Order Set Use and Education Association With Pediatric Acute Pancreatitis Outcomes

We conducted a retrospective chart review of those children admitted with a diagnosis of an episode of AP to 1 of the 3 hospitals that comprise a large system of freestanding children’s hospitals in the southeastern United States from January 1, 2017, to December 31, 2018. Given that 1 center lacks gastroenterology specialists and refers all cases of AP to the other 2 hospitals, analysis was performed accounting for the primary 2 hospitals. Our electronic medical record (EMR) identified AP episodes through appropriate International Classification Diseases, Ninth Revision or International Classification Diseases, 10th Revision codes for AP. Criteria for inclusion into the study include the following: (1) meeting International Study Group of Pediatric Pancreatitis: In search for a Cure criteria definition for pancreatitis, (2) age <21 years, 3) initial presentation being at our children’s hospital, and (4) required hospital admission. Patients whose AP was secondary to trauma or gallstone or was drug induced (eg, asparaginase) were excluded because of the effect of narcotic use and LOS. Consistent with previous publications related to AP in children, each AP admission that occurred >3 months from the last episode was considered an independent event in the analysis.^5–10 

A combination of EMR reports and manual chart review was used to determine demographic information, date of admission, order set usage, laboratory work, imaging, number of doses of analgesia (narcotic versus other) given, fluid rate and choice (ie, 1.5 maintenance rate versus less, lactated Ringer’s [LR] solution versus saline), early diet initiation (within 48 hours), etiology of AP (idiopathic, viral, genetic, trauma, or chemotherapy), LOS, 30-day readmission, primary service, requirement of PICU admission, and hospital site (Supplemental Figure 1). By using the NASPGHAN AP clinical report guidelines for the management of pediatric AP, a pancreatitis order set was designed to encourage the practices suggested by the guidelines. The pancreatitis order set was activated January 1, 2018; thus, 2018 represents the first year that practice guidelines for treatment management were provided to all services in the EMR. In addition to activating the order set in January 2018, the senior author (A.J.F.) led several educational talks regarding the treatment of AP and use of the order set to hospitalists, intensivists, emergency department physicians, and gastroenterologists throughout the hospital system, primarily at hospital 1.

All data were collected and recorded in a password-protected Excel spreadsheet, to which only the primary investigator had access. This study was approved by, and conducted in accordance with, the internal review board guidelines. Before analysis, all information was deidentified and given a study specific identifier with no private health information distributed.

Multiple hospitalizations per patient separated by at least 3 months were included as independent pancreatitis encounters per previous literature review.^5–10  Because of the skewed LOS distribution, the natural log transformation of LOS was used to meet the assumption of a normality. As a result, geometric mean, as well as its SD, was calculated for LOS. LOS statistics were reported after a reverse transformation as days. To test the relationship of the order set usage on patient characteristics, order set characteristics, and outcomes such as logged LOS and readmission, mixed-effects models were used to control for the clustering within hospitals. All P values are based on likelihood ratio tests from clustered analyses. Statistical analysis was performed by using R 3.6.0 (R Foundation for Statistical Computing, Vienna, Austria), with P < .05 representing statistical significance.¹¹ 

A total of 189 AP encounters were studied between January 1, 2017, and December 31, 2018. Of these encounters, 22 patients had multiple hospitalizations for pancreatitis encounters, resulting in 137 unique patients. Among the 189 encounters, 30 were excluded because of failure to meet inclusion criteria (AP was secondary to trauma, drug induced, or cardiogenic shock or ischemic injury) or because of readmission within 3 months, resulting in 159 encounters included in the analysis. A total of 5 patients initially screened had gallstone pancreatitis or drug-induced pancreatitis. The remainder of our patients analyzed had idiopathic pancreatitis. Episodes of trauma, drug-induced, gallstone, and ischemic pancreatitis were excluded because their treatment course and LOS may have been impacted by several other clinical factors. There were no differences in regard to age, race, or physician service by year, but there were significantly more female patients treated for AP in the 2017 group (72% vs 51%; P = .010) (Table 1).

The absolute reduction in mean LOS between 2017 and 2018 was 0.28 days in 2018 (Table 2). When using the log transform of LOS in a linear mixed-effects model to account for clustering by hospital, use of the AP order set was associated with a 10% decrease in LOS between 2017 and 2018, although this decrease was not statistically significant (P = .30). However, in hospital 1, in which the majority of education was focused and completed because this is the primary workplace for the educating author, there was a decrease in LOS by 0.73 days, which is clinically significant but again not statistically significant (Table 2). Despite more aggressive fluid and dietary management, when comparing clinical outcomes by year, there were no statistically significant differences between PICU admission frequency or 30-day readmission rate frequency (P = .31 and P = .25, respectively) (Table 2).

In 2018, LR solution was used significantly more often (42% vs 20%; P = .004) and narcotic use was less often (76.7% vs 61.6%; P = .05), but there were no differences among IVF rate, diet, or timing of feeding (Table 3). In addition, the mean number of narcotic doses (ie, among those who received at least 1 dose) decreased from 5.2 in 2017 to 3.6 in 2018 (P = .001) (Table 4).

