Introduction

There has been a spotlight on Emergency Laparotomy (EL) mortality in recent years, with the National Emergency Laparotomy Audit (NELA) shining a light on multiple factors that contribute to mortality among this patient population. With recent publication of their fourth report, NELA has presented an increase in pre-operative CT imaging to 87%, the presence of a surgical and anaesthetic consultant in 90% of cases that occurred within normal working hours and 66% of after-hours cases, a rate of 87% of patients admitted to critical care following surgery, and an improvement in hospital length of stay to 15.6 days. The fourth NELA summarised an overall improvement in 30-day mortality from 11.8% to 9.5% [1]. With the emergence of these new published standards, it is important to compare Australian contemporary practice outcomes and understand the factors that are associated with perioperative mortality in EL patients in the Australian population.

Few Australian studies exist detailing local trends and outcomes from EL. Ho et al. reported a 9.7% 30-day mortality in their EL population in a 5-year review at Logan Hospital (a 360 bed public hospital), and found factors impacting on mortality to include patient age, P-POSSUM score, and admission source [2]. They found no statistically significant difference between patients that survived or died when comparing antibiotic administration, the use of a perioperative medical service, or frequency of intensive care admission. Limited other Australian studies reporting on EL mortality include a Victorian review (8.8%

inpatient mortality) [3], a South Australian centre (3.8% 30-day mortality, 5.1% 90-day mortality) [4], and Western Australian reviews (5.2% in-hospital mortality [5], 5.4% 30-day mortality [6]). Gold Coast University Hospital is a 750-bed public teaching hospital. Between 2014 and 2017, the Acute Surgical Unit (ASU) was overseen by a daily alternating on call consultant. The authors review outcomes of EL at the Gold Coast University Hospital, and analyse factors that are associated with 30-day mortality in this patient population.

Methods

A retrospective review of all EL performed by the adult General Surgery service at the Gold Coast University Hospital between March 2015 and July 2017 was performed. All surgical admissions from the emergency department were reviewed, and patients who underwent an EL within 24 hours were included. Trauma laparotomies were excluded due to a potentially confounding high morbidity and mortality. Data was collected on patient demographics, American Society of Anesthesiologists (ASA) physical status classification system score, time to diagnosis, indication for surgery, timing of surgery in or out of normal working hours, consultant surgeon presence, ICU admission, and Hospital Length of Stay (HLOS). ASA score was divided into two groups for analysis; ASA I/II/III, or ASA IV/V. Outcomes included ICU admission, Hospital Length of Stay (HLOS), and 30-day mortality.

Shapiro-Wilk test for normality was used to determine which continuous variables showed a normal parametric distribution. All continuous data showed a non-parametric distribution and was thus presented as median (interquartile range, IQR). To identify predictors of 30-day mortality, a univariate logistic regression was initially used with a significance level of α=0.1. All variables which showed significance (p-value <0.1) were then analysed together as covariates in a multivariate analysis with a significance level of α=0.05. To identify predictors of HLOS, a univariate log rank analysis was first used for categorical values whilst a univariate cox regression was used for continuous values with a significance level of α=0.1 applied. All variables which showed significance (p-value <0.1) in both univariate analyses were then analysed together as covariates in a multivariate analysis with a significance level of α=0.05. All data analyses were performed using SPSS 25 (IBM Corp. Released 2017. IBM SPSS Statistics for Windows, Version 25.0. Armonk, NY: IBM Corp).

Results

30-day mortality

Demographics: 105 patients underwent EL in the 2.33 year study period. 49.5% (n=52) were male, and mean age of patients was  63+/-17 years (range 14-94 years; Table 1). Univariate analysis identified a statistically significant (p=0.010) higher mortality in older patients (mean age 75+/-10 years) when compared to younger patients (61+/-17 years), however this was not statistically significant on multivariate analysis (p=0.084). There was a higher mortality rate in males undergoing EL (53.8% vs 48.9%), however, again this was not statistically significant (p=0.739) (Table 2, Table 3).

American Society of Anesthesiologists (ASA) physical status classification system score: Of patients undergoing EL, 67.6% (n=71) were classed ASA I-III and 32.4% (n=34) were classified an ASA of IV or V (Table 1). There was a higher 30-day mortality rate in patients in the ASA IV/V group (76.9% vs 26.1%), and this was statistically significant on univariate (p=0.001) and multivariate analysis (p=0.036).

