Material and Method: This prospective, double-blind observational study included frail patients aged ≥65 years admitted to tertiary ICUs between April 2024 and March 2025. Frailty was defined as a Rockwood Clinical Frailty Score ≥5. Delirium was diagnosed using the CAM-ICU scale. Fasting triglyceride and glucose levels measured within 24 hours of ICU admission were used to calculate TGI. Categorical variables were compared using Chi-square tests, while continuous variables were analysed using Student’s t-test or Mann-Whitney U-test, depending on distribution. ROC curve analysis was performed to assess the predictive value of TGI and other parameters. Multivariate logistic regression was used to identify independent predictors of delirium. A p-value <0.05 was considered statistically significant.

Results: Of 206 included patients (mean age 72.1±4.6 years), 35.4% developed delirium, predominantly of the hyperactive subtype (84.9%). TGI values were significantly higher in patients with delirium (p <0.001). ROC analysis demonstrated a strong predictive ability for TGI (cut-off >4.85, AUC=0.909, sensitivity 86.3%, specificity 87.2%). Multivariate logistic regression identified high TGI, elevated BMI, and male sex as independent predictors of delirium. TGI was not correlated with APACHE II or SOFA scores. Sedation and physical restraint were significantly more frequent in delirious patients.

Conclusion: TGI is a strong and independent predictor of delirium in frail and older ICU patients. Its ease of calculation from routine laboratory values makes it a promising biomarker for the early identification and targeted prevention of delirium in high-risk populations.

In older patients admitted to the intensive care unit, delirium has an acute onset and fluctuating course. It is characterised by alteration in consciousness, inattention, and cognitive impairment^9,10. Its prevalence ranges from 45% to 87% in intensive care unit patients¹⁰. Despite this high prevalence, it is difficult to recognise and manage. The underlying pathophysiology is complex, and many causes have been blamed. Critical illnesses, polypharmacy, nerve conduction damage, neuroinflammation, impaired neurotransmitter balance, and metabolic disorders are just a few examples. Just hospitalisation itself can cause delirium.

The presence of delirium is known to contribute to poorer prognoses, such as higher mortality rates, longer durations of mechanical ventilation, and extended ICU and hospitalizations¹¹. In addition, accumulating evidence suggests that frailty is a major predisposing factor for delirium^12,13. Numerous biomolecules have been studied in the prediction of delirium. However, the effect of these molecules on prognosis is not yet clear. Cognitive impairment occurs for various metabolic reasons. One of them is insulin resistance (IR). It is characterised by unresponsiveness of target tissues to insulin and impairment in lipid and glucose metabolism¹⁴. Insulin resistance, a core mechanism in metabolic syndrome, has been linked to endothelial dysfunction and cognitive impairment. The Triglyceride-Glucose Index (TGI), as a reliable surrogate marker of insulin resistance as it was evaluated in recent studies¹⁵. It was reported to be a predictor of coronary artery disease, chronic kidney disease, stroke and atherosclerosis¹⁶. The TGI, as a marker of insulin resistance, thus indirectly reflects a state of heightened vulnerability to neurocognitive complications in frail patients. The disruption of the vascular endothelium by IR has caused the need to investigate the effect of this parameter on cognitive impairment. However, very few studies indicate the relationship between the TGI and delirium in intensive care units¹⁷.

Frailty, delirium, and insulin resistance may share overlapping pathophysiological mechanisms. Thus, current study aimed to investigate the development of delirium in frail older ICU patients and to assess the potential predictive role of the triglyceride-glucose index (TGI) as a surrogate marker of insulin resistance in this patient group.

Patients admitted to the intensive care unit for any reason aged 65 years and older, with RASS ≥-3, Rockwood's Clinical Frailty Score (RCFS)≥5, and speaking Turkish were included in the study. Patients in intensive care for less than 24 hours (death or discharge), All conditions with brain injury (such as traumatic brain injury, ischaemic stroke, haemorrhagic stroke, hypoxic brain injury, hepatic encephalopathy, central nervous system infections), with mental impairment, severe aphasia, coma status or RASS<-3, patients with dementia or Alzheimer's disease, with hearing loss, with visual loss, with missing data and patients or their relatives not accepting to participate in the study were excluded.

