delirium detection in intensive care patients willemijn van der kooi department of intensive care...

Delirium detection in Intensive Care patients

Willemijn van der KooiDepartment of Intensive Care MedicineUniversity Medical Center Utrecht, The Netherlands

• Orion Pharma: contributed to printing costs of my thesis

• NPK design: contributed to printing costs of my thesis

Disclosures

Delirium prevalence:•50%-80% for ICU patients•10-15% for cardiac surgery patients

ICU delirium is associated with:•Long term cognitive impairment•Increased hospital and ICU length of stay•Increased mortality

Introduction

* Actor

Delirium often (71%) missed by ICU physicians1

• questionnaires developed for screening

Daily practice• Sensitivity of questionnaire with best performance

(Cam-ICU):–47% in ICU patients2

–28% in post-operative patients3

• Cognitive screening may not fit well in the culture of the ICU

Introduction

1 Van Eijk et al. Crit Care Med 2009;37:1881-52 Van Eijk et al. Am J Respir Crit Care Med 2011;184:340-43 Neufeld et al. Br J Anaesth 2013;111:612-8

New approach: delirium detection using physiological alterations

Introduction

Ultimate goal:•2 sensors coupled to a monitor•Monitor shows on a scale the chance of having delirium

Three physiological parameters studied:•Temperature variability•Eye movements•Brain activity (EEG)

Future perspective

Content

Temperature variability during delirium in ICU patients

Van der kooi et al. PLoS One. 2013; 8:e78923

Delirium: manifestation of encephalopathy •In delirium tremens, Wernicke encephalopathy and schizophrenia: temperature regulation is disturbed•Does delirium affect thermoregulation?

Introduction

To investigate whether:

•ICU delirium is related to absolute body temperature

•ICU delirium is related to temperature variability

Aim of the study

• Subjects from 3 previous delirium studies• Daily delirium assessments by research-

nurse/physician

Temperature: measured per minute 24/7

Methods

Inclusion: •Patients with delirious + non-delirious days during ICU admission of >24 hrs

Exclusion criteria: • Disturbed body temperature regulation

(treatment/diagnoses)• Neurological/neurosurgical disease• Days with sepsis, coma or death were excluded

from analysis

*All patients received paracetamol 1000 mg 4 times daily

Methods

Methods

day 1 day 2 day 335

36

37

38te

mpe

ratu

re ( C

)

day 1 day 2 day 30

0.5

1

1.5

2

tem

p ac

c.(

C/m

in2 )

day 1 day 2 day 30.0

0.05

0.1

0.15

0.2

0.25

mea

n te

mp

acc.

(C

/min

2 )

A

B

C

Coma No Delirium Delirium

Linear Mixed models: •Univariable (unadjusted) •Multivariable (adjusted for confounders RASS and SOFA)

Outcome: •body temperature [°C]•temperature variability (absolute second derivative) [°C/min2]

Methods

Results

Results

Patient characteristics

Age: mean (SD) 68 (14)

Gender: number of males (%) 15 (63%)

Admission type: number (%)  

-internal medicine 3 (12%)

-surgery 12 (50%)

-cardiothoracic surgery 9 (38%)

Delirium type: number (%)  

-Hypoactive 6 (25%)

-Hyperactive 0 (0%)

-Mixed type 18 (75%)

Number of analyzed days: median (IQR)

-Delirium 2.0 (1.0)

-Non-delirium 1.0 (1.8)

Body Temperature:

Results

Model Variable Effect estimate 95% Confidence interval p-value

Unadjusted

Delirium [yes] -0.03 -0.17; 0.10 0.61

Adjusted

Delirium [yes] -0.03 -0.17; 0.10 0.63

Rass 0.01 -0.09; 0.10 0.90

Sofa 0.001 -0.04; 0.04 0.95

Temperature Variability:

Results

Model Variable Effect estimate 95% Confidence interval p-value

Unadjusted

Delirium [yes] 0.005 0.003; 0.008 <0.001

Adjusted

Delirium [yes] 0.005 0.002; 0.008 <0.001

Rass -0.001 -0.003; 0.001 0.20

Sofa -0.0001 -0.001; 0.001 0.71

Strengths:

