The global demographic shift toward an aging population has resulted in a steady increase in the proportion of elderly patients presenting with acute surgical conditions requiring emergency intervention. Patients aged 60 years and above currently represent over 25% of all emergency general surgical admissions in tertiary care centres across India, a figure projected to rise further with increasing life expectancy. Emergency general surgery in the elderly is a particularly high-risk clinical scenario owing to the convergence of acute physiological stress, diminished functional reserves, polypharmacy, and the frequent coexistence of multiple chronic comorbidities.

Postoperative intensive care unit (ICU) admission is frequently required following emergency abdominal surgery in older adults, with reported rates ranging from 30% to over 60% depending on the case mix and institutional practice. ICU admission in this population is associated with prolonged hospital stay, higher complication rates, increased healthcare expenditure, and elevated mortality. In resource-limited settings, where ICU beds are often at a premium, the ability to predict which patients will require intensive postoperative monitoring is of considerable practical importance for surgical planning, bed management, and informed consent discussions with patients and families.

Several factors have been implicated in the need for postoperative ICU care, including advanced age, poor preoperative physiological status as reflected by scoring systems such as the ASA physical status classification and APACHE II score, nutritional depletion indicated by low serum albumin levels, the presence of sepsis or hemodynamic instability at admission, and intraoperative variables such as prolonged operative duration and significant blood loss. However, the relative importance of these factors and their combined predictive utility in the Indian elderly population undergoing emergency general surgery remain insufficiently characterised.

The present study was therefore designed to identify the preoperative and intraoperative predictors of ICU admission in patients aged 60 years and older undergoing emergency general surgery at a tertiary teaching hospital in Bengaluru, South India. By elucidating these predictors, the study aims to contribute to the development of evidence-based triage strategies that can optimise resource utilisation and improve outcomes in this vulnerable patient cohort.

Patient Characteristics

A total of 68 patients meeting the inclusion criteria were enrolled during the study period. The mean age of the cohort was 69.1 ± 6.8 years (range 60–89 years), and 40 patients (58.8%) were male. Thirty-one patients (45.6%) required postoperative ICU admission, while 37 (54.4%) were managed in the general surgical ward. Patients in the ICU group were significantly older (72.4 ± 6.8 versus 66.3 ± 5.2 years; p=0.001), and a higher proportion were aged 70 years or above (67.7% versus 35.1%; p=0.008). No significant difference was observed in sex distribution or body mass index between the two groups.

Comorbidities and Preoperative Clinical Parameters

The burden of comorbidity was substantially greater in the ICU group. Diabetes mellitus was present in 58.1% of ICU patients compared with 32.4% of non-ICU patients (p=0.034), and the prevalence of chronic kidney disease was also significantly higher (25.8% versus 8.1%; p=0.050). Nearly two-thirds of ICU patients (67.7%) had two or more comorbidities, compared with less than one-third (29.7%) in the non-ICU group (p=0.002). The preoperative physiological burden, as reflected by the ASA physical status classification and APACHE II score, was markedly worse in the ICU group. An ASA score of III or higher was observed in 77.4% of ICU patients versus 35.1% of non-ICU patients (p<0.001), and the mean APACHE II score was nearly twice as high (18.6 ± 5.4 versus 10.2 ± 4.1; p<0.001). Preoperative serum albumin was significantly lower in the ICU group (2.6 ± 0.5 versus 3.4 ± 0.6 g/dL; p<0.001), as was haemoglobin (9.8 ± 1.6 versus 11.4 ± 1.8 g/dL; p<0.001). Serum creatinine was correspondingly elevated in the ICU group (1.8 ± 0.9 versus 1.1 ± 0.4 mg/dL; p<0.001). Sepsis at presentation (61.3% versus 21.6%; p<0.001) and hemodynamic instability (51.6% versus 13.5%; p<0.001) were both substantially more common among patients who required ICU care. The demographic and clinical characteristics of both groups are detailed in Table 1.

