Clinical profile, outcomes and mortality determinants among emergency general surgery patients in a Latin American tertiary hospital: a retrospective registry-based cross-sectional study

Abstract

Background

Emergency General Surgery (EGS) patients are clinically complex and frequently underreported, particularly in Latin America, which hampers benchmarking and quality-improvement efforts. We implemented a hospital-based structured data collection system to characterize EGS patients and outcomes in a tertiary university hospital.

Methods

We conducted a retrospective cross-sectional study of a prospectively collected database including all patients aged ≥ 13 years who required emergency general surgery between May and November 2020 at a high-complexity public university hospital in Costa Rica, comparable to a US Level I trauma center. Sociodemographic, perioperative, and outcome variables were collected. Descriptive statistics were computed. Univariate analyses and multivariable Firth penalized logistic regression were performed to identify factors independently associated with in-hospital mortality.

Results

A total of 575 patients were included; 52.9% were male and the most frequent age group was 18–29 years (23.7%). Most procedures occurred outside ordinary hours (72.2%), and surgical residents were the primary surgeons in 51.1% of cases. Laparoscopy was the initial approach in 67.8%. The most frequent care categories were gastrointestinal emergencies (48.2%), hepatobiliary emergencies (23.7%), and trauma (8.2%). Acute appendicitis (37.4%) and gallbladder and biliary disease (23.3%) were the leading diagnostic groups. Overall in-hospital mortality was 5.6%. In the final complete-case cohort (n = 511), mechanical ventilation, ASA III-IV status, hemodynamic instability, and open abdomen were independently associated with in-hospital mortality.

Conclusions

These findings provide a contemporary description of emergency general surgery patients treated at a Latin American tertiary-care hospital and identify clinical factors associated with in-hospital mortality.

Trial registration

NCT05930574

Introduction

Worldwide, the care of emergency general surgery patients is limited compared with patients in other surgical domains because of the inherent complexity of this population, its pathological characteristics, incomplete patient information, variability in available resources as infrastructure and human resources, and the heterogeneity in their clinical management. Previous studies have shown that critical decision-making for these patients is often suboptimal [1]. This variability is also driven by underreporting frequently observed in Latin American countries [2]. Currently, no standard data collection or registry model applicable to emergency general surgery patients exists in the region. Key indicators such as morbidity and mortality ⁱⁿ emergency patients remain unclear [3], partly because of the low number of scientific publications available, which also include low case numbers [4, 5]. Even in countries of other regions such as North America and Europe that have extensive registries for data analysis, reporting of these indices is as well heterogeneous [6]. For example, a multicenter retrospective study recently published by Saunders et al. reported a mortality rate ranging from 3.6% to 41.7% [7] among patients undergoing emergency laparotomy, highlighting a high outcome variability. In contrast, a review by Stahel et al. [8] reported a 50% of complication rate in these patients and up eightfold increase in mortality compared to elective surgery patients.

To address these limitations, Structured data collection systems may facilitate characterization of emergency surgical patients to profile and understand their clinical characteristics. It has been proven to be beneficial to optimize resources and make the corresponding practical modifications that allow improving the quality of care [9]. The objective of this study was to characterize the clinical and epidemiological profile of emergency general surgery patients treated at a high-complexity Latin American university hospital using a structured institutional database.

Materials and methods

This study was a retrospective cross-sectional analysis of a prospectively maintained institutional emergency general surgery database conducted at Dr. Rafael Ángel Calderón Guardia Hospital (Costa Rican Social Security Fund), a high-complexity public university hospital comparable to a US level I trauma center. The affiliated population is approximately 1.3 million people, with 81,084 annual emergency admissions and 16,583 annual surgical emergency admissions. Eligible patients were identified and entered contemporaneously during routine clinical care between May and November 2020, whereas statistical analyses and outcome assessments were performed retrospectively after completion of the data collection period.

We included all patients aged 13 years or older who required emergency general surgery, including trauma, during the study period. We excluded patients operated primarily by other surgical specialties within the hospital (Urology, vascular surgery, gynecology, Oncologic surgery, among others) and patients whose first surgical procedure was performed at another medical center. The patient selection process and final analytical cohort are shown in (Fig. 1).

