Electroencephalography and delirium in the postoperative period

Abstract

Delirium commonly manifests in the postoperative period as a clinical syndrome resulting from acute brain dysfunction or encephalopathy. Delirium is characterized by acute and often fluctuating changes in attention and cognition. Emergence delirium typically presents and resolves within minutes to hours after termination of general anaesthesia. Postoperative delirium hours to days after an invasive procedure can herald poor outcomes. Easily recognized when patients are hyperactive or agitated, delirium often evades diagnosis as it most frequently presents with hypoactivity and somnolence. EEG offers objective measurements to complement clinical assessment of this complex fluctuating disorder. Although EEG features of delirium in the postoperative period remain incompletely characterized, a shift of EEG power into low frequencies is a typical finding shared among encephalopathies that manifest with delirium. In aggregate, existing data suggest that serial or continuous EEG in the postoperative period facilitates monitoring of delirium development and severity and assists in detecting epileptic aetiologies. Future studies are needed to clarify the precise EEG features that can reliably predict or diagnose delirium in the postoperative period, and to provide mechanistic insights into this pathologically diverse neurological disorder.

Keywords: delirium; electroencephalography; encephalopathy.

Fig 1

Visual grading scheme of EEG patterns observed during hepatic encephalopathy. (A) Eyes closed wakefulness produces waves in the occipital EEG derivations of most normal individuals. These ‘Berger waves’ represent the posterior dominant rhythm, here shown as 9–10 Hz oscillations in the α band (blue). The dominant frequency is altered during the delirium of hepatic encephalopathy. Grade A: posterior α waves are replaced by higher frequency β activity. (B) Grade B: ‘unstable α’, with periodic replacement with θ waves of 5–7 Hz. (C) Grade C: α with runs of θ waves 5–6 Hz. (D) Grade D: persistent θ activity despite eye opening. (E) Persistent oscillations in the δ band. Figures were modified from Parsons-Smith and colleagues, reproduced with permission, Elsevier.

Fig 2

EEG patterns observed during sepsis-associated encephalopathy. (A) Oscillations (5–6 Hz) in the θ band of a 51-yr-old man with confusion during a Salmonella infection. (B) Frontal intermittent rhythmic δ activity superimposed on θ oscillations. These were recorded from a 62-yr-old woman infected with streptococcal pneumonia. (C) δ oscillations (0.5–1 Hz) with little background activity in a comatose 68-yr-old man afflicted with a Pseudomonas infection. (D) Triphasic waves with minimal background activity recorded from a 74-yr-old man with bacteraemia. Shared voltage and time scales are noted. Figures were modified from Young and colleagues, reproduced with permission, Clinical and Investigative Medicine (CIM).

Fig 3

EEG spike-and-wave complexes can be observed during non-convulsive status epilepticus. (A) Spike-and-wave complexes arise at slightly higher frequencies (2–3 Hz) than triphasic waves (Fig. 2D), as shown in frontal EEG traces. These epileptiform discharges reflect underlying non-convulsive status epilepticus in this poorly responsive patient without generalized myoclonic activity. (B) Spectrograms show the distribution of EEG power as functions of frequency and time; here, high amplitude is shown as red, intermediate as yellow or green, and low as blue. Low-frequency (∼3 Hz) power, corresponding to the frequency of the spike-and-wave complexes, is abolished (*) after administration of midazolam (▾). The white trace indicates the spectral edge frequency 95%. Figures are adapted from Hernández-Hernández and Fernández-Torre, reproduced with permission, Elsevier.

Fig 4

Low-frequency EEG predominance during delirium appears to resolve after recovery. (A) During delirium in the postoperative period, high-amplitude low-frequency activity can be seen in the EEGs of frontal (F7-Cz, red) channels, particularly with eyes closed. High-frequency activity present with eyes open may be related to activity of the frontalis muscle underneath the recording electrodes. A predominance of low-frequency activity is present in the occipital EEG (O1-Cz, green) during both eyes open (top panel) and eyes closed conditions (bottom panel). (B) After recovery from delirium, low-frequency activity becomes less prominent, and higher frequency components return during eyes open in both occipital and frontal channels. (C and D). Occipital EEGs in Fig. 3A and B are shown at a magnified scale. Low-frequency EEG oscillations can be seen during delirium (C), for eyes open (left panel) and eyes closed recordings (right panel). After resolution of delirium (D), higher frequency activity dominates during eyes open conditions (left panel). Note the return of occipital α activity during eye closure (right panel). Diagnosis of delirium was evaluated using the Confusion Assessment Method. This 62-yr-old man did not have an obvious aetiology of delirium after a mitral valve replacement and Maze procedure. Clinical details of this patient are available in Supplementary Data. All EEG signals have undergone bandpass filtering for 0.3–50 Hz. Shared voltage and time scales are noted for a and b and for c and d.

Fig 5

Power spectrograms of eyes closed occipital EEGs during and after delirium. (A) Comparison spectra of relative power are provided for two different patients in the postoperative period. Delirium was present in one patient (red) and absent in another (blue), as diagnosed with the Confusion Assessment Method. Relative power is the proportion of power in a given frequency range relative to the total power. This metric is commonly used in quantitative EEG evaluation for both postoperative delirium, , , and hepatic encephalopathy., A peak (→) in occipital α (8–13 Hz) is clear in the patient without delirium but is not prominent in the patient with delirium. Comparisons of the two spectra show a greater proportion of power in the δ (<4 Hz) band during delirium. Differences in the θ (4–8 Hz) band are unclear. The EEGs of the patient with delirium also had a lower proportion of power in the α band and at frequencies >13 Hz. (B and C) The consistency of these power distributions over time is shown in the spectrograms of EEGs acquired in the presence (B) or absence of delirium (C). The rhythm is in the low α range (→) and attenuates with eye opening, and is referred to as the posterior dominant rhythm. Clinical details of these patients are available in Supplementary Data. (D) Power spectra of longitudinal EEG recordings of a third patient (same as in Fg. 3) demonstrate a shift of power during the resolution of delirium from postoperative day 1 (red); signals acquired upon recovery (blue) show a reduction in δ and increase in θ and α band relative power. (E and F) Spectrograms of relative power in the EEGs acquired while delirium was present (E) and after resolution (F). Black markers above the spectrograms show the times of traces presented in Fig. 4C and D. Despite greater power in the θ and α bands, a distinct peak is not apparent after recovery. Spectrograms were generated using Chronux Matlab subroutines and multitaper spectral analysis to estimate power spectra (1–20 Hz bandwidth, 10 s moving window at 1 s increments, five tapers, time bandwidth of 3 s). For purposes of presentation, the relative power has been transformed by log base 10.