Validation of a termination of

resuscitation algorithm in out-of-

hospital cardiac arrest patients with

a non-shockable initial rhythm

receiving advance life support

MD, PhD Markus Skrifvars

Liverpool Hospital ICU, SSWAHS

Study Group

Markus Skrifvars, MD, PhD

Taneli Vayrynen MD

Markku Kuisma MD, PhD

Michael Parr MD, PhD

Maaret Castren MD, PhD

Johan Silfverstolpe MD, PhD

Leif Svensson MD, PhD

Johan Herlitz MD, PhD

Background

Survival after out-of-hospital cardiac arrest (OHCA) with a non-shockable rhythm is poor [1]

In order to identify patients in whom chance of survival is minimal termination of resuscitation (TOR) algorithms have been developed [2]

Most of these algorithms have been developed for emergency medical service (EMS) systems capable of basic life support [3]

[1] Iwami et al. Circulation 2009, [2] Bailey et al. Prehospital Emergency Care 2000, [3] Morrison et al. NEJM 2006

Background

Recently a TOR algorithm developed for EMS systems capable of advance life support (ALS) was been developed and validated [4]

1. Not witnessed by EMS personnel

2. No shock was delivered

3. No return of spontaneous circulation (ROSC) at any point during

the resuscitation

4. No bystander cardiopulmonary resuscitation performed

5. Arrest not witnessed by bystander

If none of the above criteria are present resuscitation can

be terminated

[4] Morrison et al. Resuscitation 2009, [5] Sasson et al. JAMA 2008

Survival from asystole and

PEA in Helsinki, FinlandHelsinki; Finland is a medium sized urban city with a population of 500,000

Prospective Utstein Style data collection of OHCAs since 1994

In two previous papers we examined survival to hospital discharge in patients resuscitated from asystole and PEA [4,5]

Based on these findings we develope a TOR algorithm for patients with OHCA and PEA or

asystole as the initial rhythm[5] Vayrynen et al. Acta Anaesth Scand 2008, [6] Vayrynen et al. Resuscitation 2008

Apply TOR in asystole if…

Unwitnessed arrest

Delay to arrival of

ambulance exceeds

10 minutes

No ROSC within

20 minutes from start

of ALS

Delay to arrival of ambulance exceeds

15 minutes

No ROSC within

20 minutes from start of ALS

In unwitnessed PEA if no ROSC within

10 minutes of ALS

Apply TOR in pulseless electrical

activity if…

Exclusion criteria to the TOR!!

Accidental hypothermia

Drowning

Penetrating trauma

Purpose

To retrospectively validate this Helsinki TOR algorithm and the algorithm by Morrison et al. in a

large OHCA database with survival with good neurological status at 30 days as the end-point

Study material

The Swedish Cardiac Arrest Registry

includes prospectively collected OHCAs

occurring in Sweden since 1990

Approximately 70% of all OHCA are

included annually and presently includes

about 44.000 patients

Data is collected prospectively and

entered into the database centrally

Outcome data is obtained from hospital

charts

Methods

We identified patients with asystole and

PEA as the initial rhythm

Patients with drowning, trauma, drug

overdose or accidental hypothermia were

excluded

Sensitivity, specificity, positive and

negative predictive values including 95%

confidence intervals were calculated for

both TOR algorithms

Results Asystole17.707

patients with Asystole

7.423 Unwitnessed Survival 0.4%

8.237Witnessed

Survival 1.4%

2.901 patients withdelay to ambulance

> 10 minSurvival 0.3%

5.336 patients with delay to ambulance

<10 minSurvival 2.0%

1171 patients with ROSC

Survival 8.2%

518 patients with ROSC delay < 20 min

from start of CPRSurvival 13.7%

653 patients with ROSC delay > 20 min

from start of CPRSurvival 4.4%

Conversion to VF

and

bystander

initiated life

support

Results PEA3.104 patients

with PEA

790 patients withUnwitnessed Survival 0.6%

14 patients with ROSC within 10 min from

start of CPRSurvival 7%

2.050 patients with Witnessed

Survival 1.5%

40 patients with ROSC exceeding 10 min from

start of CPRSurvival 0.2%

1.493 patients with delay to ambulance

less than 15 minSurvival 1.9%

460 patients with delay to ambulance exceeding 15 min

Survival 0.2%

209 patients with ROSC

Survival 8.6%

163 patients with ROSC delay < 20 min

Survival 11%

46 patients with ROSC > 20 min

Survival 0%

Performance of the Helsinki

TORAlive at 1 month Died before 1 month

Discontinue CPR 45 11,074

Continue CPR 162 9,631

Sensitivity 54% (53-55)

Specificity 78% (72-84)

Negative predictive value 1.7

Positive predictive value 99.6

Survival when algorithm recommends terminating 0.4%

Proportion of continued resuscitations 54%

95% Confidence intervals in parenthesis

Performance of TOR by

Morrison et al.

Alive at 1 month Died before 1 month

Discontinue CPR 5 5,465

Continue CPR 202 15,033

Sensitivity 26.7 (26.1-27.2)

Specificity 97.4 (94.4-99.0)

Negative predictive value 1.5

Positive predictive value 99.9

Survival when algorithm recommends terminating 0.1%

Proportion of continued resuscitations 74%

95% Confidence intervals in parenthesis

Discussion

The present TOR algorithm including special time

limits depending on the corresponding rhythm

worked well in patients with PEA (survival 0.1%)

In patients with asystole the survival of patients in

whom the TOR recommended discontinuing CPR

was low (0.4%) but some survivors where seen; The majority of these received bystander BLS or had conversion into

VF during the course of CPR

This study implies that survival of some patients is due to

other factors than just time delays

Problemce with rhythm interpretation?

Pseudo PEA or asystole?

Discussion

The TOR algorithm by Morrison et al. worked

very well in this large OHCA database with

only patients with non-shockable rhythms

(survival 0.1%)

However, the amount of continued

resuscitations was also higher (74% vs. 54%)

which will translate into more patients needing

intensive care

ConclusionTHe Morrison et al. algorithm seems to be

more applicable in practise and can be

recommended for use

The Helsinki TOR seems clinically justified

with limitations Correct identification of the initial rhythm

Conversion of asystole into VF?

Accuracy of measured time delays?

Bystander initiated basic life support?

Which one to use will depend on the EMS

system management protocols