality rates despite unchanging population mortality rates, the cause of which is unknown [5]. The trends in pancreatitis complication rates and provider specialty treatment bias has yet to be thoroughly examined. Most recently, the emergence of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) has complicated the care and treatment of patients with pancreatitis who also test positive for the virus, possibly because of overstimulation of the inflammatory cascade, a well-described sequelae of coronavirus disease 2019 (COVID-19) [6]. This has led to an increase in morbidity and mortality as well as multiorgan and persistent organ failure in this patient population [2]. Because future investigations will undoubtedly attempt to further unravel the impact of the pandemic on pancreatitis, a clear understanding of pancreatitis in the pre-COVID-19 era will be crucial to these efforts. Furthermore, the lack of a universal medical electronic reporting system poses challenges to analyzing disease incidence, mortality, complications, and provider specific treatment strategies regarding pancreatitis. The practice of surgically treating gallbladder disease to alleviate pancreatitis further complicates reporting of trends in patient care.
This study focuses on the trends in the incidence, mortality, and complication rates of pancreatitis from 2008–2015 (pre-SARS-CoV-2 era) using comprehensive data from ten states included in the Healthcare Cost and Utilization Project's (HCUP) databases serving as a nationally representative sample. We also investigated whether providers differed in their treatment strategy based on the provider's specialty (medical vs. surgical) when treating patients with pancreatitis in Wisconsin, which was the only state providing this information. We hypothesize that pancreatitis incidence in the United States is increasing, however, with more advanced treatment strategies as the field of medicine evolves, we expect that mortality, complication rates, and provider specialty biases are decreasing.
Patients and Methods
Data
The HCUP consists of a family of databases that provides information on nearly all hospital admissions in the United States. This includes data on individual patients' age, gender, health, and insurance status as well as hospital procedures and outcomes. We used ICD-9 codes to identify individuals within this database who were admitted with pancreatitis in one of ten states (Arkansas, Arizona, California, Florida, Kentucky, Maryland, Michigan, North Carolina, New York, and Wisconsin) from 2008 to 2015 (Supplementary Table S1). We similarly determined if the identified individuals had comorbidities on admission, what treatment strategy they received, whether they experienced complications after admission, and whether they had gallbladder disease-associated pancreatitis. Comorbidities considered on admission were hematological failure, hepatic failure, neurologic failure, acute renal failure, cardiovascular failure, and sepsis. Complications included myocardial infarction, pneumonia, venous thromboembolism, acute renal failure, gastrointestinal (GI) bleed, hemorrhage, respiratory failure, sepsis, and surgical site infection. Patients were considered to have gallbladder disease-associated pancreatitis (herein referred to as gallbladder disease etiology) if they had a cholangitis, cholecystitis, or choledocholithiasis diagnosis, or at least one of these conditions present on admission. Supplementary Table S1 includes the full list of how procedures, diagnoses, and surgeries were classified. Only Wisconsin reported provider medical specialty, which we dichotomized into medical or surgical. As reflected in Table 1, Wisconsin only reported data for the years 2012–2014.
Table 1.
Pancreatitis Admissions and Patient Characteristics by Year
| 2008 | 2009 | 2010 | 2011 | 2012 | 2013 | 2014 | 2015 | |
|---|---|---|---|---|---|---|---|---|
| n | 131,388 | 91,695 | 15,1563 | 14,7261 | 118,066 | 108,217 | 120,488 | 54,781 |
| Age, mean (SD) | 54.33 (17.87) | 54.77 (17.48) | 54.27 (17.71) | 54.25 (17.62) | 54.27 (17.26) | 54.28 (17.30) | 54.12 (17.14) | 54.44 (16.80) |
| Female (%) | 65,462 (50.0) | 44,769 (48.8) | 75,667 (50.1) | 72,212 (49.2) | 57,000 (48.3) | 52,126 (48.2) | 57,941 (48.1) | 26,073 (47.7) |
| Insurance status (%) | ||||||||
| Medicare | 46,279 (35.2) | 33,452 (36.5) | 53,500 (35.4) | 52,307 (35.6) | 43,133 (36.6) | 39,621 (36.7) | 43,240 (35.9) | 20,297 (37.1) |
| Medicaid | 22,484 (17.1) | 15,841 (17.3) | 28,903 (19.1) | 27,956 (19.0) | 22,189 (18.8) | 20,769 (19.2) | 27,956 (23.2) | 11,033 (20.1) |
| Private | 40,798 (31.1) | 27,073 (29.5) | 43,486 (28.7) | 41,936 (28.5) | 32,937 (27.9) | 29,610 (27.4) | 34,132 (28.4) | 15,491 (28.3) |
| Self-paid | 14,143 (10.8) | 10,922 (11.9) | 16,995 (11.2) | 16,184 (11.0) | 14,568 (12.4) | 13,244 (12.3) | 10,612 (8.8) | 5,475 (10.0) |
| No charge | 1,431 (1.1) | 1,787 (2.0) | 1,920 (1.3) | 1,994 (1.4) | 1,986 (1.7) | 1,994 (1.8) | 1,397 (1.2) | 1,096 (2.0) |
| Other | 6,210 (4.7) | 2,558 (2.8) | 6,509 (4.3) | 6,569 (4.5) | 3,089 (2.6) | 2,851 (2.6) | 2,974 (2.5) | 1,367 (2.5) |
| CCI, mean (SD) | 1.22 (1.73) | 1.25 (1.77) | 1.25 (1.74) | 1.26 (1.75) | 1.25 (1.74) | 1.29 (1.77) | 1.30 (1.76) | 1.34 (1.76) |