Fig. 1.

STROBE flow diagram of patient inclusion, outcome ascertainment, and final analytical cohort

Case ascertainment was performed through daily operating-room schedules and emergency surgery logs. Surgical residents notified the study team when eligible patients were scheduled for emergency surgery, and operating-room records were reviewed to verify inclusion. Because case notification depended partially on rotating residents, missed cases cannot be completely excluded and this was considered a potential source of selection bias.

Data were obtained from the National Single Digital Patient Record (EDUS) and operating-room logs. Variables included demographic characteristics, comorbidities, ASA physical status classification, admission unit, hemodynamic stability at admission, timing of surgery, surgical approach, primary surgeon, damage-control surgery, open abdomen, temporary abdominal closure technique, mechanical ventilation, postoperative complications, and in-hospital mortality. Hemodynamic instability was defined as systolic blood pressure less than or equal to 90 mmHg or vasopressor-supported shock at admission. ASA III-IV status was analyzed as a dichotomous marker of baseline severity.

Data abstraction was performed by the principal investigator (JJML) using the National Single Digital Patient Record (EDUS). A second investigator (LCB) independently reviewed database completeness, internal consistency, and outcome coding. Discrepancies were resolved through consensus review of the original medical record. Formal inter-rater reliability testing was not prospectively planned and therefore was not calculated.

The manuscript was prepared in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement.

Statistical analysis

Data were analyzed by using Statistical Package for Social Sciences (SPSS), version 28.0, (SPSS Inc., Chicago, USA) and Microsoft Excel, version 16.69.1, (Microsoft Corporation, Redmond, USA). All statistical tests were two-sided with significance level set at α = 0.05. Categorical variables are reported as counts and percentages. Associations were assessed using Fisher’s exact test because of sparse cells. The following dichotomized variables were examined: damage-control surgery, hemodynamic instability at admission defined SBP 90 mmHg, open abdomen technique, mechanical ventilation requirement and ASA III–IV status (defined as ASA III or IV versus ASA I–II). Variables demonstrating statistical association with mortality at the p < 0.05 threshold in univariate analysis were selected as candidates for inclusion in the multivariable model, following established methodology to avoid exclusion of potentially important confounders and effect modifiers. A binary logistic regression model was designed using variables selected from univariate analysis. The model building strategy employed simultaneous (enter) inclusion of candidate variables that met the p < 0.05 threshold to avoid stepwise selection bias. The outcome variable was dichotomized in-hospital mortality (yes/no).

The final multivariable model included ASA III-IV status, hemodynamic instability at admission, damage-control surgery, open abdomen, and mechanical ventilation. Missing outcome data reflected incomplete discharge disposition documentation in the institutional database rather than uncertainty regarding vital status. Descriptive analyses used all available observations. Multivariable regression was performed using complete-case analysis. 64 patients (11%) were excluded from the regression model because of incomplete information in one or more predictor variables, resulting in a final analytical sample of 511 patients for regression modeling. Firth penalized logistic regression was used to reduce small-sample and separation-related bias. Results are reported as adjusted odds ratios (ORs) with 95% confidence intervals (CIs). Discrimination was assessed using the area under the receiver operating characteristic curve (AUC). Calibration was assessed using a Hosmer-Lemeshow goodness-of-fit statistic based on ranked deciles of predicted risk (Fig. 1).

Results

Among 575 patients, 52 were excluded because of duplicate records, missing key variables, or failure to meet inclusion criteria. The final analytic cohort consisted of 523 patients (Fig. 1). Of those, 304 (52.9%) were male. The most represented age group was 18–29 years (136/575, 23.7%). Comorbidities were present in 44.2% (n = 254); most patients were ASA I (56.7%, n = 326). Hemodynamic instability at admission (SBP < 90 mmHg) was observed in 4.3% (n = 25). Baseline characteristics are summarized in Table 1 and our institution’s characteristics in Table 2.

Table 1.