Data Collection
Admission reasons, age, gender, educational status, marital status, body mass index (BMI), smoking, alcohol, and substance abuse history of patients were recorded. According to the referenced criteria, individuals are considered to have severe hearing loss if they cannot follow a conversation at a distance of under one meter, and severe visual loss if they fail to identify two fingers held at that distance¹⁸. The comorbidities of the patients were assessed using the Charlson Comorbidity Index. Acute Physiology and Chronic Health Evaluation II (APACHE II), Sequential Organ Failure Assessment (SOFA) scores, mechanical ventilation, renal replacement therapy, sedation or analgesia usage, physical restriction, and the presence of a central catheter and urinary catheter were recorded. The daily physical activities (Activity of Daily Living - ADL) of the patients before coming to the intensive care unit were calculated using the Barthel Index (0-20 fully dependent, 21-61 highly dependent, 62-90 moderately dependent, 91-99 slightly dependent, 100 independent) and then recorded.

In our tertiary intensive care units, laboratory parameters such as biochemistry, blood gas, haemogram, coagulation, acute phase reactants are tested routinely. Fasting glucose and triglyceride levels were recorded once, at ICU admission, and used to calculate the TGI. Triglyceride-glucose index (TGI): Ln was calculated as [Fasting triglyceride (mg/dl) x Fasting glucose (mg/dl)]/2. The routine intensive care management of the patients was not intervened.

Determination of Fragility
The frailty status of the patients was recorded on admission using the Turkish version of Rockwood's Clinical Frailty Score (RCFS) with information obtained from the patients' relatives. The Rockwood's Clinical Frailty Score is scored on a scale of 1-9 depending on the patient's physical reserve. 1: Extremely fit, 9: Terminal stage patient. In our study, patients with RCFS values of 5 and above were considered frail, in accordance with the literature¹⁹.

Detection of Delirium
Since deep sedation or a comatose state should not be present when evaluating patients for delirium, the Richmond Agitation Sedation Scale (RASS; -5 cannot be awakened → +4 exhibits aggressive behaviors) was used, and patients with RASS ≥ -3 were considered with regard to delirium development. Delirium was assessed using the Confusion Assessment Method-Intensive Care Units (CAM-ICU) (1-Acute change or fluctuations in mental status, 2-Attention deficit/distraction, 3-Change in level of consciousness, 4-Disordered thoughts). Delirium is diagnosed when criteria 3 or 4 are present alongside criteria 1 and 2. Patients were defined as delirium-positive if the first episode of delirium was detected during their intensive care unit stay. Delirium was considered negative if it did not develop before discharge or death.

Subtypes of delirium were diagnosed using the RASS score. If RASS was between -3 and -1, it was defined as hypoactive delirium; if RASS was between +1 and +4, it was defined as hyperactive delirium²⁰. We used the Turkish version of CAM-ICU in the assessment. The reliability and applicability of this version was validated in a previous study²¹. The physicians who worked shifts in the intensive care units received comprehensive training on delirium assessment and administered the CAM-ICU test twice a day at the same times in the morning and evening. If delirium was suspected, CAM-ICU was applied at that time.

The practitioners had no information about the demographic data and laboratory parameters of the patients. Therefore, our study is a double-blind observational study. The development of delirium was monitored with the CAM-ICU test until the development of delirium was observed or, if not, to the last day of ICU.

Endpoints
In this study, the primary endpoint was the development of delirium in frail older patients. The secondary endpoints were the relationship between TGI level and delirium development, the rate of delirium development in intensive care, the relationship between delirium, its subtypes and mortality.

Statistical analysis
To determine the number of patients who should be included in the study in line with the main purpose of our research, sample size analysis was performed using the G*power (Version 3.1) package programme before the study. In this direction, the literature was utilised. As a result of the sample size analysis, it was found that a total of 188 patients would be sufficient for sampling for 80% power (1-β=0.80) and α=0.05 error value (95% confidence interval).

