What if You Need More Than One? More Acute Care Surgery Procedures Are Associated with Mortality

Abstract

Background:

It is unknown whether having multiple acute care surgery (ACS) procedures performed in one admission confers additional risk. We hypothesized that having multiple procedures (for example, hernia repair plus bowel resection) would be associated with higher mortality.

Patients and Methods:

We identified all 2017 National Inpatient Sample admissions with ACS procedures including: colon, small bowel/appendix (SB), hernia, adhesiolysis, peptic ulcer procedures, gallbladder, debridement, other laparotomy, other laparoscopy. The total number of procedures for each admission and common dyad (two-procedure) and triad (three-procedure) combinations were identified. Logistic regression estimated the odds of in-hospital mortality for increasing procedure count and specific dyad and triad combinations, using patients with one procedure as the reference.

Results:

A total of 216,317 ACS patients (median age, 57, interquartile range [IQR], 43–70; 50.6% female) were included; 2.8% died. Patients with multiple procedures were more likely to die than patients with one procedure (7.4% vs. 1.9%). An increasing number of procedures was associated with higher odds of death (two procedures: odds ratio [OR], 3.0; 95% confidence interval [CI], 2.9–3.2] to six or more procedures, OR, 9.5; 95% CI, 4.9–18.5); having more than three procedures was associated with at least fivefold higher odds of death. Specific dyads/triads were associated with particularly high risk of mortality, including ulcer/laparotomy (OR, 15.5; 95% CI, 13.7–17.5) and laparotomy/SB (OR, 8.31; 95% CI, 5.15–13.40).

Conclusions:

Multiple ACS procedures in one hospitalization confer increased odds of in-hospital mortality. This knowledge enables the ACS providers to better counsel patients by giving more specific expectations regarding mortality based on the number of procedures required or anticipated during an admission.

Acute care surgery (ACS) diseases account for approximately 8% of all U.S. hospitalizations, and an estimated 1%–3% of all hospitalizations include a surgery for these diseases.^1–3 It is well described that ACS patients have a disproportionately increased risk for complications, high hospitalization cost, and even death.^2,4,5 Acute care surgery patients are up to eight times more likely to die than a patient undergoing the same operation electively; additionally, one-third of ACS patients have a major peri-operative complication, compared with less than 13% of patients with non-emergent surgery.⁶

Among ACS procedures, seven main procedure types have been most identified and studied including colon, small bowel, gallbladder, peptic ulcer disease, adhesiolysis, appendix, and laparotomy operations.⁷ These procedures represent more than 75% of the operation burden under emergency general surgery. Many ACS procedures (such as cholecystectomy, appendectomy, small bowel and colon procedures, and debridement) are either primarily surgically managed infectious pathologies or have complications that would inherently be infectious that subsequently then require surgical management. Risk stratification for most ACS procedures can be performed using publicly available calculators.^8–11 However, for some patients with complex diseases or who develop complications, multiple operative interventions can occur during a single hospital admission; for example, a patient with a perforated ulcer may require a relaparotomy and washout for post-operative abdominal sepsis. Currently available risk estimation calculators do not account for the cumulative risk of additional emergency surgeries, and little is known about how added procedures and specific combinations of ACS procedures affect mortality in the post-operative setting.

Our study sought to first assess whether having multiple ACS procedures during a single admission was associated with higher odds of death. Additionally, we aimed to quantify the odds of death of multiple procedures and determine whether specific combinations of procedures conferred particularly high risk. We hypothesized that having more than one ACS procedure on a single admission would be associated with increased odds of death.

Patients and Methods

Study population

We utilized the 2017 National Inpatient Sample (NIS) database of the Healthcare Cost and Utilization Project (HCUP), supported by the Agency for Healthcare Research and Quality.¹² The NIS includes a weighted sample of hospital admissions across the United States and approximates a 20% stratified sample that is representative of hospitals throughout the United States.¹³ In addition to demographics, the NIS includes up to 40 diagnosis codes and 25 procedure codes per inpatient admission, allowing for extraction of multiple procedures. The NIS includes data from inpatient stays, not individual patients, and therefore records of events and diagnoses before or after the stay are not available and not included in this analysis.