Table 1. Demographic and clinical characteristics of study participants stratified by ICU admission status

SD: standard deviation; BMI: body mass index; COPD: chronic obstructive pulmonary disease; ASA: American Society of Anesthesiologists; APACHE: Acute Physiology and Chronic Health Evaluation. *Statistically significant (p < 0.05).

Operative Characteristics

The mean operative duration was significantly longer in the ICU group (138.4 ± 42.6 versus 104.2 ± 36.8 minutes; p=0.001), with 58.1% of ICU patients having operations exceeding 120 minutes compared with 24.3% in the non-ICU group (p=0.005). Estimated intraoperative blood loss was also significantly greater in the ICU group (486.3 ± 274.5 versus 298.6 ± 182.4 mL; p=0.002), and a blood loss exceeding 500 mL was observed in 45.2% of ICU patients versus 16.2% of non-ICU patients (p=0.009). The time interval from presentation to surgical incision did not differ significantly between the groups (Table 2).

Table 2. Operative characteristics stratified by ICU admission status

SD: standard deviation. *Statistically significant (p < 0.05).

Distribution of Surgical Diagnoses

The most common surgical diagnoses were intestinal obstruction (n=18, 26.5%) and hollow viscus perforation (n=16, 23.5%), followed by complicated appendicitis (n=10, 14.7%), strangulated hernia (n=8, 11.8%), mesenteric ischaemia (n=6, 8.8%), and acute cholecystitis (n=5, 7.4%). The remaining five patients (7.4%) had miscellaneous conditions including sigmoid volvulus and intra-abdominal abscess. The distribution of diagnoses across the ICU and non-ICU groups is presented in Figure 1. Patients with mesenteric ischaemia and hollow viscus perforation had proportionally higher ICU admission rates.

Multivariate Logistic Regression Analysis

Variables achieving a significance threshold of p<0.10 on univariate analysis — including age ≥70 years, diabetes mellitus, chronic kidney disease, COPD, presence of ≥2 comorbidities, ASA score ≥III, APACHE II score ≥15, serum albumin <3.0 g/dL, haemoglobin, serum creatinine, sepsis at presentation, hemodynamic instability, operative duration >120 minutes, and blood loss >500 mL — were entered into a multivariate binary logistic regression model. Four variables emerged as independent predictors of ICU admission: APACHE II score ≥15 (adjusted OR 5.34; 95% CI 1.72–16.58; p=0.004), ASA physical status ≥III (adjusted OR 4.82; 95% CI 1.56–14.89; p=0.006), preoperative serum albumin <3.0 g/dL (adjusted OR 3.67; 95% CI 1.21–11.12; p=0.022), and sepsis at presentation (adjusted OR 3.41; 95% CI 1.08–10.75; p=0.036). The results of the multivariate analysis are presented in Table 3, and the corresponding forest plot is shown in Figure 3.

Table 3. Independent predictors of ICU admission: multivariate logistic regression analysis

CI: confidence interval; ASA: American Society of Anesthesiologists; APACHE: Acute Physiology and Chronic Health Evaluation. *Statistically significant (p < 0.05).

Predictive Model Performance

The composite predictive model incorporating the four independent variables demonstrated excellent discriminative ability, with an area under the ROC curve (AUC) of 0.891 (95% CI 0.812–0.970) (Figure 2). At the optimal cutoff determined by the Youden index, the model achieved a sensitivity of 84.0% and a specificity of 81.1%, with a positive predictive value of 78.8% and a negative predictive value of 85.7%. The Hosmer–Lemeshow goodness-of-fit test was non-significant (χ²=6.42; p=0.601), indicating adequate model calibration. The Nagelkerke R² was 0.574, suggesting that the model accounted for approximately 57% of the variance in ICU admission. Detailed model performance metrics are summarised in Table 4.

Table 4. Performance metrics of the multivariate predictive model

ROC: receiver operating characteristic; AUC: area under the curve; CI: confidence interval.