| CCI (%) | ||||||||
| 0 | 61,718 (47.0) | 42,087 (45.9) | 69,902 (46.1) | 67,566 (45.9) | 53,801 (45.6) | 48,338 (44.7) | 53,300 (44.2) | 23,432 (42.8) |
| 1–2 | 47,779 (36.4) | 33,920 (37.0) | 55,190 (36.4) | 53,292 (36.2) | 43,631 (37.0) | 40,317 (37.3) | 44,886 (37.3) | 20,759 (37.9) |
| 3–4 | 14,031 (10.7) | 10,077 (11.0) | 17,303 (11.4) | 17,349 (11.8) | 13,561 (11.5) | 12,762 (11.8) | 14,677 (12.2) | 7,080 (12.9) |
| 5+ | 7,860 (6.0) | 5,611 (6.1) | 9,168 (6.0) | 9,054 (6.1) | 7,073 (6.0) | 6,800 (6.3) | 7,625 (6.3) | 3,510 (6.4) |
| Present on admission | ||||||||
| Acute renal failure (%) | 12,473 (9.5) | 10,365 (11.3) | 17,280 (11.4) | 17,361 (11.8) | 13,894 (11.8) | 13,462 (12.4) | 15,587 (12.9) | 7,631 (13.9) |
| MI (%) | 1,541 (1.2) | 988 (1.1) | 1,510 (1.0) | 1,483 (1.0) | 1,053 (0.9) | 972 (0.9) | 1,080 (0.9) | 534 (1.0) |
| Pneumonia | 1,889 (1.4) | XXX (%) | 2,058 (1.4) | 2,146 (1.5) | 1,465 (1.2) | 1,348 (1.2) | 1,571 (1.3) | 742 (1.4) |
| VTE (%) | 669 (0.5) | 516 (0.6) | 718 (0.5) | 814 (0.6) | 600 (0.5) | 609 (0.6) | 704 (0.6) | 336 (0.6) |
| CV failure (%) | 4,724 (3.6) | 3,648 (4.0) | 5,963 (3.9) | 5,959 (4.0) | 4,651 (3.9) | 4,476 (4.1) | 5,140 (4.3) | 2,454 (4.5) |
| Heme failure (%) | 6,648 (5.1) | 5,319 (5.8) | 9,755 (6.4) | 10,318 (7.0) | 8,379 (7.1) | 7,922 (7.3) | 8,842 (7.3) | 4,040 (7.4) |
| Hemorrhage (%) | 293 (0.2) | 226 (0.2) | 314 (0.2) | 294 (0.2) | 197 (0.2) | 169 (0.2) | 196 (0.2) | 90 (0.2) |
| Hepatic failure (%) | 1,183 (0.9) | 878 (1.0) | 1,552 (1.0) | 1,537 (1.0) | 1,190 (1.0) | 1,177 (1.1) | 1,388 (1.2) | 682 (1.2) |
| Neurologic failure (%) | 1,952 (1.5) | 1,538 (1.7) | 2,518 (1.7) | 2,767 (1.9) | 2,131 (1.8) | 2,120 (2.0) | 2,715 (2.3) | 1,436 (2.6) |
| Sepsis (%) | 28,868 (22.0) | 2,1623 (23.6) | 35,268 (23.3) | 34,567 (23.5) | 32,376 (27.4) | 29,843 (27.6) | 32,903 (27.3) | 13,576 (24.8) |
| Gallbladder etiology (%) | 28,509 (21.7) | 18,633 (20.3) | 31,973 (21.1) | 30,958 (21.0) | 22,976 (19.5) | 20,834 (19.3) | 22,819 (18.9) | 9,524 (17.4) |
| State (%) | ||||||||
| Arkansas | 4,258 (3.2) | 4,472 (4.9) | 4,811 (3.2) | 4,523 (3.1) | 4,651 (3.9) | 4,340 (4.0) | 4,239 (3.5) | 0 (0.0) |
| Arizona | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 9533 (8.1) | 9,513 (8.8) | 10,098 (8.4) | 0 (0.0) |
| California | 45,099 (34.3) | 0 (0.0) | 48,094 (31.7) | 46,677 (31.7) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) |
| Florida | 28,551 (21.7) | 30,164 (32.9) | 31,191 (20.6) | 29,757 (20.2) | 29,740 (25.2) | 30,039 (27.8) | 32,446 (26.9) | 26,591 (48.5) |
| Kentucky | 7,672 (5.8) | 0 (0.0) | 8,390 (5.5) | 7862 (5.3) | 7852 (6.7) | 8,153 (7.5) | 8,054 (6.7) | 6,266 (11.4) |
| Maryland | 0 (0.0) | 8,870 (9.7) | 9,160 (6.0) | 9,032 (6.1) | 8,845 (7.5) | 8,997 (8.3) | 8,469 (7.0) | 0 (0.0) |
| Michigan | 13,917 (10.6) | 14,953 (16.3) | 15,132 (10.0) | 15,015 (10.2) | 15,177 (12.9) | 14,807 (13.7) | 15,158 (12.6) | 11,675 (21.3) |
| North Carolina | 12,432 (9.5) | 12,920 (14.1) | 13,622 (9.0) | 13,192 (9.0) | 13,366 (11.3) | 3,650 (3.4) | 13,008 (10.8) | 10,249 (18.7) |
| New York | 19,459 (14.8) | 20,316 (22.2) | 21,163 (14.0) | 21,203 (14.4) | 21,737 (18.4) | 21,719 (20.1) | 21,930 (18.2) | 0 (0.0) |
| Wisconsin | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 7,165 (6.1) | 6,999 (6.5) | 7086 (5.9) | 0 (0.0) |
SD = standard deviation; CCI = Charlson comorbidity index; MI = myocardial infarction; VTE = venous thromboembolism; CV = cardiovascular; Heme = hematologic.
Statistical methods
All analyses were conducted in R version 3.6.3 and p values less than 0.05 were considered statistically significant. Demographic characteristics, comorbidities, treatment strategy (whether patients received endoscopy, surgery, both, or neither), length of stay (LOS), mortality and complications were summarized using the R package “tableone” [7]. Crude incidence proportions and mortality rates were estimated by year across all included states and reported with 95% Wald confidence intervals. We defined incidence proportion as the number of patients admitted with pancreatitis divided by the total number of admissions in a given year. Mortality rates by year were similarly defined as the number of deaths among patients admitted for pancreatitis divided by the total number of pancreatitis admissions for which mortality was reported.
All models controlled for the following risk factors: age, gender, continuous Charlson comorbidity index (CCI), specific comorbidities on admission, gallbladder disease etiology, and insurance status. Mortality and complication rates were assessed using mixed effects logistic regression models implemented by the R package “lme4” [8] with random intercepts for each hospital. Treating year as continuous, we used an interaction term between year and gallbladder etiology to investigate whether the trend in mortality over time differed by etiology. The mortality model was re-fit after the addition of endoscopy, surgery, and all three-way and pairwise interactions between treatment strategy and etiology to determine the association between treatment strategy and mortality and whether these associations differed by etiology.
