Abstract
Objective
This study aimed to identify independent predictors of under-triage among patients with non-traumatic acute abdominal pain (NAAP) and quantify their impact on emergency department length of stay (EDLOS).
Methods
A multicenter retrospective study was conducted at three tertiary hospitals in China from December 2023 to May 2024, enrolling 769 patients with NAAP. Data collected included patient demographics, environmental factors, nurse characteristics, and clinical presentations. Binary logistic regression was employed to identify risk factors for under-triage, and multivariate linear regression was used to assess its association with EDLOS.
Results
The incidence of under-triage was 16.38% (126/769). Multivariate analysis revealed that a history of hyperlipidemia (OR = 7.944), scleral icterus (OR = 5.731), persistent pain (OR = 4.116), and advanced age (OR = 2.447) were significant independent risk factors for under-triage (all P < 0.05). Conversely, higher nursing seniority (Senior Nurse: OR = 0.295; Charge Nurse: OR = 0.311) served as a protective factor. Furthermore, under-triage independently predicted operational inefficiency, prolonging EDLOS by approximately 310 min (B = 309.765, P < 0.001) after adjustment for confounders.
Conclusion
This multicenter study demonstrates that under-triage in NAAP is associated with patient age, history of hyperlipidemia, higher nurse professional titles, and abdominal pain characteristics. Under-triage significantly prolongs EDLOS. These findings provide a foundation for understanding and mitigating under-triage through competency-based training and risk-stratified assessment protocols in emergency departments.
Keywords: Triage, Abdominal pain, Under-triage, Influencing factors, Emergency nursing
Introduction
Nontraumatic acute abdominal pain (NAAP), also referred to as acute abdomen (AA) or acute abdominal pain (AAP), is defined by AAP arising from intra‑abdominal lesions or extra‑abdominal conditions, including thoracic and systemic diseases. The condition typically develops within one week and may require urgent surgical or medical intervention [1]. NAAP accounts for approximately 5%-10% of ED visits [2]. Its etiologies are diverse and complex, including acute appendicitis, acute pancreatitis, gastritis, cholelithiasis, ureteral stones, perforated peptic ulcer, ectopic pregnancy, and intestinal obstruction. Multiple organs and systems are often involved. In severe cases, NAAP can be life‑threatening [2–4], posing a substantial challenge to emergency triage.
Triage is the initial step in emergency care and involves rapid patient assessment and prioritization based on illness severity [5]. However, under‑triage frequently occurs due to factors such as ED crowding and the limited time available for triage [6, 7]. Under‑triage is defined as the assignment of a triage level lower than the patient’s actual clinical severity. This may result in delayed treatment and intervention and can increase the risk of adverse or fatal outcomes [8]. Previous studies have shown that underestimation of NAAP severity may lead to delayed diagnosis, prolonged operative time, and increased morbidity and mortality [9]. Factors associated with under‑triage have been investigated in patients with cardiovascular disease, trauma, and general abdominal pain [10–12]. However, evidence specific to NAAP remains limited. Therefore, this study aimed to describe the prevalence of under‑triage in NAAP, identify factors associated with nurse under‑triage, and examine its association with EDLOS, with the goal of informing and improving triage management for NAAP patients.
Methods
Study subjects
A convenience sampling method was employed to select NAAP patients who visited the EDs of two hospitals in Shanghai and one hospital in Suzhou, Jiangsu Province, from December 2023 to May 2024. Inclusion criteria were as follows: (1) age ≥ 18 years; (2) diagnosis of NAAP based on the “2015 Guidelines for Primary Care of Acute Abdomen” [1]; and (3) availability of complete patient data (including general information, vital signs, triage content, resuscitation, and death records). Exclusion criteria included: (1) patients with abdominal trauma, surgical operations, or other invasive abdominal procedures within the past 30 days; (2) patients returning for follow-up, transferred from outpatient departments, or previously diagnosed by other hospitals; and (3) confirmed pregnant patients. The sample size was estimated based on binary logistic regression analysis requirements, suggesting 5 to 10 times the number of explanatory variables. This study involved 26 explanatory variables; thus, considering a 20% dropout rate, a sample size of 163–325 patients was required. Ultimately, 769 cases were included. Ethical approval was obtained from the Tongren Hospital Non-interventional Research Ethics Committee (approval no. 2024-001-01). Written and verbal informed consent was obtained from all participants. The study was conducted according to relevant guidelines and regulations.
Investigative tools
Definition of triage results
Currently, no gold standard exists for assessing triage accuracy. A systematic review by Lentz et al. [13] revealed that 76% of under-triage definitions rely on expert opinions, whereas 24% utilize patient outcomes and other clinical indicators. Triage accuracy is closely associated with clinical outcomes; thus, evaluating triage based on patient prognosis should be the preferred method. Through literature review [14–17] and expert consultation, under-triage in this study was defined based on patient prognosis, as follows: (1) patients initially triaged at level IV (non-urgent) who required medical intervention within 24 h and could not be discharged directly from the ED. “Cannot be discharged” included patients under emergency observation (≥ 24 h), requiring admission, emergency interventions or surgery, death, or those leaving against medical advice (abnormal discharge); (2) patients initially triaged at level III or IV who subsequently required transfer to a resuscitation room or intensive care unit. Any patient meeting these conditions was classified as under-triaged. Over-triage referred to patients initially assigned level I (resuscitation) or level II (emergent) but who left the ED without medical intervention within 24 h.
