Clinical EEG slowing correlates with delirium severity and predicts poor clinical outcomes

Abstract

Objective: To determine which findings on routine clinical EEGs correlate with delirium severity across various presentations and to determine whether EEG findings independently predict important clinical outcomes.

Methods: We prospectively studied a cohort of nonintubated inpatients undergoing EEG for evaluation of altered mental status. Patients were assessed for delirium within 1 hour of EEG with the 3-Minute Diagnostic Interview for Confusion Assessment Method (3D-CAM) and 3D-CAM severity score. EEGs were interpreted clinically by neurophysiologists, and reports were reviewed to identify features such as theta or delta slowing and triphasic waves. Generalized linear models were used to quantify associations among EEG findings, delirium, and clinical outcomes, including length of stay, Glasgow Outcome Scale scores, and mortality.

Results: We evaluated 200 patients (median age 60 years, IQR 48.5-72 years); 121 (60.5%) met delirium criteria. The EEG finding most strongly associated with delirium presence was a composite of generalized theta or delta slowing (odds ratio 10.3, 95% confidence interval 5.3-20.1). The prevalence of slowing correlated not only with overall delirium severity (R ² = 0.907) but also with the severity of each feature assessed by CAM-based delirium algorithms. Slowing was common in delirium even with normal arousal. EEG slowing was associated with longer hospitalizations, worse functional outcomes, and increased mortality, even after adjustment for delirium presence or severity.

Conclusions: Generalized slowing on routine clinical EEG strongly correlates with delirium and may be a valuable biomarker for delirium severity. In addition, generalized EEG slowing should trigger elevated concern for the prognosis of patients with altered mental status.

Figure 1. Prevalence of EEG features by delirium severity for all patients

Patients were stratified by delirium severity as indicated in the top row (3-Minute Diagnostic Interview for Confusion Assessment Method severity [3D-CAM-S] scores 0–7). The figure below represents EEG feature data from all patients, with each feature represented by a row and each patient represented by a single column. A black cell indicates the presence of the EEG feature for that patient, whereas a white cell indicates the absence of the feature for that patient. Within each stratum, for display purposes, patients are sorted according to the presence of generalized slowing. GPD = generalized periodic discharges; LPD = lateralized periodic discharges; PDR = posterior dominant rhythm.

Figure 2. Prevalence of generalized EEG slowing was correlated with delirium severity

(A) Patients were stratified by 3-Minute Diagnostic Interview for Confusion Assessment Method delirium severity (3D-CAM-S) scores, and the prevalence of generalized EEG slowing was calculated at each score. Black line indicates fit by linear regression (adjusted R² = 0.907, p < 0.001). Gray vertical lines indicate the bootstrap confidence intervals for each stratum (2.5%–97.5% percentiles of 1,000 bootstraps). (B–E) Patients were additionally stratified by the severity score of each individual 3D-CAM-S delirium feature (1–4), and the prevalence of generalized EEG slowing was calculated at each score. Logistic regression was used to quantify the relationship between delirium feature severity (dependent variable) and EEG slowing (independent variable). Severity of each delirium feature was significantly associated with EEG slowing (all odds ratio [OR] > 1, p < 0.05).

Figure 3. EEG slowing was increased in delirium even in patients with normal levels of arousal

(A) Prevalence of EEG slowing was calculated for patients with 3-Minute Diagnostic Interview for Confusion Assessment Method–defined delirium within 4 levels of arousal. Arousal was stratified by Richmond Agitation Sedation Scale (RASS) scores: −5 to −4 represent coma-like states; −3 to −1 represent hypoactive delirium states; 0 represents an alert and calm state; and +1 to +4 represent hyperactive delirium states. Proportions of EEG slowing did not differ significantly among these 4 strata (Pearson χ² = 6.69, p = 0.076). More specifically, proportion of EEG slowing did not differ between patients with hypoactive and hyperactive levels of arousal (χ² = 0.01, p = 0.921). (B–C) We also compared the prevalence of EEG slowing between patients with and without delirium at normal levels of arousal (B, RASS value of 0; C, Glasgow Coma Scale [GCS] value of 15). EEG slowing was more prevalent among patients who screened positive for delirium than those who screened negative with either measure of normal arousal (B, RASS = 0: χ² = 14.0, p < 0.001; C, GCS = 15: χ² = 5.6, p = 0.018).

Figure 4. EEG slowing and delirium were associated with poor clinical outcomes

Clinical outcomes are shown for patients stratified by delirium status (gray = no delirium, red = delirium) and generalized EEG slowing (lighter shade/− = no EEG slowing; darker shade/+ = with EEG slowing). (A) Both EEG slowing and delirium were associated with increased length of stay (robust rank estimation for linear models/analysis of variance: main effects of EEG slowing F = 17.9, p < 0.001; and delirium F = 11.9, p < 0.001; no significant interaction F = 2.9, p = 0.092). Horizontal black lines depict medians; bars depict interquartile ranges; and thin vertical lines depict ranges (minimum–maximum). Length of stay is plotted on a log scale given the long-tailed distribution. (B) Both EEG slowing and delirium were associated with worse functional outcomes as measured by the Glasgow Outcome Scale (main effects of EEG slowing F = 7.2, p = 0.008; delirium F = 8.8, p = 0.003; no significant interaction F = 0.08, p = 0.774). (C) Rates of mortality differed depending on delirium status and EEG slowing (4-sample test for equality of proportions: χ² = 14.1, p = 0.003). EEG slowing was associated with increased mortality in patients both with and without delirium (χ² and p values reflect post hoc χ² tests between the indicated groups).