Trends in complication rates over time and differences by etiology were also examined using an interaction term between year and etiology. For the acute renal failure and sepsis complication models, the corresponding variables for whether these were present on admission were not included as covariates. This was because of complications being defined as having not been present on admission. Although we recognize organ failure on admission is likely a complication of the disease process as well, we are differentiating concomitant problems at presentation (present at admission) versus problems developed during admission (complications). Differences in treatment strategy by primary provider specialty (surgical vs. medical) were analyzed using the baseline category logit model implemented by the R package “mclogit” [9] with random intercepts for each hospital. Patients were categorized as receiving endoscopy only, surgery only, both, or neither with those receiving neither serving as the baseline group. An interaction term between primary provider specialty and etiology was used to determine whether the association between primary provider and treatment strategy differed for those with and without gallbladder etiology.
We evaluated trends in LOS over time and the association between treatment strategy and LOS by fitting negative binomial regression models using the R package “glmmTMB” [10] with random intercepts for each hospital. Similar to mortality, we fit two nested LOS models. The first included an interaction term between year and etiology to assess time trends and whether they differed by gallbladder etiology. The second model additionally incorporated endoscopy, surgery, and all three-way and pairwise interactions between treatment strategy and etiology. We also performed a sensitivity analysis to assess the impact of mortality as a competing risk by re-fitting the LOS models after removing the 21,532 patients who died.
Results
Demographic characteristics
We identified 923,459 admissions for pancreatitis between 2008 and 2015 across the ten included states. Age, gender, insurance status, and CCI were similar across each of the years (Table 1). Comorbidities on admission were also similar across years, with the observed proportion of patients with acute renal, hematologic, hepatic, and neurological failure increasing slightly over time. There was also some variability in the observed proportion of patients with sepsis on admission, with the observed proportion of patients in the years 2012–2014 being slightly higher than the other years. A total of 186,226 patients (20.2%) were classified as having gallbladder etiology, with the observed proportion of cases with gallbladder etiology decreasing slightly with time (Table 1). However, as shown in Table 1, not all states were represented in each year, therefore, it is possible that this inconsistency is contributing to these trends.
Crude incidence proportions and mortality rates
The crude incidence proportions and mortality rates for each year, along with corresponding 95% confidence intervals, are presented in Table 2 and plotted in Figure 1 and Supplementary Figure S1. The observed mortality rates decreased slightly each year, with the highest and lowest rates being 2.85% in 2008 and 2.01% in 2015, respectively. The crude incidence proportions range from 1.11% in 2015 to 1.37% in 2014. While the observed rates were higher in 2010–2014 than in 2008 and 2009, the incidence rate was lowest in 2015.
Table 2.
Pancreatitis Outcomes and Treatment by Year
| 2008 | 2009 | 2010 | 2011 | 2012 | 2013 | 2014 | 2015 | |
|---|---|---|---|---|---|---|---|---|
| n | 131,388 | 91,695 | 151,563 | 14,7261 | 118,066 | 108,217 | 120,488 | 54,781 |
| Mortality rates (%; 95% CI) | 2.85 (2.76, 2.94) | 2.54 (2.44, 2.64) | 2.45 (2.37, 2.52) | 2.36 (2.28, 2.43) | 2.11 (2.03, 2.19) | 2.07 (1.99, 2.16) | 2.04 (1.96, 2.12) | 2.01 (1.89, 2.13) |
| Incidence proportions (%; 95% CI) | 1.26 (1.26, 1.27) | 1.23 (1.22, 1.24) | 1.36 (1.36, 1.37) | 1.33 (1.33, 1.34) | 1.31 (1.30, 1.32) | 1.33 (1.32, 1.33) | 1.37 (1.36, 1.38) | 1.11 (1.11, 1.12) |
| Procedures | ||||||||
| Surgery (%) | 21,676 (16.5) | 14,061 (15.3) | 24,452 (16.1) | 23,595 (16.0) | 17,755 (15.0) | 16,133 (14.9) | 17,567 (14.6) | 7,640 (13.9) |
| Endoscopy (%) | 13,496 (10.3) | 8,921 (9.7) | 15,147 (10.0) | 14,969 (10.2) | 11,505 (9.7) | 10,818 (10.0) | 11,385 (9.4) | 4,602 (8.4) |
| Outcomes | ||||||||
| Died (%) | 3,739 (2.8) | 2,329 (2.5) | 3,706 (2.4) | 3,468 (2.4) | 2,489 (2.1) | 2,242 (2.1) | 2,457 (2.0) | 1,102 (2.0) |
| Length of stay, mean (SD) | 6.84 (9.79) | 6.71 (9.43) | 6.35 (9.04) | 6.28 (9.16) | 6.05 (8.27) | 6.03 (8.83) | 5.90 (8.23) | 5.78 (8.24) |
| Complications | ||||||||
| MI (%) | 618 (0.5) | 407 (0.4) | 644 (0.4) | 567 (0.4) | 483 (0.4) | 379 (0.4) | 419 (0.3) | 140 (0.3) |
| Pneumonia (%) | 1,430 (1.1) | 921 (1.0) | 1,501 (1.0) | 1,498 (1.0) | 1,281 (1.1) | 1,122 (1.0) | 1,193 (1.0) | 554 (1.0) |
| VTE (%) | 551 (0.4) | 448 (0.5) | 694 (0.5) | 718 (0.5) | 616 (0.5) | 582 (0.5) | 629 (0.5) | 285 (0.5) |
| Acute renal failure (%) | 2,956 (2.2) | 2,192 (2.4) | 3,525 (2.3) | 3,241 (2.2) | 2,743 (2.3) | 2,574 (2.4) | 2,795 (2.3) | 1,188 (2.2) |
| Gastrointestinal bleed (%) | 480 (0.4) | 368 (0.4) | 505 (0.3) | 481 (0.3) | 630 (0.5) | 567 (0.5) | 539 (0.4) | 268 (0.5) |
| Hemorrhage (%) | 688 (0.5) | 433 (0.5) | 678 (0.4) | 624 (0.4) | 492 (0.4) | 446 (0.4) | 478 (0.4) | 160 (0.3) |
| Pulmonary gailure (%) | 4,139 (3.2) | 3,010 (3.3) | 4,936 (3.3) | 4,712 (3.2) | 4,263 (3.6) | 3,860 (3.6) | 4,235 (3.5) | 1,814 (3.3) |
| Sepsis (%) | 3,903 (3.0) | 2,550 (2.8) | 4,400 (2.9) | 4,485 (3.0) | 2,684 (2.3) | 2713 (2.5) | 3,353 (2.8) | 1,650 (3.0) |
| SSI (%) | 579 (0.4) | 357 (0.4) | 565 (0.4) | 498 (0.3) | 386 (0.3) | 347 (0.3) | 354 (0.3) | 135 (0.2) |
CI = confidence interval; SD = standard deviation; MI = myocardial infarction; VTE = venous thromboembolism; SSI = surgical site infection.