Survey on factors influencing under-triage in patients with AA
A literature search was conducted to identify factors associated with under-triage in patients experiencing AAP. The research team used English and Chinese keywords, including “influencing factor/risk factor/factor” and “under-triage/triage error,” to search databases such as PubMed, Cochrane Library, Web of Science, Embase, CINAHL, CNKI, Wanfang Database, VIP Database, and China Biomedical Literature Database.
Initially, 522 records were retrieved. After deduplication using NoteExpress 3.2, 491 records remained. Following title and abstract screening and subsequent full-text review, 13 studies were included: one guideline, one evidence summary, and eleven original research articles. Data extraction from the included literature resulted in 26 potential influencing factors, categorized as follows: Patient characteristics (7 items): age, sex, mode of admission, underlying diseases, history of abdominal surgery, past medical history, and Modified Early Warning Score (MEWS). Environmental characteristics (8 items): triage department, triage level, duration of triage, number of ED visits within one hour before the index triage, number of ED visits within 30 min after the index triage, number of patients triaged as level I or II within two hours before the index triage, peak ED visit period (18:00–22:00), and time period of triage. Nurse characteristics (6 items): age, sex, total years of service, years of ED experience, educational level, and professional title. Abdominal pain characteristics (7 items): precipitating factors, frequency, location, accompanying symptoms, pain quality, duration, and Numeric Rating Scale (NRS) pain score. Based on these factors, the research team developed a draft questionnaire titled “Survey on Influencing Factors of Emergency Triage for Patients with NAAP.” The draft was reviewed, revised, and supplemented by emergency physicians, emergency nursing experts, and triage nurses. The final questionnaire provided comprehensive coverage of relevant variables, ensuring a robust foundation for accurately analyzing under-triage factors in this patient population.
Clinical data collection methods
The research team leader contacted department heads at the three selected hospitals and obtained their consent. Two nursing graduate students, who received standardized training, conducted data collection using the electronic pre-triage system and electronic medical record system. Basic information of patients presenting to the ED with the chief complaint of “abdominal pain” from December 2023 to May 2024 (including emergency ID number, name, age, visit time, and gender) was imported directly into Excel via the electronic pre-triage system. Subsequently, the two graduate students used patients’ emergency ID numbers to extract clinical data from electronic medical records, performing validity and completeness checks. Data entry was independently conducted by two individuals, followed by cross-verification.
Statistical methods
Statistical analyses were conducted using SPSS version 26.0. Continuous variables with non-normal distributions were expressed as medians and interquartile ranges (IQR) and compared using the Mann-Whitney U test. Categorical variables were presented as frequencies and percentages, and between-group comparisons were performed using the Chi-square (χ²) test or Fisher’s exact test as appropriate. To identify independent predictors of under-triage, a backward stepwise binary logistic regression analysis was performed with variables that showed statistical significance in univariate analysis. Multicollinearity among candidate predictors was assessed prior to modeling, confirming that all Variance Inflation Factors (VIF) were below 5. The association between under-triage and EDLOS was first evaluated using the Mann-Whitney U test. Subsequently, a multivariate linear regression model was constructed to quantify this relationship after controlling for the initial triage level. All statistical tests were two-tailed, and a P-value < 0.05 was considered statistically significant.
Results
Current triage status and univariate analysis of under-triage in NAAP patients
A total of 769 adult patients presenting with NAAP were retrospectively enrolled from three tertiary hospitals. Accurate triage occurred in 556 cases (72.30%), whereas under-triage occurred in 126 cases (16.38%) and over-triage in 87 cases (11.31%). To identify risk factors specifically associated with under-triage and avoid selection bias, patients were divided into two groups for analysis: the under-triage group and the non-under-triage group (comprising accurately triaged and over-triaged patients). Comparative analyses were conducted regarding patient characteristics, environmental characteristics, nurse characteristics, and abdominal pain characteristics.
Results revealed significant differences in patient characteristics, including age, sex, mode of admission, underlying diseases, and history of hyperlipidemia (P < 0.05). Regarding environmental characteristics, differences were found in triage levels, number of ED visits within 1 h before the index triage, number of ED visits within 30 min after the index triage, and number of patients triaged as level I or II within 2 h before the index triage (P < 0.05). For nurse characteristics, significant differences were identified in age, total years of service, ED experience, sex, and professional titles (P < 0.05). Regarding abdominal pain characteristics, significant differences were observed in frequency, vomiting, decreased bowel movements or flatus, back pain, scleral icterus, and pain quality (P < 0.05). Detailed results are presented in Table 1.
Table 1.