•Delirium diagnoses prospectively

•Within subjects comparisons

•Easy method temperature variability

Limitations

•Possible effect of medication

•Natural circadian rhythm bias

Discussion

Temperature variability: increased during delirium in ICU patients • encephalopathy that underlies delirium

Future studies: •Monitoring temperature variability in total ICU population •Combine with EEG for objective tool to detect delirium

Discussion

Delirium detection based on monitoring of blinks and eye movements

Van der kooi et al. Am J Geriatr Psychiatry. 2014

Delirium associated with change in motor level activity•Actigraphy not practical•Eye movements less affected by muscle weakness, restraints, pain

Introduction

Goal

Determine whether eye blinks and eye movements differ in patients with delirium compared to patients without delirium.

Methods

Population: post-cardiac surgery patientsReference: psychiatrist, geriatrist, neurologist using DSM

4 criteria

Methods

Standard 21 electrode EEG recording (30 minutes) with periods of eyes open and closed

First artifact free minute selected with eyes closed and open

Methods: Eye movements

Eye movements compared between delirium and non-delirium

Number (per min) and duration (sec) of:•Blinks

•Vertical eye movements

•Horizontal eye movements

Results: study population

Delirious patients (n=28)

Non-delirious patients (n=28)

p-value

Age, mean (SD) 76 (5.6) 74 (8.6) 0.16

Gender: male, n (%) 16 (57%) 16 (57%) 1

Apache IV score, median (IQR) 58 (45-65) 43 (35-51) <0.01

Charlson comorbidity index,

median (IQR)

2 (1-3) 1 (0-1) 0.02

Haloperidol use past 24 hours

n (%)

17 (61%) 2 (7%) <0.01

Postsurgical day of EEG,

median (IQR)

3 (2-5) 3 (2-4) 0.78

Results: eye movements

Variable DeliriumMedian (IQR)

Non-deliriumMedian (IQR)

p-value

Number of Vertical eye movements

(min-1)

1 (0-13) 15 (2-54) 0.01

Number of Blinks (min-1) 12 (5-18) 18 (8-25) 0.02

Duration of Blinks (s) 0.50 (0.36-

0.96)

0.34 (0.23-

0.53)

<0.01

Eyes Open

Results: eye movements

Variable DeliriumMedian (IQR)

Non-deliriumMedian (IQR)

p-value

Duration of Horizontal eye

movements (s)

0.41 (0.15-

0.75)

0.08 (0.06-

0.22)

<0.01

Eyes Closed

Results: Eye movements haloperidol

Eyes Variable Delirium with haloperidolMedian (IQR)

Delirium without haloperidol Median (IQR)

p-value

Open  Number of vertical eye

movements

2 (0-17) 0 (0-17) 0.69

Open  Number of blinks

12 (4-19) 12 (6-17) 0.87

 Open Duration of blinks (s) 0.49 (0.39-1.01) 0.52 (0.34-0.93) 0.81

 Close

d

Duration of horizontal of

eye movements (s)

0.59 (0.23-1.40) 0.27 (0.13-0.69) 0.19

Conclusion

Especially blinks are affected in delirious patients

Strengths: •non-invasive •Only 1 minute of data necessary

Limitations:•22 electrodes needed for eye movement measurement, except for blinks• Difference in Apache and Charlson Comorbidity score

Future studies: •Detection of eye movements in general population of ICU patients•Determining whether eye movements can detect delirium at early stage

Delirium detection using EEG: what and how to measure?

Van der kooi et al. Chest. 2014

Delirium characterized by EEG abnormalities•EEG not practical

Introduction

Without Delirium With Delirium

Goal

Determine the electrode derivation and EEG characteristic that have the best capability of discriminating delirium from non-delirium

Methods

Standard 21 electrode EEG recording (30 minutes) with periods of eyes open and closed

First artifact free minute selected with eyes closed

Methods: EEG

Eyes closed= 210 different derivations

Methods: EEG

For every derivation 6 parameters: 1

Relative delta power (0.5-4 Hz), Relative theta power (4-8 Hz),Relative

alpha power (8-13 Hz), Relative beta power (13-20 Hz), Peak frequency, Slow-

fast ratio

1van der Kooi, et al. J Neuropsychiatry Clin Neurosci 2012; 24: 472-477.