FIG. 1.
Crude mortality by year for pancreatitis admissions. Crude mortality (blue dots) with 95% confidence intervals (blue lines).
Mortality
In-hospital mortality decreased significantly over time. A one-year difference was associated with 9.5% (OR, 0.90; 95% CI, 0.89–0.92) and 10.5% (OR, 0.90; 95% CI, 0.89–0.90) decreases in the odds of mortality for those with and without gallbladder disease etiology, respectively, after controlling for age, gender, CCI, etiology, insurance status, and selected comorbidities on admission (Table 3). Although patients without gallbladder disease etiology exhibited larger decreases in mortality over time, this difference was not statistically significant (p = 0.26). However, having gallbladder disease etiology was associated with a 40.3% (OR, 0.60; 95% CI, 0.57–0.62) decrease in the odds of mortality.
Table 3.
Temporal Trend of Mortality in Pancreatitis Admissions by Gallbladder Etiology
| OR (95% CI) | p | |
|---|---|---|
| (Intercept) | −5.467 (−5.52, −5.413) | <0.001 |
| Age, mean (SD) | 1.036 (1.034, 1.037) | <0.001 |
| Female (%) | 0.878 (0.852, 0.905) | <0.001 |
| Insurance status (%) | ||
| Medicare | Ref | Ref |
| Medicaid | 1.23 (1.168, 1.295) | <0.001 |
| Private | 1.054 (1.007, 1.103) | 0.024 |
| Self-paid | 1.223 (1.137, 1.316) | <0.001 |
| No charge | 0.867 (0.708, 1.062) | 0.167 |
| Other | 1.37 (1.247, 1.504) | <0.001 |
| CCI, mean (SD) | 1.222 (1.215, 1.23) | <0.001 |
| Hematologic failure (%) | 1.838 (1.763, 1.916) | <0.001 |
| Hepatic failure (%) | 4.164 (3.907, 4.438) | <0.001 |
| Neurologic failure (%) | 2.161 (2.047, 2.281) | <0.001 |
| Acute renal failure (%) | 2.377 (2.3, 2.457) | <0.001 |
| CV failure (%) | 4.802 (4.626, 4.985) | <0.001 |
| Sepsis (%) | 4.983 (4.822, 5.15) | <0.001 |
| Year (non-GB) | 0.895 (0.888, 0.903) | <0.001 |
| Year (GB) | 0.905 (0.89, 0.921) | <0.001 |
| Gallbladder etiology (%) | 0.597 (0.573, 0.623) | <0.001 |
Test for interaction between year and gallbladder etiology was not statistically significant (p = 0.258)
OR = odds ratio; CI = confidence interval; SD = standard deviation; CCI = Charlson comorbidity index; CV = cardiovascular; GB = gallbladder disease.
Increased age, CCI, and being male were each associated with increased odds of mortality. Hematologic failure, hepatic failure, neurologic failure, acute renal failure, cardiovascular failure, and sepsis on admission were each associated with large increases in the odds of mortality. In particular, cardiovascular failure (OR, 4.80; 95% CI, 4.63–4.99), sepsis (OR, 4.98; 95% CI, 4.82–5.15), and hepatic failure (OR, 4.16; 95% CI, 3.91–4.44) on admission were each associated with 380%, 398%, and 316% increases in the odds of mortality, respectively.
Notably, the overall rate of mortality for patients admitted with pancreatitis in the years 2008–2015 in the included states was low (2.33%). Because a mortality rate of 2.33% still amounted to 21,532 patients, we were not concerned with sparse data bias affecting the model coefficients. However, this low event rate was largely attributable to roughly 25% of the hospitals in the dataset reporting no deaths. The substantial proportion of no-event hospitals may have led to underestimation of the predicted probabilities from the mixed-effects model and motivated the decision to focus on the crude mortality rates per year rather than the model-based adjusted mortality rates. However, we confirmed that the large proportion of no-event hospitals did not affect the interpretability of the model coefficients via a sensitivity analysis in which we fit the same logistic regression model without random effects and observed that the model coefficients were congruous with the mixed-effects model (Supplementary Table S2).
Treatment (surgery and endoscopy)
Of the 923,459 patients with pancreatitis, 50,625 (5.48%) had only endoscopic procedures, 102,661 had only surgery, and 40,218 (4.36%) had both endoscopy and surgery; 729,955 (79.05%) patients had neither endoscopy nor surgery. After adding endoscopy, surgery, and all three-way and pairwise interactions between treatment strategy and gallbladder disease etiology, the estimated associations between mortality and each of the other covariates remained similar apart from etiology (Table 4). This model did not include an interaction between year and gallbladder etiology, because this was not statistically significant in the previous model. Unlike in the first mortality model, gallbladder disease etiology was associated with a 9.7% (OR, 1.01; 95% CI, 1.04–1.16) increase in the odds of mortality after adjusting for treatment strategy.
Table 4.
