An Evidence-Based Care Protocol Improves Outcomes and Decreases Cost in Pediatric Appendicitis

Abstract

Background:

Appendicitis is a common indication for urgent abdominal surgery in the pediatric population. The post-operative management varies significantly in time to discharge and cost of care. The objective of this study was to investigate whether implementation of an evidence-based protocol after an appendectomy would lead to decreased length of stay and cost of care.

Study Design:

In 2014 at the Children’s Hospital of Pittsburgh, an initiative to develop an evidenced-based protocol to treat appendicitis was undertaken. A work group was formed of pediatric surgeons and other important personnel to determine best practices. Treatment pathways were created. Pathways differed with recommendation on post-operative antibiotic choice and duration, diet initiation, and discharge criteria. Data was prospectively gathered from all patients (ages 0-18 years) with acute appendicitis from January 2015 to December 2016. Primary outcomes were length of stay and cost of care. Secondary outcomes were surgical site infection, readmission rate, and duration of post-operative antibiotics.

Results:

Among the 1289 patients, 481 patients were in the pre-protocol cohort and 808 patients were in the post-protocol cohort. 27% of patients had an intraoperative diagnosis of complicated appendicitis. There was a significantly shorter length of stay in the post-protocol cohort(p < 0.001). Median costs for the whole cohort decreased 0.6% and 24.6% for patients with complicated appendicitis after protocol initiation (p < 0.01).

Conclusions:

This study has demonstrated that introduction of an evidence-based clinical care protocol for pediatric patients with appendicitis leads to shorter hospital stay and decreased hospital costs.

Introduction

Appendicitis continues to be one of the most common indications for urgent abdominal surgery in the pediatric population with overall incidence of 3.2% by age 20 (1). The management can vary with from antibiotics alone, antibiotics with interval appendectomy, percutaneous drainage to definitive appendectomy based upon clinical presentation and surgeon preference. With about 60,000-80,000 pediatric appendectomies performed annually, appendicitis accounts for a significant portion of surgery-related costs (2). This translates to a mean cost of about $9,000 per patient and an annual aggregate cost of over 3 billion dollars in the United States (2–5). Although our understanding of the appropriate management of appendicitis has improved in recent years, there continues to be substantial variability among surgeons regarding the care of these patients (5).

In patients who undergo appendectomies, the greatest variabilities in clinical care lie in the time to discharge as well as the choice and duration of antibiotic treatment (6–8). Implementation of data-driven protocols for the management of patients with disease processes such as low birth weight and congenital diaphragmatic hernias improves clinical outcomes(9–13). In January 2015, an evidence-based protocol was set in place at the UPMC Children’s Hospital of Pittsburgh (CHP) for patients undergoing appendectomy for appendicitis. In this study, we aim to examine the outcomes of these patients with a specific focus on length of stay (LOS), healthcare cost, the type and course of antibiotic use, and postoperative complications. We hypothesized that the introduction of an evidence-based clinical care protocol would decrease LOS and health care cost in the post-intervention group without compromising clinical outcomes.

Methods

Protocol Development

In 2014, an initiative to develop an evidenced-based protocol to treat appendicitis was undertaken at UPMC Children’s Hospital of Pittsburgh (CHP), a free-standing children’s hospital and the only tertiary pediatric hospital serving Western Pennsylvania. A work group was formed of pediatric surgeons, nurses, pharmacists and infectious disease specialists to review the literature and hold discussions with other key stakeholders including emergency medicine physicians, anesthesiologists, and operating room staff, in order to determine best practices. Treatment pathways were created to include specific guidelines regarding pre-operative antibiotics, blood work, and imaging requirements. After surgery, the protocol pathways were separated based on intraoperative findings of uncomplicated versus complicated appendicitis. Pathways differed with recommendation on post-operative antibiotic choice and duration, diet initiation, and discharge criteria. Protocols were initially used solely for patients with acute appendicitis who then underwent an appendectomy. Once the treatment pathways were finalized (Figure 1), order sets with all components of the protocols were created in the electronic medical record (EMR) ordering system to facilitate ease of use. Additionally, an eRecord dashboard was established, allowing us to prospectively track outcomes and pathway compliance in real time. The protocol was implemented on January 1, 2015.

