Longitudinal serum albumin trajectories and neurological outcomes after aneurysmal subarachnoid hemorrhage

Background Serum albumin is routinely measured after aneurysmal subarachnoid hemorrhage (aSAH), but most prognostic studies have relied on single measurements. We investigated whether longitudinal albumin patterns during the acute phase were associated with neurological outcome. Methods This retrospective cohort study screened 490 patients with intracranial aneurysms and 498 admissions. After restricting to the index hospitalization for strictly defined aSAH, 370 patients remained. The trajectory cohort included 318 patients with serum albumin measured within 48 h of admission and on at least three different days during days 0–14. Latent class mixed models identified albumin trajectories. The primary outcome was unfavorable neurological status at discharge (modified Rankin Scale [mRS] ≥ 3). Multivariable logistic regression, restricted cubic spline analyses, landmark analyses using early albumin windows, and multiple sensitivity analyses addressing severity, treatment modality, missing data, hospitalization duration, and latent-class uncertainty were performed. Results Three trajectories were identified: high-preserved ( n = 90, 28.3%), high-declining ( n = 148, 46.5%), and low-persistent ( n = 80, 25.2%). Overall, 114 patients (35.8%) had unfavorable neurological outcomes. In the principal complete-case model ( n = 258), compared with the high-preserved group, the adjusted odds ratios were 3.65 (95% CI, 1.57–8.47; p = 0.003) for the high-declining trajectory and 2.48 (95% CI, 1.02–6.02; p = 0.044) for the low-persistent trajectory (global p = 0.007). Early-window landmark analyses showed that higher day-3 and day-7 nadir albumin levels were associated with lower odds of subsequent unfavorable discharge status (OR per 1 g/L increase, 0.86 and 0.84, respectively; both p < 0.001). Treatment-stratified analyses showed no evidence of trajectory-by-treatment interaction ( p = 0.453). Accounting for latent-class uncertainty yielded consistent results, although the low-persistent estimate became less precise. Adding trajectory class to the clinical model resulted in only modest improvement in discrimination (C-statistic 0.754 to 0.776). Conclusion Longitudinal serum albumin patterns were associated with neurological status at discharge after aSAH. However, because trajectory assignment incorporates post-baseline measurements and class separation was moderate, these trajectories should be interpreted as markers of evolving clinical course rather than validated independent predictors. Early landmark analyses improved temporal interpretation but did not eliminate residual confounding or reverse causation. As standardized 90-day mRS was unavailable, these findings should not be extrapolated to long-term functional prognosis.