Pancreatitis Mortality Risk by Treatment Model
| OR (95% CI) | p | |
|---|---|---|
| (Intercept) | −5.452 (−5.505, −5.398) | <0.001 |
| Age, mean (SD) | 1.036 (1.034, 1.037) | <0.001 |
| Female (%) | 0.881 (0.854, 0.908) | <0.001 |
| Insurance status (%) | ||
| Medicare | Ref | Ref |
| Medicaid | 1.221 (1.16, 1.286) | <0.001 |
| Private | 1.051 (1.004, 1.1) | 0.032 |
| Self-paid | 1.208 (1.122, 1.299) | <0.001 |
| No charge | 0.857 (0.700, 1.049) | 0.135 |
| Other | 1.354 (1.233, 1.487) | <0.001 |
| CCI, mean (SD) | 1.218 (1.21, 1.226) | <0.001 |
| Hematologic failure (%) | 1.814 (1.74, 1.891) | <0.001 |
| Hepatic failure (%) | 4.108 (3.854, 4.379) | <0.001 |
| Neurologic failure (%) | 2.114 (2.003, 2.232) | <0.001 |
| Acute renal failure (%) | 2.323 (2.248, 2.402) | <0.001 |
| CV failure (%) | 4.647 (4.476, 4.825) | <0.001 |
| Sepsis (%) | 5.102 (4.936, 5.274) | <0.001 |
| Year | 0.896 (0.889, 0.903) | <0.001 |
| Gallbladder etiology (%) | 1.097 (1.040, 1.158) | 0.001 |
| Treatment (by GB etiology) | ||
| Endoscopy only (no GB) | 0.467 (0.415, 0.525) | <0.001 |
| Surgery only (no GB) | 1.404 (1.309, 1.507) | <0.001 |
| Both (no GB) | 0.671 (0.527, 0.855) | 0.001 |
| Endoscopy only (GB) | 0.407 (0.366, 0.454) | <0.001 |
| Surgery only (GB) | 0.334 (0.303, 0.367) | <0.001 |
| Both (GB) | 0.227 (0.198, 0.26) | <0.001 |
Test for three-way interaction between endoscopy, surgery, and gallbladder etiology was statistically significant (p = 0.004).
OR = odds ratio; CI = confidence interval; SD = standard deviation; CCI = Charlson comorbidity index; CV = cardiovascular; GB = gallbladder disease.
Among patients without gallbladder etiology, endoscopy only and both endoscopy and surgery were associated with 53.3% (OR, 0.47; 95% CI, 0.42–0.53) and 32.9% (OR, 0.67; 95% CI, 0.53–0.86) decreases in the odds of mortality, respectively, compared to having neither procedure. However, surgery only was associated with a 40.4% (OR, 1.40; 95% CI, 1.31–0.51) increase in the odds of mortality compared with having neither procedure. Among patients with gallbladder disease etiology, endoscopy only, surgery only, and both endoscopy and surgery were associated with 59.3% (OR, 0.41; 95% CI, 0.37–0.45), 66.6% (OR, 0.33; 95% CI, 0.30–0.37), and 77.3% (OR, 0.23; 95% CI, 0.20–0.26) decreases in the odds of mortality, respectively, compared with having neither procedure. The association between treatment strategy and mortality differed by gallbladder etiology (p = 0.004). Most notable is the difference in the association between surgery only and mortality; whereas surgery appears to be associated with mortality benefits for those having gallbladder etiology, there is an inverse association for those without gallbladder etiology. Given that the interaction was significant, a secondary analysis was completed by running the same models on each gallbladder disease and non-gallbladder disease population. The estimates did not change substantially, and these data are included in Supplementary Table S3 and Supplementary Table S4.
Complication rates
We observed substantial decreases over time in the odds of myocardial infarction, pneumonia, acute renal failure, hemorrhage, and surgical site infection when controlling for demographic characteristics, etiology, comorbidities on admission, and procedures conducted (Supplementary Figure S2 and Supplementary Table S5). Etiology substantially modified the association between year and the following complications: pneumonia (p = 0.04), venous thromboembolism (p < 0.001), acute renal failure (p = 0.004), hemorrhage (p < 0.001), pulmonary failure (p < 0.001), sepsis (p = 0.001), and surgical site infection (p < 0.001).
Although year was not associated with venous thromboembolism among those without gallbladder etiology, there was a 3.3% (OR, 0.97; 95% CI, 0.94–0.99) decrease in the odds of venous thromboembolism per year among those with gallbladder etiology. We observed increases in the odds of GI bleeding for those with and without gallbladder disease etiology (non-gallbladder disease OR, 1.03; 95% CI, 1.02–1.05; gallbladder disease OR, 1.03; 95% CI, 0.99–1.07), however, the association for those without gallbladder disease etiology was not statistically significant. A one-year difference was also associated with a 3.2% decrease and 1.5% increase in the odds of sepsis (non-gallbladder disease OR, 0.97; 95% CI, 0.96–0.98; gallbladder disease OR, 1.02; 95% CI, 1.003–1.03) for those with and without gallbladder etiology, respectively.
Sepsis on admission was common among those who later experienced complications; all patients who experienced acute renal failure as a complication had sepsis on admission. We were unable to estimate the odds ratio corresponding to sepsis on admission in the acute renal failure model because of quasi-complete separation, where all patients with acute renal failure as a complication had sepsis on admission. However, this should not impact the validity of the other estimated model coefficients. Because of near-quasi-complete separation in the other models, the odds ratios for sepsis on admission are large. Although having sepsis on admission was associated with a large increase in the odds of these complications (Supplementary Table S5), the overall complication rates are low (Table 1).
Length of stay
Length of stay decreased substantially over time. Each subsequent year was associated with a 3.6% (relative risk [RR], 0.96; 95% CI, 0.96–0.97) and 3.3% (RR, 0.97; 95% CI, 0.97–0.97) decrease in LOS for those with and without gallbladder etiology, respectively, after controlling for age, gender, CCI, etiology, insurance status, and selected comorbidities on admission (Supplementary Table S6). The association of year with LOS differed substantially by etiology (p < 0.001). Having gallbladder disease etiology was associated with a 12.1% (RR, 1.12; 95% CI, 1.12–1.13) increase in LOS, whereas increases in age and CCI and having any comorbidities present on admission were also associated with increased LOS.
We re-fit the LOS model after the addition of endoscopy, surgery, and all three-way and pairwise interactions between treatment strategy and gallbladder etiology and found the estimates for the other covariates to be similar between the two models (Supplementary Table S7). Endoscopy only (non-gallbladder disease OR, 1.26; 95% CI, 1.25–1.27; gallbladder disease OR, 1.06; 95% CI, 1.05–1.07) surgery only (non-gallbladder disease OR, 1.66; 95% CI, 1.65–1.67; gallbladder disease OR, 1.12; 95% CI, 1.11–1.13), and both procedures (non-gallbladder disease OR, 1.82; 95% CI, 1.78–1.86, gallbladder disease OR, 1.20; 95% CI, 1.19–1.22) were associated with increased LOS for those with and without gallbladder etiology. These increases were more pronounced for those without gallbladder etiology for all three treatment strategies (p < 0.001).