Univariate analysis of under-triage in patients with NAAP (N = 769)
| Project | Non-under-triage (n = 643) |
Under-triage (n = 126) |
Statistic | P |
|---|---|---|---|---|
| Patient Characteristics | ||||
| Age | 46.00 (33.00, 65.50) | 60.00 (39.00, 71.00) | -3.72a) | < 0.001 |
| Sex | 4.91b) | 0.027 | ||
| Male | 293 (80.49) | 71 (19.51) | ||
| Female | 350 (86.42) | 55 (13.58) | ||
| Mode of admission | 7.33b) | 0.007 | ||
| Non-120 | 473 (81.55) | 107 (18.45) | ||
| 120 | 170 (89.95) | 19 (10.05) | ||
| Presence of underlying diseases | 9.30b) | 0.002 | ||
| No | 433 (86.60) | 67 (13.40) | ||
| Yes | 210 (78.07) | 59 (21.93) | ||
| History of abdominal surgery | 0.03b) | 0.866 | ||
| No | 506 (83.50) | 100 (16.50) | ||
| Yes | 137 (84.05) | 26 (15.95) | ||
| Past medical history | ||||
| Hypertension | 115 (83.94) | 22 (16.06) | 0.01b) | 0.909 |
| Diabetes | 50 (78.12) | 14 (21.88) | 1.54b) | 0.215 |
| Heart Disease | 36 (78.26) | 10 (21.74) | 1.02b) | 0.312 |
| Cancer | 57 (76.00) | 18 (24.00) | 3.52b) | 0.061 |
| Cerebrovascular Disease | 14 (73.68) | 5 (26.32) | 0.76b) | 0.384 |
| Hyperlipidemia | 5 (50.00) | 5 (50.00) | 6.06b) | 0.014 |
| MEWS | - | 0.125 | ||
| 0~4 | 639 (83.75) | 124 (16.25) | ||
| 5~8 | 4 (80.00) | 1 (20.00) | ||
| ≥ 9 | 0 (0.00) | 1 (100.00) | ||
| Environmental Characteristics | ||||
| Triage Department | 3.05b) | 0.218 | ||
| Internal Medicine | 200 (86.96) | 30 (13.04) | ||
| Surgery | 401 (82.51) | 85 (17.49) | ||
| Gynecology | 42 (79.25) | 11 (20.75) | ||
| Triage Level | 86.69b) | < 0.001 | ||
| Level III | 194 (68.31) | 90 (31.69) | ||
| Level IV | 288 (88.89) | 36 (11.11) | ||
| Number of ED visits within 1 h before the index triage | 39.00 (26.00, 53.00) | 49.50 (28.25, 71.75) | -3.40a) | < 0.001 |
| Number of ED visits within 30 min after the index triage | 20.00 (12.00, 28.50) | 23.00 (14.25, 34.00) | -2.72a) | 0.006 |
| Number of patients triaged as level I or II within 2 h before the index triage | 5.00 (2.00, 8.00) | 6.00 (4.00, 9.00) | -3.03a) | 0.002 |
| Time required to complete triage | 49.00 (32.00, 87.00) | 48.00 (30.00, 86.25) | -0.62a) | 0.533 |
| ED peak-visit period (18:00–22:00) | 133 (87.50) | 19 (12.50) | 2.09b) | 0.149 |
| Triage Time Period | 3.25b) | 0.196 | ||
| 8:01–16:00 | 236 (80.55) | 57 (19.45) | ||
| 16:01 − 00:00 | 289 (85.50) | 49 (14.50) | ||
| 00:01–08:00 | 118 (85.51) | 20 (14.49) | ||
| Nurse Characteristics | ||||
| Age | 32.00 (29.00, 35.00) | 29.00 (24.00, 32.00) | -5.64a) | < 0.001 |
| Total years of service | 49.02b) | < 0.001 | ||
| ≤ 4 | 173 (70.04) | 74 (29.96) | ||
| 5–9 | 178 (89.45) | 21 (10.55) | ||
| ≥ 10 | 292 (90.40) | 31 (9.60) | ||
| Years of ED experience | 8.00 (1.00, 11.00) | 1.00 (1.00, 8.00) | -6.09a) | < 0.001 |
| Sex | 24.29b) | < 0.001 | ||
| Male | 174 (73.73) | 62 (26.27) | ||
| Female | 469 (87.99) | 64 (12.01) | ||
| Education level | 2.84b) | 0.241 | ||
| Vocational School | 9 (100.00) | 0 (0.00) | ||
| Junior College | 149 (80.98) | 35 (19.02) | ||
| Bachelor’s Degree | 485 (84.20) | 91 (15.80) | ||
| Professional Title | 49.86b) | < 0.001 | ||
| Nurse | 172 (69.92) | 74 (30.08) | ||
| Senior Nurse | 313 (90.72) | 32 (9.28) | ||
| Charge Nurse | 158 (88.76) | 20 (11.24) | ||
| Abdominal Pain Characteristics | ||||
| Precipitating factors | 0.45b) | 0.504 | ||
| No | 563 (83.28) | 113 (16.72) | ||
| Yes | 80 (86.02) | 13 (13.98) | ||
| Frequency | 24.20b) | < 0.001 | ||
| Paroxysmal | 295 (91.33) | 28 (8.67) | ||
| Persistent | 348 (78.03) | 98 (21.97) | ||
| Location | 13.62b) | 0.058 | ||
| Left Upper Abdomen | 34 (70.83) | 14 (29.17) | ||
| Middle Abdomen | 57 (83.82) | 11 (16.18) | ||
| Left Lower Abdomen | 72 (87.80) | 10 (12.20) | ||
| Upper Middle Abdomen | 190 (84.82) | 34 (15.18) | ||
| Lower Middle Abdomen | 34 (87.18) | 5 (12.82) | ||
| Right Upper Abdomen | 108 (83.08) | 22 (16.92) | ||
| Right Lower Abdomen | 123 (86.62) | 19 (13.38) | ||
| Whole Abdomen | 25 (69.44) | 11 (30.56) | ||
| Accompanying Symptoms | ||||
| Nausea | 87 (82.08) | 19 (17.92) | 0.21b) | 0.645 |
| Vomiting | 139 (78.53) | 38 (21.47) | 4.34b) | 0.037 |
| Diarrhea | 70 (87.50) | 10 (12.50) | 0.98b) | 0.321 |
| Decreased bowel movements or flatus | 12 (57.14) | 9 (42.86) | 9.15b) | 0.002 |
| Lower Back Pain | 33 (82.50) | 7 (17.50) | 0.04b) | 0.845 |
| Fever | 34 (82.93) | 7 (17.07) | 0.01b) | 0.903 |
| Abdominal Distension | 32 (86.49) | 5 (13.51) | 0.23b) | 0.629 |
