populations TFQO: Professor Judith Finn EVREVs: Dr Janet Bray and - - PowerPoint PPT Presentation

Dallas 2015

TFQO: Professor Judith Finn EVREVs: Dr Janet Bray and Marion Leary #COI Taskforce: ETI

BLS training for high risk populations

Dallas 2015

COI Disclosure

(specific to this systematic review)

Commercial/industry

Leary co-ownership of Resuscor, LLC

Potential intellectual conflicts

Leary co-author on Blewer 2012

Dallas 2015

2010 CoSTR

In lay providers requiring BLS training (P), does focusing training on high risk populations (I) compared with no such targeting (C) increase

• utcomes (eg. bystander CPR, survival etc) (O)

16 studies met criteria for review. Of these, 1 was published in Korean, with no English translation available. Overall, it seems as though CPR training in family members of high risk patients may improve the rate of bystander CPR seen when the bystanders are relatives of the victims, but it doesn’t improve it to beyond the levels seen by strangers. However, it may not be feasible to train the numbers of people potentially required to take CPR in order to effect an improvement in survival. In addition, there is very conflicting data on the potential effects (either positive or negative) on psychological adjustment in relatives of high-risk patients who take CPR training.

Dallas 2015

C2015 PICO

Population: people at high-risk of OHCA Intervention: focused training of likely rescuers (eg family or care-givers) Comparison: no such targeting Outcomes: survival with favorable neurological outcome at discharge, ROSC, bystander CPR performance, number of people trained in CPR, willingness to provide CPR

Dallas 2015

Inclusion/Exclusion & Articles Found

Inclusions: adults and children

Excluded studies: that did not directly address the PICO question, only examined

psychological outcomes, reviews, editorials, commentaries, abstracts only, duplicate data Reviewed 1563 abstracts with 29 articles included in the review (1 additional paper in Korean excluded –authors did not reply to contact and no English translation was available)

Dallas 2015

2015 Proposed Treatment Recommendations

Based on moderate-low quality of evidence, we recommend training likely rescuers (e.g. family or care-givers) of high-risk populations in CPR – based on the willingness to be trained and the fact that there is low harm and high potential benefit.

There is low to moderate quality of evidence in 29 studies related

to CPR training in likely rescuers (e.g. family or care-givers) of high-risk OHCA groups, with no strong evidence of improved patients outcomes.

The recommendation for providing training in this group places

higher value on the potential benefits of patients receiving bystander CPR by a family-member or caregiver in the case of cardiac arrest, and the willingness of this group to be trained.

We place lesser value on associated costs, and the potential that

skills may not be retained without on-going CPR training. The cost

• f training are potentially reduced with CPR training self-

instruction kits (e.g. DVD training).

Dallas 2015

Risk of Bias in studies

Study Year Design Total ¡ Patients Industry ¡ Funding Allocation: ¡Generation Allocation: ¡Concealment Blinding: ¡Participants Blinding: ¡Assessors Outcome: ¡Complete ¡ Outcome: ¡Selective Other ¡Bias

Blewer 2012 RCT 406 no Low Low Low Unclear Low High Low Brannon 2009 RCT 23 unclear High High Low Low Low Low Unclear Dracup ¡ 2000 RCT 335 no unclear High High Unclear High Unclear Low Dracup ¡ 1986 RCT 134 no unclear High High Unclear Low ¡ High Low Dracup ¡ 1998 RCT 480 no High High High Unclear Low unclear Low Greenberg ¡ 2012 RCT 162 no Low Low High Unclear Low Unclear Low Moser 1999 RCT 335 unclear unclear high high Unclear High High Low