The results were analysed using SPSS (Statistical Package for Social Sciences; SPSS Inc., Chicago, IL) 22 software package. Descriptive data were presented as n, % values for categorical data and mean±standard deviation (mean±standard deviation) values for continuous data. Chi-square analysis (Pearson Chi-square) was used to compare categorical variables between groups. The suitability of continuous variables for normal distribution was analysed by the Kolmogorov-Smirnov test. In the comparison of paired groups, Student t-test was used for normally distributed variables and Mann Whitney U-test was used for non-normally distributed variables. In the comparison of more than two groups, One Way ANOVA analysis was used for variables with normal distribution, and Kruskal Wallis analysis was used for variables without normal distribution.

Receiver operating characteristic (ROC) curve was drawn to measure the contribution of TGI to the development of delirium. Logistic regression analysis was performed to calculate the risk of delirium and mortality. The significant results in pairwise comparison were included in the multivariate model and the Enter method was used. A 1:1 nearest-neighbor propensity score matching (PSM) analysis was performed to reduce confounding effects and balance covariates between delirium and non-delirium groups. The propensity scores were calculated using a logistic regression model including the following variables: age, sex, APACHE II, SOFA, RCFS, ADL, Charlson Comorbidity Index, body mass index (BMI), presence of sepsis, sedation use, and physical restraint. Based on the availability of eligible patients, 73 individuals with delirium were successfully matched with 73 individuals without delirium, resulting in a matched cohort of 146 patients. Matching was performed without replacement, using the nearest neighbor method without applying a caliper. The statistical significance level was accepted as p <0.05 in the analyses.

Click Here to Zoom

Figure 1: Flow diagram of patient selection and analyse.

A total of 206 patients, 95 (46.1%) females and 111 (53.9%) males, were included in the study with a mean age of 72.1±4.6 years. Thirty-five percent of the patients were primary school graduates, 42.7% were high school graduates, and 22.3% were university graduates. Of the 52.4% of the patients were smokers, and 18% were alcohol users. Delirium was observed in 35.4% of patients, of which 57.5% were hyperactive, 15.1% hypoactive and 27.4% mixed type. A total of 64.1% of the patients died. Additional patient characteristics are presented in table 1.

Click Here to Zoom

Table 1: All characteristics of patients.

Clinical and Biochemical Parameters for Delirium
Body mass index (p =0.004), RASS (p <0.001), CCI (p =0.008), glucose (p =0.001), triglyceride (p =0.001), and TGI (p <0.001) values of patients with delirium were significantly higher than those without delirium.

The rate of sedation in patients with delirium (68.5%) was significantly higher than the rate of sedation in patients without delirium (19.5%) (p <0.001). The rate of restriction in patients with delirium (68.5%) was significantly higher than the rate in patients without delirium (30.1%) (p <0.001) (Table 2).

Click Here to Zoom

Table 2: Comparison of the presence of delirium for all parameters.

ROC and Regression Analyses
The predictive value of the Triglyceride-Glucose Index (TGI) for the presence of delirium was evaluated using ROC analysis, and an optimal cut-off value was identified. At a cut-off value of >4.85, TGI demonstrated a sensitivity of 86.3% and a specificity of 87.2%, with a positive predictive value of 78.7% and a negative predictive value of 92.1%. The area under the curve (AUC) was 0.909 (95% CI: 0.862-0.945), indicating that TGI is a strong predictor of delirium (Figure 2).

Click Here to Zoom

Figure 2: ROC Curve of TGI value for the presence of delirium.

According to the multivariate logistic regression analysis, being male, high BMI, and high TGI were associated with a risk for the presence of delirium (Table 3).

Click Here to Zoom

Table 3: Logistic regression analysis of the presence of delirium.

Following propensity score matching based on age, sex, APACHE II, SOFA, RCFS, ADL, CCI, BMI, presence of sepsis, use of sedation, and physical restraints, 73 patients with delirium were matched to 73 patients without delirium. In the matched cohort, the median TGI value was significantly higher in the delirium group compared to the non-delirium group (5.02 (IQR: 4.94-5.13) vs. 4.74 (IQR: 4.52-4.92), respectively; p <0.0001, Mann-Whitney U tes)) (Figure3, Table 4).

Click Here to Zoom

Figure 3: TGI Distribution by Delirium Status.

Click Here to Zoom

Table 4: TGI Comparison After Propensity Score Matching.