| Back Pain | 13 (59.09) | 9 (40.91) | 8.18b) | 0.004 |
| Bloody Stool | 9 (69.23) | 4 (30.77) | 1.07b) | 0.301 |
| Scleral Icterus | 3 (27.27) | 8 (72.73) | 21.85b) | < 0.001 |
| Nature | 41.99b) | < 0.001 | ||
| Colicky Pain | 150 (86.71) | 23 (13.29) | ||
| Dull Pain | 134 (90.54) | 14 (9.46) | ||
| Distending Pain | 173 (91.53) | 16 (8.47) | ||
| Stabbing Pain | 77 (74.04) | 27 (25.96) | ||
| Migratory | 109 (70.32) | 46 (29.68) | ||
| Pain Duration (hours) | 10.71b) | 0.058 | ||
| < 6 | 170 (88.54) | 22 (11.46) | ||
| 6~12 | 95 (77.87) | 27 (22.13) | ||
| 13~24 | 195 (85.53) | 33 (14.47) | ||
| 25~48 | 68 (78.16) | 19 (21.84) | ||
| 49~96 | 75 (85.23) | 13 (14.77) | ||
| 97~168 | 40 (76.92) | 12 (23.08) | ||
| NRS | 3.87b) | 0.145 | ||
| 1~3 | 318 (85.95) | 52 (14.05) | ||
| 4~6 | 173 (79.72) | 44 (20.28) | ||
| 7~10 | 152 (83.52) | 30 (16.48) | ||
Note: a): Mann-Whitney U test; b): χ² test; -: Fisher’s exact test
Multivariate analysis of under-triage in NAAP
A backward stepwise binary logistic regression analysis was conducted to identify independent predictors of under-triage (Table 2). Risk factors significantly associated with under-triage included advanced age (OR = 2.45), history of hyperlipidemia (OR = 7.94), persistent pain frequency (OR = 4.12), and migratory pain (OR = 2.82), as well as accompanying symptoms such as vomiting, decreased bowel movements, back pain, and scleral icterus (all P < 0.05). Conversely, protective factors against under-triage included higher nurse professional titles (Senior Nurse/Charge Nurse), admission via ambulance, and dull pain characteristics (all P < 0.05).
Table 2.
Binary logistic regression analysis of under-triage among patients with NAAP
| Variable | B | S.E. | Wald χ² | P | OR (95% CI) |
|---|---|---|---|---|---|
|
Patient Age (≥ 60 vs. < 60) |
0.895 | 0.259 | 11.919 | < 0.001 | 2.447 (1.472–4.066) |
|
Mode of Admission (120 vs. Non-120) |
-2.274 | 0.396 | 33.042 | < 0.001 | 0.103 (0.047–0.223) |
|
Hyperlipidemia (Yes vs. No) |
2.072 | 0.865 | 5.742 | 0.017 | 7.944 (1.458–43.269) |
|
Nurse Professional Title (Ref: Nurse) |
15.185 | < 0.001 | |||
| Senior Nurse | -1.221 | 0.331 | 13.613 | < 0.001 | 0.295 (0.154–0.564) |
| Charge Nurse | -1.169 | 0.384 | 9.271 | 0.002 | 0.311 (0.146–0.659) |
|
Triage Level (Level III vs. IV) |
0.816 | 0.272 | 9.013 | 0.003 | 2.261 (1.327–3.850) |
|
No. of I-II patients (> 5 vs. ≤ 5) |
0.476 | 0.264 | 3.251 | 0.071 | 1.610 (0.959–2.701) |
|
Nature of Pain (Ref: Colicky) |
31.130 | < 0.001 | |||
| Dull Pain | -0.964 | 0.433 | 4.942 | 0.026 | 0.381 (0.163–0.892) |
| Distending Pain | -0.689 | 0.400 | 2.966 | 0.085 | 0.502 (0.229–1.100) |
| Stabbing Pain | 0.135 | 0.405 | 0.110 | 0.740 | 1.144 (0.517–2.532) |
| Migratory Pain | 1.035 | 0.354 | 8.556 | 0.003 | 2.815 (1.407–5.632) |
| Accompanying Symptoms | |||||
| Vomiting | 0.572 | 0.274 | 4.362 | 0.037 | 1.771 (1.036–3.029) |
| Decreased bowel movements | 1.299 | 0.600 | 4.686 | 0.030 | 3.666 (1.131–11.888) |
| Back Pain | 1.195 | 0.543 | 4.841 | 0.028 | 3.303 (1.139–9.573) |
| Scleral Icterus | 1.746 | 0.826 | 4.466 | 0.035 | 5.731 (1.135–28.931) |
|
Frequency (Persistent vs. Paroxysmal) |
1.415 | 0.282 | 25.189 | < 0.001 | 4.116 (2.369–7.152) |
| Constant | -2.731 | 0.481 | 32.169 | < 0.001 | 0.065 |
Correlation between Under-triage and EDLOS
To evaluate the impact of under-triage on ED process efficiency and avoid circular reasoning, patients who met outcome-based under-triage criteria (e.g., hospitalization, ICU admission, death, or EDLOS > 24 h) were excluded from this analysis. Stratified analyses based on initial triage levels revealed a significant association between under-triage and prolonged EDLOS (Table 3). For Level III patients, the median EDLOS in the under-triage group was more than three times longer than in the accurately triaged group (295.00 min vs. 91.50 min, Z = -6.106, P < 0.001). Similarly, under-triaged Level IV patients experienced significantly longer EDLOS compared with accurately triaged patients (171.00 min vs. 82.00 min, Z = -2.270, P = 0.020). A multiple linear regression was subsequently performed to adjust for confounding from initial triage level (Table 4). The regression model was statistically significant (F = 56.670, P < 0.001, R² = 0.198). Results indicated that under-triage remained an independent predictor of prolonged EDLOS (B = 309.765, P < 0.001).