Ruwe EEG

δ 0-4 Hz

θ 4-8 Hz

α 8-13 Hz

β 13-20 Hz

Methods: EEG

210 derivations x 6 parameters = 1260 combinations

All 1260 combinations • Compared between delirium and non-delirium (Mann-whitney U)

• P-values ranked

• smallest p-value is optimal combination (Bonferoni correction )

1van der Kooi, et al. J Neuropsychiatry Clin Neurosci 2012; 24: 472-477.

Results: EEG

Eyes closedRank

p-value* Deriviation

Parameter

1 1.8e-12 F8-Pz Relative δ2 3.7e-12 F8-P3 Relative δ3 1.1e-11 F8-O2 Relative δ4 1.5e-11 Fp2-O1 Relative δ5 1.7e-11 F8-F4 Relative δ6 2.2e-11 F8-O1 Relative δ7 2.4e-11 F8-Cz Relative δ8 2.4e-11 F8-C3 Relative δ9 2.9e-11 Fp2-Pz Relative δ10 3.0e-11 Cz-O1 Relative δ

*p< 4.0*10-5 is significant

Delirium Non-delirium

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Rel. d

elta

pow

er

F8-Pz

Results: EEG

Most optimal

electrode

locations, based

on first 4

rankings.

Conclusion

EEG easily detects delirium from non-delirium using •2 electrodes in frontal-parietal derivation and relative delta power

Strengths: new approach, non-invasive, only 2 electrodes and 1 minute data necessary

Future studies: •Validation study in unselected population of postoperative- and critically ill patients•Determine whether it recognizes delirium at an early stage

Future Directions

Overall Conclusion

EEG most promising method for delirium detection.

Project started: Development of delirium monitor

Product development

Product and algorithm

Validation study

Goal: To determine sensitivity, specificity and predictive values of the delirium monitor when compared to reference standard (specialized geriatric nurse) in elderly postoperative patients (n=154).

Usability study

• Practical?• Easy to Use?• Opinion of nurses of different medical departments

Extra slides

Results: EEG eyes open

Ogen OpenRang

p-waarde*

Afleiding

Parameter

1 2.0e-07 P7-P4 Relative alpha2 4.2e-07 P3-P4 Relative alpha3 1.6e-06 P7-O1 Relative delta4 3.2e-06 P7-O1 Relative alpha5 3.5e-06 P3-P4 Slow Fast ratio6 4.0e-06 P4-O1 Relative alpha7 6.1e-06 P7-P8 Relative alpha8 7.9e-06 P7-P4 Slow Fast ratio9 9.4e-06 P3-P8 Relative alpha10 1.1e-05 P7-O2 Relative alpha

*p< 5.6*10-4 is significant

Delirium met/zonder haloperidol geen

verschil (p=0.37)

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

0.55

D NDre

l. al

pha

powe

r

0.1

0.2

0.3

0.4

0.5

0.6

D ND

rel.

alph

a po

wer

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

D ND

rel.

delta

pow

er

1: P7-P4 2: P3-P4 3: P7-O1

Results: Eye movements eyes openEyes

Variable DeliriumMedian (IQR)

Non-deliriumMedian (IQR)

p-value

AUC (95% CI) 

Ope

n

Number of eye

movements

  Horizontal

6 (0-51)

n=23

26 (0-55)

n=28

0.54 0.55 (0.39-

0.71)

  Vertical

1 (0-13)

n=23

15 (2-54)

n=28

0.01 0.70 (0.55-

0.85)

  Blinks

12 (5-18)

n=23

18 (8-25)

n=27

0.02 0.65 (0.50-

0.80)

Ope

n

Duration of eye

movements (s)

  Horizontal 0.24 (0.10-

0.56)

n=14

0.14 (0.04-

0.27)

n=17

0.14 0.66 (0.47-

0.85)

  Vertical 0.14 (0.06-

0.49)

n=10

0.07 (0.04-

0.60)

n=18

0.46 0.59 (0.37-

0.81)

  Blinks 0.50 (0.36-

0.96)

n=20

0.34 (0.23-

0.53)

n=27

<0.01 0.74 (0.59-

0.88)

Results: Eye movements eyes closedEyes Variable Delirium

Median (IQR)

Non-deliriumMedian (IQR)

p-value

AUC (95% CI) 