We performed a sensitivity analysis to assess the impact of mortality as a competing risk by re-fitting the LOS models after removing the 21,532 patients who died. The estimated effect sizes were similar when excluding these patients (Supplementary Table S8 and Supplementary Table S9). Whereas gender was not associated with LOS in the model not accounting for treatment strategy (p = 0.9), the association became significant after removing the 21,532 patients who died from the analysis (p = 0.03). Even so, the odds ratios corresponding to gender in both models only differed by 0.003.
Medical specialty
These results are based only on data from Wisconsin reported from 2012–2014 because it was the only state to report medical provider specialty. We considered the following four treatment strategies as a categorical outcome: endoscopy only, surgery only, both endoscopy and surgery, and neither endoscopy nor surgery. After adjusting for the aforementioned risk factors, we found that having a surgical provider as opposed to a medical provider was associated with a 20.5-fold (OR, 20.50; 95% CI, 16.81–25.01) increase in the odds of having surgery alone versus neither procedure among those without gallbladder disease etiology (Supplementary Table S10). This association attenuated to a 7.11-fold (OR, 7.11; 95% CI, 5.46–9.25) increase among those with gallbladder disease etiology. Similarly, having a surgical provider as opposed to a medical provider was associated with 7.0-fold (OR, 7.00; 95% CI, 5.22–9.37) and 8.8-fold (OR, 8.84; 95% CI, 5.61–13.96) increases in the odds of receiving both procedures versus neither procedure for those with and without gallbladder disease etiology, respectively. Surgical providers also appeared to have a higher likelihood of treating patients with endoscopy, albeit to a lesser extent than surgery. Having a surgical provider as opposed to a medical provider was associated with 1.2-fold (OR, 1.22; 95% CI, 0.87–1.72) and 1.6-fold (OR, 1.60; 95% CI, 1.18–2.16) increases in the odds of receiving endoscopy alone versus neither procedure for those with and without gallbladder etiology, respectively. The difference in the associations between provider specialty and treatment strategy by gallbladder disease etiology was statistically significant for surgery versus neither (p < 0.001), but not both versus neither (p = 0.39) or endoscopy versus neither (p = 0.25). Given the different clinical algorithms between gallbladder disease and non-gallbladder disease pancreatitis, the interaction term was included in the above model, however, we further investigated how medical specialty affected treatment decisions within the non-gallbladder disease population in isolation. Compared with an internal medicine provider, admission to a surgical provider was associated with a 63% increased odds of having an endoscopy only (OR, 1.63; 95% CI, 1.20–2.21) and 16-fold increased odds of surgery (with or without endoscopy) compared with no intervention (Supplementary Table S11).
Discussion
In our study, we observed a 10% decrease on average, per year, from 2008 to 2015 in the odds of hospital mortality among patients admitted with pancreatitis. The decline in mortality was significant and similar among those with and without gallbladder disease etiology, however, pancreatitis caused by gallbladder disease etiology was found to be associated with a 40% decrease in the odds of mortality compared with the non-gallbladder disease etiology cohort. Certain types of organ failure on admission, specifically cardiovascular, hepatic, and sepsis, were associated with three-fold increases in the odds of death. In addition to mortality, LOS and complications declined over time apart from GI bleeding, which was found to be increasing. Unsurprisingly, in gallbladder disease-associated pancreatitis, all interventions were associated with lower odds of mortality compared with no intervention. In non-gallbladder disease pancreatitis, only endoscopy and endoscopy with surgery were associated with lower odds of mortality, whereas surgery alone was associated with increased odds of mortality. Finally, in contrast to gallbladder disease-associated etiology that is likely to result in surgical interventions, the interventions may vary by admitting provider specialty in non-gallbladder disease-associated pancreatitis. Among these patients, we found that being admitted under the care of surgical providers had higher odds of performing surgery and endoscopy compared with care by medical providers.
Consistent with prior research, the in-hospital mortality for patients admitted with pancreatitis continues to decrease over time. In general, overall hospital [11] and intensive care unit (ICU) [12] mortality have been declining over decades, however, more recent data, specifically in the ICU, would suggest a plateau in these improvements whereas morbidity is increasing [13,14]. Overall estimates are not representative of each individual illness and thus evaluating entities in isolation is key to determining future targets for improvement. We found that this decrease in mortality coincides with an incidence rate that has fluctuated between 1.37% and 1.11% over time. Although there are spikes in incidence during the eight-year period of this study, mortality consistently decreased. Mandalia et al. [3] have also observed conflicting trends for pancreatitis incidence in the United States, with some studies showing an increase while others demonstrating a decrease. The difficulty appreciating a clear signal may stem from the multiple etiologies of pancreatitis and the heterogenous sequelae associated with each culprit. Furthermore, we speculate that different databases, time periods of study, and measuring metrics (acute pancreatitis vs. chronic pancreatitis vs. both) may account for the variance in reporting. To evaluate this concern, we separated gallbladder disease associated pancreatitis from those without gallbladder disease associated pancreatitis. Admission with gallbladder disease-associated pancreatitis was associated with 40% lower odds of mortality compared with non-gallbladder disease-associated pancreatitis. These findings are supported by Yadav et al. [15], who suggested that gallstones are the leading cause of acute pancreatitis and early cholecystectomy reduces the risk of future flares. Cholecystectomy in patients with pancreatitis caused by gallbladder disease etiology decreases mortality because of fixing the root cause of the acute pancreatitis. In regards to variation in temporal changes within each cohort, overall mortality declines were present in both subgroups. The degree of decline was not different between the two, suggesting that differences in gallbladder etiology are unlikely to be responsible for declining mortality over time. Luckhust et al. [1] reported that the decrease in overall mortality may be due to the switch from open surgical necrosectomy (OSN) to minimally invasive surgery (MIS) regarding pancreatic necrosectomy, which has substantially decreased new post-operative organ failure. These interventions may occur in both cohorts and affect patients later during their disease course. The switch is also correlated with a decreased hospital LOS, a trend also noticed in our study [1]. We observed a decrease in LOS from 6.8 (IQR, 9.8) days to 5.8 (IQR, 8.2) days from 2008–2015 which correlates with the decreasing mortality rate. Again, this decrease over time was observed in each subgroup, however, gallbladder disease etiology was associated with a longer LOS than those with non-gallbladder disease etiology. Although some studies have found population mortality of pancreatitis in the United States has remained unchanged, our findings along with other studies estimated decreasing case fatality rates [5,16,17]. Our study is limited to the ten states included in the HCUP databases that could account for the discrepancy, however, our studies are consistent with previously observed decrease in case fatality rates over time and few have assessed this trend in more recent years. Despite the discrepancies in reporting, the mortality trend for pancreatitis, with acute and chronic combined, remains low (approximately 2%–3%).