We included any hospitalization of an adult patient (18 years old or older), with a non-elective admission for emergency general surgery. Acute care surgery was identified using the International Classification of Diseases, Tenth Revision (ICD-10) procedure codes, developed by Smith et al.¹⁴ Acute care surgery procedure types were: colon, small bowel or appendix (referred to as small bowel), hernia, lysis of adhesions, peptic ulcer procedures of the stomach and duodenum, gallbladder surgery, debridement, other laparotomy, and other laparoscopy. We did not implement code to identify traumatic injuries or mechanisms, and therefore did not systematically exclude trauma patients.

Combinations of ACS procedures

Data in the NIS does not include the order of procedures such that if more than one procedure were to be performed within a given admission, the order of procedures would remain unknown. We assumed that multiple procedures in a single admission would likely either occur because of a single pathology requiring multiple procedures to treat ongoing contamination or complications or would signify the co-occurrence of multiple pathologies. We believe that the very act of needing multiple operations, rather than the time frame in which procedures were performed, would be meaningful for clinical impact.

Therefore, we examined the question of multiple procedures in a variety of ways. We first identified the total number of unique procedure types that a patient had during the hospitalization and created a total summary value. As we examined nine unique procedure types, there are 36 unique possible dyads (two-procedure combinations), 84 unique triads, and 126 unique four-procedure combinations. Because of the large possible number of combinations, we used association rule mining to identify the most common combinations of emergency general surgery conditions. Association rule mining (ARM) is a machine learning method originally developed for consumer research to find items commonly purchased together in the same transaction; this is one algorithm behind a store that suggests “if you are buying pencils, you also may want an eraser.”^15,16 Association rule mining has since been applied to a variety of applications in medicine and bioinformatics.^17–19 We applied the method, treating each hospitalization as the “transaction” and the surgical procedure types as the “items” to find the most frequently occurring combinations.

Covariates

The covariates included in this study are: age at hospitalization (in years); gender (male/female); primary payer (Medicare, Medicaid, private insurance, self-pay, no charge, or other); race/ethnicity (white, black, Hispanic, Asian/Pacific Islander, Native American, and other); ZIP code income quartile; the Elixhauser comorbidities were also identified during the hospitalization from the NIS as covariates.²⁰ We included race as a covariate as a sociodemographic variable to adjust for mortality, as existing literature on race suggests that ACS outcomes vary between race categories.^21,22 For the Elixhauser comorbidities, a commonly used set of physical and psychiatric comorbidities, the total number an individual had was counted to estimate the overall comorbidity burden.

Statistical analyses

First, we performed bivariable comparisons between patients who had just one ACS procedure during their hospitalization and those who had more than one. Continuous variables were compared using the Wilcoxon rank sum test, and categorical variables were compared using a χ² test.

We then used multivariable logistic regression to examine the association between ACS procedures and in-hospital mortality. Our first model evaluated the relation between the number of ACS procedures and mortality, with one ACS procedure serving as the reference level. Our second set of models evaluated the relation between all specific single procedure and all two-procedure combinations and mortality. For single procedure types, the reference group was the other admissions with one procedure. For dyads, the reference group was all patients with a single procedure type. Our third set of models evaluated the relation between the most common combinations of three or more ACS procedures and mortality, with the reference group limited to patients with only one ACS procedure. For this third set of models, we performed a regression for any combination of three or more procedures that occurred in more than 0.1% of the population. All regression models were adjusted for age, gender, race, payer, and ZIP code income quartile.

An α of 0.05 was set as the significance level for all analyses. Data analysis was conducted in R version 3.6.3 and RStudio v1.2.5033, along with the “arules” package version 1.6, 8 and the “arulesViz” package version 1.5, 0) for association rule mining.^23,24 The remainder of the analysis was performed using STATA MP, version 17.0 (StataCorp, College Station, TX). This study was deemed exempt by our institution's Institutional Review Board.