Figure 1 –

Operative appendicitis treatment pathway

Patients and Indications

Data was prospectively gathered from all patients with acute appendicitis presenting to CHP from January 2015 to December 2016. This data was then compared to retrospective data collected from patients presenting to CHP with acute appendicitis in the time immediately preceding this, spanning from November 2013 to December 2014 prior to establishing protocols. Patients were excluded if they underwent nonsurgical treatment of appendicitis or percutaneous abscess drainage followed by an interval appendectomy. This study was performed under the approval of the University of Pittsburgh institutional review board, protocol number PRO16120214.

Data Collection

Patient data was collected for both the pre and post intervention cohorts including patient age, sex, race, length of stay (LOS), preoperative antibiotic use, postoperative antibiotic duration, use of postoperative imaging, post-operative complications, the use of peripherally inserted central venous catheters, as well as 30-day readmission (Table 1). Patient were separated by the severity of appendicitis. Uncomplicated appendicitis was defined by simple or gangrenous appendicitis, meanwhile complicated appendicitis was defined as appendicitis with an abscess or perforation. The diagnosis of perforation was based off of intraoperative finds of either a visible hole or presence of an extra-luminal fecalith in the peritoneal cavity (14).

Table 1:

Demographic Characteristics

	Pre (n=481)	Post (n=808)	P value
Age, years, mean(SD)	10.8(3.9)	10.9(3.5)	0.69
Sex, female, n(%)	189(39.3)	316(39.1)	0.95
Race, white, n(%) Race, nonwhite, n(%)	416(86.5) 65(13.5)	680(84.2) 128(15.8)	0.26
Uncomplicated appendicitis, n(%) Complicated appendicitis, n(%)	352(73.2) 129(26.8)	591(73.1) 217(26.9)	0.99

Cost Analysis

A brief cost analysis was performed based upon data for each patient obtained from an internal cost management system in the Department of Finance at CHP. The costs are reflective of the actual cost of care provided to each patient with appendicitis. This includes all the costs to our hospital for the initial admission including the components of care that occurred in the Emergency Department, any medications administered, supplies, radiology tests, blood work obtain, as well as room and board (which includes nursing care hours and unit costs). This data does not include the costs incurred in regard to any readmissions or subsequent Emergency Department visits.

The total cost of care does include all of the care that occurred in the initial episode, including the components of care that occurred while that patient was in the Emergency Department (radiology exams, labs, pharmaceuticals, etc.). The cost data does not include subsequent encounters of care for the management of appendicitis (readmission, subsequent ED visits, clinic visits).

Data Analysis

Pathway adherence was determined by whether patients were ordered and received all components of the pre-specified clinical pathway in our hospital’s electronic medical record system (Figure 1). Patients who were found to have deviated from the pathway were deemed “off pathway.” Continuous data were compared using an unpaired, unequal variance t-test; the Wilcoxon rank-sum test was utilized for non-normally distributed continuous data. Dichotomous outcomes were compared using a Chi-squared test or Fisher’s Exact Test, as appropriate. Odds ratios and exact 95% confidence intervals were computed as a ratio of the post-protocol group to pre-protocol group. Homoscedasticity was assessed using the Brown-Forsythe Test. Correlation was assessed using Spearman’s rho.

The primary outcome variable of length of stay (in hours) was analyzed in an unadjusted fashion by Wilcoxon rank-sum test followed by multivariable linear regression of natural log transformed length of stay. Using group (pre or post), age, sex, race, and appendicitis status (uncomplicated or complicated) as available variables for the model, we selected the final model using backwards elimination. The dichotomous variable of “long stay” was analyzed using a Chi-squared test followed by multivariable logistic regression, retaining the same covariates as the primary multivariable linear regression model. All possible interaction terms within the final models were tested, and appropriate model diagnostics were employed.

Analysis of data was performed in an intention-to-treat fashion comparing pre- and post-intervention groups, including all patients in the protocol group regardless of complete clinical protocol adherence. All statistical tests were two-sided with α = 0.05. All calculations were performed using Stata 14 (StataCorp, College Station, TX).

Results

Patient Characteristics

The study included 1289 patients admitted to CHP with a diagnosis of acute appendicitis who then underwent an appendectomy between November 2013 and December 2016. The protocol was implemented on January 1^st 2015. The pre-protocol cohort was composed of 481 patients and the post-protocol cohort with 808 patients. Average age was 10.8 or 10.9 years respectively. About 39% of both cohorts were female and 13-15% of patients were non-white. In both cohorts, about 27% of patients were diagnosed with complicated appendicitis. There were no significant differences among any of these characteristics (Table 1).