Table 3.
Comparison of EDLOS between accurate triage and under-triage groups stratified by initial triage level
| Triage Level | Group | n | EDLOS (min), M (Q1, Q3) | Z | P |
|---|---|---|---|---|---|
| Level III | Accurate Triage | 159 | 91.50 (64.00, 162.50) | -6.106 | < 0.001 |
| Under-triage | 29 | 295.00 (184.00, 700.00) | |||
| Level IV | Accurate Triage | 269 | 82.00 (64.00, 130.00) | -2.270 | 0.020 |
| Under-triage | 5 | 171.00 (107.50, 265.50) |
Table 4.
Multiple linear regression analysis of factors associated with EDLOS
| Model | Unstandardized Coefficient | Standardized Coefficient | t | P | VIF | ||
|---|---|---|---|---|---|---|---|
| B | Standard Error | Beta | |||||
| (Constant) | 344.300 | 67.457 | 5.104 | < 0.001 | |||
| Independent Variable | Under-triage | 309.765 | 34.659 | 0.386 | 8.938 | < 0.001 | 1.070 |
| Control Variable | Triage Level | -60.881 | 18.421 | -0.143 | -3.305 | 0.001 | 1.070 |
| R2 | 0.198 | ||||||
| F | 56.670 | ||||||
| P | < 0.001 | ||||||
Dependent Variable: Emergency Department Length of Stay
Discussion
Under-triage status of NAAP patients
In this multicenter cohort, the under-triage rate for NAAP was 16.38%, exceeding the widely accepted safety benchmark of < 5% established in quality evaluation frameworks [18]. Although this rate is lower than the 31.0% reported by studies employing the Korean Triage and Acuity Scale (KTAS) [12], the discrepancy likely reflects structural differences in algorithm design rather than clinical variations alone. Our study hospitals utilized a four-level consensus protocol. In contrast, international standards such as the Emergency Severity Index (ESI), Manchester Triage System (MTS), and Canadian Triage and Acuity Scale (CTAS) apply five-level algorithms incorporating specific resource predictions or discriminators. Despite these systemic differences, NAAP remains a universal triage challenge. Unlike trauma or cardiac arrest, severe intra-abdominal conditions, such as early-stage appendicitis or intestinal obstruction, often present with stable vital signs. This stability creates a discrepancy between physiological parameters and actual acuity [19]. Clinicians may mistakenly underestimate urgency based solely on the patient’s initial general appearance.
Critically, our stratified analysis identified Triage Level III as the primary vulnerability, demonstrating the highest under-triage rate (31.7%). This finding aligns with the “grey zone” phenomenon described in ESI and CTAS literature [20, 21]. Level III represents patients who appear stable yet require multiple resources or have conditions prone to deterioration [15]. For NAAP patients, Level III individuals lack the overt hemodynamic instability of Levels I/II but present significantly higher acuity than minor ailments typical of Level IV [22]. The elevated error rate in this subgroup indicates inadequate sensitivity of current protocols in identifying potential deterioration within intermediate-acuity cases. Moreover, as demonstrated by our multivariate analysis, ED crowding exacerbates this issue by increasing cognitive load. Nurses under stress may default to a “standard” Level III categorization without adequately assessing subtle high-risk signs [23].
These observations highlight the need for targeted procedural safeguards rather than general training alone. Immediate changes in practice should mandate timely re-evaluation of Level III abdominal pain patients to detect early deterioration. Additionally, integrating objective “red flag” prompts into triage workflows may help address misleadingly normal vital signs. While emerging technologies, such as AI-based decision support systems, represent promising future research directions to reduce human variability, the current clinical priority should be refining manual protocols to more effectively identify high-risk patients within the “grey zone” of Level III.
