Malnutrition and Postoperative Delirium

Evidence

This retrospective observational cohort study examined the association between preoperative malnutrition and postoperative delirium in consecutive patients aged 65 years or older who underwent elective or emergency surgery under general anesthesia at one institution from January 2022 through December 2025. The exposure was malnutrition classified retrospectively with the Global Leadership Initiative on Malnutrition framework, using available phenotypic and etiologic criteria. Contemporaneous patients who did not meet those criteria served as potential controls. Eligibility required electronic medical record information sufficient to determine preoperative nutritional status and to ascertain postoperative delirium during the index hospitalization. Source: pmid:42898307

The investigators created a 1:2 propensity score–matched comparison between patients with and without Global Leadership Initiative on Malnutrition–defined malnutrition. The matched cohort contained 495 patients: 165 with malnutrition and 330 without malnutrition. This matching approach was intended to compare groups with measured propensity-score covariates rather than simply contrast the original exposure groups. The reported primary analysis used conditional logistic regression stratified by matched set and included multivariable adjustment. The study also reported analyses in the unmatched cohort, alternative matched analyses, and sensitivity analyses that changed exclusions, covariates, observation windows, or postoperative delirium definitions. Source: pmid:42898307

Postoperative delirium during hospitalization was identified through Confusion Assessment Method assessments, Confusion Assessment Method for the Intensive Care Unit assessments, or an explicit physician diagnosis. Thus, the outcome represented delirium documented in routine clinical records during the hospitalization, rather than an outcome collected through a uniformly scheduled study-specific assessment protocol. Data extraction was performed independently by two investigators using a standardized form; discrepancies were reviewed against original records and adjudicated through discussion and, when needed, a third investigator. Missing, implausible, and extreme values were checked against source records before the analytical dataset was finalized. Source: pmid:42898307

In the matched cohort, postoperative delirium occurred in 42 patients with Global Leadership Initiative on Malnutrition–defined malnutrition (25.5%) and 37 patients without malnutrition (11.2%; p < 0.001). In the primary matched-set conditional logistic regression, Global Leadership Initiative on Malnutrition–defined malnutrition was associated with higher odds of postoperative delirium, with an adjusted odds ratio of 2.36 (95% confidence interval, 1.36–4.10; p = 0.002). This is an association between the retrospectively classified exposure and the recorded hospital outcome in the matched older surgical population; it is not a direct estimate of an intervention effect. Source: pmid:42898307

The authors reported that the association was comparable in the unmatched cohort, alternative matched analyses using generalized estimating equations, and several sensitivity analyses. Results were described as remaining in the same direction when the analysis excluded patients admitted to an intensive care unit, excluded patients receiving intraoperative blood transfusion, or substituted operative duration for anesthesia duration. Analyses using fixed postoperative observation windows and a restricted-length-of-stay cohort were also reported as yielding comparable associations. These consistency checks address whether the observed pattern changed under the specified analytic choices, but they remain analyses of retrospectively collected observational data. Source: pmid:42898307

The report framed Global Leadership Initiative on Malnutrition classification as a multidimensional measure because it integrates phenotypic and etiologic information rather than relying only on an isolated laboratory index or screening score. Preoperative C-reactive protein was used as supportive evidence for the inflammatory etiologic component of the classification. Accordingly, the study’s discussion cautioned that any association involving preoperative C-reactive protein should not be treated as independent of nutritional-inflammatory vulnerability. The exposure therefore reflects the framework and the available medical-record variables used to apply it, not a single biochemical measurement. Source: pmid:42898307

Analysis — Observational Association

The observed pattern is that older patients classified as malnourished before surgery had more documented postoperative delirium during hospitalization and higher adjusted odds in the matched analysis than their matched counterparts without that classification. Propensity score matching, matched-set conditional logistic regression, and the reported alternative analyses make the comparison more structured than an unadjusted contrast, while still leaving it dependent on variables available in the records and on the study’s selection process. The study population was restricted to patients with sufficient electronic medical record information for both exposure classification and outcome ascertainment at one institution, which shapes the population represented by the estimate. The reported statistical association does not establish clinical significance or patient benefit. It also does not establish that Global Leadership Initiative on Malnutrition–defined malnutrition causes postoperative delirium, because this was not a randomized exposure and unmeasured or incompletely measured characteristics may be related to both nutritional classification and delirium documentation. Source: pmid:42898307

Limitations

The retrospective design cannot establish causation. The single-center design may have introduced selection bias, and residual confounding may remain despite propensity score matching, multivariable adjustment, and sensitivity analyses. Selection also depended on availability of complete information needed to classify nutritional status and ascertain postoperative delirium in the electronic medical record. Consequently, the comparison may not capture older surgical patients whose records could not support those determinations, and the measured covariates cannot account for every relevant preoperative or perioperative difference between groups. Source: pmid:42898307

Formal cognitive and frailty assessments were sparse and nonuniform, while standardized muscle-mass measurements were unavailable; these variables were not included in the primary regression model. Those extraction and measurement constraints leave uncertainty around characteristics that could be related to both Global Leadership Initiative on Malnutrition classification and postoperative delirium. In addition, delirium ascertainment relied on routine clinical documentation rather than a uniformly scheduled protocol. The authors noted that this approach may under-recognize transient or hypoactive episodes, so the recorded outcome may not fully represent all postoperative delirium occurring during hospitalization. Source: pmid:42898307

The analyses describe associations in older patients receiving general anesthesia at the participating institution and should remain scoped to that setting and record-based measurement approach. The reported robustness analyses show that the association persisted across specified analytic variations, but they do not remove the limitations of retrospective exposure classification, nonuniform assessment of several patient characteristics, or routine-documentation outcome capture. The available evidence therefore supports a description of the observed association and its uncertainty, not a causal or patient-specific conclusion. Source: pmid:42898307