General characteristics of patients who required emergency surgery in our center

Characteristic	Number of patients n (%)
Sex: male	304 (52.9)
Age group
Age group: 13–17	35 (6.1)
Age group: 18–29	136 (23.7)
Age group: 30–39	104 (18.1)
Age group: 40–49	82 (14.3)
Age group: 50–64	107 (18.6)
Age group: 65 and up	108 (18.8)
Age missing	3 (0.5)
Comorbidities (presence of ≥ 1 comorbidity according to Clavien–Dindo classification)	254 (44.2)
No comorbidity recorded	321 (55.8)
ASA Rating
ASA I	326 (56.7)
ASA II	202 (35.1)
ASA III	30 (5.2)
ASA IV	4 (0.7)
ASA missing	13 (2.3)
Hemodynamic instability at admission (SBP < 90mmHg)	25 (4.3)
Timing of surgery
Ordinary shift	160 (27.8)
Non-ordinary shift	415 (72.2)
Main Surgeon
Attending primary surgeon	281 (48.9)
Surgical resident primary surgeon	294 (51.1)
Surgical Approach
Initial laparoscopic approach	391 (68.0)
Initial open approach	184 (32.0)
Damage-control surgery	20 (3.5)
Open abdomen	50 (8.7)
Mechanical ventilation	27 (4.7)
ICU admission	9 (1.6)
Overall in-hospital mortality	32 (5.6)

Table 2.

Characteristics of the center where the study was conducted

Characteristic	Value
Total beds	457
Surgery beds	175
ICU beds	34
Annual emergency admissions	81,084
Annual admissions to surgical emergencies	16,583
Affiliated population	1,300,000

Most procedures occurred outside ordinary hours (n = 415, 72.2%). Surgical residents were the primary surgeons in 294 cases (51.1%). Laparoscopy was the initial approach in 391 cases (68.0%), and conversion was required in 32 of laparoscopic cases (8.2%).

The most frequent care categories were gastrointestinal emergencies (n = 277, 48.2%), hepatobiliary emergencies (n = 136, 23.7%), and trauma (n = 47, 8.2%) (Table 3). The leading diagnostic groups were appendicitis (n = 215, 37.4%) and gallbladder and biliary disease (n = 134, 23.3%) (Table 4).

Table 3.

Major categories of care in surgical emergencies

Category	Number and percentage of patients (%)
Gastrointestinal emergencies	277 (48.2)
Hepatobiliary emergencies	136 (23.7)
Trauma	47 (8.2)
Abdominal wall emergencies	32 (5.6)
Neoplasms	25 (4.3)
Proctological emergencies	25 (4.3)
Soft tissue infections	5 (0.9)
Other	28 (4.9)

Table 4.

Main diagnoses

Diagnosis	Number and percentage of patients (%)
Appendicitis	215 (37.4)
Gallbladder and biliary disease	134 (23.3)
Trauma	47 (8.2)
Hernias	32 (5.6)
Neoplasms	25 (4.3)
Other diagnosis combined	122 (21.2)

Open abdomen management was required in 50 patients (8.7%), and damage-control surgery was performed in 20 patients (3.5%). Mechanical ventilation was required in 27 patients (4.7%). Overall in-hospital mortality was 5.6% (n = 32). However, only 24 deaths were included in the complete-case regression model because eight deaths had incomplete information in one or more predictor variables.

In univariate analysis, ASA III-IV status, hemodynamic instability at admission, damage-control surgery, open abdomen, and mechanical ventilation were all associated with in-hospital mortality (all p < 0.001). In multivariable Firth penalized regression, mechanical ventilation (OR 11.03, 95% CI 2.71–44.85), hemodynamic instability (OR 6.26, 95% CI 1.60-24.43), ASA III-IV status (OR 9.20, 95% CI 2.63–32.20), and open abdomen (OR 3.90, 95% CI 1.01–14.97) remained independently associated with mortality. The model demonstrated good discrimination (AUC = 0.898) and acceptable calibration by the Hosmer-Lemeshow test (chi-square = 11.48, df = 8, p = 0.176) (Table 5). Given the limited number of mortality events, model performance estimates should be interpreted cautiously and require external validation.

Table 5.