Results

We identified 216,317 nonelective adult admissions in the 2017 NIS in which at least one ACS procedure was performed. The median age was 57 (interquartile range [IQR], 43–70], and 5,966 (2.8%) died during the hospitalization. Of these patients, 33,744 (15.6%) had more than one major procedure documented during the hospital stay. Debridement was the most common procedure included and was included in 89,581 (41%) of admissions. Gallbladder (26.1%), hernia (11.1%), colon surgery (10.7%), adhesiolysis (10.4%), and small bowel/appendix procedures (8.7%) were also common.

Bivariable comparisons

Patients with multiple procedures were older, more likely to be female and had a higher Elixhauser comorbidity count (Table 1). Race also differed, with more white patients having multiple procedures. We saw a greater proportion of patients with multiple ACS procedures from ZIP codes in the lowest quartile. Patients with multiple procedures had a longer median length of stay (5 [3–9] vs. 9 [5–15]) and a higher proportion of patients who died (7.4% vs. 1.9%).

Table 1.

Two-Group Comparisons of Admissions with One versus Multiple Procedures

n (%) or median [IQR]	One procedure n = 182,573	Multiple procedures n = 33,744	p
Age	57 [42, 69]	62 [50, 74]	< 0.0001
Female	91,903 (50.3%)	17,610 (52.2%)	< 0.0001
Elixhauser comorbidity count	3 [1, 4]	3 [2, 5]	0.0001
Race category			< 0.0001
White	114,171 (64.5%)	22,135 (68.0%)
Black	25,751 (14.5%)	4,867 (14.9%)
Hispanic	26,270 (14.8%)	3,735 (11.5%)
Asian/Pacific Islander	3,958 (2.2%)	709 (2.2%)
Native American	1,419 (0.8%)	208 (0.6%)
Other	5,292 (3.1%)	917 (2.8%)
Insurance			< 0.0001
Medicare	73,590 (40.4%)	16,547 (49.1%)
Medicaid	37,299 (20.5%)	5,164 (15.3%)
Private insurance	51,475 (28.2%)	9,449 (28.1%)
Self-pay	12,743 (7.0%)	1,492 (4.4%)
No charge	1,179 (0.7%)	105 (0.3%)
Other	5,959 (3.3%)	920 (2.7%)
ZIP code income			< 0.0001
Highest income quartile	32,267 (18.0%)	6,418 (19.4%)
Second highest quartile	41,289 (23.1%)	7,892 (23.8%)
Third highest quartile	47,473 (26.5%)	8,825 (26.6%)
Lowest income quartile	57,984 (32.4%)	9,993 (30.2%)
Surgery type^a
Gallbladder	48,720 (26.7%)	7,830 (23.2%)	< 0.0001
Ulcer	2,989 (1.6%)	6,562 (19.5%)	< 0.0001
Small bowel/appendix	5,029 (2.8%)	13,794 (40.9%)	< 0.0001
Colon	12,675 (6.9%)	10,489 (31.1%)	< 0.0001
Lysis	9,092 (5.0%)	13,322 (39.5%)	< 0.0001
Hernia	13,011 (7.1%)	11,055 (32.8%)	< 0.0001
Debridement	86,935 (47.6%)	2,646 (7.8%)	< 0.0001
Other laparoscopy	2,606 (1.4%)	5,238 (15.5%)	< 0.0001
Other laparotomy	1,516 (0.8%)	6,588 (19.5%)	< 0.0001
Outcomes
Death	3,463 (1.9%)	2,503 (7.4%)	< 0.0001
Discharge to home	107,987 (59.2%)	16,100 (47.7%)	< 0.0001
Hospital LOS	5 [3, 9]	9 [5, 15]	< 0.0001

IQR = interquartile range; LOS length of stay.

Median [IQR] for continuous and % for categorical. Wilcoxon rank sum for continuous and χ² for categorical.