Multiple factors affecting Under-triage in NAAP patients
The results of this study indicate that advanced age (≥ 60 years) was a robust independent risk factor (OR = 2.447, P < 0.001), consistent with previous findings [12, 15, 19]. Elderly patients with NAAP commonly present with atypical symptoms due to underlying vascular disease and other complications. Additionally, elderly patients often delay medical consultation due to reduced independence or anxiety regarding hospitalization or mortality, leading to increased complexity and severity [24, 25]. Related studies have shown [26] that elderly NAAP patients are triaged as Level II (critical) approximately half as frequently as patients without abdominal pain. However, they are 4.5 times more likely to require emergency surgery, reflecting a systematic underestimation of severity in this population.
A particularly notable finding was the significant association between a history of hyperlipidemia and under-triage, exhibiting the highest odds ratio among patient history variables (OR = 7.944, P = 0.017). Hyperlipidemia serves as a sentinel marker for metabolic syndrome [27], intrinsically linked to severe abdominal pathologies with high mortality risks. Hypertriglyceridemia is a well-established precipitant of acute pancreatitis, often presenting with vague epigastric pain, masking severity until rapid deterioration occurs [28]. Additionally, dyslipidemia is a key risk factor for chronic mesenteric ischemia, characterized by “pain out of proportion to exam,” yet initially normal vital signs [29]. The substantial magnitude of this association indicates that triage protocols should prioritize patients with metabolic comorbidities, treating hyperlipidemia as an indicator of vascular or pancreatic risks despite stable initial physiological parameters.
The regression model demonstrated that persistent abdominal pain had the strongest association among pain characteristics (OR = 4.116, P < 0.001). Persistent pain likely signals ongoing organ dysfunction that transcends subjective pain scores, indicating conditions such as bowel obstruction or peritonitis. Objective predictors, such as scleral icterus (OR = 5.731, P = 0.035) and decreased bowel movements or flatus (OR = 3.666, P = 0.030), also emerged as significant [27]. These symptoms reflect specific organ dysfunctions, such as biliary obstruction or paralytic ileus, that may not immediately alter vital signs but imply high acuity. Similarly, migratory pain (OR = 2.815, P = 0.003) and back pain (OR = 3.303, P = 0.028) were significant risk factors. Migratory pain is pathognomonic of appendicitis, while back pain can indicate retroperitoneal conditions such as pancreatitis or aortic dissection. The strong predictive value of these symptoms emphasizes the necessity of incorporating objective, symptom-based discriminators into triage algorithms, counterbalancing the misleading appearance of normal vital signs.
Conversely, higher nursing professional titles were protective factors, reflecting experienced nurses’ capacity to recognize subtle deterioration signs that novice nurses might overlook. Senior nurses can better integrate nonspecific cues with objective parameters, achieving a gestalt assessment of patient acuity [30]. Although ED crowding (number of Level I-II patients > 5 within 2 h) showed a trend toward increased under-triage (OR = 1.610), this did not reach statistical significance in the adjusted model (P = 0.071). Thus, environmental stressors might contribute to triage errors, but experienced staff may partly mitigate this impact. Additionally, admission via ambulance (120 emergency system) was protective (OR = 0.103, P < 0.001), likely due to pre-hospital alerts raising nurses’ awareness of higher acuity. Pain quality also modulated risk: compared to colicky pain, dull pain was protective (OR = 0.381, P = 0.026), likely representing lower-acuity conditions like gastritis. In contrast, migratory pain significantly increased risk, as previously noted.
These findings underscore the need for multilevel interventions. At the protocol level, triage systems should mandate re-evaluation for elderly patients and those with metabolic comorbidities (particularly hyperlipidemia), even when initial vital signs appear stable. Objective symptom-based “red flags,” including persistent pain, scleral icterus, decreased bowel movements, migratory pain, or back pain, should be explicit discriminators within triage algorithms. At the organizational level, ensuring adequate senior nursing staff during high-volume periods may help offset the cognitive load associated with ED crowding. Finally, competency-based training emphasizing illness script development and simulation-based exercises to recognize subtle clinical deterioration in metabolically complex patients may help reduce disparities arising from limited nursing experience.
The impact of Under-triage on EDLOS
Multiple linear regression analysis indicated that under-triage was an independent predictor of prolonged EDLOS, extending it by approximately 310 min (B = 309.765, P < 0.001). This finding is consistent with previous literature [31]. Accurate assessment by healthcare personnel during the pre-triage process is crucial. Proper triage optimizes resource allocation, enhances emergency service efficiency, and ensures patient safety.
Several management tools and methods could improve triage accuracy and efficiency. Firstly, the Six Sigma management method standardizes the triage process by reducing variability and minimizing human error [32]. Secondly, structured clinical pathways offer clear triage guidance, ensuring timely and appropriate treatment [33]. Additionally, comprehensive assessment tools enable healthcare personnel to thoroughly evaluate patient conditions, supporting more accurate triage decisions [34]. Furthermore, regular professional training and education are essential to enhance healthcare providers’ knowledge and skills in recognizing the severity of AAP.
In conclusion, implementing targeted management strategies and tools can significantly improve triage accuracy and efficiency, reducing EDLOS and enhancing patient experiences and overall emergency service quality. Future research should explore the effectiveness of various triage tools and strategies across diverse patient populations to generate further evidence-based improvements in emergency triage management.
