Group-based trajectory modeling identifies distinct glycemic patterns predictive of mortality in aneurysmal subarachnoid hemorrhage
摘要
Dysglycemia is common after aneurysmal subarachnoid hemorrhage (aSAH) and may reflect transient stress or sustained metabolic dysfunction. Traditional glucose metrics may not fully capture these dynamic changes. We aimed to investigate whether distinct glucose trajectory patterns during hospitalization are associated with 90-day mortality in aSAH patients.
MethodsWe retrospectively included 2,182 patients with aSAH admitted to a tertiary center. Blood glucose levels over the first 14 days of hospitalization were analyzed using group-based trajectory modeling (GBTM). The primary outcome was 90-day all-cause mortality. Secondary outcomes included functional outcomes, rebleeding, delayed cerebral ischemia, and intracranial infection. Multivariable Cox and logistic regression models were used to assess associations. Model performance was evaluated using AUC, net reclassification improvement, and integrated discrimination improvement.
ResultsFour distinct glucose trajectory groups were identified: low stable (28.8%), moderate declining (39.3%), rising to plateau (22.9%), and highly fluctuating with extreme values (9.0%). Ninety-day mortality increased across groups, from 1.6% in the low stable group to 14.7% in the highly fluctuating group. Rising-to-plateau (HR 4.31; 95% CI 2.13–8.76) and highly fluctuating (HR 6.31; 95% CI 2.91–13.67) patterns were independently associated with increased mortality, whereas the moderate-declining group was not (adjusted HR, 1.93; 95% CI, 0.95–3.95). These groups also had higher risk of adverse secondary outcomes. Incorporating glycemic trajectories improved mortality prediction beyond admission glucose.
ConclusionDistinct glycemic trajectories following aSAH were independently associated with 90-day mortality. Dynamic glucose trends, particularly rising-to-plateau and highly fluctuating pattern, have prognostic significance and may inform individualized glycemic management.