Cumulative procedure types and mortality

Having an increasing number of procedure types performed during one hospitalization led to increasing odds of death (Table 2). There was an association between the number of procedures and the odds of death. Compared with patients who had only one procedure, the odds of death increased as the number of procedures increased (Table 3): two procedures (OR, 3.01; 95% CI, 2.86–3.24), three procedures (OR, 3.92; 95% CI, 3.54–4.33), four procedures (OR, 5.74; 95% CI, 4.83–6.83), five procedures (OR, 7.64; 95% CI, 5.38–10.86), and more than five procedures (OR, 9.53; 95% CI, 4.92–18.49; Table 3)

Table 2.

Multivariable Logistic Regression: Outcome = Death

Variable	Adjusted OR (95% CI)
Number of ACS procedures
1 procedure	Ref.
2 procedures	3.04 (2.86–3.24)
3 procedures	3.92 (3.54–4.33)
4 procedures	5.74 (4.83–6.83)
5 procedures	7.64 (5.38–10.86)
>5 procedures	9.53 (4.92–18.49)
Age, y	1.03 (1.02–1.03)
Female	0.86 (0.81–0.91)
Number of Elixhauser comorbidities	1.38 (1.37–1.40)
Payer
Medicare	Ref.
Medicaid	1.07 (0.97–1.19)
Private Insurance	0.95 (0.87–1.03)
Self-pay	1.49 (1.27–1.74)
No charge	0.93 (0.51–1.70)
Other	1.23 (1.02–1.48)
Race
White	Ref
Black	1.06 (0.98–1.15)
Hispanic	0.86 (0.78–0.95)
Asian/Pacific Islander	1.26 (1.05–1.50)
Native American	1.14 (0.81–1.59)
Other	1.23 (1.05–1.46)
Zip code income
Highest income quartile	Ref
Second highest income quartile	0.96 (0.88–1.04)
Third highest income quartile	1.03 (0.95–1.12)
Lowest income quartile	1.09 (1.01–1.19)

OR = odds ratio; CI = confidence interval; ACS = acute care surgery.

Model pseudo r² = 0.15.

Hosmer Lemeshow goodness of fit = 0.000.

Table 3.

Specific Combinations, with Odds Ratio and 95% Confidence Intervals

Combination	n (%)	Adjusted OR (95% CI)
Triad
Colon/lysis/SB	794 (0.4%)	3.6 (2.78–4.67)
Lysis/SB/hernia	599 (0.3%)	2.21 (1.52–3.22)
Laparotomy/lysis/ulcer	350 (0.2%)	3.63 (2.36–5.60)
Laparotomy/SB/ulcer	343 (0.2%)	9.37 (6.73–13.03)
Laparotomy/hernia/ulcer	297 (0.1%)	6.52 (4.23–10.03)
Colon/SB/hernia	280 (0.1%)	3.68 (2.35–5.76)
Laparoscopy/lysis/gallbladder	256 (0.1%)	0.42 (0.10–1.69)
Laparotomy/colon/ulcer	232 (0.1%)	9.67 (6.69–14.00)
Other triad	276 (0.1%)	3.61 (3.14–4.14)
Any combination of 4	1,353 (0.6%)	5.68 (4.77–6.77)
Any combination of 5	275 (0.1%)	7.6 (5.33–10.84)
Any combination of 6 or 7	66 (0.03%)	9.62 (4.93–18.79)

Each line represents a multivariable logistic regression that includes the noted combination with single procedure group as the reference group, adjusted for age, gender, Elixhauser comorbidities, payer, race, and ZIP code income quartile. Logistic regression was performed for any triad that occurred in 0.1% or more of the total population (combination occurred at least 216). Other triad are the triad types that were less common, aggregated together.

OR = odds ratio; CI = confidence interval; SB = small bowel/appendix.