Limitations
Several limitations of this study should be noted. First, the retrospective design introduces potential information bias and missing data. Subjective indicators, such as pain characteristics, heavily depend on the accuracy of initial documentation by triage nurses, potentially introducing variability. Second, the study sample originated exclusively from three tertiary hospitals in the Yangtze River Delta region, limiting generalizability to other geographic areas, rural settings, or primary healthcare facilities with different resources or triage protocols. Third, although 26 potential covariates were adjusted in multivariate analysis, residual confounding from unmeasured variables remains possible. Future prospective studies with broader geographic coverage and mixed-method designs are needed to address these limitations.
Conclusion
This multicenter retrospective study identified an under-triage incidence of 16.38% among patients with NAAP. Multivariate regression analysis demonstrated that advanced age, hyperlipidemia history, and specific abdominal pain characteristics, including persistent pain, migratory pain, and scleral icterus, were significant independent predictors of under-triage. Conversely, higher nursing seniority served as a protective factor. Critically, under-triage independently prolonged EDLOS by approximately 310 min, underscoring its substantial impact on patient safety and operational efficiency. These findings emphasize the urgent need for competency-based training, targeted risk-stratification protocols incorporating metabolic and symptom-specific “red flags,” and strategic staffing with senior nurses during peak periods. These measures will help optimize triage accuracy and reduce preventable delays in emergency care delivery.
Acknowledgements
The successful completion of this study was facilitated by the invaluable support and collaborative efforts of all medical staff at Shanghai Tongren Hospital, the Main Campus of The First Affiliated Hospital of Suzhou University, and the Shanghai Renji Hospital. We extend our heartfelt gratitude to these institutions and their dedicated professionals for their essential contributions.
Author contributions
WXH, LR, ZLJ, and LXQ conceived the research question and study design. WXH, FY, and ADN collected and validated the data. WXH performed data analysis and prepared all figures, drafted the initial manuscript and prepared tables. ZLJ, LXQ and LR critically revised the manuscript. All authors read and approved the final manuscript.
Funding
This work was supported by the Key Research Project of the Shanghai Nursing Association, under project number 2023SD-B01, and by the Changning District Health Commission of Shanghai, under project number CNKW2024Y15.
Data availability
The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.
Declarations
Ethics approval and consent
Before the research commenced, ethics approval was obtained from the Tongren Hospital Non-interventional Research Ethics Committee (approval no. 2024-001-01). All participants provided written and verbal informed consent to participate in the study. The study was carried out in accordance with the applicable guidelines and regulations.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rui Li, Li-Jin Zhou and Xiao-Qin Li contributed equally to this work.
Contributor Information
Xiao-Hui Wei, Email: weixh0108@163.com.
Xiao-Qin Li, Email: 13812792585@163.com.
References
- 1.Mayumi T, Yoshida M. The practice guidelines for primary care of acute abdomen 2015 [J]. Jpn J Radiol. 2016;34(1):80–115. [DOI] [PubMed] [Google Scholar]
- 2.Chanana L, Jegaraj M A, Kalyaniwala K, et al. Clinical profile of non-traumatic acute abdominal pain presenting to an adult emergency department [J]. J Family Med Prim Care. 2015;4(3):422–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Agboola J O, Olatoke S A, Rahman GA. Pattern and presentation of acute abdomen in a Nigerian teaching hospital [J]. Niger Med J. 2014;55(3):266–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Minten L, Messiaen P, Van Der Hilst J. Acute abdominal pain: a challenging diagnosis [J]. Acta Gastroenterol Belg. 2022;85(4):646–7. [DOI] [PubMed] [Google Scholar]
- 5.Expert Consensus Group on emergency pretest triage. Expert consensus on emergency predetection triage [J]. Chin J Emerg Med. 2018;27(6):599–604. [Google Scholar]
- 6.Di Somma S, Paladino L, Vaughan L, et al. Overcrowding in emergency department: an international issue [J]. Intern Emerg Med. 2015;10(2):171–5. [DOI] [PubMed] [Google Scholar]
- 7.Hitchcock M, Gillespie B, Crilly J, et al. Triage: an investigation of the process and potential vulnerabilities [J]. J Adv Nurs. 2014;70(7):1532–41. [DOI] [PubMed] [Google Scholar]
- 8.Yurkova I. Under-triage as a significant factor affecting transfer time between the emergency department and the intensive care unit [J]. J Emerg Nurs. 2011;37(5):491–6. [DOI] [PubMed] [Google Scholar]
- 9.Banz V M, De Moya Paulk. Ignoring non-specific abdominal pain in emergency department patients May be related to decreased quality of life. A follow up of an underestimated problem [J]. Swiss Med Wkly. 2011;141:w13167. [DOI] [PubMed] [Google Scholar]
- 10.Choi Y, Lee H. Factors related to under-triage of patients with acute coronary syndrome in the emergency department: A retrospective study [J]. Int Emerg Nurs. 2023;69:101316. [DOI] [PubMed] [Google Scholar]