These findings indicate that even after balancing for key clinical confounders, elevated TGI remains significantly associated with delirium development in frail ICU patients.

Clinical and Biochemical Parameters for Mortality
The age (p =0.012), CCI (p <0.001), APACHE2 (p <0.001), and SOFA (p <0.001) values of the deceased were significantly higher than those of the survivors, while the ADL (p =0.002) values were significantly lower. The mortality rate of patients hospitalized in the ICU for renal reasons (73.2%) was significantly higher than that of others (59.3%) (p =0.047). The mortality rate among ventilated patients (78.8%) was significantly higher than that of non-ventilated patients (50.5%) (p <0.001). Additionally, the mortality rate of patients requiring renal replacement therapy (76.7%) was significantly higher than that of those who did not need it (58.9%) (p =0.016) The mortality rate of those with sepsis (78.7%) was significantly higher than that of those without sepsis (51.8%) (p <0.001) (Table 5).

Click Here to Zoom

Table 5: Presence of mortality according to all parameters.

In the comparison of delirium subtypes, no statistically significant differences were found between the groups in terms of age (p =0.520), sex (p =0.891), BMI (p =0.584), education level (p =0.880), marital status (p =0.189), ADL score (p =0.970), CCI (p =0.614), APACHE II score (p =0.902), SOFA score (p =0.734), RCFS (p =0.891), presence of sepsis (p =0.864), ICU length of stay (p =0.705), mortality (p =0.975), sedation (p =0.457), use of mechanical ventilation (p =0.147), presence of central venous catheter (p =0.298), physical restraint (p =0.293), and blood glucose level (p =0.149). However, a statistically significant difference was found in triglyceride levels between the subtypes (p =0.040), with median (IQR) values of 204.5 (198-242) mg/dL in hyperactive, 187.0 (160-204) mg/dL in hypoactive, and 197.5 (179.5- 233.5) mg/dL in mixed delirium. TGI values did not differ significantly between the groups (p =0.833). RASS scores showed a statistically significant difference between the subtypes (p <0.001), with median (IQR) values of +1.0 (0.0-2.0) in hyperactive, -1.0 (-2.0--1.0) in hypoactive, and +1.0 (1.0-1.0) in mixed type (Table 6).

Click Here to Zoom

Table 6: Comparison of all parameters according to delirium type.

The TGI is a validated surrogate marker of insulin resistance and is increasingly recognized as a systemic metabolic risk indicator. Its clinical usefulness has been demonstrated in cardiovascular disease, stroke, and neurocognitive disorders¹⁷. Delirium, a multifactorial syndrome characterized by sudden cognitive impairment, results from a complex interplay of neuroinflammation, oxidative stress, metabolic disturbances, and endothelial dysfunction-all of which are associated with insulin resistance. Therefore, our findings are physiologically plausible and align with existing literature emphasizing insulin resistance and metabolic dysregulation in the development of delirium.

Delirium is a common and serious complication in intensive care, particularly among older and frail individuals. Advanced age and frailty increase vulnerability to delirium due to impaired physiological reserve, reduced cognitive capacity, and heightened inflammatory responses (22-24, 29). In a large prospective study, frailty increased the risk of delirium by 60% and was associated with higher mortality¹². In our study, we included patients with an RCFS score -3 to +4, forming a homogeneously frail cohort with a mean age of 72.1 years and a delirium incidence of 35.4%. Additionally, significantly lower ADL scores in delirious patients further support the association between functional dependence and heightened delirium risk. Delirium not only results in temporary or permanent cognitive decline in frail older adults but also delays recovery, impedes return to independent daily living, and increases healthcare costs²⁵. Therefore, early identification and prevention of delirium are critical priorities in the ICU setting²⁶.

Previous studies have demonstrated that the incidence of delirium can be reduced by up to 40% through the recognition of modifiable risk factors and the implementation of targeted preventive strategies²⁷. Multicomponent interventions such as the ABCDEF bundle-which includes spontaneous awakening and breathing trials, sedation coordination, regular delirium monitoring, early mobilization, and family engagement-have been shown to decrease delirium rates²⁸. However, the comprehensive application of such bundles requires significant clinical resources. Therefore, there is a need for simple, accessible, and objective biomarkers to help identify high-risk patients early and support preventive decision-making. Our study addresses this need by exploring the role of the TGI as a potential predictor of delirium in this vulnerable population. To our knowledge, this is one of the few studies to evaluate the role of TGI in ICU delirium among frail older adults, a population particularly susceptible to both metabolic and cognitive disturbances.