Our study additionally focused on acute renal failure, myocardial infarction (MI), pneumonia, venous thromboembolism, cardiovascular (CV) failure, hematologic failure, hemorrhage, hepatic failure, neurologic failure, and sepsis as comorbidities present on admission for pancreatitis. We found that hematologic, hepatic, neurologic, acute renal, and CV failure along with sepsis were associated with increased odds of mortality, which is to be expected. Most other studies have focused on environmental and behavioral risk factors in conjunction with pancreatitis such as alcohol, cannabis, tobacco smoking, etc., but we expect future studies to produce similar results to those shown here [3,15]. Importantly, the presence of sepsis, hepatic, and CV failure were clear among the higher risk presenting co-conditions and may represent specific high-yield subgroups for targeted research to further improve outcomes in patients with pancreatitis.
The complications and rate of complications of patients treated in the hospital for pancreatitis were also analyzed. We found decreases in the odds of myocardial infarction, pneumonia, acute renal failure, hemorrhage, respiratory failure, sepsis, and surgical site infection during the timeframe of this study whereas the odds of GI bleeding consistently increased. The increase in GI bleeding has been observed in other studies recently. One study explains this phenomenon as having two possible etiologies: pancreatic enzymatic degradation of local vessels and variceal bleeding caused by splenic vein thrombosis causing portal hypertension [18]. We suspect these are the underlying reasons we found increases in GI bleeding, because many patients with pancreatitis also had comorbidities for risks of bleeding events.
Last, over the past few decades there has been a focus on prevention of hospital-acquired venous thromboembolism. This has led to a large number of successful interventions increasing the rate of hospitalized patients on prophylaxis that may have been a contributing factor as well [19]. Although GI bleeding has increased in incidence, a general trend toward decreased complication rates was observed. Mandalia et al. [3] reviewed the risk stratification tools for identifying individuals at risk for developing severe negative outcomes caused by pancreatitis. The review found that accurate identification of the severity of acute pancreatitis within 48–72 hours led to decreases in morbidity and mortality because of proper initiation of treatment. The Bedside Index of Severity in Acute Pancreatitis (BISAP) (blood urea nitrogen [BUN], impaired mental status, systemic inflammatory response syndrome, age, and pleural effusion), Acute Physiology and Chronic Health Evaluation II (APACHE II), CT severity index, and Ranson predictive models all show similar accuracy in predicting the severity of pancreatitis [3]. We suspect that advances and compliance in the use of these risk stratification tools is associated with the decrease in complication and mortality rates observed in our study.
Treatment effects patients with gallbladder disease versus non-gallbladder disease etiology pancreatitis differed in our present study. Gallbladder disease-associated pancreatitis had lower odds of mortality if they received surgery (OR, 0.31; 95% CI, 0.28–0.34), endoscopy (OR, 0.39, 95% CI 0.35–0.44), or both (OR, 0.20, 95% CI 0.18–0.23) in comparison to no intervention at all. This is not surprising considering many studies show the recommended treatment for gallstones or other gallbladder diseases responsible for causing acute pancreatitis is early cholecystectomy or endoscopy [15,20]. Interestingly, the interaction was significant in our model, which warrants further consideration as to why those with both interventions appeared to have the lowest odds of mortality. Intuitively, more interventions suggest higher illness severity, however, there are exceptions (i.e., necrotizing pancreatitis) in which interventions are intentionally delayed until the degree of inflammation lessens and the disease process has time to “wall off.” Although we were unable to adjust for this through the lack of patient level laboratory and vital sign data, these findings warrant further investigation. Non-gallbladder etiology pancreatitis patients have a 43% increased odds of mortality with surgical intervention, whereas endoscopy alone or both were associated with decreased odds of mortality compared with neither intervention. Buchler et al. [21] explains this phenomenon as an artifact of the high incidence of severe acute necrotizing sterile and infected pancreatitis that have historically been treated with surgery at a rate of 60%–70%. The team discovered that using early conservative treatments, such as antibiotic agents, for sterile necrotizing pancreatitis is safe and effective [21]. In lieu of these findings, it could be expected that the surgical interventions for non-gallbladder etiology pancreatitis may slowly decrease in the future.
Although endoscopy and surgery are well known to be indicated in most cases of gallbladder disease associated pancreatitis, the intervention needed will likely inform which team cares for the patient during the admission [22]. In contrast, the bias in treatment strategies for non-gallbladder disease etiology pancreatitis according to provider specialty has yet to be fully studied. This is not unique. For example, epidemiologic studies assessing critically ill patients commonly separate surgical and medical patients given the differences in disease processes [13,14,23]. One aspect that is not well studied is the effect of the provider's specialty (medical vs. surgical) on outcomes when caring for the same disease process. We speculate this deficiency is because of the lack of specialty reporting in databases, especially large national databases as used in our study. Similarly, despite having access to multiple states' data, Wisconsin was the only state in the HCUP database that included the attending provider's specialty. Admission to a surgical provider, compared with a medical provider, was associated with increased odds of endoscopy alone (OR, 1.63; 95% CI: 1.20–2.21) and surgery (OR, 16.56; 95% CI: 13.62–20.17) compared with neither intervention during admission for non-gallbladder etiology. One could argue two reasons for this phenomenon. First, patients in need of surgery for pancreatitis are more likely to be referred to a surgical specialty than a medical specialty as in the case of acute necrotizing pancreatitis. Second, a specialty bias is present. Sah et al. [24] studied the effects of surgical physician bias, patient trust, and treatment choice in depth and found that specialty bias is inevitable. Physician knowledge of treatment options within their own specialty far outweigh their knowledge of alternative treatment options [24]. This same group found that even with physician disclosure of surgical specialty bias, patients were more likely to choose surgical treatments and displayed greater trust in the physician [24]. Further research would be needed to distinguish the actual root cause of our findings, however, these two hypotheses are equally as likely and realistically are both partially contributing to some degree. Furthermore, these findings also support crucial next steps in assessing if provider specialty affects patient outcomes given the variations in treatment and outcomes seen here, specifically in necrotizing pancreatitis where specific guidelines are lacking as to when endoscopy, surgery, or both are appropriately timed and balanced.
