This retrospective study used purposive sampling to select 481 elderly patients (aged ≥75 years) with community-acquired pneumonia (CAP) admitted to the Second People’s Hospital of Lianyungang between June 2020 and June 2024.

1. Diagnosis of CAP based on Infectious Diseases Society of America/American Thoracic Society (IDSA/ATS) guidelines (Metlay et al., 2019).

1. Records of delirium within 30 days prior to admission.

This study did not involve direct interventions in laboratory tests or clinical treatments. Ethical approval was obtained from the Ethics Committee of the Second People’s Hospital of Lianyungang (No. 2022K040), and the study adhered to the Declaration of Helsinki. Informed consent was waived due to the retrospective nature of the study.

Clinical and laboratory data were extracted from the hospital’s electronic medical record system. Collected data included demographic characteristics, underlying diseases, admission and discharge details, discharge outcomes, and clinical results. Laboratory data focused on routine indicators obtained within the first 24 h of admission, including: (1) Complete blood count: Hemoglobin (Hb), white blood cell count (WBC), platelet count (PLT). (2) Biochemical tests: Liver function (ALT, AST), kidney function (BUN, SCr), glucose (Glu), albumin (Alb), triglycerides (TG), total cholesterol (TC). (3) Inflammatory markers: C-reactive protein (CRP).

The Mini Nutritional Assessment-Short Form (MNA-SF) was used to evaluate nutritional status (Liu et al., 2022). It comprises six questions that collectively generate a score assessing nutritional status, addressing aspects such as appetite loss, weight changes over the past 3 months, mobility, psychological stress or acute illness, neuropsychological problems, and Body Mass Index (BMI). Scores range from 0 to 14, categorizing nutritional status as: normal (12–14 points), at risk of malnutrition (8–11 points), and malnutrition (0–7 points).

Delirium was assessed using the Confusion Assessment Method (CAM) (Green et al., 2019). The CAM focuses on four core features: (1) acute onset and fluctuating course, (2) inattention, (3) disorganized thinking, and (4) altered level of consciousness. Diagnosing delirium requires the presence of both Feature one and Feature 2, plus either Feature 3 or Feature 4.

The FI-Lab quantifies frailty by assessing abnormalities in routine laboratory and vital sign indicators (Metlay et al., 2019; Blodgett et al., 2016). Blood pressure, a critical marker of cardiovascular health, when abnormal may indicate risks such as hypertension or hypotension, thereby affecting both the overall health and functional status of older individuals. Similarly, heart rate is a direct indicator of cardiac function; deviations from the norm may suggest arrhythmias or heart diseases impacting the quality of life in the elderly (Williams et al., 2018). The WBC count reflects immune system efficiency, where abnormalities could point to infections, inflammation, or hematological disorders—particularly significant for older adults with potentially compromised immunity. Hemoglobin levels illustrate the blood’s oxygen-carrying capacity, with aberrations potentially indicating anemia, malnutrition, or chronic diseases. Platelet levels are indispensable for coagulation, and irregularities may denote coagulation disorders (Patel, 2008). ALT and AST are hepatic function markers; abnormal values suggest liver damage or diseases like hepatitis or cirrhosis. Lipid profile indicators, total cholesterol and triglycerides, when abnormal, may highlight risks for cardiovascular diseases such as atherosclerosis. Blood glucose levels are vital markers for diabetes and metabolic syndrome, with deviations suggesting insulin resistance or diabetes. Renal function is evaluated through creatinine and BUN levels; abnormalities may signal renal dysfunction. Albumin levels indicate nutritional and liver status, where low values could suggest malnutrition, chronic illness, or hepatic dysfunction (Moreno et al., 2019). CRP serves as an inflammation and infection marker, with elevated levels pointing to acute or chronic inflammatory conditions (Ridker, 2014). The integration of these indicators allows for a comprehensive health evaluation, assisting in the identification of potential health risks and functional decline within the elderly demographic. Consequently, the FI-Lab provides clinicians with an objective metric for assessing health status and frailty in elderly patients. In our study, we selected 15 variables comprising three vital signs and 12 laboratory indicators. The normal reference ranges are as follows: SBP: 90–140 mmHg; DBP: 60–90 mmHg; Heart rate: 60–99 bpm; WBC count: 4.0–11.0 × 10⁹ cells/L; Hemoglobin: Males: 12–16 g/dL, Females: 11–15 g/dL; Platelet count: 100 - 300 × 10⁹ cells/L; ALT: 7–56 IU/L; AST: 10–40 IU/L; TC: <200 mg/dL; TG: <150 mg/dL; Glu: 70–110 mg/dL; Scr: Males: 0.74–1.35 mg/dL, Females: 0.59–1.04 mg/dL; BUN: 6–20 mg/dL; ALB: 3.5–5 g/dL; CRP: 0.42–5.2 mg/L. Values outside these ranges were scored as 1, while normal values were scored as 0. The FI-Lab index was calculated as the ratio of abnormal indicators to the total number of indicators assessed. For example, if three out of ten indicators were abnormal, the FI-Lab score would be 0.3, indicating moderate frailty. Scores range from 0 to 1, with higher values reflecting greater frailty.