Specific single procedures and dyad combinations

Odds ratios and 95% confidence intervals for specific single and dyad procedure types are presented in Figure 1. The diagonal of the figure are the singletons. Procedure combinations that included gallbladder surgery tended have lower odds, whereas procedure combinations that included a laparotomy had higher odds (even when the combination was gallbladder plus laparotomy). The dyad with the highest odds ratio was laparotomy plus ulcer surgery (OR, 15.47; 95% CI, 13.72–17.46), followed by laparoscopy with laparotomy (OR, 9.3; 95% CI, 4.03–21.49). Two dyads, gallbladder plus hernia and gallbladder plus lysis had odds of mortality less than one, likely a reflection of the generally lower mortality for gallbladder operations overall compared with other ACS procedures.

FIG. 1.

Adjusted odds ratios for specific single procedure and dyad combinations. For single procedures (presented on the diagonal), the reference group is all other patients with single procedures. For dyads, the reference group is all patients with one acute care surgery procedure. SB = small bowel or appendix procedure. Adjusted odds ratios and 95% confidence intervals are presented. *Indicates p < 0.05. Blue indicates lower odds and red indicates increased odds; dark red notes odds ratios with a point estimate of 5 or greater. Color image is available online.

Specific combinations of three or more

Eight triad combinations were present in the population at 0.1% prevalence or higher (at least 216 occurrences) and are presented in Table 3. No procedure type combinations that included four or more procedures occurred at a rate higher than 0.1%, and therefore all four-procedure, five-procedure, or six-/seven-procedure type combinations are aggregated for regression analysis. No individual admission included more than seven procedure types. Similar to findings with dyads, combinations that included laparotomy had higher odds of death; in particular, combinations that included laparotomy plus ulcer conferred particularly higher odds (laparotomy/colon/ulcer: OR, 9.67; 95% CI, 7.13–14.62), and laparotomy/small bowel/ulcer (OR, 9.37; 95% CI, 6.73–13.03). Table 3 also shows that patients with four, five, or more procedure types have increased odds of in-hospital death. Table 3 estimates vary slightly from Table 2, as slightly different comparisons groups were utilized (Table 3 models limited to only the combination in question and the reference group; Table 2 represents a single logistic regression which included all ACS patients).

Discussion

Acute care surgery carries a significant increased risk of morbidity and mortality in addition to increases in cost to the U.S. healthcare system. As demonstrated by Havens et al.,⁶ ACS is an independent risk factor for negative outcomes independent of patient pre-operative comorbidities and physiologic condition. It is well known that given the complex nature of ACS diseases, patients may undergo more than one operation during the same admission. Our study, designed to investigate if specific combinations of ACS procedures result in an increased odds of in-hospital mortality in the post-operative period, showed that there is a statistically significant association between increasing numbers of procedure types performed at one admission and increased odds of death. In addition, specific combinations of procedure types seem to be particularly risky, including disease that requires treatment with laparotomy, ulcer disease, and bowel surgery. Knowledge of these results will enable the ACS practitioner to provide better pre-operative counseling to patients and families and give more specific expectations regarding mortality based on the expected necessities of treatment, specifically number of procedures required during a single admission.

It is important to identify and better characterize patients who will most likely undergo more than one operation to coordinate and accommodate their care around the known higher risk of complications and mortality. It is possible, for example, that a patient or family might elect to pursue palliative care in lieu of undergoing numerous procedures with a higher chance of mortality if this were known at the beginning of a course of illness. Lunardi et al.²⁵ reported that among ACS patients from the National Readmissions Database, 18.6% were re-admitted within six months, and 11.6% of re-admitted patients had another ACS procedure at a subsequent admission. The majority of those patients had their second procedure within 90 days. Similar to our findings, patients in the study by Lunardi et al.²⁵ who were treated for ulcer disease at the index admission had the highest risk and were most likely to have another procedure at the time of repeat admission.

