The Impact of an Acute Care Surgical Service on the Quality and Efficiency of Care Outcome Indicators for Patients with General Surgical Emergencies

Background Acute care surgery (ACS) models address high volumes of emergency general surgery and emergency room (ER) overcrowding. The impact of ACS service model implementation on the quality and efficiency of care (EOC) outcomes in acute appendicitis (AA) and acute cholecystitis (AC) cohorts was evaluated. Methods A retrospective chart review (N=1,229) of adult AA and AC patients admitted prior to (pre-ACS; n=507; three hospitals; 2007) and after regionalization (R-ACS; n=722; one hospital; 2011). Results R-ACS time to ER physician assessment was significantly longer for AA (3.4 ± 2.3 versus 2.4 ± 2.6 hr; p ≤ 0.001). Surgical response times (1.3 ± 1.2 vs 2.6 ± 4.3 hr for AA; 1.8 ± 1.5 vs 4.1 ± 5.0 hr for AC; p ≤ 0.0001) and acquisition of imaging (4.1 ± 4.1 vs 6.9 ± 9.9 hr for AA, p ≤ 0.0001; 7.8 ± 1.9 vs 13.2 ± 18.5 hr for AC, p ≤ 0.008) occurred significantly faster with R-ACS. R-ACS resulted in a significant increase in night-time appendectomies (21.7% vs 11.1%; p ≤ 0.002), perforated appendices (29.1 % vs 18.9 %; p ≤ 0.006), 30-day readmissions (4.56% vs 0.82%; p ≤ 0.01), and lower rate of intraoperative complications for AC patients (2.78% vs 7.69%; p ≤ 0.02). Conclusions Despite the increased volume of patients seen with the implementation of R-ACS, surgical assessments and diagnostic imaging were significantly more prompt. EOC measures were maintained. Worse AA outcomes highlight areas for improvement in delivering R-ACS.


Introduction
General surgeons face the challenge of managing high volumes of emergency general surgery (EGS) patients with advanced age and increasingly complex conditions [1]. Often, 1 1 1 1 urgent intervention to prevent rapid patient deterioration is necessary [2] and compounded by limits in emergency department (ED) access due to overcrowding. Such delays may be associated with higher rates of major complications and death [3]. ACS was developed as an extension of trauma surgery services, including emergency surgery while maintaining operative skills through increased operative volume. It also benefited patients by improving the timeliness of care [4]. In Canada, the impetus to create acute care surgery services (ACS) also arose from the high volume of EGS cases, increasing surgeon sub-specialization, facilitation of the separation of elective practice from emergency surgery cases, and the regionalization of subspecialty surgical care [5].
There are 13 ACS services in Canada with significant variation in organization and function [1,5]. The lack of standardization for ACS outcome measures and reporting makes comparisons among ACS services challenging [17][18]. Acute appendicitis (AA) and acute cholecystitis (AC) are the most common EGS diagnoses requiring admission [7], with standard, measurable points of care along the patient trajectory.
The lack of agreement and utilization of standard metrics to report EOC and patient outcomes within ACS service delivery, in addition to the high degree of variability in the definitions of each variable, make reporting and comparisons challenging. Standardization and consensus is required among ACS surgeons (and the health care teams/stakeholders involved in delivering care) to create an organized system of data capture and analysis to establish national standards for ACS service implementation and delivery, including support for evidence-based practice, quality improvement initiatives, as well as research, training/education, and innovation. Metcalfe et al. [17] reported that approximately 27% (6/22) of studies of ACS implementation provided surgical response time and included five different definitions for this metric. Time to operating room was reported in 91% (20/22) of studies, with between five to seven distinct definitions while LOS was reported in 86% of studies with three separate definitions.
Three Winnipeg hospitals' EGS services were consolidated into a regionalized ACS service (R-ACS) [6], resulting in a 221% increase in patient volume. Approximately 56% of ACS service inpatient admissions are transferred from another institution; 39 provincial hospitals (five within and 34 outside Winnipeg), as well as 19 other institutions (nursing stations, personal care homes, health centers, home care/physician's offices). Interhospital transfer has been shown to impact the surgical quality and patient outcome metrics, as well as the utilization of hospital and health system delivery resources [19][20].
The average annual number of cases admitted to the ACS service was 4,024; depending on the hospital, between 32% to 53% are medically managed while 47% to 68% are surgically managed. Per capita, the Winnipeg ACS service receives approximately 566 ACS admissions per 100,000 people. Research evaluating whether R-ACS in Winnipeg leads to improved efficiency and patient outcomes is crucial in facilitating evidence-based care, creating benchmark standards, and highlighting areas for system improvement. The study objective was to determine the impact of implementing a regionalized ACS on EOC and patient outcomes for common general surgical emergencies (acute appendicitis and acute cholecystitis).