Frailty is characterized by diminished physiological reserve and heightened susceptibility to homeostatic disruption. The combination of frailty and elevated TGI may represent a "metabolic frailty phenotype" that predisposes to acute brain dysfunction under the stress of critical illness. The strong predictive value of TGI in our cohort supports this hypothesis. Since TGI is a practical and indirect measure of insulin resistance, it is strongly correlated with standard methods such as the hyperinsulinaemic euglycaemic clamp test (HEC) and HOMA³⁰. Insulin resistance has been shown to negatively affect neurological function and may contribute to cognitive impairment through mechanisms involving tau phosphorylation, Aβ accumulation, and disrupted insulin signaling^31-33. Associations between preoperative IR and postoperative delirium, as well as Alzheimer's disease biomarkers, have also been reported³⁴. Moreover, IR-related endothelial dysfunction has been linked to cognitive decline via atherosclerosis and dyslipidemia³⁵. Sun et al. identified TGI as an independent risk factor for postoperative delirium in patients with type 2 DM^36, while Huang et al. showed that elevated TGI predicted both delirium and mortality in elderly ICU patients¹⁷. Consistent with these findings, our study confirmed the independent predictive value of TGI for delirium in frail ICU patients using multivariate regression model. To strengthen the validity of our findings, we performed propensity score matching to balance key confounding variables between the delirium and non-delirium groups. Even after matching, the TGI remained significantly higher in patients with delirium, supporting its potential role as an independent metabolic biomarker associated with acute cognitive dysfunction. This reinforces the hypothesis that systemic metabolic dysregulation, particularly insulin resistance, may contribute to the pathophysiology of delirium in frail ICU populations.

In ROC analysis, we observed that TGI with a cut-off value >4.85 had a high diagnostic accuracy with an area under the curve (AUC) of 0.909, showing a sensitivity of 86.3% and specificity of 87.2%. Our findings align with previous studies that identified metabolic dysregulation as a key contributor to acute cognitive impairment. However, further large-scale prospective studies are warranted to validate the diagnostic thresholds and explore the integration of TGI into comprehensive delirium risk prediction models.

In addition to the strong predictive performance of TGI, our study identified other relevant clinical associations. Body mass index (BMI) was significantly higher in patients with delirium, suggesting that obesity-related systemic inflammation may contribute to cognitive vulnerability¹⁴. Interestingly, well-established ICU severity scores such as APACHE II and SOFA did not show significant associations with delirium occurrence in this cohort, which further highlights the potential prognostic utility of specific metabolic markers like TGI. Furthermore, delirious patients were significantly more likely to receive sedation and physical restraint compared to non-delirious patients. This observation may reflect both reactive clinical management in response to agitation, and possibly, the iatrogenic role of physical restraint in delirium onset. Although causality cannot be established from this observational data, these findings underscore the importance of cautious sedation practices and minimizing the use of restraints in high-risk patients.

In addition to the primary analysis, we conducted a detailed subgroup analysis based on delirium subtypes-hyperactive, hypoactive, and mixed. In our study cohort, hyperactive delirium was the most common subtype (46.7%), followed by mixed (29.3%) and hypoactive (24.0%) types. Interestingly, this distribution differs from most of the existing literature, where hypoactive delirium is typically reported as the predominant subtype, especially among older ICU patients. The discrepancy may stem from methodological factors such as closer monitoring and structured delirium assessments in our study, which may have led to better detection of agitated behavior associated with hyperactive delirium. Additionally, cultural and clinical differences in sedation practices, restraint use, or ICU staffing may influence delirium subtype prevalence across settings. The predominance of hyperactive delirium in our cohort may partly reflect detection bias, as clinicians more easily recognize agitated behaviors compared to hypoactive presentations. Standardization of assessment training and the use of continuous observation tools may reduce this bias in future research.