Peri-operative Outcomes

There was a significant decrease in surgical site infections (p = 0.01, Table 2) and postoperative antibiotic use in the post-protocol cohort (Uncomplicated: median [IQR] 1 [0 – 1] vs. 0 [0 – 0] days, p < 0.001; Complicated: median [IQR] 6 [5 – 9] vs. 5 [3 – 6] days, p < 0.001 Table 3). The odds ratio (95% CI) for readmission for any patient with complicated appendicitis was 1.97 (1.24 – 3.11) compared to patients with uncomplicated appendicitis. There was no significant difference in readmission rates between the pre and post protocol cohorts (p = 0.37).

Table 2:

Post-operative Outcomes

	Pre (n=481)	Post (n=808)	OR (95% CI)	P value
Length of stay, hours, median[IQR]	38.8[28.4-81.5]	35.5[24.2-61.8]	-	<0.001
Uncomplicated	34.1[26.6-41.6]	29.0[21.9-38.3]	-	<0.001
Complicated	147.4[128.4-219.1]	110.5[75.9-150.8]	-	<0.001
Long stay, n(%)	247(51.4)	275(34.0)	0.49 (0.39 – 0.62)	<0.001
Uncomplicated	148(42.1)	184(31.1)	0.62 (0.47 – 0.83)	0.001
Complicated	99(76.7)	91(41.9)	0.22 (0.13 – 0.37)	<.001
Readmissions, n(%)	34(7.1)	47(5.8)	0.81 (0.50 – 1.32)	0.37
Uncomplicated	20(5.7)	28(4.7)	0.83 (0.44 – 1.57)	0.52
Complicated	14(10.9)	19(8.8)	0.79 (0.36 – 1.77)	0.52
Surgical site infections, n(%)	55(11.4)	57(7.1)	0.59 (0.39 – 0.89)	0.01
Uncomplicated	7(2.0)	3(0.5)	0.25 (0.04 – 1.11)	0.046
Complicated	48(37.2)	54(24.9)	0.56 (0.34 – 0.92)	0.02

Table 3:

Resource utilization

	Pre (n=481)	Post(n=808)	OR (95% CI)	P value
Postoperative antibiotic duration, days, median[IQR]	1[0-3]	0[0-2]	-	<0.001
Uncomplicated	1[0-1]	0[0-0]	-	<0.001
Complicated	6[5-9]	5[3-6]	-	<0.001
Postoperative imaging, n(%)	67(13.9)	89(11.0)	0.76 (0.54 – 1.09)	0.120
Uncomplicated	13(3.7)	24(4.1)	1.10 (0.53 – 2.39)	0.780
Complicated	54(41.9)	65(30.0)	0.59 (0.37 – 0.96)	0.020
Peripherally inserted central venous catheter use, n(%)	39(8.1)	14(1.7)	0.20 (0.10 – 0.38)	<0.001
Uncomplicated	1(0.3)	0(0.0)	0 (undefined)	0.370
Complicated	38(29.5)	14(6.5)	0.17 (0.08 – 0.33)	<0.001

Resource Utilization

There was no significant difference in postoperative imaging studies ordered between the two cohorts (p=.12, Table 3). However, when comparing patients with complicated appendicitis only, there was a significant decrease in the utilization of imaging in the postoperative setting in the post-protocol cohort when compared to the pre-protocol cohort (30.0 vs. 41.9%, p = 0.02). There was a significant decrease in peripherally inserted central venous catheter (PICC) use in the post-protocol cohort (1.7 vs. 8.1%, p < 0.001, Table 3).

Length of Stay Analysis

Patients in the post-protocol cohort overall had a significantly shorter LOS when compared to the pre-protocol cohort (median [IQR] 35.5 [24.2 – 61.8] vs. 38.8 [28.4 – 81.5] hours, p < 0.001, Table 2). After adjusting for appendicitis status (uncomplicated vs. complicated) and race (white vs. nonwhite) mean LOS in the post-protocol cohort was decreased by 18.5% (95% CI 14.3 – 22.5%). Although LOS decreased for patients with either uncomplicated or complicated appendicitis, patients with complicated appendicitis gained a much larger benefit in hospital after protocol implementation when compared to their pre-protocol cohorts, 147.4 vs. 110.5 hrs. (p=<.001,Table 2). Multivariable linear regression model coefficients for natural log transformed LOS are summarized in Table 3 (model overall F-test p < 0.001, adjusted R² = 0.6671). There were no significant interaction terms in the model.