Statistical analyses were performed using IBM SPSS Statistics version 21 and R 4.4.1. All statistical tests were two-sided, with a significance level set at p < 0.05.

To ensure the robustness of the analyses, missing data were addressed using Multiple Imputation by Chained Equations (MICE). This method accounts for uncertainty associated with missing data and provides unbiased parameter estimates. Imputation was conducted in R using the mice package with five iterations to stabilize imputed values. Variables with more than 20% missing data were excluded to prevent potential bias.

Descriptive statistics were used to summarize the demographic and clinical characteristics of the study population. Continuous variables were expressed as mean ± standard deviation (SD) or median (interquartile range, IQR), depending on the normality of their distribution. Categorical variables were presented as frequencies and percentages.

For continuous data following a normal distribution, the Student’s t-test was applied to compare means between two groups. For non-normally distributed data, the Mann-Whitney U test was used to compare medians. Chi-square tests were employed to examine associations between categorical variables.

To assess the impact of FI-Lab and MNA-SF scores on delirium, logistic regression analyses were conducted in three models: Model one without adjustments, Model two adjusted for age, gender, and Charlson Comorbidity Index (CCI), and Model 3 further adjusted for additional variables, including vital signs, as well as laboratory indicators such as white blood cell count, hemoglobin, albumin, blood urea nitrogen, glucose, and C-reactive protein. Odds ratios (OR) and 95% confidence intervals (CI) were calculated for each model.

Receiver operating characteristic (ROC) curves and the area under the ROC curve (AUC) were calculated to assess the predictive value of the FI-Lab, MNA-SF, and their combined detection. Patients were categorized into two groups based on the cut-off value of the FI-Lab. The cut-off value for the FI-Lab was determined through ROC curve analysis, selecting the value that maximized both sensitivity and specificity.

Kaplan-Meier survival analysis was conducted to evaluate the impact of the FI-Lab on survival outcomes in patients with different nutritional statuses. A log-rank test was used to determine whether there was a statistically significant difference in survival between these groups.

Using the Bootstrap approach with 5,000 repeated samples, we assessed the mediating effect of the MNA-SF and FI-Lab. The mediating effect was considered statistically significant if the 95% confidence interval did not include zero.

The study enrolled 481 elderly CAP patients, adhering to specific inclusion and exclusion criteria as outlined in Figure 1. The participants had an average age of 84.39 ± 5.14 years, with a delirium incidence rate of 21.4% and a 28-day mortality rate of 9.6%. Comparative analysis revealed that the delirium group had significantly higher values than the non-delirium group in terms of age (P< 0.001), mortality rate (P< 0.001), incidence of SCAP (P< 0.001), CCI score (P< 0.001), malnutrition (P< 0.001), polypharmacy (p = 0.018), blood glucose (P< 0.001), total cholesterol (p = 0.027), triglycerides (p = 0.004), blood urea nitrogen (p = 0.002), serum albumin (P< 0.001), hemoglobin (p = 0.002), C-reactive protein (P< 0.001), and FI-Lab (P< 0.001) (Table 1).

