Author(s): Steven C Brooks, Laurie J Morrison
Date:
Question: Mechanical chest compressions compared to manual chest compressions for cardiac arrest
Quality assessment
№ of
studies
Study
design
Risk of
bias
Inconsistency
Indirectness
№ of patients
Imprecision
Other
considerations
Mechanical
chest
compressions
n/N (%)
manual chest
compressions
n/N (5%)
Effect
Relative
Risk
(95% CI)
Quality
Importance
Absolute
(95% CI)
Survival to hospital discharge with good neurological function (assessed with: CPC scores or MRS )
3
randomised
trials
serious
1
not serious
not serious
not serious
none
not
pooled
⨁⨁⨁
◯
CRITICAL
MODERATE
Hallstrom
2006
12/391 (3.07)
28/371 (7.55)
0.41
(0.210.79)
Wik 2014
87/2099 (4.14)
112/2132
(5.25)
0.79
(0.601.03
108/1300
(8.31)
100/1289
(7.76)
1.07
(0.831.39)
105/1300
(8.1%)
94/1289
(7.3%)
Rubertsson
2013
Survival to 30 days with good neurological outcome (assessed with: CPC scores)
1
randomised
trials
serious
2
not serious
not serious
not serious
none
Rubertsson
2013
1.11
(0.841.45)
⨁⨁⨁
◯
CRITICAL
MODERATE
Survival to 60 days with good neurological function (assessed with: CPC scores)
0
see comment
not
pooled
98/1289
(7.6%)
1.11
(0.861.45)
see
comment
CRITICAL
Survival to 180 days with good neurological outcome
1
randomised
trials
serious
2
not serious
not serious
not serious
none
110/1300
(8.5%)
Rubertsson
2013
⨁⨁⨁
◯
CRITICAL
MODERATE
Survival to 1 year with good neurological outcome
0
Survival to hospital discharge
see comment
not
pooled
see
comment
CRITICAL
Quality assessment
№ of
studies
Study
design
Risk of
bias
Inconsistency
Indirectness
№ of patients
Imprecision
Other
considerations
Mechanical
chest
compressions
n/N (%)
manual chest
compressions
n/N (5%)
Effect
Relative
Risk
(95% CI)
Quality
Importance
Absolute
(95% CI)
Survival to hospital discharge with good neurologic outcome (GCS 1-2 or mRS 0-3)
5
randomised
trials
serious
34
not serious
5
not serious
not serious
none
not
pooled
⨁⨁⨁
◯
CRITICAL
MODERATE
Hallstrom
2006
23/394 (5.84)
37/373 (9.92)
0.59
(0.360.97)
Lu 2010
25/76 (32.89)
11/74 (14.86)
2.21
(1.174.17)
Rubertsson
2013
117/1300 (9)
118/1289
(9.15)
0.98
(0.771.25)
6/75 (8)
7/72 (9.72)
0.82
(0.292.33)
196/2099
(9.34)
233/2132
(10.93)
0.85
(0.711.02
Smekal
2011
Wik 2014
Survival to 30 days
2
randomised
trials
serious
27
not serious
not serious
not serious
none
Not
pooled
⨁⨁⨁
◯
CRITICAL
MODERATE
Perkins
2014
Rubertsson
2013
104/1652 (6.3)
193/2819
(6.85)
0.92
(0.731.16)
112/1300
(8.62)
109/1289
(8.07)
1.02
(0.791.31)
see comment
not
pooled
see
comment
104/1289
(8.1%)
not
estimable
not
estimable
Survival to 60 days
0
CRITICAL
Survival to 180 days
1
Rubertsson
randomised
trials
serious
2
not serious
not serious
not serious
none
111/1300
(8.5%)
⨁⨁⨁
◯
CRITICAL
Quality assessment
№ of
studies
Study
design
Risk of
bias
Inconsistency
№ of patients
Indirectness
Imprecision
Other
considerations
Mechanical
chest
compressions
n/N (%)
manual chest
compressions
n/N (5%)
Effect
Relative
Risk
(95% CI)
Quality
Importance
Absolute
(95% CI)
2013
MODERATE
Survival to 1 year
1
randomised
trials
serious
7
not serious
not serious
not serious
none
89/1652
(5.4%)
175/2819
(6.2%)
not
estimable
not
estimable
Perkins
2014
⨁⨁⨁
◯
CRITICAL
MODERATE
Return of spontaneous circulation
7
randomised
trials
serious
137
serious
6
not serious
7
not serious
none
not
pooled
see
comment
⨁⨁◯
◯
IMPORTANT
LOW
Dickinson
1998
1/7 (14.29)
0/10 (0)
Halperin
1993
8/17 (47.06
3/17 (17.65)
2.67
(0.858.37)
42/76 (55.26)
28/74 (37.84)
1.46
(1.022.08)
446/1289
(34.60)
1.02
(0.921.14)
23/72 (31.94)
1.27
(0.821.96)
Lu 2010
460/1300
(35.38)
Rubertsson
2013
30/74 (40.54)
Smekal
2011
n/a
Wik 2014
600/2099
(28.59)
689/ 2132
(32.32)
0.88
(0.810.97)
Perkins
2014
522/1652
(31.60)
885/2819
(31.39)
1.01
(0.921.10)
MD – mean difference, RR – relative risk
1.