Dichotomized clinical characteristics of operated patients and penalized multivariable logistic regression for in-hospital mortality

Variable	Alive n (%)	Deceased n (%)	Univariate p	Adjusted OR (95% CI)	p	VIF
ASA III–IV
Yes	17 (3.3)	10 (2.0)	< 0.001	9.20 (2.63–32.20)	< 0.001	1.11
No	470 (92.0)	14 (2.7)	Reference	Reference	—	—
Hemodynamic instability at admission
Yes	8 (1.6)	11 (2.2)	< 0.001	6.26 (1.60–24.43)	0.008	1.50
No	479 (93.7)	13 (2.5)	Reference	Reference	—	—
Damage-control surgery
Yes	5 (1.0)	10 (2.0)	< 0.001	1.95 (0.33–11.61)	0.461	1.77
No	482 (94.3)	14 (2.7)	Reference	Reference	—	—
Open abdomen
Yes	27 (5.3)	13 (2.5)	< 0.001	3.90 (1.01–14.97)	0.048	1.40
No	460 (90.0)	11 (2.2)	Reference	Reference	—	—
Mechanical ventilation
Yes	8 (1.6)	13 (2.5)	< 0.001	11.03 (2.71–44.85)	< 0.001	1.69
No	479 (93.7)	11 (2.2)	Reference	Reference	—	—

OR odds ratio, CI confidence interval, VIF variance inflation factor

Percentages are calculated using the complete-case cohort as denominator

Univariate p-values were calculated using Fisher’s exact test because of sparse cells. Adjusted odds ratios were estimated using Firth penalized logistic regression to reduce small-sample and separation-related bias. Model discrimination: AUC = 0.898. Hosmer-Lemeshow goodness-of-fit test based on ranked deciles of predicted risk: χ² = 11.48, df = 8, p = 0.176

Discussion

During the six-month study period, 575 emergency general surgery patients were recorded. This database allowed characterization of the patients treated at our institution and provided a contemporary description of emergency surgical practice in a Latin American tertiary-care center. Regarding diagnoses, the findings of this study differ from those reported in studies conducted in low- and middle-income countries. For example, a retrospective study by Shah et al [10], which included data from 13,893 emergency general surgery patients across resource-limited settings, reported biliary diseases in 20.2%, soft tissue pathologies in 15.7%, and hernias in 14.9% as the most common diagnoses. In contrast, our structured data collection system identified acute appendicitis as the leading diagnosis (37.4%), followed by biliary pathologies (23.3%) and trauma (8.2%). These findings provide a local epidemiological profile of emergency general surgery patients treated at our institution. Regarding postoperative outcomes, the complication rate observed in our cohort was comparable to rates reported in previous emergency surgery studies [11–13]. Mullen et al [12]. reported morbidity rates of 12.3% among urgent procedures and 13.8% among emergent procedures using data from the American College of Surgeons National Surgical Quality Improvement Program database. These findings are consistent with the complication profile observed in our population, although direct comparisons should be interpreted cautiously because differences in patient selection, operative complexity, and outcome definitions may exist across studies. The overall in-hospital mortality of 5.6% was higher than that reported in some broader EGS cohorts, Shah et al [10]. reported an overall mortality of 2.7%, and Mullen et al [12]. reported mortality rates of 2.3% for urgent patients and 3.7% for emergent patients, both of which are lower than the mortality rate observed in our study. This finding may reflect differences in case mix, pandemic-period healthcare utilization, referral patterns, and the inclusion of high-acuity operative emergencies and may also reflect variation in EGS definitions and operative inclusion criteria. We acknowledge however, that direct comparison with other studies should be cautious because definitions of EGS, operative inclusion criteria, and outcome ascertainment vary considerably across systems [14]. Mechanical ventilation was independently associated with mortality. This association should not be interpreted as causal. Rather, mechanical ventilation likely reflects severe physiologic derangement, critical illness, and greater need for organ support. Similarly, ASA III-IV status captured baseline physiologic reserve and comorbidity burden, both of which are clinically plausible determinants of adverse postoperative outcomes. Hemodynamic instability and open abdomen also remained associated with mortality in the adjusted analysis. These findings are consistent with the clinical expectation that patients presenting with shock or requiring complex abdominal management represent a higher-risk subgroup. Damage-control surgery was associated with mortality in univariate analysis but did not retain statistical significance after adjustment, suggesting confounding by severity and related operative variables. One of the major challenges faced by surgical services worldwide is the standardization and optimization of care for emergency general surgery patients, who exhibit significantly higher complication and mortality rates than elective surgical patients8. This underscores the importance of establishing database collections based on standardized classification systems, as proposed in this study. This perspective is aligned with the current recommendations of the American College of Surgeons (ACS) and the American Association for the Surgery of Trauma (AAST) as outlined in the recently published General Emergency Surgery Verification Program [13]. Although, robust evidence on the systemic impact of database implementation is limited, several studies have reported improvements in surgical care associated with structured data collections [15–17].