Limitations
The findings presented in this study should be appropriately taken in the context of the following limitations. This is a retrospective study and despite including multiple key covariates in all our models, unknown confounders could not be accounted for in this analysis. ICD-9 codes and Current Procedural Terminology codes were used for this analysis, which are subject to administrative errors. Providers were only documented in Wisconsin that limited the data used to assess admitting provider effects, and results may not be generalizable outside of the population studied. Moreover, the assumption was made that patients who were admitted to a surgical or medical provider continued to receive care by this provider, or at least other providers from the same specialty, and we were not able to confirm that this assumption was true in every patient encounter. This analysis was exploratory, and we did not correct for multiple comparisons. These results should not be interpreted causally and could be subject to residual confounding. For example, although we found that the provider having a surgical specialty was associated with being treated with endoscopy, surgery, or both, it is possible that patients who were more likely to need a surgical procedure were assigned to a surgeon, even after adjusting for comorbidities. These data are from inpatient admissions and thus long-term and post-hospitalization outcomes were not evaluated. Finally, these models were developed for hypothesis testing and may not perform well in prediction.
Conclusions
We observed decreased length of stay, mortality, and complication rates from pancreatitis over time after controlling for demographic characteristics, gallbladder etiology, and comorbidities on admission. Significant differences in outcomes and treatment options depends on whether the etiology was gallbladder related or not. These findings support the separation of these populations when assessing pancreatitis in future studies or at least adjusting for the etiology in analysis. We also noted specialty differences in treatment preference. Additional work is also needed to determine whether the observed decrease in mortality and increase in LOS associated with having an endoscopy or surgery is the result of the procedure themselves.
Supplementary Material
Authors' Contributions
All authors contributed substantially to developing, writing, and revising this manuscript.
Funding Information
This work was supported by NIH NHLBI T32HL07741 (to N.E.I.); the Agency for Healthcare Research and Quality (AHRQ) and Patient-Centered Outcomes Research Institute (PCORI), grant K12HS026379 (to C.J.T.); the National Institutes of Health's National Center for Advancing Translational Sciences, grant KL2TR002492 and UL1TR002494; the NIH NHLBI T32HL129956 (to J.P., L.S.); and AHRQ R01HS026732 (to M.G.U.).
Author Disclosure Statement
No competing financial interests exist for all authors.
Supplementary Material
References
- 1. Luckhurst CM, El Hechi M, Elsharkawy AE, et al. Improved mortality in necrotizing pancreatitis with a multidisciplinary minimally invasive step-up approach: Comparison with a modern open necrosectomy cohort. J Am Coll Surg 2020;230:873–883. [DOI] [PubMed] [Google Scholar]
- 2. Dirweesh A, Li Y, Trikudanathan G, et al. Clinical outcomes of acute pancreatitis in patients with coronavirus disease 2019. Gastroenterology 2020;159:1972–1974. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Mandalia A, Wamsteker EJ, DiMagno MJ. Recent advances in understanding and managing acute pancreatitis. F1000Res 2018;7:F1000. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Lew D, Afghani E, Pandol S. Chronic pancreatitis: Current status and challenges for prevention and treatment. Dig Dis Sci 2017;62:1702–1712. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. 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]
- 6. Ingraham NE, Lotfi-Emran S, Thielen BK, et al. Immunomodulation in COVID-19. Lancet Respir Med 2020;8:544–546. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Yoshida K. Package ‘tableone’. 2020. [Google Scholar]
- 8. Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. J Stat Softw 2015;67:1–48. [Google Scholar]
- 9. Elff M. mclogit: Mixed conditional logit models. R package. 2018. [Google Scholar]
- 10. Magnusson A, Skaug H, Nielsen A, Berg C, Kristensen K. Package ‘glmmTMB’. [Google Scholar]
- 11. Joynt KE, Orav EJ, Zheng J, Jha AK. Public reporting of mortality rates for hospitalized Medicare patients and trends in mortality for reported conditions. Ann Intern Med 2016;165:153–160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Zimmerman JE, Kramer AA, Knaus WA. Changes in hospital mortality for United States intensive care unit admissions from 1988 to 2012. Crit Care 2013;17:R81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Ingraham NE, Vakayil V, Pendleton KM, et al. National trends and variation of functional status deterioration in the medically critically ill. Crit Care Med 2020;48:1556–1564. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Ingraham NE, Vakayil V, Pendleton KM, et al. Recent trends in admission diagnosis and related mortality in the medically critically ill. J Intensive Care Med 2020:885066620982905. [DOI] [PubMed] [Google Scholar]
- 15. Yadav D, Lowenfels AB. The epidemiology of pancreatitis and pancreatic cancer. Gastroenterology 2013;144:1252–1261. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Munigala S, Yadav D. Case-fatality from acute pancreatitis is decreasing but its population mortality shows little change. Pancreatology 2016;16:542–550. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Yadav D, Lowenfels AB. Trends in the epidemiology of the first attack of acute pancreatitis: a systematic review. Pancreas 2006;33:323–330. [DOI] [PubMed] [Google Scholar]
- 18. Bugiantella W, Rondelli F, Boni M, et al. Necrotizing pancreatitis: A review of the interventions. Int J Surg 2016;28:S163–S171. [DOI] [PubMed] [Google Scholar]
- 19. Kahn SR, Morrison DR, Cohen JM, et al. Interventions for implementation of thromboprophylaxis in hospitalized medical and surgical patients at risk for venous thromboembolism. Cochrane Database Syst Rev 2013(7):CD008201. [DOI] [PubMed] [Google Scholar]
- 20. Janisch NH, Gardner TB. Advances in management of acute pancreatitis. Gastroenterol Clin North Am 2016;45:1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Buchler MW, Gloor B, Muller CA, et al. Acute necrotizing pancreatitis: treatment strategy according to the status of infection. Ann Surg 2000;232:619–626. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Greenberg JA, Hsu J, Bawazeer M, et al. Clinical practice guideline: management of acute pancreatitis. Can J Surg 2016;59:128–140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Vakayil V, Ingraham NE, Robbins AJ, et al. Epidemiological trends of surgical admissions to the intensive care unit in the United States. J Trauma Acute Care Surg 2020;89:279–288. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Sah S, Fagerlin A, Ubel P. Effect of physician disclosure of specialty bias on patient trust and treatment choice. Proc Natl Acad Sci USA 2016;113:7465–7469. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.