The present study identified four independent predictors of ICU admission following emergency general surgery in elderly patients: ASA physical status ≥III, APACHE II score ≥15, preoperative serum albumin below 3.0 g/dL, and the presence of sepsis at the time of surgical presentation. The predictive model constructed from these variables demonstrated excellent discriminative ability with an AUC of 0.891, suggesting that a combination of physiological reserve assessment, illness severity scoring, nutritional status, and sepsis evaluation can reliably stratify elderly patients for the likelihood of postoperative intensive care requirement.

The ICU admission rate of 45.6% observed in our cohort is consistent with published data from comparable settings. A multicentre study by Parmar et al. (2022) reported an ICU transfer rate of 42% among elderly patients undergoing emergency laparotomy in India, while international series have documented rates ranging from 35% to 58% depending on the acuity of the case mix and institutional ICU admission criteria. This consistently high proportion underscores the physiological vulnerability of the elderly surgical population and the clinical importance of anticipating ICU needs before operating.

The ASA classification emerged as the strongest preoperative predictor in our model (adjusted OR 4.82), a finding that aligns with extensive prior evidence. A large Danish registry analysis by Vester-Andersen et al. (2021) demonstrated that ASA ≥III was independently associated with a fourfold increase in the odds of unplanned ICU admission after emergency abdominal surgery in patients over 65 years. The prognostic utility of the ASA score lies in its capacity to integrate multiple dimensions of chronic health impairment into a single, clinically intuitive assessment. Importantly, it is universally available, requires no laboratory data, and can be determined rapidly in the emergency setting — features that enhance its practical value as a triage tool.

The APACHE II score was the most powerful predictor in our analysis (adjusted OR 5.34), reflecting the score's incorporation of acute physiological derangement alongside chronic health status. Singh et al. (2022), in a prospective Indian study of 120 elderly emergency surgery patients, similarly found that an APACHE II threshold of 15 predicted ICU requirement with a sensitivity of 82% and specificity of 76%. The present study corroborates these findings and extends them by demonstrating that APACHE II retains its predictive significance even after adjustment for other independently informative variables. While the score requires laboratory inputs that may delay computation, its measurement is generally feasible within the first hours of emergency presentation and provides a quantitative severity benchmark that complements the subjective ASA assessment.

Preoperative hypoalbuminemia (serum albumin <3.0 g/dL) was identified as an independent predictor of ICU admission, with an adjusted OR of 3.67. Serum albumin serves as a composite marker of nutritional reserve and systemic inflammatory burden, both of which are critically relevant to surgical outcomes in the elderly. A recent meta-analysis by Zhang et al. (2023), encompassing over 28,000 patients across 14 studies, confirmed that preoperative hypoalbuminemia was independently associated with increased postoperative complications, ICU admission, and mortality in elderly surgical patients. In the context of emergency surgery, where the opportunity for preoperative nutritional optimisation is limited, a low albumin at presentation effectively signals a patient whose physiological margin for coping with the combined insults of acute illness and major surgery is substantially diminished.

Sepsis at presentation, present in over 60% of patients who required ICU care, was the fourth independent predictor identified (adjusted OR 3.41). The relationship between preoperative sepsis and the need for intensive postoperative monitoring is intuitively clear: septic patients frequently require ongoing haemodynamic support, fluid resuscitation, and antimicrobial optimisation that exceeds the capabilities of a general surgical ward. Kumar et al. (2021) demonstrated in a single-centre Indian cohort that sepsis at the time of emergency laparotomy in elderly patients was associated with a threefold increase in ICU admission and a twofold increase in 30-day mortality. The Sepsis-3 criteria used in the present study provide a standardised and widely adopted framework for identifying this high-risk subgroup.

Several variables that showed significant differences on univariate analysis — including haemoglobin, serum creatinine, hemodynamic instability, operative duration, and intraoperative blood loss — did not achieve independent predictive significance in the multivariate model. This likely reflects the considerable collinearity among these variables: for instance, hemodynamic instability and elevated creatinine are frequent manifestations of sepsis, while prolonged operative duration and greater blood loss are often consequences of more advanced pathology in patients with higher ASA and APACHE II scores. The loss of these variables from the final model does not diminish their clinical relevance; rather, it suggests that their prognostic information is substantially captured by the four retained predictors.