RCTs

Dallas 2015

Risk of Bias in studies

Study Year Design Total ¡ Patients Industry ¡ Funding Eligibility ¡Criteria Exposure/Outcome Confounding Follow ¡up Barr 2013 Non-­‑RCT 126 no high low high low Dracup 1989 Non-­‑RCT 83 unclear unclear low low low Dracup 1994 Non-­‑RCT 238 no low low unclear uncler Dracup 1998 Non-­‑RCT 94 no unclear unclear low low Haugk 2006 Non-­‑RCT 115 Yes, ¡partly unclear low unclear High Higgins 1998 Non-­‑RCT no low ¡ high unclear unclear Khan 2010 Non-­‑RCT 300 no low unclear unclear unclear Kliegal 2000 Non-­‑RCT 195 no unclear low unclear unclear Knight 2013 Non-­‑RCT 117 unclear low low low High Komelasky 1993 Non-­‑RCT 87 unclear Low low low unclear Komelasky 1990 Non-­‑RCT 55 No low low unclear Low McDaniel ¡ 1988 Non-­‑RCT 40 Yes, ¡partly Unclear High unclear unclear McLauchlan 1992 Non-­‑RCT 49 no unclear High high unclear Messmer 1993 Non-­‑RCT 30 no unclear unclear unclear low Moore 1997 Non-­‑RCT 34 Yes, ¡partly low Low unclear low Pane ¡ 1989 Non-­‑RCT 1388 no low low unclear low Pierick 2012 Non-­‑RCT 311 no low low low High Sanna 2006 Non-­‑RCT 89 unclear low low low High Schneider 2014 Non-­‑RCT 85 unclear low low low low Sharieff 2001 Non-­‑RCT 18 no low low unclear low Sigsbee 1990 Non-­‑RCT 50 no low low low low Wright ¡ 1989 Non-­‑RCT 41 no low low unclear low

Non-RCTs

Dallas 2015

Evidence profile table(s)

Quality assessment № of patients Effect Quality Importance № of studies Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations focussed CPR training no such targeting Relative (95% CI) Absolute (95% CI) survival with favourable neurologic outcome at discharge and ROSC 2 randomised trials serious 1 not serious not serious serious 2 none

3

see comment not pooled see comment ⨁⨁◯◯ LOW CRITICAL survival with favourable neurologic outcome at discharge and ROSC 7

• bservational

studies very serious 4 not serious serious 5 serious 2 none

6

see comment not pooled see comment ⨁◯◯◯ VERY LOW CRITICAL bystander CPR performance (subsequent utilisation of skills) 2 randomised trials serious 1 not serious not serious serious 2 none

3

see comment not pooled see comment ⨁⨁◯◯ LOW IMPORTANT bystander CPR performance (subsequent utilisation of skills) 7

• bservational

studies very serious 4 not serious serious 5 serious 2 none

6

see comment not pooled see comment ⨁◯◯◯ VERY LOW IMPORTANT

¡

1 .Studies were not blinded, used temporal randomization, survival was self-reported, with large loss to follow up (Dracup 2000) or small sample size (Dracup 1986).

• 2. Large loss to follow-up and/or scant number of events.
• 3. The heterogeneous nature of RCT data prevents pooling. Dracup 1986 followed-up 65 adult cardiac patients six months after intervention. They did not report overall number of patients

requiring CPR, but reported 4/65 patients died (2/24 control and 2/41 CPR groups) –none received CPR by trained individuals (unknown if present at time of arrest). Dracup 2000 conducted follow-up at 12-months in high-risk infants (with high rate of loss to follow up). They reported 13 arrests in this period (13 intervention arms and 0 in control), 100% were successfully resuscitated.

• 4. Most studies were subject to high loss to follow-up, did not adjust for confounders and used self-reported outcomes. Three studies did not provided adequate information about screening

and eligibility (Dracup 1989, McDaniel 1988, McLauchlan 1992) and two studies were conducted on very small sample size McDaniel 1988, Mclauchlan 1992). Higgins 1998 unclear whether cardiac arrest case ascertainment was consistent across centres.

