The analysis used data collected prospectively for an observational cohort between July 2023 and July 2024 and included all 361 eligible participants, 165 of whom had appeared in an earlier publication. Patients underwent general anesthesia and were classified by American Society of Anesthesiologists physical status as I or II. Individuals were excluded if they had preoperative cognitive impairment, major psychiatric or neurological disorders, serious sensory or language impairments, alcohol or substance abuse, pre-existing systemic inflammatory disease, emergency surgery, severe hepatic or renal dysfunction, or other conditions that could interfere with cognitive or inflammatory measurements. Trained clinicians who were blinded to nutritional status assessed cognition before surgery and on postoperative days 1 and 2 with the Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA).

Early postoperative cognitive decline was predefined as a drop of at least three points from the preoperative baseline in either the MMSE or the MoCA. Sixteen patients, 4.4%, met the criterion on postoperative day 1 and nine, 2.5%, met it on day 2; one patient met the criterion on both days and was counted only once in the overall total. The researchers therefore studied changes confined to the immediate postoperative period rather than persistent postoperative neurocognitive disorder. Such early changes may be influenced by residual anesthetic effects, pain, fatigue, sleep disruption and acute physiological stress, and the study did not include a non-surgical retest control group. It also followed patients only through the first two postoperative days, leaving the longer-term course unknown.

GNRI provides a compact assessment of nutritional risk in older adults by combining serum albumin concentration with body weight relative to ideal body weight. The researchers calculated it as 1.489 multiplied by albumin in grams per liter, plus 41.7 multiplied by the ratio of actual to ideal body weight. Patients with GNRI values of at least 98 were classified as low risk, those from 92 to below 98 as moderate risk, and those below 92 as high risk. The cohort included 152 low-risk, 128 moderate-risk and 81 high-risk patients. Albumin concentrations fell progressively across these groups, from 40.1 grams per liter in the low-risk group to 37.0 in the moderate-risk group and 33.8 in the high-risk group, while most demographic and perioperative characteristics were similar.

In statistical models that penalized estimates for the small number of events, each one-point increase in GNRI was associated with lower unadjusted odds of early cognitive decline, with an odds ratio of 0.88 and a 95% confidence interval of 0.84 to 0.97. The association remained after adjustment for age, ASA class and preoperative MoCA score: the adjusted odds ratio was 0.91 per GNRI point, with a 95% confidence interval of 0.86 to 0.96 and a P value of 0.001. In a separate categorical model, moderate-risk patients had 39% lower adjusted odds than high-risk patients, while low-risk patients had 90% lower adjusted odds. Age and preoperative MoCA score were also independently associated with the outcome. These estimates describe odds within this observational cohort and should not be interpreted as a validated probability of cognitive decline for an individual patient.

The model’s apparent ability to distinguish patients with early cognitive decline was moderate but declined after internal validation. The apparent area under the receiver operating characteristic curve was 0.81, but 1,000 bootstrap resamples indicated an optimism of 0.07, producing an optimism-corrected AUC of 0.74. The corrected calibration slope was 0.89, suggesting some overfitting. In an exploratory analysis focused on postoperative day 2, GNRI alone had an apparent AUC of 0.731, which fell to 0.707 after bootstrap correction. A GNRI cutoff of 95.6 maximized the Youden index in that analysis, corresponding to sensitivity of 77.8% and specificity of 76.4%. Because the model was developed and assessed in the same single-center cohort, with only nine day-2 events, its discrimination requires confirmation in independent populations.

Perioperative blood measurements offered a possible biological correlate for the association. Patients with higher nutritional risk generally had higher concentrations of inflammatory, intestinal-barrier and neuronal-injury markers before surgery. Interleukin-6 (IL-6) rose after operation and then declined by day 2, although it remained above baseline. After adjustment for preoperative IL-6 and prespecified clinical covariates, the high-risk group had higher IL-6 than the low-risk group on both postoperative days. High-risk patients also had higher baseline-adjusted S100β, a marker used to indicate possible neuronal injury, on day 1. Preoperative GNRI was negatively correlated with day-2 IL-6, with a Spearman rho of -0.12, a 95% confidence interval of -0.22 to -0.02 and a P value of 0.022.

These biomarker patterns are compatible with the hypothesis that limited physiological reserve and nutritional impairment may influence immune regulation, inflammatory responses and recovery from surgical stress. Greater inflammatory activity has been proposed as one contributor to postoperative cognitive changes, while albumin and body weight capture only part of a patient’s nutritional and physiological condition. However, the study measured associations among nutritional risk, biomarkers and cognition without demonstrating a mechanism. The analyses did not establish that inflammation mediated the relationship between GNRI and cognitive decline, that lower GNRI caused the biomarker differences, or that any biomarker directly produced the observed score changes. Formal mechanistic and mediation analyses would be required to evaluate those explanations.

Several limitations constrain how far the findings can be generalized. The study was a secondary analysis of one hospital’s observational cohort, had only 24 early cognitive-decline events, and relied on parsimonious regression models to limit overfitting. Residual confounding, including unmeasured differences in health, frailty or perioperative stress, cannot be excluded. Cognitive testing was restricted to days 1 and 2, so the findings say nothing definitive about persistent postoperative neurocognitive disorder, later cognition or daily function. Performance estimates were internally validated but lacked external validation, and exploratory subgroup analyses included multiple comparisons, zero-event cells and a nominal age interaction. The results also do not show that nutritional optimization before surgery changes cognitive outcomes; prospective multicenter studies and interventional trials would be needed to establish whether such an intervention has an effect.

Subject of Research: Association of preoperative nutritional risk with early postoperative cognitive decline in older adults undergoing laparoscopic cholecystectomy

Article Title: Association between preoperative geriatric nutritional risk index and early postoperative cognitive decline after laparoscopic cholecystectomy

Article References: Fu, Y., Wei, Q., & Shi, W. (2026). Association between preoperative geriatric nutritional risk index and early postoperative cognitive decline after laparoscopic cholecystectomy. Frontiers in Nutrition, 13, Article 1905999. https://doi.org/10.3389/fnut.2026.1905999

Image Credits: AI Generated

DOI: 10.3389/fnut.2026.1905999

Keywords: geriatric nutrition, postoperative cognition, laparoscopic cholecystectomy, GNRI, older adults, inflammatory biomarkers, albumin, observational study, surgical outcomes, Association, between, preoperative