Our analysis showed that there is increasing odds of death for each additional procedure, with a clear relation whereby the odds of death increased as the number of procedure types increased. We note that the model fit is poor; in other words, the covariates included in our model are not enough to explain mortality, but the effect sizes are large and the relations are clearly present. In addition, our data also suggest that it is not just the number of procedures but also the specific combinations of procedures. Biliary disease, which occurred in 26.7% of single procedure admissions and 23.2% of multiple procedure admissions had generally lower odds of mortality even in combinations, unless that admission also included a laparotomy. Admissions in which a laparotomy occurred in addition to another procedure had the highest odds of mortality. For most open procedures such as an open colectomy, a laparotomy procedure code would not be used concurrently, because these codes are bundled together; for these, a laparotomy code at a separate procedure would likely be a sign of a severe complication such as severe recurrent intra-abdominal sepsis from an anastomotic leak or abscesses. Non-directed relaparotomy for sepsis has poor outcomes, explaining the higher proportion of mortality for combinations that include laparotomy.^26–29 Our data help to quantify the odds of mortality for ACS patients with multiple surgical emergencies and multiple procedure type combinations.

In our analysis, patients who had single procedure types differed from patients who needed multiple procedure types. Patients who needed multiple procedures were older, were more likely to be white, had more comorbidities, and were more likely to be from lower income ZIP codes. All of these factors are important for risk stratification; in particular, our data can be used to help counsel patients and families who return to the operating room during the same admission. In particular, it is known that older adults with ACS pathologies, particularly those who are frail or who have multiple comorbidities, have high risk of mortality and a high burden of palliative care needs.^17,30 For patients with high and increasing odds of death with each subsequent procedure, it is important to continue to iteratively address goals of care as risk changes. We hope the analysis provided here will provide clinicians with estimations of odds of death in these difficult cases.

There are several limitations to our analysis. It was not possible to extract if a procedure was performed concurrently or sequentially because of limitations with the NIS. We suspect that there may be a difference in odds between patients who, for example, have a laparoscopic cholecystectomy and a hernia repair done at the same operation and a laparoscopic cholecystectomy who develops a post-operative port site hernia and needs an emergency hernia repair during the same admission. However, we expect that concurrent operations would be a relatively uncommon occurrence, especially given that some procedures cannot be billed together (such as open abdominal procedures such as ulcer, small bowel, or colon operations, which cannot be billed with a laparotomy code). We also believe that in general, the ACS admission would be to treat one pathology emergently, and subsequent operations would be undertaken because they were urgently necessary because of a complication; it would be rare to develop an unrelated pathology in the same admission that would also require operative management. In addition, as we grouped multiple procedures ICD-10 codes into each category, we would not have identified patients who had two separate procedures of the same type, for example a patient with a small bowel resection who needed a re-operation for a second small bowel resection or a patient who had multiple unspecified laparotomies without other procedures. Additionally, because of limitations in the NIS data, we cannot account for coding errors in comorbidities, surgeries, or re-admissions. We also did not implement code for traumatic injuries, and therefore did not systematically exclude trauma patients. Despite these limitations, our data shows a remarkable increase in mortality when multiple procedure types are performed. Use of the NIS allowed an assessment of mortality for combinations of procedures that would have too rare to analyze in smaller datasets.

Emergency general surgery confers a high risk of mortality, and odds of mortality increase as patients require and undergo increasing numbers of ACS procedures in a single admission. Combinations of procedure types that include gastrointestinal tract procedures such as ulcer surgery and bowel surgery are particularly associated with increased mortality. However, it is the addition of an unspecified laparotomy that is associated with the highest odds of mortality when combined with other procedures. Future studies can explore additional specific risk factors to help further risk-stratify and illuminate outcomes in these high, risk patients. For patients at the highest risk of in-hospital death, communication with the patient and their surrogates and caregivers about additional risk accrued when complications require additional surgery is of the utmost importance.

Footnotes

Acknowledgments

This publication was made possible by the Clinical and Translational Science Collaborative of Cleveland, KL2TR002547 from the National Center for Advancing Translational Sciences (NCATS) component of the National Institutes of Health and NIH roadmap for Medical Research. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.

Parts of this work were presented as a poster presentation at the 80th Annual Meeting of the American Association for the Surgery of Trauma, September and October, 2021, Atlanta, Georgia.