Closed Number of eye

movements

  Horizontal

0 (0-42)

n=27

0 (0-51)

n=27

0.37 0.57 (0.41-

0.72)

  Vertical

5 (0-47)

n=27

10 (0-52)

n=27

0.40 0.56 (0.41-

0.72)

Closed Duration of eye

movements (s)

  Horizontal 0.41 (0.15-

0.75)

n=12

0.08 (0.06-

0.22)

n=13

<0.01 0.81 (0.64-

0.99)

  Vertical 0.15 (0.07-

0.29)

n=15

0.07 (0.03-

0.27)

n=17

0.19 0.64 (0.44-

0.84)

Results: Eye movements haloperidolEyes Variable Delirium with

haloperidolMedian (IQR)

Delirium without haloperidol Median (IQR)

p-value

Number of eye

movements

Open  Vertical  2 (0-17)

n=14

0 (0-17)

n=9

0.69

Open  Blinks

12 (4-19)

n=14

12 (6-17)

n=9

0.87

Duration of eye

movements (s)

 Open Blinks 0.49 (0.39-1.01)

n=14

0.52 (0.34-0.93)

n=9

0.81

 Close

d

Horizontal 0.59 (0.23-1.40)

n=6

0.27 (0.13-0.69)

n=6

0.19

Stap1 Van onderzoek naar klinische prakti

• Ontwikkeling van delirium monitor– Product– Algoritme

• Validatie studie• Gebruiksvriendelijkheids-

studie

Validatie studie

Doel:

Het bepalen van de sensitiviteit, specificiteit en voorspellende waarden van de delirium monitor in vergelijking met de referentie standaard in oudere postoperatieve patiënten (n=154).

Validatie studie

Inclusie:•≥ 70 jaar•Opname voor grote operatie (min. 2 opname dagen ZH na operatie)•Preoperatieve verhoogde kwetsbaarheid en/of verhoogd risico op delirium

Exclusie:•Geen communicatie mogelijk•Neurologische chirurgische ingreep•Eerdere deelname studie•Patient in isolatie vanwege resistente bacterie

Validatie studie - Studie verloop

OperatieT0 T1 T2 T3

= Delirium monitor

= Referentie standaard

= POS Geriatrische screening

Validatie studie

Delirium monitor•4 elektrodes•5 minuten EEG meting OD•Relatieve δ power

Referentie standaard onderzoeker/vpk•DRS-R-98 Ernst van delirium•VAS (0-10) Kans dat patiënt delirant is•Classificatie Deliriant/Mogelijk delirant/Niet delirant

(Op basis van DSM-V criteria)

Validatie studie - Analyses

• 1e artefact vrije minute relatieve δ power• ROC curve relatieve δ power vs. classificatie

van referentie standaard

Validatie studie - Secundaire doelen

1) Schaal voor ernst van delirium (relatieve δ vs. DRS-R-98)

2) Vroegtijdig herkennen van delirium?

0 1 2 4 5 6 7 8 9 103

OperatieT0 T1 T2 T3

= Delirium monitor

= Referentie standaard

= Geriatrische screening

Stap2 Van onderzoek naar klinische praktijk

• Gebruiksvriendelijkheidsonderzoek– Handig product?– Ervaring verpleegkundige

Stap3 Van onderzoek naar klinische praktijk

• Delirium monitor bredere doelgroep– Dementie– Neurotrauma– IC: Effect sedatie op EEG

Samenvatting

1) EEG in delirium studie = het idee– Relatieve δ power– Frontaal- Pariëtaal

2) Ontwikkeling prototype3) Validatiestudie 4) Gebruiksvriendelijkheidsstudie5) Hoe krijgen we het naar de IC

Delirium monitor project

UMCU - IC•Arjen Slooter•Willemijn van der Kooi•Tianne Numan•Annemieke Hoekman

Pontes Medical•Rutger van Merkerk

NPK design•Tessa Souhoka•Marlies van Dullemen•Jos Oberdorf

Medische Techniek•Leonard van Schelven•Rene van de Vosse•Bert Westra•Maurice Konings

Geriatrie•Marielle Emmelot-Vonk•Jolanda Peijster- de Waal•Marcel Weterman

KNF•Geert-Jan Huiskamp•Frans Leijten