- 11.Huabbangyang T, Rojsaengroeng R, Tiyawat G, et al. Associated factors of under and over-triage based on the emergency severity index; a retrospective cross-sectional study [J]. Arch Acad Emerg Med. 2023;11(1):e57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Oh B Y Kim k. Factors associated with the undertriage of patients with abdominal pain in an emergency room [J]. Int Emerg Nurs. 2021;54:100933. [DOI] [PubMed] [Google Scholar]
- 13.Lentz B A Jensona, Hinson Js, et al. Validity of ED: addressing heterogeneous definitions of over-triage and under-triage [J]. Am J Emerg Med. 2017;35(7):1023–5. [DOI] [PubMed] [Google Scholar]
- 14.Gans S L, Pols M A, Stoker J, et al. Guideline for the diagnostic pathway in patients with acute abdominal pain [J]. Dig Surg. 2015;32(1):23–31. [DOI] [PubMed] [Google Scholar]
- 15.Hinson Js, Martinez D A, Schmitz P S K, et al. Accuracy of emergency department triage using the emergency severity index and independent predictors of under-triage and over-triage in brazil: a retrospective cohort analysis [J]. Int J Emerg Med. 2018;11(1):3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Sunyoto T Fau - Van Den Bergh R, Van Den Bergh R Fau - Valles P, Valles P Fau - Gutierrez R, et al. Providing emergency care and assessing a patient triage system in a referral hospital in somaliland: a cross-sectional study [J]. BMC Health Serv Res. 2014;14:531. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Jin Jingfen C, Yuguo Z, Huadong, et al. Triage criteria for emergency preexamination (adult part). Chin J Acute Crit Care Nurs. 2020;1(01):45–8. [Google Scholar]
- 18.Liu Wei H, Qingbo D. Construction of quality evaluation criteria for pre-detection triage system in emergency department [J]. J Nurs. 2016;31(13):45–7. [Google Scholar]
- 19.Grossmann F F, Zumbrunn T. Undertriage in older emergency department patients–tilting against windmills? [J]. PLoS ONE. 2014;9(8):e106203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Wandl B, Kellerer J D, Fuhrmann V, et al. Diagnostic test accuracy of the emergency severity index: a systematic review and meta-analysis [J]. Eur J Emerg Med. 2025;32(5):325–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hinson Js, Martinez D A, Cabral S, et al. Triage performance in emergency medicine: A systematic review [J]. Ann Emerg Med. 2019;74(1):140–52. [DOI] [PubMed] [Google Scholar]
- 22.Savioli G, Ceresa I F, Bressan M A, et al. Five level triage vs. Four level triage in a quaternary emergency department: National analysis on waiting Time, Validity, and Crowding-The CREONTE (Crowding and RE-Organization National TriagE) study group [J]. Med (Kaunas), 2023, 59(4). [DOI] [PMC free article] [PubMed]
- 23.Cremonesi P, Di Bella E, Montefiori M, et al. The robustness and effectiveness of the triage system at times of overcrowding and the extra costs due to inappropriate use of emergency departments [J]. Appl Health Econ Health Policy. 2015;13(5):507–14. [DOI] [PubMed] [Google Scholar]
- 24.BISCHOFF A. Geriatric emergencies. Often the status is more serious than it seems. MMW Fortschr Der Medizin. 2013;155(20):18–20. J. [DOI] [PubMed] [Google Scholar]
- 25.Lyon C, Clark Dc. Diagnosis of acute abdominal pain in older patients [J]. Am Fam Physician. 2006;74(9):1537–44. [PubMed] [Google Scholar]
- 26.Friedman A B, Chen A T, Wu R, et al. Evaluation and disposition of older adults presenting to the emergency department with abdominal pain [J]. J Am Geriatr Soc. 2022;70(2):501–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Wang Y, Zhao H, Zhou Z, et al. Establishment and application of early risk stratification method for acute abdominal pain in adults [J]. Chin Med J (Engl). 2017;130(5):530–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Hajibandeh S, Jurdon R. The risk of recurrent pancreatitis after first episode of acute pancreatitis in relation to etiology and severity of disease: A systematic review, meta-analysis and meta-regression analysis [J]. J Gastroenterol Hepatol. 2023;38(10):1718–33. [DOI] [PubMed] [Google Scholar]
- 29.Ma T, Zhao H, Zhang Q, et al. Mesenteric vein thrombosis following sleeve gastrectomy: A case report and review of the literature [J]. Obes Facts. 2024;17(2):211–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Chen Ss, Chen J C, Ng C J, et al. Factors that influence the accuracy of triage nurses’ judgement in emergency departments [J]. Emerg Med J. 2010;27(6):451–5. [DOI] [PubMed] [Google Scholar]
- 31.Ausserhofer D, Zaboli A, Pfeifer N, et al. Errors in nurse-led triage: an observational study [J]. Int J Nurs Stud. 2021; 113(1873-491X (Electronic)):103788. [DOI] [PubMed]
- 32.Zan HY, Xie XP, Tan XM, et al. Application of six Sigma management in reducing the error rate of triage in emergency abdomen [J]. Chin J Practical Nurs. 2020;36(10):757–60. [Google Scholar]
- 33.Guo Shasha H, Yongchao F, Jingjing, et al. Evidence-based practice of pre-examination triage management for patients with non-traumatic acute abdomen [J]. Chin J Practical Nurs. 2022;38(7):492–500. [Google Scholar]
- 34.Zhao Ting Z, Min Z, Jiandong. Application of clinical pathway in the treatment of acute abdomen [J]. Int J Nurs. 2012;31(6):1060–2. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.