Authors' Contributions

Literature review: S.V., J.A.F., E.C., C.W.T., J.A.C., V.P.H. Study design: S.V., W.P.B., C.W.T., V.P.H. Data collection: W.P.B., E.C., V.P.H. Analysis: S.V., W.P.B., C.W.T., J.A.C., V.P.H. Manuscript writing: all authors. Critical revisions: all authors

Funding Information

Dr. Ho is supported by the Clinical and Translational Science Collaborative of Cleveland (KL2TR002547) from the National Center for Advancing Translational Sciences (NCATS) component of the National Institutes of Health and NIH roadmap for Medical Research. Wyatt P. Bensken is supported by the National Institute on Minority Health and Health Disparities of the National Institutes of Health (1F31MD015681).

Author Disclosure Statement

Dr. Towe is a consultant for Medtronic, Zimmer Biomet, Atricure, and Astra Zeneca. Dr. Ho's spouse is a consultant for Medtronic, Zimmer Biomet, Atricure, and Astra Zeneca. The remainder of authors have no conflicts of interest to disclose.

References

Ogola

, Gale

, Haider

, Shafi

. The financial burden of emergency general surgery: National estimates 2010 to 2060. J Trauma Acute Care Surg, 2015; 79(3):444–448; doi: 10.1097/TA.0000000000000787.

Gale

, Shafi

, Dombrovskiy

, et al. The public health burden of emergency general surgery in the United States: A 10-year analysis of the Nationwide Inpatient Sample—2001 to 2010. J Trauma Acute Care Surg, 2014; 77(2):202–208; doi: 10.1097/TA.0000000000000362.

Bensken

, Pieracci

, Ho

. Basic introduction to statistics in medicine, Part 1: describing data. Surg Infect, 2021; 22(6):590–596; doi: 10.1089/sur.2020.429.

Shafi

, Aboutanos

, Agarwal

Jr , et al. Emergency general surgery: Definition and estimated burden of disease. J Trauma Acute Care Surg, 2013; 74(4):1092–1097; doi: 10.1097/TA.0b013e31827e1bc7.

Shah

, Haider

, Zogg

, et al. National estimates of predictors of outcomes for emergency general surgery. J Trauma Acute Care Surg, 2015; 78(7):482–491; doi: 10.1016/S2352-4642(19)30123-3.

Havens

, Peetz

, Do

, et al. The excess morbidity and mortality of emergency general surgery. J Trauma Acute Care Surg, 2015; 78(2):306–311; doi: 10.1097/TA.0000000000000517.

Scott

, Olufajo

, Brat

, et al. Use of national burden to define operative emergency general surgery. JAMA Surg 2016 Jun 15;151(6):e160480; doi: 10.1001/jamasurg.2016.0480.

Bilimoria

, Liu

, Paruch

, et al. Development and evaluation of the universal ACS NSQIP surgical risk calculator: A decision aid and informed consent tool for patients and surgeons. J Am Coll Surg, 2013; 217(5):833–842; doi: 10.1016/j.jamcollsurg.2013.07.385.

Nandan

, Bohnen

, Sangji

, et al. The Emergency Surgery Score (ESS) accurately predicts the occurrence of postoperative complications in emergency surgery patients. J Trauma Acute Care Surg, 2017; 83(1):84–89; doi: 10.1097/TA.0000000000001500.

10.

Sangji

, Bohnen

, Ramly

, et al. Derivation and validation of a novel Physiological Emergency Surgery Acuity Score (PESAS). World J Surg, 2017; 411782–411789.

11.

Sangji

, Bohnen

, Ramly

, et al. Derivation and validation of a novel Emergency Surgery Acuity Score (ESAS). J Trauma Acute Care Surg 2016 Aug;81(2):213-20. doi: 10.1097/TA.0000000000001059.

12.

HCUP National Inpatient Sample (NIS). Healthcare Cost and Utilization Project (HCUP): Agency for Healthcare Research and Quality, Rockville, MD; 2017.

13.