Based off of published literature, long LOS was defined as LOS greater than 36 hours for uncomplicated appendicitis and greater than 5 days for patients with complicated appendicitis (15). Patients in the protocol group were less likely to experience a “long” LOS (34.0% vs. 51.4%, p < 0.001). This difference was present in patients with either uncomplicated or complicated patients (Table 2). Logistic regression yielded an overall adjusted odds ratio of “long” LOS for the post-protocol group of 0.47 (95% CI 0.37 – 0.60, p < 0.001).

Pathway Compliance and Cost Analysis

After implementation of the protocol, 81.3% of patients were ordered all components of the preoperative protocol. 32.7% of patients were ordered all components of the postoperative pathway, and 26.5% of patients had perfect compliance with both preoperative and postoperative pathway components (Table 4). Cost data was strongly correlated with length of hospital stay (rho = 0.86, p < 0.001). Median costs for the whole cohort decreased 0.6% after protocol initiation with a significant decrease in variance of cost (p < 0.01, Table 5). Median cost after introduction of the protocol decreased 0.2% (p=0.9) for patients with uncomplicated appendicitis and 24.6% (p < 0.01) for patients with complicated appendicitis. The components of cost that had the greatest reduction after implementation of this protocol included the cost of medications (antibiotics and opioids), nursing care hours, and utilization of radiology.

Table 4 –

Multivariable Linear Regression Model of Natural Log Transformed Length of Stay

Variable	Coefficient	Standard Error	P value	95% CI
Treatment Group = Protocol	−0.204	0.025	<0.001	(−0.254, −0.155)
Complicated Status	1.388	0.028	<0.001	(1.333, 1.442)
Nonwhite Race	0.078	0.034	0.020	(0.011, 0.146)
Intercept	3.541	0.022	<0.001	(3.489, 3.584)

Table 5–

Controllable Cost Analysis

	Pre	Post	PercentChange	Ranksum p	Brown-Forsythe p
Cost, USD, median[IQR]	5911[4811-10473]	5874[4882-8551]	−0.6%	.56	<.010
Uncomplicated	5286[4503-6219]	5277[4666-6137]	−0.2%	.90	.090
Complicated	17074[13014-23788]	12877[9585-19147]	−24.6%	<.01	.110

Discussion

Acute appendicitis continues to be the one of the most common indications for urgent surgery in pediatric patients. Despite being a common disease process for which surgery is routinely performed, the management varies significantly. This investigation was a pre-post study to examine the effects of implementing an evidence-based clinical care protocol for patients with acute appendicitis undergoing appendectomy. This study reports that the implementation of an evidence-based care protocol resulted in a significantly decreased LOS in the post-protocol cohort (p<.001). Reduction of median LOS in the uncomplicated patient cohort was over 5 hours, bringing the entire median hospitalization length to slightly over 24 hours. Decreases in LOS were even more substantial in patients with complicated appendicitis, with an almost 37 hour decrease in their median LOS (p<.001, Table 2). Multivariable linear regression modeling found that each individual patient at our facility could expect, on average, a nearly 20% shorter hospital stay in the time-period following introduction of the care protocol (p<.001). Decreased time to discharge fortunately did not negatively affect the 30-day readmission rates in the post-protocol cohort (p=.37) and according to prior studies has been shown to be associated with increased parental satisfaction (16).