The practical implications of our findings merit emphasis. The four identified predictors are all measurable at or shortly after the time of emergency presentation, before the patient enters the operating theatre. This temporal window is precisely when ICU bed allocation decisions must be made. A clinician encountering an elderly patient with an ASA score of III or above, an APACHE II score exceeding 15, a serum albumin below 3.0 g/dL, and clinical evidence of sepsis can anticipate with high confidence that postoperative ICU admission will be necessary and can initiate the logistical steps — ICU bed reservation, anaesthetic team briefing, communication with critical care colleagues — that facilitate a seamless transition of care. Conversely, the absence of these predictors, while not eliminating ICU risk entirely, provides some reassurance that ward-level care may suffice.

The strengths of this study include its prospective design, the use of standardised definitions for all variables, and the application of a rigorous multivariate analytical framework with assessment of model calibration and discrimination. However, several limitations must be acknowledged. First, the sample size of 68 patients, while adequate for the identification of strong predictors, limits the number of variables that can be simultaneously evaluated in a logistic regression model and precluded the derivation of a validated scoring system. Second, the study was conducted at a single tertiary centre, and the case mix may not be representative of all practice settings. Third, the decision to admit to the ICU was based on clinical judgment rather than a protocolised algorithm, introducing the possibility of subjective variation. Finally, the study did not assess longer-term outcomes such as functional status at discharge or quality of life, which are of particular importance in the elderly population.

Future multicentre studies with larger sample sizes are warranted to validate the predictive model developed in this study, to refine the identified thresholds, and to explore additional variables such as frailty indices, sarcopenia assessment, and point-of-care lactate measurements that may further improve risk stratification. The integration of these predictors into a bedside decision-support tool or scoring system could enhance the consistency and efficiency of ICU triage in the elderly emergency surgical population.

This prospective observational study demonstrates that ASA physical status classification ≥III, APACHE II score ≥15, preoperative serum albumin <3.0 g/dL, and the presence of sepsis at the time of surgical presentation are robust independent predictors of ICU admission following emergency general surgery in patients aged 60 years and above. The composite predictive model incorporating these four variables exhibits excellent discriminative performance (AUC 0.891) and adequate calibration. These readily assessable clinical parameters can be deployed at the time of emergency presentation to facilitate proactive ICU bed allocation, guide perioperative management planning, and support informed discussions with patients and their families. Multicentre validation studies with larger cohorts are recommended to confirm the generalisability of these findings and to explore the development of a formal risk-scoring instrument.