While TGI values did not significantly differ between delirium subtypes, serum triglyceride levels were significantly lower in patients with hypoactive delirium compared to those with hyperactive delirium. This finding may reflect divergent metabolic or inflammatory profiles between subtypes. The clinical parameters such as age, sex, BMI, illness severity (APACHE II, SOFA), comorbidity burden, functional status, ICU length of stay, and mortality were comparable across subtypes.

Although TGI was significantly associated with the presence of delirium, we did not observe a statistically significant difference in TGI levels between survivors and non-survivors in our cohort. This finding suggests that TGI may not independently predict mortality among frail ICU patients. One possible explanation is that TGI reflects a metabolic vulnerability, such as insulin resistance and endothelial dysfunction, that primarily contributes to neurocognitive complications like delirium rather than directly influencing survival outcomes. Additionally, mortality in the ICU is a multifactorial endpoint influenced by various acute physiological insults, disease severity, and comorbid conditions. In our study, classical severity scores such as APACHE II and SOFA were significantly higher in patients who died.

Several strengths of our study should be noted. The prospective design and systematic delirium assessment using the CAM-ICU enhance the internal validity. We focused specifically on a frail elderly population, in whom delirium has both higher prevalence and greater prognostic impact. The use of validated measurement tools such as CAM-ICU, RCFS, and Barthel ADL supports methodological reliability. Confounding factors were controlled using multivariate logistic regression, and robustness was further improved through propensity score matching. Even after matching, TGI remained significantly higher in delirium patients, reinforcing the potential of TGI as an independent biomarker. The calculation of TGI from routine laboratory data provides an additional advantage for its clinical use. Our findings suggest that TGI may be an easily available and reliable parameter for the prediction of neurocognitive complications in intensive care conditions.

On the other hand, the single-center nature of the study and the relatively small sample size which may not fully represent the broader ICU population and may limit its generalizability. Multicenter studies with larger sample sizes are warranted to validate the predictive value of TGI and confirm its external applicability across diverse ICU settings. Another limitation is that the long-term outcomes and costs of delirium were not investigated. However, this did not affect our results because we examined whether TGI is an independent risk factor for the development of delirium. Another important limitation is the lack of control for potential confounding factors affecting glucose levels. In this study, patients with diabetes mellitus, those receiving corticosteroids, or those under glucose infusion were not excluded. These interventions and conditions may have influenced fasting glucose values and, consequently, the TGI. However, we intentionally retained these patients in order to reflect the metabolic and therapeutic complexity of real-life ICU populations. This approach enhances the external validity of our findings but may have introduced some variability in the interpretation of TGI values. In addition, insulin resistance was assessed only by TGI, and direct methods such as the HEC test or HOMA-IR were not used. Another limitation is that the TGI was only measured on ICU admission, and dynamic changes in metabolic status over time were not assessed. Given that glucose and triglyceride levels can be influenced by stress responses, medications (e.g., steroids), and feeding strategies, a single measurement may not fully reflect ongoing metabolic risk. The other limitation is, TGI was measured only once at ICU admission. Serial monitoring throughout the ICU stay could have provided a better understanding of metabolic fluctuations related to delirium risk. The observation of hyperactive delirium as the predominant subtype in our study may be due to the fact that hypoactive delirium could be mistaken for depression, fatigue, or a general state of illness, making it more difficult to recognize clinically with the CAM-ICU scale we used. The fluctuating course of delirium, especially in mixed-type delirium, leads to the alternation of both hypoactive and hyperactive features of motor symptoms, complicating subtype classification. Glucose-modifying factors such as diabetes mellitus, corticosteroid therapy, and glucose infusions were not excluded, which might have influenced fasting glucose and TGI values. Future studies should consider subgroup analyses adjusting for these confounders.

Ethics approval and consent to participate
This study was approved by the Fırat University Non-Interventional Research Ethics Committee on 27 March 2024 (approval no: 23375). Written informed consent to participate was obtained from all patients or their legally authorised representatives. The research protocol followed ethical standards the Declaration of Helsinki, and was reported in line with the STROBE recommendations for observational studies.

Consent for publication
Not applicable

Availability of data and materials
The datasets used and analysed during the current study are available from the corresponding author upon reasonable request.

Competing Interests
The authors declare that they have no competing interests.