Indications for emergency laparotomy: The etiology of each patient’s acute abdomen was recorded as the pathology diagnosed intra-operatively. The most common indication for EL in our study population was perforated viscus (39%, n=41), followed by small bowel obstruction (17.7%, n=18), large bowel obstruction (9.1%, n=10), incarcerated hernia (13.3%, n=14), and ischaemic bowel (9.5%, n=10) (Table 1). Other indications included peptic ulcer disease, post-operative complications (eg. anastomotic leak), caecal mass, and sepsis in enterocutaneous fistula disease 2.9% (n=3) of patients underwent a negative EL, and indications included haemodynamic instability and sepsis concerning for ischaemic bowel, and pneumoperitoneum or peritonitis concerning for perforation. EL performed in patients with ischaemic bowel had the highest mortality (38.5%, n=5; Table 2), and were secondary to superior mesenteric artery occlusion, global small bowel ischaemia secondary to a necrotic mass, and embolic infarcts. The next highest mortality was EL performed for perforated viscus (23.1%, n=3), and these included esophageal perforation, perforated diverticular disease, and perforated stercoral colitis.

Time to diagnosis: Data was collected for time from patient presentation to CT study (ED-CT), and time from initiation of the imaging study to the EL start time recorded in the operating theatre (CT-OT). The time from patient presentation to EL start time was also recorded (ED-OT). Note was made of the time taken for radiology reporting of an imaging study to be completed. The median time from ED-CT was 179 (110-248) minutes with an additional 282 (176-404) minutes between CT-OT, and the median duration for all operations was 171 (128-205) minutes. There was a similar median ED-CT time between the 30-day mortality and non-30-day mortality groups, although a wider distribution was noted (179mins, IQR 83-565 vs 180mins, IQR 111-246). This was statistically significant on both univariate (p=0.089) and multivariate analyses (p=0.044). Patients in the 30-day mortality group were found to have longer CT-OT, CT reporting, and total ED-OT times, however these were not statistically significant on univariate analysis. Patients in the 30-day mortality group were also noted to have a shorter OT duration (161mins, IQR 136-182 vs 171mins, IQR 127-206), however this was again not statistically significant on univariate analysis.

Timing of emergency laparotomy - Normal working hours vs. After-hours: Normal working hours were defined as weekdays between 0800 to 1800 hours, and after-hours defined as between 1801 to 0759 hours on weekdays, and 0000 to 2359 hours on weekends. There was no statistically significant difference in mortality in patients undergoing EL during normal working hours in comparison to EL performed after-hours (Table 1), which has also been found in other studies critically assessing emergency surgery [7,8].

Presence of surgical consultant: NELA data reported on consultant presence in patients with high mortality risk >5% based on P-POSSUM scores. On review of all patients in our study period who underwent EL, the authors found a similar proportion of consultant presence between the 30-day mortality (69.2%) and non-30-day mortality (70.7%) groups.

Hospital length of stay (HLOS) and ICU admission: Median HLOS was 10 days (range 6-19 days) and 59% of all patients were admitted to ICU with a median LOS of 1 day (range 0-4). 59 total patients were admitted to ICU post operatively, 64.4% of those were in weekdays compared to weekends. Among 30-day mortality patients, 84.6% were admitted to ICU with a median LOS of 2 days (range 2-12).  Univariate analyses identified age (p=0.067), ASA (p<0.001), ICU admission (p<0.001), ICU LOS (p<0.001), and operation duration (p=0.007) as potential predictors of HLOS (Table 4, Table 5). When analysed as part of a multivariate analysis, only age (p=0.048, OR 0.986) and ICU LOS (p<0.001; OR 0.853) were found to be independent predictors of HLOS (Table 6).

A higher proportion of patients in the 30-day mortality group were admitted to ICU (84.6% vs 52.2%). This was significant on univariate analysis (p=0.042) but not on statistically significant on multivariate analysis (p=0.587). Patients in the 30-day mortality group were also found to have a longer median ICU LOS (2 days, IQR 2-12 vs 1days, IQR 0-3), however this was not statistically significant on univariate analysis.

Discussion

There is currently a huge focus worldwide on factors affecting 30-day mortality in EL patients with multiple studies in the literature reviewing their own variables and evaluating outcomes [2-6,9], largely spurred by the publication of annual NELA reports.

Overall mortality in the authors’ study population was 12.4%. The authors identified age (p=0.01), ASA IV/V (p=0.001), ICU admission (p=0.042), and ED-CT time (p=0.089) as potential predictors of 30-day mortality on univariate analysis. ICU admission was not routinely sort, and dependent on clinical concern and clinical parameters, therefore patients with higher morbidity may have confounded the ICU admission group. When analysed as part of a multivariate analysis, only ASA IV/V (p=0.036; OR 9.44) and ED-CT time (p=0.044; OR 1.005) were found to be independent predictors of 30-day mortality. Limitations included the moderate sample size, which only included patients who underwent EL within 24 hours of admission from the emergency department and excluded trauma patients, in order to reduce confounding factors.