• 5. One study (Higgins 1998) examined whether hospitals provided CPR training and did not report the number of individuals trained
• 6. The heterogeneous nature of data prevents pooling. Follow-up periods varied from 3-months to 10 years. Dracup 1998 – at 6-months, 7/94 CPR-trained parents self-reported using CPR,

with 100% survival. Dracup 1994 – at 21±6 months 11/172 adult cardiac patients died (1 out of hospital –no CPR, trained individual not present). Higgins 1998 –examined admission rates

• f pediatric cardiac arrests over 10-years in 41 centers. In centers teaching CPR to parents, CPR was attempted in 28/41 children with a 46% survival rate. In centers not teaching CPR 24

events occurred with no CPR attempt and no survivors. McDaniel 1988 at 3-months, 1/16 adult cardiac patients arrested out of hospital –no CPR attempted, trained individual not present. Mc Lauchlan 1992 at 24-months, 1/27 adult VT patients arrested, wife physically unable to perform CPR, patient died. Pierick 2012 at 12-months, 8/311 events occurred, seven parents performed CPR, one event was unknown; six infants survived with good or stable neurologic status (as reported by parents). Sanna 2006 –at 12-months, no events in 33 adult cardiac patients.

Dallas 2015

Evidence profile table(s)

• 7. Studies were not blinded and one was subject to loss to follow-up (Blewer 2012).
• 8. The heterogeneous nature of data prevents pooling. Blewer 2012: CPR training kits were shared with 2.0 ± 3.4 additional family members in the continuous chest compression CPR

group vs. 1.2 ± 2.2 in the standard CPR group (p = .03). Greenberg 2012 12/177 followed medical advice and were CPR trained –all were in DVD trained group (0 underwent traditional CPR class).

• 9. All studies did not adjust for confounders or unclear. Loss to follow-up high or unclear
• 10. The heterogeneous nature of data prevents pooling. Kliegel 2000 targeted 190 OHCA survivors, with 50/101 responding and 20 patients and 71 family/friends) subsequently trained.

Pane 1989 targets recruitment of senior citizens and cardiac patients and families to attend free mass CPR training -particpants training because of heart disease significantly increased from 5.6% to 13.2%. Barr 2013– 49% shared CPR DVD with family/friends. Knight 2013 79% shared kit with at least 2 family members/friends.

• 11. Studies were subject to loss of follow-up.
• 12. The heterogeneous nature of data prevents pooling. Blewer 2012 trainees in the continuous chest compression CPR group were more likely to rate themselves “very comfortable” with

the idea of using CPR skills in actual events than the standard CPR trainees (34% v. 28%, p = .08). Moser 1999, asked subjects asked about their willingness to perform CPR, and the majority stated they would “absolutely” be willing to perform

• 13. Studies did not adjust for confounders and were subject to loss of follow-up. Did not provide sufficient detail in methods.
• 14. The heterogeneous nature of data prevents pooling. In three of the studies (Dracup 1994, Haugk 2006, Kliegal 2000), the vast majority of subjects stated they would use CPR if needed

(82%, 90%, 99%) or in one study (Knight 2013) subjects felt neutral to somewhat confident in their comfort with providing CPR. One study (Kliegal 2000) compared a non-trained control group with the trained intervention group and found only 68% of the control group were confident in their ability to perform CPR (p=0.02). Another of the studies (Dracup 1994) did not report on willingness to use CPR in the non-trained control group. Haugk 2006 found that at 1-year follow-up 98% of those trained stated they “agreed” or that they “maybe” would perform first aid (including CPR) correctly (p=0.03). Knight 2013, found a slight decrease in comfort level with CPR use at 1-month 3.8 (3.5-4.0), 3.7 (3.4-3.9) at 3 months, and 3.5 (3.2-2.9) at 6 months.