Problems with allocation concealment, blinding of outcome assesors. Hallstrom study: 99.3% of participants entered into the study had full follow-up data reported. Cointervention bias: Site C of the study changed the protocol halfway through the study; this involved applying the device to participants after a period of CPR and rhythm
analysis. This change in CPR technique is likely to have had an impact on outcomes for participants treated at this site with mechanical CPR. Rubertsson study: 1.
Control arm follows 2005 resuscitation guidelines. Outdated comparator. 2. Co-intervention: Treatment arm includes a novel study-specific protocol including defibrillation
without pulse/rhythm checks not scientifically validated and may have an impact on the outcomes measured independent of the mechanical CPR intervention. 3. CPR
quality measured in only 10% of the sample. This makes it difficult to interpret the nature of the control group and whether this was a valid control reflective of state-ofthe-art contemporary care. 4. 5% of enrolled patients had a protocol violation related to the inclusion or exclusion criteria but were included in the intension-to-treat
analysis. 5. 5% of patients had some type of contamination (i.e. received treatment for which they were not allocated).
2.
3.
4.
5.
6.
7.
In the Rubertsson study -1. Control arm follows 2005 resuscitation guidelines. Outdated comparator. 2. Co-intervention: Treatment arm includes a novel study-specific
protocol including defibrillation without pulse/rhythm checks not scientifically validated and may have an impact on the outcomes measured independent of the
mechanical CPR intervention. 3. No measure of quality of manual CPR in either group. This makes it difficult to interpret the nature of the control group and whether this
was a valid control reflective of state-of-the-art contemporary care. 4.. The 4-hour end-point lacks rationale for the being the primary outcome. It is not a commonly
excepted outcome and not validated in other prehospital arrest trials. 5. 5% of enrolled patients had a protocol violation related to the inclusion or exclusion criteria but
were included in the intension-to-treat analysis. 6. 5% of patients had some type of contamination (i.e. received treatment for which they were not allocated).
Issues of co-intervention with novel and unconventional CPR sequences in the mechanical chest compression arms . Blinding of the providers and patients not possible
given the nature of intervention. Outcome assessors were generally not blinded.
Dickinson study: Quasi-randomisation based on odd/even days. Quasi-randomisation did not allow for allocation concealment. Co-intervention issue: Both groups of
participants placed on Thumper back-board. HCPR (control) group had machine running but not applied to participant to allow human compressions to be delivered at the
same rate. Halperin study: Allocation sequence concealment: Sequenced envelopes used. Unclear whether these envelopes were opaque; no mention of a randomisation
log to track any subterfuge of the randomisation process.
Although there was considerable inconsistency in the direction of results from included studies when considering individual study results, the inconsistency was generated
mostly by inclusion of smaller studies and studies with recognized sources of bias. The larger studies, Wik and Rubertsson demonstrate equivalence and no difference,
respectively. The Hallstrom study demonstrated harm but has methods issues that likely contributed to this (see explanation 1)
Dickinson, Halperin and Lu suggest benefit, Rubersson, Smekal, Wik and Perkins do not observe a difference.
In the Perkins study - 1. 60 % of the intervention group did not receive any mechanical chest compressions. 2. No CPR quality measures were instituted. 3. Unclear if
training and quality assurance was sufficient for EMS crews using the mechanical device and for manual CPR in control group.