Although it is noted that some studies have reviewed the use of EL protocols at their institution [10], there was no formalised EL care pathway or protocol established in the authors’ institution, which is not uncommon between institutions. With the limitations of a retrospective study in mind, independently reviewing patient variables and outcomes in this study population may derive a better understanding of the true day-to-day outcomes in a tertiary Australian hospital, in comparison to a well publicised prospective study’s potential procedural or interviewer bias.

Further prospective blinded studies assessing these variables over a longer study period would be crucial in auditing and reviewing long term progress in reducing 30-day mortality in EL patients. ASA scores can at times be rudimentary and biased by the scoring physician, and implementation of other risk stratifying tools in co-morbid EL patients such as analysis based on P-POSSUM scores [11,12] used in other studies [2,7,10,11,13] would add depth to these findings.

As a result of these findings the Gold Coast University Hospital has now implemented a consultant led ASU with a permanent consultant during normal working hours. Further prospective data collection is currently being undertaken to allow for comparison of the different care models, as published literature worldwide has suggested better outcomes with a consultant lead ASU [14,15], including studies with supportive findings in Australia [13,16,17].

Tables

References

1. NELA Project Team (2018) Fourth patient report of the National Emergency Laparotomy Audit. London: RCoA, 2018.
2. Ho YM, Cappello J, Kousary R, McGowan B, Wysocki AP (2018) Benchmarking against the National Emergency Laparotomy Audit recommendations. ANZ J Surg 88: 428-433.
3. Stevens CL, Brown C, Watters DAK (2018) Measuring outcomes of clinical care: Victorian emergency laparotomy audit using quality investigator. World J Surg 42: 1981-1887.
4. Tocaciu S, Thiagarajan J, Maddern GJ, Wichmann MW (2018) Mortality after emergency abdominal surgery in a non-metropolitan Australian centre. Aust J Rural Health 26: 408-415.
5. Burmas M, Aitken RJ, Broughton KJ (2018) Outcomes following emergency laparotomy in Australian public hospitals. ANZ J Surg 88: 998-1002.
6. Broughton KJ, Aldridge O, Pradhan S, Aitken RJ (2017) The Perth Emergency Laparotomy Audit. ANZ J Surg 87: 893-897.
7. Nageswaran H, Rajalingam V, Sharma A, Joseph AO, Davies M, et al. (2019) Mortality for emergency laparotomy is not affected by the weekend effect: a multicenter study. Ann R Coll Surg Engl 101: 366-372.
8. O’Leary JD, Wunsch H, Leo A-M, Levin D, Siddiqui A, et al. (2019) Hospital admission on weekends for patients who have surgery and 30-day mortality in Ontario, Canada: A matched cohort study. PLoS Med 16: e1002731.
9. Saunders DI, Murray D, Pichel AC, Varley S, Peden CJ (2012) Variations in mortality after emergency laparotomy: the first report of the UK Emergency Laparotomy Network. British Journal of Anaesthesia 109: 368-375.
10. Vashistha N, Singhal D, Budhiraja S, Aggarwal B, Tobina R, et al. (2018) Outcomes of emergency laparotomy (EL) care protocol at tertiary care centre from low-middle-income country (LMIC). World J Surg 42: 1278-1284.
11. Echara AL, Singh A, Sharma G (2019) Risk-adjusted analysis of patients undergoing emergency laparotomy using POSSUM and P-POSSUM score: A prospective study. Niger J Surg 25: 45-51.
12. High-risk emergency laparotomy in Australia: Comparing NELA, P-PSSOUM, and ACS-NSQIP calculators. Journal of Surgical Research 246: 300-304.
13. Guy S, Lisec C (2018) Emergency laparotomy outcomes before and after the introduction of an acute surgical unit. International Journal of Surgery Open 10: 61-65.
14. Nagaraja V, Eslick GD, Cox MR (2014) The acute surgical unit model verses the traditional “on call” model: a systematic review and meta-analysis. World J Surg 38: 1381e7.
15. Vergis A, Metcalfe J, Stogryn SE, Clouston K, Hardy K (2019) Impact of acute care surgery on timeliness of care and patient outcomes: a systematic review of the literature. Can J Surg 62: 281-288.
16. Kinnear N, Britten-Jones P, Hennessey D, Lin D, Lituri D, et al. (2017) Impact of an acute surgical unit on patient outcomes in South Australia. ANZ J Surg 87: 825-829.
17. Suhardja TS, Bae L, Seah EZ, Cashin P, Croagh DG (2015) Acute surgical unit safely reduces unnecessary after-hours cholecystectomy. Annals of the Royal College of Surgeons England 97: 568-573.