Funding: This manuscript did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Acknowledgements
The authors thank the intensive care unit staff of Fırat University for their support during data collection.

1) Athari F, Hillman KM, Frost SA. The concept of frailty in intensive care. Aust Crit Care 2019; 32: 175-8.

2) Bagshaw SM, Stelfox HT, McDermid RC et al. Association between frailty and short- and long-term outcomes among critically ill patients: a multicentre prospective cohort study. CMAJ 2014; 186: E95-102

3) Clegg A, Young J, Iliffe S, Rikkert MO, Rockwood K. Frailty in elderly people. Lancet 2013; 381: 752-62 11p North American Edition.

4) Darvall JN, Bellomo R, Paul E et al. Frailty in very old critically ill patients in Australia and New Zealand: a population-based cohort study. Med J Aust 2019; 211: 318-23.

5) Hoogendijk EO, Aflalo J, Ensrud KE, Kowal P, Onder G, Fried LP. Frailty: implications for clinical practice and public health. Lance. 2019; 394: 1365-75.

6) Muscedere J, Waters B, Varambally A, Bagshaw SM, Boyd JG, Maslove D et al. The impact of frailty on intensive care unit outcomes: a systematic review and meta-analysis. Intensive Care Med 2017; 43: 1105-22.

7) Bock JO, König HH, Brenner H et al. Associations of frailty with health care costs-results of the ESTHER cohort study. BMC Health Serv Res 2016; 16: 128.

8) Collard RM, Boter H, Schoevers RA, Oude Voshaar RC. Prevalence of frailty in community-dwelling older persons: a systematic review. J Am Geriatr Soc 2012; 60: 1487-92.

9) Inouye SK, Westendorp RG, Saczynski JS. Delirium in elderly people. Lancet 2014; 383: 911-22.

10) Moon KJ, Lee SM. The effects of a tailored intensive care unit delirium prevention protocol: a randomized controlled trial. Int J Nurs Stud 2015; 52: 1423-32.

11) Mazzola P, Tassistro E, Di Santo S, Rossi E, Andreano A, Valsecchi MG et al. The relationship between frailty and delirium: insights from the 2017 Delirium Day study. Age Ageing 2021; 50: 1593-9. doi: 10.1093/ageing/afab042. PMID: 33792656.

12) Sanchez D, Brennan K, Al Sayfe M, Shunker SA, Bogdanoski T, Hedges S et al. Frailty, delirium and hospital mortality of older adults admitted to intensive care: the Delirium (Deli) in ICU study. Crit Care 2020; 24: 609.

13) Zhang XM, Jiao J, Xie XH, Wu XJ. The association between frailty and delirium among hospitalized patients: an updated meta-analysis. J Am Med Dir Assoc 2021; 22: 527-34.

14) Di Pino A, DeFronzo RA. Insulin resistance and Atherosclerosis: implications for insulin-sensitizing agents. Endocr Rev 2019; 40: 1447-67.

15) Fritz J, Bjørge T, Nagel G et al. The triglyceride glucose index as a measure of insulin resistance and risk of obesity-related cancers. Int J Epıiemiol 2020; 49: 193-204.

16) Liao Y, Zhang R, Shi S, Zhao Y, He Y, Liao L et al. Triglyceride-glucose index linked to all-cause mortality in critically ill patients: a cohort of 3026 patients. Cardiovasc Diabetol 2022; 21: 128.

17) Huang X, Cheng H, Yuan S et al. Triglyceride-glucose index as a valuable predictor for aged 65-years and above in critical delirium patients: evidence from a multi-center study. BMC Geriatr 2023; 23: 701. doi: 10.1186/s12877-023-04420-0. PMID: 37904099; PMCID: PMC10617052.

18) Morandi A, Inzitari M, Udina C et al. Visual and hearing impairment are associated with delirium in hospitalized patients: results of a multisite prevalence study. J Am Med Dir Assoc 2021; 22: 1162-7.

19) Church S, Rogers E, Rockwood K, Theou O. A scoping review of the Clinical Frailty Scale. BMC Geriatr 2020; 20: 393. doi: 10.1186/s12877-020-01801-7

20) Ely EW, Inouye SK, Bernard GR et al. Delirium in mechanically ventilated patients: validity and reliability of the confusion assessment method for the intensive care unit (CAM-ICU) JAMA 2001; 286: 2703-46. doi: 10.1001/jama.286.21.2703.