Materials And Methods
A retrospective chart review of AA and AC patients was performed to compare EOC and patient outcome measures between the Pre-ACS and R-ACS cohorts (University of Manitoba Health Research Ethics Board number H2012:166; HS15311; individual institutional ethics approvals obtained). A flow diagram of the patient chart extraction process is provided in Figure 1. AA (n=244 and n=329 for the Pre-ACS and R-ACS groups, respectively) and AC (n=157 and n=254 for the Pre-ACS and R-ACS groups, respectively) diagnoses were chosen, as they are the most common reasons for EGS admission [7] with standard, measurable points of care. Adult (>17 years) patients diagnosed with AA or AC during the January to December (2007) and January to December (2011) study periods were included. Charts were identified by diagnostic and procedural ICD-10 codes and included regardless of whether or not they underwent an operation during admission. Exclusion criteria included patients who were admitted to a service other than general surgery or ACS, admission with a diagnosis not meeting the case definition, and patients undergoing appendectomy or cholecystectomy as an elective procedure or secondary to another operation. Biliary cases with a diagnosis other than acute cholecystitis were excluded (including gallstones, choledocholithiasis, and biliary colic) due to the significant variability in course of care and surgical decision-making.  Figure 2. Five baseline hospital admission variables were measured, including (1) admission diagnosis, (2) origin hospital (if transferred), (3) Regional Health Authority of the origin hospital, (4) admission hospital and service, and (5) previous admissions information for the same diagnosis. The patient outcome variables measured included intra-operative, intervention-related, and postoperative complications, time to operating room, total length of stay (LOS), 30-day hospital readmission rate, 30-day emergency room (ER) visits, risk of perforated appendicitis, and pathology for appendectomy and cholecystectomy specimens. Post-operative complications were graded according to the Clavien-Dindo Classification System on a scale of I through V; minor grades I to II, major grades III to V [9].
Study-specific reference documents, specifying the chart location for each outcome variable being extracted, facilitated reproducible and uniform data extraction (Supplemental Files 1, 2, and 3 in the Appendix). If the variable was not found in the pre-specified location, it was recorded as "not specified." Uniform and consistent data extraction methods were facilitated through joint data extracted by two research team members (HG and JM); 50 charts at both Concordia and Victoria General Hospitals. Where differences in assessment existed, discussion and resolution were attained and, if required, arbitrated by KH. SAS procedures were utilized to handle missing data. For data missing completely at random, listwise deletion was used (the number of missing values was low). Data missing at random were dealt with through imputation.
A sample size/power calculation was not performed for this research study, as all of the charts for each cohort (Pre-ACS and R-ACS) for both the acute appendicitis (AA) and acute cholecystitis (AC) groups were retrieved and reviewed.
Data were analyzed with SAS® statistical software (SAS Institute, North Carolina, US). The AA and AC groups within the Pre-ACS and R-ACS cohorts were analyzed separately. Continuous data were analyzed with the two-tailed student's t-test. Chi-square tests were used to test categorical variables; Fisher's exact test was used if sample numbers were small (<5). A logistic regression analysis was performed for postoperative complications, readmissions, risk of perforated appendicitis, time to OR, and total length of stay (LOS). Logistic regression model covariates were chosen by a clinical decision and the backward selection method. Age and gender were deemed significantly relevant and included in all logistic models. The backward selection method started with all covariates included in the analysis plan. Variables were deleted from the model based on p-value; p-values > 0.02 were deleted. Cook's D statistics were plotted for each regression model to determine the influence of each observation. Tolerance values to test for multicollinearity among the covariates was performed and absence confirmed among the variables tested; all tolerance values were greater than 0.1 and had Variance Inflation Factor (VIF) less than 10.