This study has several limitations that complicate the interpretation of its findings. Some limitations are inherent to the methodology and study design, including the retrospective interpretation of collected information and the fact that data were obtained from a single hospital service. Additionally, data collection was conducted during 2020, a period marked by significant social restrictions due to the SARS-CoV-2 pandemic, which altered healthcare system operations and likely produced an atypical epidemiological distribution of pathologies. Findings from this single-center, retrospective study conducted during the COVID-19 pandemic may not be generalizable to other healthcare settings or time periods, in fact this circumstance highlights the need for comparative studies to evaluate potential clinical and epidemiological differences in the post-pandemic period. However, our findings suggest that even within a pandemic crisis, the systematic data collection for emergency general surgery is feasible in a resource-constrained Latin American setting and provides a framework that can be adapted to other institutions in the region. Specific methodological limitations are also noteworthy. Patient recruitment depended on notification by rotating surgical residents at the time patients were scheduled for the operating room, making inclusion contingent upon such notifications and potentially leading to missed cases. Operating-room records were reviewed to verify inclusion. Furthermore, database completion was performed by the principal investigator using information extracted from the Single Digital Patient Record of the Costa Rican Health System (EDUS); in some instances, relevant variables had to be obtained from narrative progress notes, introducing potential bias because data quality depended on the accuracy and completeness of resident documentation. Additionally, missing outcome data reflected incomplete discharge disposition coding in the institutional database and not uncertainty regarding patient vital status. Finally, despite the use of penalized regression techniques, the number of mortality events remained relatively small, limiting model complexity and precision of effect estimates.

Conclusions

Emergency general surgery patients treated at this Latin American tertiary-care hospital represented a heterogeneous and clinically complex population. This study demonstrates that systematic collection of emergency general surgery data was achievable within routine clinical practice at our institution.

This structured database facilitated characterization of the local EGS population and may support future quality-improvement initiatives. Mechanical ventilation, ASA III-IV status, hemodynamic instability, and open abdomen were associated with mortality in the final penalized model. These findings should be interpreted as hypothesis-generating and require external validation in multicenter cohorts.

Authors’ contributions

JJML contributed to the conceptualization, methodology, software development, validation, formal analysis, investigation, resource management, data curation, original draft writing, review and editing, visualization, supervision, project administration, and funding acquisition. LCB participated in the methodology, software development, validation, formal analysis, and data curation. APPH contributed to validation, review and editing of the manuscript and visualization. MREM was involved in the review and editing of the manuscript and provided supervision. All authors read and approved the final manuscript.

Funding Statement

The authors declare that no external funding was received for the development, analysis, or publication of this study.

Data availability

The datasets generated and analysed during the current study are not publicly available due to institutional restrictions from the Costa Rican Social Security Fund (CCSS). Access to the data is limited to authorized personnel. However, de-identified data may be available from the corresponding author upon reasonable request and with permission from the CCSS.

Declarations

Ethical approval and consent

The local Bioethics Committee at Dr. Rafael Ángel Calderón Guardia Hospital approved the study (CEC-CNCDYCP-01-2023). Given its retrospective study design, informed consent was waived. The study was conducted in accordance with Good Clinical Practice guidelines and applicable national regulations and was registered at ClinicalTrials.gov on June 30, 2023 (NCT05930574).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.