Increased recognition and prevalence of AP led to the first published clinical report with recommendations for AP management in children by the NASPGHAN Pancreatitis and Cystic Fibrosis Committee in 2018. However, the impact of these recommendations, which are based largely on expert opinion, has yet to be assessed. The purpose of this study was to test the effectiveness of these recommendations on clinical outcomes. Given the more aggressive treatment approach proposed by the clinical report, including higher IVF rates, earlier initiation of oral and enteral intake, and nonnarcotic pain medication choices, we focused on the impact of these recommendations on LOS, PICU admission rates, and 30-day readmission rates as the primary outcomes of this study. The primary exposure in this study was year (2017 or 2018), and this was chosen because January 1, 2018, was the first date at which the order set was implemented with simultaneous education surrounding pancreatitis to various practice physicians in the hospital. Using year as the exposure variable is more encompassing of a change in practice patterns rather than simply using order set use or lack thereof. The authors believe that some providers did not use the order set in 2018 but were able to provide appropriate care on the basis of the education they had received.

Importantly, our data revealed that implementation of the AP order set was associated with a 10% decrease in LOS. Although this failed to meet statistical significance, these data were largely impacted by the increase in hospital stay from 2017 to 2018 in hospital 2, in which far less education regarding AP occurred. When solely evaluating hospital 1, LOS decreased by 25%, which is clinically significant. It is no surprise that earlier discharge from the hospital has been associated with improved quality of life and is an important outcome measure because many pediatric hospitals look to evaluate ways to decrease LOS to combat overcrowding within their hospitals.^12,13  Additionally, the cost of a pancreatitis admission with a mean LOS of 5 days is estimated to be ∼$15 000.¹⁴  The decrease in LOS observed in our overall cohort would result in decrease in cost occurred of ∼$840 per admission, or $87 200/year at our center. If looking at hospital 1 only where the decrease in LOS was greater, the decrease would be an approximate cost of $2190 per admission and $175 200/year at our center. Critically, the more aggressive treatment approach with earlier discharge in this study did not result in any adverse clinical outcomes, including increase in PICU admissions or in 30-day readmission rates.

Additionally, appropriate use of narcotic medications remains a significant concern in pediatric AP management. Pain control is one of the most challenging aspects of clinical management of this illness. In a single-center study, researchers reported that opiate medications are used more frequently than nonsteroidal antiinflammatory drugs, specifically in the emergency department, and increased opioid doses were prescribed if an opioid was the initial analgesic provided.⁵  The same study also revealed that patients who received opioids as initial management of an episode of AP were more likely to require further doses of any type of analgesia than those whose initial pain medication was nonopioid with accompanying increased hospital LOS and increased health care costs.⁵  The current order set encouraged providers to use nonopiate medications for pain control. This study’s data indicate a decrease in overall narcotic use after order set implementation and physician education (Table 3). Simultaneously, our results indicate that among those patients who received at least a single dose of narcotic analgesic, the median number of narcotic doses decreased by 2 per AP episode (P = .001) (Table 4). Given the increasing rates of prescription opioid use in the adolescent population and long-lasting impact on neuropsychological and cognitive development of a child, order set use may encourage physicians to elect nonopiate pain control, thereby ultimately decreasing opiate usage.¹⁵  Although this study was not designed or powered to assess the relationship between LOS and narcotic use, it is suggestive that decreasing narcotic use may aid in earlier discharge in a subset of pediatric patients who are experiencing more severe pain and require admission for AP.

In addition to the evaluation of these outcomes, further aspects of the order set that were evaluated include data on treatment decisions focused on fluid resuscitation and dietary choice collected before and after order set implementation. These variables are important to evaluate because many of these AP treatment options are still being studied in the pediatric population to optimize disease management. Small, single-center studies in pediatric AP suggest that aggressive fluid resuscitation is associated with improved clinical outcomes, including decreased multiorgan failure, mortality, and progression to severe AP and decreased LOS.^8,16,17,18  Adult literature largely favors the use of LR solution over normal saline, but it has not been well studied in pediatrics.^17,19  Unlike in adult AP, a low-fat diet has not been shown to be protective in pediatric AP.¹⁸  Although LR solution use was significantly higher in 2018, IVF rate, timing of initiation of a diet, and type of diet (low fat versus normal) were not different in this patient population. This may be suggestive that LR solution may be superior to normal saline in decreasing LOS, but causality cannot be determined from our data and deserves additional study.

Together, our results reveal that the management approach recommended by the NASPGHAN clinical report and implemented through an AP order set along with focused provider education, specifically with respect to early feeding, aggressive LR solution hydration, and pain management, is safe and effective. This management strategy is associated with limiting narcotic use without increasing PICU admission or 30-day readmission rates. These data support the use of the recommendations provided by the NASPGHAN clinical report by pediatric gastroenterologists, hospitalists, and other providers for the treatment of pediatric AP.

Our study is limited in its retrospective nature, which prevents the ability to assess causation between the reported associations. Patients with AP encounters that were separated by >3 months were included in this study. Although it is understandable that these repeat episodes within a single patient may be considered linked and thus are a potential limitation to our study, including such patients was done to compare data with previous studies that took a similar approach.^5–10  Additionally, there are a number of variables we are unable to control for including initial care provided in the emergency department, previous treatment during a previous admission, and interphysician and interdepartment variation in practice. As a single-network study, the external validity of our results deserves further exploration.

Implementation of an AP order based on the management strategy recently recommended by the NASPGHAN clinical report on AP is safe and effective in decreasing narcotic use without increase PICU admission or 30-day readmission rates. Researching the impact of these recommendations in a larger, multicenter study is still needed. Furthermore, the AP order set may serve as an example of how order set use and hospital-wide education may be used to standardized care and improve clinical outcomes in less common pediatric conditions.