Results
Data were extracted from a total of 1,229 charts and represents two cohorts: the Pre-ACS (n=507) and R-ACS (n=722) cohorts. The Pre-ACS cohort includes patients from three hospitals (n=243 SBGH, n=168 VGH, n=96 CGH). The AA and AC procedures were utilized to compare the Pre-ACS and ACS surgical outcomes. Table 1 outlines the baseline demographic variables for both AA and AC groups for the Pre-ACS versus R-ACS cohorts. No significant differences were found in the baseline demographic variables between the Pre-ACS and R-ACS cohorts for AC. All baseline demographic variable outcomes were similar for AA patients in both the Pre-and R-ACS cohorts, except mean age at admission, which was significantly higher in the R-ACS group as compared to the Pre-ACS group (39. 26    EOC and patient outcome measures were compared between the Pre-ACS and R-ACS cohorts for both the AA and AC patient groups ( Table 3 and Table 4; Figure 3 and Figure 4)    For patients with AA, regionalization of the ACS service did not result in significant improvements from the Pre-ACS service for transfer time, time from operating room to discharge, or total length of stay (LOS). Time to emergency room physician (ERP) for AA patients was significantly higher in the R-ACS as compared to the Pre-ACS cohort (3.4 ± 2.3 hours vs 2.4 ± 2.6 hours, p<0.001). However, these same patients experienced a significantly shorter time to surgical assessment/surgical response time (1.3 ± 1.2 hours vs 2.6 ± 4.3 hours, p<0.0001) as well as time from ERP or surgical assessment to imaging (4.1 ± 4.1 hours vs 6.9 ± 9.9 hours, p<0.0001). In addition, the frequency of pre-operative imaging (ultrasound (US), computed tomography (CT), magnetic resonance imaging (MRI)) was significantly higher among R-ACS AA patients as compared to Pre-ACS AA patients (93.3% vs 70.1%; p<0.0001). Similarly, the R-ACS service significantly reduced both time to surgical assessment/surgical response time (1.8 ± 1.5 hours vs 4.1 ± 5.0 hours, p<0.0001) and time from ERP or surgical assessment to imaging for those patients with AC (7.8 ± 1.9 vs 13.2 ± 18.5, p<0.0001). All other EOC outcome variables were unaffected under R-ACS services. Data collected from multiple time points pre-and post-ACS implementation were used to retrospectively assess for temporal trends in the data, similar to an interrupted time series analysis. Evaluation of length of stay (LOS) for the AA and AC patient groups confirmed the absence of pre-existing trends towards decreased LOS prior to the implementation of ACS and no additional temporal trends were identified. Table 4 presents a summary of the findings for the patient outcome variables assessed. Under both the Pre-ACS and R-ACS service models, no differences were found in the time to the operating room for patients with perforated and non-perforated AA. However, the R-ACS cohort demonstrated significantly higher rates of risk of perforated appendicitis (29.1% vs 18.9%, p<0.006) and 30-day readmission (4.6% vs 0.8%, p<0.01) when compared to the Pre-ACS service model ( Table 4). Risk of perforated appendicitis was based on appendectomy pathology findings. The frequency of intraoperative complications for AA patients was the same for both the Pre-and R-ACS service cohorts. For AC patients, the R-ACS service significantly reduced the rate of intraoperative complications (2.8% vs 7.7%, p<0.02). The frequency of 30day emergency department visits was similar between the Pre-ACS and ACS service models for both AA and AC patient groups.
Multivariate logistic regression demonstrated that post-operative complications and/or readmission for AA was associated with age, laparoscopic surgical procedure, transfer from another hospital, and admission to VGH or CGH as compared to SBGH ( Table 5). Perforated appendicitis was associated with patient age and being transferred from another hospital versus direct admission (

Discussion
This study assessed multiple EOC and patient-related outcomes for acute appendicitis and acute cholecystitis following the implementation of an R-ACS service in Winnipeg (Manitoba, Canada).
The R-ACS in Winnipeg resulted in not only some improvements in efficiency but also highlighted some areas for improvement. Time to ER physician assessment increased but time to surgical assessment/surgical response time and time to imaging decreased, indicating improved efficiency by a dedicated surgical team. With R-ACS, time to surgery was slightly increased for AA and did not improve for AC, highlighting the challenge of timely access to operating room resources for EGS patients even with the presence of a dedicated team. Time to discharge and LOS trended towards improvement. Postoperative care was already efficient under the traditional model, with the average LOS between 1.4 to 2.8 days (depending on simple versus complex appendectomy) for AA and 2.9 days for AC, and improvements in this area were not demonstrated. Mean LOS for AA and AC from 2011 to 2017 were one and three days, respectively; one day less than the two additional Winnipeg hospitals performing these procedures. Multiple factors have been shown to impact LOS for AA patients, including time of presentation, perforation, and time of surgery [21][22].
Currently, there are no established EOC benchmarks for common EGS diagnoses. The development of wait-time benchmarks for common EOC measures is required to promote patient safety and inform program planning and evaluation.
Winnipeg has historically reported some of the longest ER wait times in Canada. The Canadian Institute for Health Information (CIHI) reported that four Winnipeg emergency departments were among the top five, with the longest wait times in Canada [23]. Multiple competing factors influence ERP assessment time, including ER patient flow, delays in admissions/ER clearance, and increased patient volumes. It is difficult to weigh the influence of various factors on the overall increase in ERP assessment time shown in this study without assessing the impact of each individual factor in detail. A comprehensive assessment of ER flow was beyond the scope of this study. However, significant restructuring of emergency care is underway in Winnipeg as part of an overall regional reorganization strategy. The impact of these changes will need to be assessed in the future.