Logistic regression was used to investigate the impact of FI-Lab and MNA-SF nutritional scores on delirium. Three models were established: Model one without adjusting for confounders, Model two adjusted for age, gender, and CCI, and Model 3 further adjusted for polypharmacy, SBP, DBP, HR, WBC, HB, PLT, Scr, BUN, ALT, AST, ALB, TC, TG, Glu, and CRP. Model 1: The logistic regression analysis indicated that FI-Lab (OR = 7.095, 95% CI: 4.291–11.730) is a significant risk factor for delirium (P< 0.05). Conversely, the MNA-SF score was significantly negatively associated with delirium (OR = 0.800, 95% CI: 0.714–0.898), meaning that for each one-point increase in the MNA-SF score, the probability of delirium occurrence decreases by approximately 20%. Model 2: After adjusting for age, gender, and comorbidity index, FI-Lab remained significantly positively associated with delirium (OR = 6.758, 95% CI: 4.005–11.402), while the MNA-SF score remained negatively associated with delirium (OR = 0.847, 95% CI: 0.751–0.956), although its influence was slightly weakened compared to Model 1. Additionally, CCI was also significantly positively associated with delirium (OR = 1.597, 95% CI: 1.301–1.961). Model 3: After further adjusting for additional variables, FI-Lab remained positively associated with delirium (OR = 3.969, 95% CI: 2.123–7.419), although its effect weakened with the inclusion of more variables. The MNA-SF score became non-significant (p = 0.337), suggesting that when considering more variables, the predictive ability of MNA-SF for delirium decreases. Other significant variables included CCI, polypharmacy, SCAP, ALB, BUN, Glu, and CRP (Table 2).

Receiver operating characteristic (ROC) analysis assessed the sensitivity, specificity, and predictive value of the FI-Lab and its combination with MNA-SF for predicting delirium in elderly CAP patients. The AUC was 0.834 (95% CI: 0.794–0.875) for FI-Lab, with a cut-off value of 0.367 yielded 70.90% sensitivity and 78.20% specificity. Combining FI-Lab and MNA-SF increased the AUC to 0.847 (95% CI: 0.808–0 0.886), providing 82.50% sensitivity and 74.00% specificity (Figure 2).

Based on the similarity in survival patterns and increased statistical power, we merged the MNA-SF 12-14 and MNA-SF 8–11 groups. This merging helped us more clearly demonstrate the impact of FI-Lab and nutritional status on prognosis. Further subgroup analysis revealed 28-day mortality rates of 0.8% for the group with FI-Lab < 0.367 and MNA-SF 8-14 points, 8.4% for the group with FI-Lab ≥ 0.367 and MNA-SF 8-14 points, 12.1% for the group with FI-Lab < 0.367 and MNA-SF ≤ 7 points, and 40.3% for the group with FI-Lab ≥ 0.367 and MNA-SF ≤ 7 points (Log Rank test, χ² = 113.474, p < 0.001). Figure 3 presents the Kaplan-Meier survival curves.

The simple mediational model, after adjusting for all potential confounding variables, revealed the following: MNA-SF mediated the relationship between FI-Lab and delirium, with an indirect effect of 0.988 (95% CI: 0.236–1.862), accounting for 7.73% of the total effect. FI-Lab also mediated the relationship between MNA-SF and delirium, with an indirect effect of −0.215 (95% CI: −0.295 to −0.153), accounting for 59.23% of the total effect. These findings suggest that FI-Lab and nutritional status not only independently serve as predictors but also interact with each other (Figure 4).