1. Vester-Andersen M, Waldau T, Wetterslev J, Møller MH, Rosenberg J, Jorgensen LN. Unplanned ICU admission after emergency abdominal surgery: a nationwide cohort study. Acta Anaesthesiol Scand. 2021;65(3):345–354.
2. Kumar A, Sharma R, Mehta S, Gupta P. Predictors of morbidity and mortality following emergency laparotomy in elderly patients: a single-centre Indian experience. Indian J Surg. 2021;83(4):892–899.
3. Scott JW, Olufajo OA, Brat GA, et al. Use of national burden to define operative emergency general surgery. JAMA Surg. 2021;156(2):e205572.
4. Parmar KD, Shah NJ, Patel DM, Trivedi GJ. Emergency abdominal surgery in elderly: a prospective study of outcomes and predictors of ICU admission. Int Surg J. 2022;9(2):328–334.
5. Singh AK, Tripathi RK, Verma S. APACHE II score as a predictor of ICU admission in elderly patients undergoing emergency surgery: a prospective study. J Clin Diagn Res. 2022;16(4):PC01–PC05.
6. Desserud KF, Veen T, Søreide K. Emergency general surgery in the geriatric patient: a systematic review and meta-analysis. World J Emerg Surg. 2021;16(1):16.
7. Hewitt J, Carter B, McCarthy K, et al. Frailty predicts mortality in all emergency surgical admissions regardless of age: an observational study. Age Ageing. 2021;50(3):782–789.
8. Akinbami F, Askari R, Steinberg J, Sarani B, Cooper Z. Emergency general surgery in the elderly: risk factors for adverse outcomes. J Trauma Acute Care Surg. 2021;91(1):168–175.
9. Zhang Y, Wang L, Chen X, Liu H. Preoperative hypoalbuminemia and postoperative outcomes in elderly surgical patients: a systematic review and meta-analysis. Ann Surg Treat Res. 2023;104(1):42–53.
10. Merani S, Payne J, Padwal RS, Hudson D, Widder SL, Khadaroo RG. Predictors of in-hospital mortality and complications in very elderly patients undergoing emergency surgery. World J Emerg Surg. 2022;17(1):29.
11. Rangel EL, Cooper Z, Olufajo OA, et al. Mortality after emergency surgery continues to rise after discharge in the elderly: predictors of postdischarge death. Surgery. 2021;169(2):341–348.
12. Svenningsen P, Manoharan T, Foss NB, Lauritsen ML, Jørgensen CC. Risk factors for ICU admission after emergency laparotomy: a retrospective cohort study. Acta Anaesthesiol Scand. 2022;66(5):598–606.
13. Poulton TE, Moonesinghe SR, Raine R, Martin P, Oliver CM. Socioeconomic deprivation and mortality after emergency laparotomy: an observational study from the United Kingdom. Br J Anaesth. 2021;126(4):840–848.
14. McIsaac DI, Moloo H, Bryson GL, van Walraven C. The association of frailty with outcomes and resource use after emergency general surgery: a population-based cohort study. Anesth Analg. 2021;132(5):1328–1337.
15. Torrance ADW, Powell SL, Griffiths EA. Emergency surgery in the elderly: challenges and solutions. Open Access Emerg Med. 2021;13:109–118.
16. Patel KV, Brennan KL, Englesbe MJ, Terjimanian MN, Henke PK. Preoperative sarcopenia and mortality after emergency general surgery. J Surg Res. 2022;275:51–58.
17. Boyd-Carson H, Doleman B, Herrod PJJ, Sherwood M, Williams JP, Lund JN. Association between surgeon volume and postoperative mortality following emergency gastrointestinal surgery: a systematic review. Ann Surg. 2022;276(1):39–45.
18. Vilches-Moraga A, Fox J, Price A, Sherrington V, Pearce L. Emergency laparotomy in the older patient: factors predicting failure to rescue. Aging Clin Exp Res. 2021;33(12):3379–3386.
19. Aitken RM, Partridge JSL, Oliver CM, et al. Older patients undergoing emergency laparotomy: observations from the National Emergency Laparotomy Audit (NELA). Age Ageing. 2021;50(4):1213–1221.
20. Kork F, Balzer F, Spies CD, et al. Perioperative risk assessment in elderly patients: a critical appraisal of contemporary scoring systems. Curr Opin Anaesthesiol. 2021;34(1):49–56.
21. Cunha HFR, Lages JS, Salluh JIF, et al. Hypoalbuminemia as a predictor of clinical outcomes in critically ill elderly patients: a systematic review. J Crit Care. 2022;72:154123.
22. Jawad M, Baigi A, Gyllenborg J, Cao Y. ASA classification as a predictor of postoperative complications in elderly patients: a population-based study. BMC Anesthesiol. 2023;23(1):124.
23. Sharma N, Aggarwal S, Kumar R, Mahajan A. Prognostic value of serum lactate and albumin in predicting outcomes of emergency abdominal surgery in elderly patients. Turk J Surg. 2023;39(2):112–119.
24. Oliver CM, Bassett MG, Poulton TE, et al. Organisational factors and mortality after an emergency laparotomy: multilevel analysis of the National Emergency Laparotomy Audit. Br J Anaesth. 2021;126(4):800–811.
25. Møller MH, Vester-Andersen M, Jørgensen LN. Emergency surgery in elderly patients: current challenges and future directions. Langenbecks Arch Surg. 2022;407(6):2325–2334.