Muehlstedt et al. reported that about 48% of surgeons prefer single antibiotic coverage for acute uncomplicated appendicitis where-as about 45% prefer multiple antibiotic coverage regardless of the extent of contamination (5). The decision in regard to duration of postoperative antibiotics is also split with about 22% of surgeons giving only one dose of preoperative antibiotics to acute uncomplicated appendicitis, 50.3% continuing antibiotics for 24 hours after surgery and 7.3% continuing antibiotics for 48 hours after surgery (5). Amongst those with perforated appendicitis, about 62% of surgeons decide the antibiotics duration based on clinical parameters while 53% of surgeons use a clinical adjunct such as white blood cell count or C-reactive protein to determine time of discharge (5). This protocol mandated the use of broad-spectrum antibiotic, cefoxitin, preoperatively in all patients undergoing surgery unless known to have a cephalosporin allergy. Patients with an intraoperative diagnosis of uncomplicated appendicitis post-operatively had their antibiotics discontinued. Those with complicated appendicitis were maintained on IV ertapenem for at least 24 hours. Any time after the 24-hour mark, they could be discharged home on oral augmentin or ciprofloxacin and flagyl if penicillin allergic to complete a 5-day course, although compliance of antibiotic use at home was not tracked. After study completion, the postoperative IV ertapenem regimen was changed to ceftriaxone and flagyl. The results of this study show that after protocol implementation there was a decrease in peri-operative antibiotic duration regardless of whether the patients were diagnosed with uncomplicated or complicated appendicitis (p<.001, Table 3). The decision to discharge patients with diagnosis of complicated appendicitis on oral antibiotics has historically also been inconsistent. Anandalwar et al. showed that discharging these patients on oral antibiotics is associated with decreased organ space infections (17). Our study reports a significant decrease in surgical site infections in the post-protocol cohort (p=.01, Table 1). All of these patients were discharged with oral antibiotics. The decrease in surgical site infections is likely due to the consistent use of a specific appropriate antibiotic selection, dosing and duration of therapy. As discussed in the Joint Commission Journal of Quality and Safety, this is a key factor to reducing surgical site infections (18). Our results are consistent with several other studies in the published literature that have suggested improved outcomes in patients who received early conversion to oral antibiotic therapy compared to prolonged IV antibiotic therapy(19,20).

Resource utilization was also impacted by implementation of our protocol. We found that patients in the protocol group were 80% less likely to undergo PICC insertion which may result in increased patient comfort, decreased resource usage and avoidance of potential line complications (p<.001, Table 3). In a study by Rangel et al, the authors identified an increased rate of hospital revisits for reasons other than treatment failure in patients who received PICC lines (odds ratio 2.21 95% CI 1.33-3.70) (17). In addition to this, patients with complicated appendicitis in the protocol group had 41% lower odds of undergoing additional postoperative imaging studies after surgery (p=.02)

Although this study was not designed as a formal cost-effectiveness analysis, LOS correlated highly with cost and reductions in cost were seen for all patient groups after introduction of the protocol (p<.01). The most significant cost reduction, 24.6%, was seen in the cohort of patients with complicated appendicitis (Table 5). We attribute the decrease in LOS to the emphasis on an earlier transition to oral medications as well as the establishment of a concrete timeline for discharge after appendectomy. For example, in patients with uncomplicated appendicitis, conditional discharge instructions are included in the postoperative orders which allow nurses to discharge patient should they meet discharge criteria.

As with all studies using historical controls, a potential limitation lies in the potential bias introduced via retrospective data gathering. Conversely, aside from introduction of the patient care protocol, there have been no other major changes in the treatment of acute appendicitis within the study period. Though every effort was made to ensure the fidelity of data entered into our database, it is possible that errors occurred during data abstraction. We attempted to address this possibility via analysis of potential outliers in our modeling approach and were unable to detect any obvious issues. Finally, the true contribution of the pathway introduction may not yet be fully realized. Overall “perfect” compliance with both preoperative and postoperative pathways was just over 25% of patients. The rate of low compliance may be due in part to the stringent nature of our definition of “on pathway” which requires that patients be fully compliant with all components of the protocol order set in the electronic medical records. It is conceivable that patients who for example, received vital sign checks q6 hours instead of q4 hours as stipulated by the protocol be considered “off pathway” despite receiving the remainder of the protocol appropriately.

This study has demonstrated that introduction of an evidence-based clinical care protocol for pediatric patients with appendicitis leads to shorter hospital stay without negatively affecting patient-centered outcomes such as hospital readmission or postoperative infection rates in addition to the added benefit of decreased hospital costs. From this we learn that improvements in patient-centered outcomes without additional resource consumption are possible with an evidence-based clinical care protocol. Future directions will include a focus on enhancing complete pathway adherence. Additionally, the presently excluded cohort of patients who underwent initial percutaneous drainage of peri-appendiceal abscess with delayed interval appendectomy may also benefit from introduction of a similar protocol.