Higher Stress Hyperglycemia Ratio Marked a Clear Rise in Early ICU Delirium
In critically ill adults, acute dysglycemia is not just a metabolic footnote. It can track with cerebral vulnerability, inflammatory stress, and the abrupt neuropsychiatric changes that define ICU-acquired delirium. In this retrospective cohort from the MIMIC-IV database, stress hyperglycemia ratio emerged as a measurable signal tied to delirium within 72 hours of ICU admission, with the strongest risk concentrated in the highest quartile of SHR.
The cohort included 7397 critically ill adults, and 1259 developed delirium within 72 hours. Median onset was 11.25 hours, a short interval that fits the clinical pattern of early brain dysfunction in the ICU rather than a slow, downstream complication. Patients in the highest SHR quartile had a higher hazard of delirium than those in the lowest quartile, with a hazard ratio of 1.23 and a 95% confidence interval of 1.04 to 1.45. The association reached statistical significance, and the dose-response pattern followed an approximately linear curve rather than a threshold-only effect.
A single sentence captures the bedside relevance: rising stress hyperglycemia moved delirium risk upward in a graded fashion.
Why Stress Hyperglycemia May Track with Brain Dysfunction
SHR is appealing because it does more than reflect a one-time glucose value. By incorporating chronic glycemic status, it estimates the degree of acute metabolic stress relative to baseline physiology. In critical illness, that distinction matters because a modest glucose elevation in a person with chronically normal glycemia may represent a larger biologic shock than the same value in someone with longstanding diabetes and higher baseline HbA1c.
That framework is clinically plausible for delirium. Acute hyperglycemia can amplify oxidative stress, endothelial dysfunction, and inflammatory signaling, all of which may impair cerebral autoregulation and disrupt neurotransmission. The ICU brain is already exposed to sedatives, hypoxemia, hemodynamic instability, infection, and sleep fragmentation. A marker that captures the extra metabolic burden of illness may therefore identify patients whose neurocognitive reserve is eroding fastest.
The analysis did not stop at a crude association. Restricted cubic spline modeling supported a linear dose-response relationship, strengthening the case that risk rose progressively with SHR rather than clustering at a single inflection point. The association also held across multiple subgroups, including different Glasgow Coma Scale thresholds, glucose-measurement windows, and diabetes or HbA1c strata. In practical terms, the signal persisted whether the patient had known diabetes or not.
Delirium Sat Between Stress Hyperglycemia and Mortality
The mortality analysis adds the most clinically consequential layer. Delirium significantly mediated the indirect effect of SHR on both 28-day and 365-day all-cause mortality, while no significant direct effect of SHR on mortality was observed. In other words, the pathway from acute glycemic stress to death appears to run through brain dysfunction rather than through an independent mortality signal from SHR itself.
That pattern is important for intensivists because it reframes the biomarker. SHR is not merely a metabolic score attached to an adverse prognosis; it may identify a state in which metabolic stress translates into acute neurologic injury, and that neurologic injury then marks a trajectory toward worse survival. The mediation analysis makes the delirium link more than associative noise. It positions delirium as a biologically and clinically meaningful intermediary outcome.
The study also applied E-value analysis to probe robustness against unmeasured confounding. In observational ICU research, that step matters because severity, sedative exposure, infection burden, and organ failure can all complicate causal interpretation. Even so, the central finding remained stable enough to support the authors’ conclusion that SHR is independently associated with delirium risk.
What the Numbers Suggest at the Bedside
The result set points toward a practical use case. SHR is readily obtainable from routine glucose data and chronic glycemic estimates, which means it can be calculated early, before delirium fully declares itself. In a setting where delirium prevention depends on timely recognition of risk, an accessible biomarker that stratifies vulnerability has obvious appeal.
This does not turn SHR into a stand-alone decision rule. Delirium in the ICU remains multifactorial, and no single metabolic measure can replace clinical surveillance, sedation minimization, reorientation strategies, or treatment of precipitating illness. But SHR may help identify patients who deserve closer neurologic monitoring in the first day of ICU care, when the median onset of delirium in this cohort occurred within hours rather than days.
The cohort’s size gives the findings weight, but the design still carries familiar limits. Retrospective modeling cannot prove causality, and the database structure cannot fully eliminate residual confounding. Timing of glucose measurement, illness severity, and unmeasured aspects of ICU practice may still influence the observed association. Even so, the consistency across subgroup analyses and the linear dose-response pattern make the signal difficult to dismiss as random variation.
A Biomarker That Connects Metabolism, Delirium, and Survival
The clinical message is straightforward: elevated stress hyperglycemia ratio was associated with a higher risk of 72-hour delirium in critically ill adults, and delirium accounted for the observed mortality pathway at both 28 days and 365 days. The findings fit a model in which acute metabolic stress destabilizes the brain early in critical illness, with delirium serving as the visible endpoint of that injury.
For clinicians, the value lies in early stratification. SHR may help flag patients who are most likely to become delirious soon after ICU admission, before cognitive dysfunction becomes obvious at the bedside. The unanswered clinical question now is whether targeting stress hyperglycemia more aggressively can change that neurologic trajectory, or whether SHR is best used as a risk marker rather than a modifiable treatment target. That distinction will require prospective testing, not retrospective inference.