Stulberg

, Haut

. Practical guide to surgical data sets: Healthcare cost and utilization Project National Inpatient Sample (NIS). JAMA Surg, 2018; 153(6):586–587; doi: 10.1001/jamasurg.2018.0542.

14.

Smith

, Knight Davis

, Quatman-Yates

, et al. Loss of community-dwelling status among survivors of high-acuity emergency general surgery disease. J Am Geriatr Soc, 2019; 67(11):2289–2297; doi: 10.1111/jgs.16046.

15.

Agrawal

, Imielinski

, Swami

. Mining association rules between sets of items in large databases. Proceedings of the 1993 ACM SIGMOD International Conference on Management of Data. Washington, D.C.;, 1993:207–216.

16.

Agrawal

, Srikant

. Fast algorithms for mining association rules. Proceedings of the 20th International Conference on Very Large Data Bases; Santiago de Chile, Chile;, 1994:487–499.

17.

, Schiltz

, Reimer

, et al. High-risk comorbidity combinations in older patients undergoing emergency general surgery. J Am Geriatr Soc, 2019; 67(3):503–510; doi: 10.1111/jgs.15682.

18.

Tai

, Chiu

. Comorbidity study of ADHD: Applying association rule mining (ARM) to National Health Insurance Database of Taiwan. Int J Med Inform, 2009; 78(12):e75–83; doi: 10.1016/j.ijmedinf.2009.09.005.

19.

Schiltz

, Warner

, Sun

, et al. Identifying specific combinations of multimorbidity that contribute to health care resource utilization: An analytic approach. Med Care, 2017; 55(3):276–284; doi: 10.1097/MLR.0000000000000660.

20.

Elixhauser

, Steiner

, Harris

, Coffey

. Comorbidity measures for use with administrative data. Med Care, 1998; 36(1):8–27; doi: 10.1097/00005650-199801000-00004.

21.

Zogg

, Olufajo

, Jiang

, et al. The need to consider longer-term outcomes of care: Racial/ethnic disparities among adult and older adult emergency general surgery patients at 30, 90, and 180 days. Ann Surg, 2017; 266(1):66–75; doi: 10.1097/SLA.0000000000001932.

22.

Hall

, Hashmi

, Zafar

, et al. Racial/ethnic disparities in emergency general surgery: Explained by hospital-level characteristics?. Am J Surg, 2015; 209(4):604–609; doi: 10.1016/j.amjsurg.2014.11.005.

23.

Hahsler

, Grun

, Hornik

. arules–A computational environment for mining association rules and frequent item sets. J Stat Softw, 2005; 14(15):1–25; doi.org/ 10.18637/jss.v014.i15.

24.

Hahsler

. arulesViz: Interactive visualization of association rules with R. R J, 2017; 9(2):163–175; doi: 10.32614/RJ-2017-047.

25.

Lunardi

, Mehta

, Ezzeddine

, et al. Recurring emergency general surgery: Characterizing a vulnerable population. J Trauma Acute Care Surg, 2019; 86(3):464–470; doi: 10.1097/TA.0000000000002151.

26.

Hutchins

, Gunning

, Lucas

, et al. Relaparotomy for suspected intraperitoneal sepsis after abdominal surgery. World J Surg, 2004; 28(2):137–141; doi: 10.1007/s00268-003-7067-8.

27.

Bunt

. Non-directed relaparotomy for intra-abdominal sepsis. A futile procedure. Am Surg, 1986; 52(6):294–298.

28.

Unalp

, Kamer

, Onal

. Analysis of early relaparotomy after lower gastrointestinal system surgery. Surg Today, 2008; 38(4):323–328; doi: 10.1007/s00595-007-3621-6.

29.

Harbrecht

, Garrison

, Fry

. Early urgent relaparotomy. Arch Surg, 1984; 119(4):369–374; doi: 10.1001/archsurg.1984.01390160005001.

30.

Lilley

, Cooper

. The high burden of palliative care needs among older emergency general surgery patients. J Palliat Med, 2016; 19(4):352–353; doi: 10.1089/jpm.2015.0502.