We assessed intraoperative and perioperative patient outcomes related to AA and AC using commonly reported measures and classifications. Despite trainee involvement in the R-ACS services model, post-operative complications did not increase and there were significantly fewer intra-operative complications for AC and no change for AA. The same surgeons were involved in the pre-ACS and R-ACS cohorts but the reduction in complications for AC might be explained by increased experience with laparoscopy during the study period or possibly by the conduct of these procedures at a tertiary teaching institution. Variation in study definitions and outcome measures for EGS conditions are common throughout the scientific literature [17]. This makes a direct comparison among study results challenging. The standardization of common outcome measures both intraoperatively and perioperatively is necessary.
In this study, R-ACS resulted in a substantial increase in patient transfers, with 67.1% of appendectomies being transferred, as compared to only 14.3% in the Pre-ACS cohort. These patients were at higher risk of pathologically confirmed perforation and post-operative complications (or readmission). At 29%, the R-ACS AA perforation rate is at the high end of published national rates of 20%-30% [24]. Other ACS studies have reported rates of 4.6% [14] and 17% [15] and a reduction of 10% [25]. Other studies have reported increased perforation rates with ACS service model implementation [15][16].
The risk of perforation might be influenced by delays in imaging or consultation requests at the origin hospital, inadequate initial care (antibiotics and fluid resuscitation), or as the result of more serious disease. We did not capture index hospital treatment and efficiencies as part of the current study. This would involve significant resources to review paper-based emergency department data at multiple sites, including transfers from outside of Winnipeg. This has been considered for future research.
Short delays in surgical intervention for AA are well-tolerated, with several studies reporting no increase in rupture when surgery is delayed until daytime hours [26][27]. One study reported the risk of rupture as negligible within the first 36 hours after symptom onset, with the risk rising 5% for each 12-hour period with untreated symptoms [28]. R-ACS assessment of patients was faster but the number of daytime operations did not increase. This is likely as a result of competing surgical procedures, such as complex emergency bowel cases, which were more likely to be scheduled in daytime operating hours. In addition, fewer AA patients had evening surgery and more had night-time surgery with R-ACS. It is likely that appendectomies were performed at night out of necessity due to an increased volume of cases rather than the influence of surgeon remuneration. Within the ACS literature, most studies report increased daytime operating with ACS services while only one study found no increase in daytime operating [12][13][14]. Our logistic regression analysis did not show the operative time of day to be associated with patient outcomes. The increased 30-day readmission rate for AA in the R-ACS cohort is likely associated with the increased rate of appendicular rupture. While some of these patients were treated with antibiotics, a significant number required either percutaneous or operative drainage.
Our study has a few limitations. Firstly, while transfer time was similar among the Pre-ACS and R-ACS cohorts, we did not capture the time spent at the referring hospital or treatments administered at the initial hospital. Secondly, our pre/post-study design lacked a concurrent control/comparison group, making it difficult to establish a cause and effect relationship between the intervention (ACS service implementation) and the outcomes measured. Thirdly, our results may be confounded by temporal and secular changes independent of the intervention. Surgeons who participated in EGS calls under the traditional (Pre-ACS) service model of care were the same as those involved in the R-ACS model and their laparoscopic experience increased over time.
Regression analysis showed that laparoscopic cases were associated with a decreased risk of postoperative complications, decreased readmissions, and shorter LOS. In the last five years, early surgical intervention for AC has been advocated and the hospitals providing the R-ACS service adopted this change. However, our results show no significant differences in time to OR or operative rate for AC between the Pre-ACS and R-ACS cohorts. Other temporal trends in practice patterns have not been captured by this study (favored antibiotic regimens, changes in antibiotic resistance). Also, competing procedures for the emergency OR were not assessed and could have impacted the R-ACS daytime OR measure.
The introduction of formal ACS services, along with changes in surgeon practice patterns across Canada, make the further regionalization of EGS likely, and its impact must be highlighted. As with other conditions with standardized protocols for care (myocardial infarctions or traumas) [28], pre-transfer guidelines, and standardized inter-hospital documentation could potentially improve the timeliness and quality of care. Some Canadian centers have developed clinical care pathways for patients with common EGS diagnoses [29]. A "Suspected Appendicitis Care Map" for adults has been developed to streamline the flow of patients from the ED to the OR [28].

Conclusions
EGS patients are a uniquely challenging population, particularly prone to poor outcomes. Improved efficiency and quality of EGS care can significantly impact the emergency department, operating room, and hospital resources. We must continue to define, measure, and monitor specific outcomes in order to identify opportunities for improvement.