21) Akıncı SB, Rezaki M, Özdemir Rezaki H, Çelikcan A, Kanbak M, Yorgancı K, Aypar Ü. Validity and reliability of the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) in Turkish. Turk J Anaesth Reanim 2005; 33: 333-41.

22) Pavone KJ, Jablonski J, Junker P, Cacchione PZ, Compton P, Polomano RC. Evaluating delirium outcomes among older adults in the surgical intensive care unit. Heart & Lung 2020; 49: 578-84.

23) Xiao LI, Zhang L, Fang GO, NG, Yuhang AI. Incidence and risk factors for delirium in older patients following intensive care unit admission: a prospective observational study. J Nurs Res 2020; 28: e101.

24) Panza F, Solfrizzi V, Barulli MR, Santamato A, Seripa D, Pilotto A, Logroscino G. Cognitive frailty: a systematic review of epidemiological and neurobiological evidence of an age-related clinical condition. Rejuven Res 2015; 18: 389-412.

25) Caplan GA, Teodorczuk A, Streatfeild J, Agar MR. The financial and social costs of delirium. Eur Geriatr Med. 2020; 11: 105-12. doi:10.1007/s41999-019-00257-2.

26) Miao L, Shen X, Du Z, Liao J. Nutritional status and triglyceride-glucose index influence delirium in elderly heart failure patients. Sci Rep. 2024; 14: 25235. doi: 10.1038/s41598-024-77573-6.

27) Hayhurst CJ, Marra A, Han JH et al. Association of hypoactive and hyperactive delirium with cognitive function after critical illness. Crit Care Med 2020; 48: e480-8.

28) Devlin JW, Skrobik Y, Gélinas C et al. Clinical practice guidelines for the prevention and management of pain, agitation/sedation, delirium, immobility, and sleep disruption in adult patients in the ICU. Crit Care Med 2018; 46: e825-73.

29) Sahle BW, Pilcher D, Litton E et al. Association between frailty, delirium, and mortality in older critically ill patients: a binational registry study. Ann Intensive Care 2022; 12: 108. doi: 10.1186/s13613-022-01080-y.

30) Wang M, Mei L, Jin A, Cai X, Jing J, Wang S et al. Association between triglyceride glucose index and atherosclerotic plaques and Burden: findings from a community-based study. Cardiovasc Diabetol 2022; 21: 204. doi: 10.1186/s12933-022-01638-x.

31) Micci MA, Krishnan B, Bishop E et al. Hippocampal stem cells promotes synaptic resistance to the dysfunctional impact of amyloid beta oligomers via secreted exosomes. Mol Neurodegener 2019; 14: 25. doi: 10.1186/s13024-019-0322-8.

32) Wakabayashi T, Yamaguchi K, Matsui K et al. Differential effects of diet- and genetically-induced brain insulin resistance on amyloid pathology in a mouse model of Alzheimer’s Disease. Mol Neurodegener 2019; 14: 15. doi: 10.1186/s13024-019-0315-7.

33) Ma L, Wang J, Li Y. Insulin resistance and cognitive dysfunction. Clin Chim Acta 2015; 444: 18-23. doi: 10.1016/j.cca.2015.01.027.

34) Wang J, Shuang P, Li Z, Zhao L, Wang X, Liu P. Association of insulin resistance with delirium and CSF biomarkers of Alzheimer’s disease in elderly patients with hip fracture. Aging Clin Exp Res 2023; 35: 1521-9.

35) Fang Y, Dou A, Shen Y et al. Association of triglyceride-glucose index and delirium in patients with sepsis: a retrospective study. Lipids Health Dis 2024; 23: 227. doi: 10.1186/s12944-024-02213-x.

36) Sun M, Liu M, Zhang F et al. Triglyceride-glucose index predicts postoperative delirium in elderly patients with type 2 diabetes mellitus: a retrospective cohort study. Lipids Health Dis 2024; 23: 107. doi: 10.1186/s12944-024-02084-2.