Quality assessment № of patients Effect Quality Importance № of studies Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations focussed CPR training no such targeting Relative (95% CI) Absolute (95% CI) number of people trained 2 randomised trials very serious 7 not serious not serious not serious none

• see

comment ⨁⨁◯◯ LOW IMPORTANT number of people trained 4

• bservational

studies very serious 9 not serious not serious not serious none

• see

comment ⨁◯◯◯ VERY LOW IMPORTANT willingness to provide CPR 2 randomised trials serious 11 not serious not serious not serious none

see comment not pooled see comment ⨁⨁⨁◯ MODERATE IMPORTANT willingness to provide CPR 6

• bservational

studies very serious 13 not serious not serious not serious strong association

see comment not pooled see comment ⨁◯◯◯ VERY LOW IMPORTANT

¡

Dallas 2015

Evidence profile table(s)

• 15. Studies were subject to loss to follow-up and did not always report sufficient details of sample.
• 16. The heterogeneous nature of data prevents pooling.Dracup 1998, enrolled 480 subjects. Most parents of infants at high risk for sudden death can demonstrate successful CPR skills at

the completion of 1 class (83%). One RCT with moderate quality of evidence: Brannon 2009, enrolled 23 subjects. 80% of subjects who viewed a CPR instructional video before attending an infant CPR class were rated as good on all three sections of the CPR test, versus 19% in the control group (no pre-instructional video) (p=0.012). Blewer 2013, describe little skill degradation (chest compression rate and depth) in the friends and family of adult cardiac patients 3- to 6-months after receiving CPR training (data not given).

• 17. Studies did not adjust for confounders. Studies were subject to loss of follow-up. Did not provide sufficient detail in methods.
• 18. The heterogeneous nature of data prevents pooling. Although different methods of training and assessment were used, these studies consistently showed improvements to CPR

performance in those trained versus control and high scores for CPR performance immediately following training. Of the studies who performed pre- and post-training knowledge and/or proficiency testing (Barr 2013, Khan 2010) they found that subjects did better after CPR training than before CPR training: 89% v 44% (p=.001; Barr 2013), 68% v 37% (p < 0.001; Khan 2010). The four studies looked at CPR skills initially after training and at follow up (Komelasky 1990, Komelasky 1993, Moore 1997, Wright 1989) and three of the studies found that CPR skills degraded overtime (Komelasky 1990, Moore 1997, Wright 1989). However, Komelasky 1990 showed no difference between the treatment group (CPR instruction) versus the control group (CPR review) (p=0.63) with regard to skill retention. Four non-RCTs (Dracup 1989, Messmer 1993, Sharieff 2001, Sigsbee 1990) found that most family members were able to learn and retain CPR skills (81%, Dracup 1989; Messmer 1993; 78% (infant and child CPR), Sharieff 2001; 94% hospitalized family member subjects, 96% non-hospitalized family member subjects, 94% control group, Sigsbee 1990). Sharieff 2001 looked at retention of skills and found that 71% (infant CPR) and 82% (child CPR) adequacy of CPR skill performance at 6 months.

Quality assessment № of patients Effect Quality Importance № of studies Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations focussed CPR training no such targeting Relative (95% CI) Absolute (95% CI) CPR skills performance and retention 3 randomised trials very serious 15 not serious not serious not serious strong association

16

see comment not pooled see comment

⨁⨁⨁◯

MODERATE IMPORTANT CPR skills performance and retention 11

• bservational

studies very serious 17 not serious not serious not serious strong association

18

see comment not pooled see comment

⨁◯◯◯

VERY LOW IMPORTANT

¡

Dallas 2015

Proposed Consensus on Science statements

For the critical outcomes of survival with favourable neurologic outcome at discharge and ROSC we have identified low quality of evidence (downgraded for risk of bias and imprecision) from two RCTs (Dracup 1986, Dracup 2000) and very low quality of evidence for seven non-RCTs (Dracup 1998, Higgins 1998, Pierick 2012, Dracup 1994, McDaniel 1988, McLaughlan 1992, Sanna 2006). The heterogeneous nature of the studies prevents pooling of data.

Dallas 2015

Proposed Consensus on Science statements

For the important outcome of bystander CPR performance (subsequent utilisation of skills) we identified low quality

• f evidence (downgraded for risk of bias and imprecision)

from two RCTs (Dracup 1986, Dracup 2000) and very low quality of evidence for seven non-RCTs (Dracup 1994, Dracup 1998 219, Higgins 1998, McClaughlan 1992, McDaniel 1988, Pierick 2012, Sanna 2006). The heterogeneous nature of the studies prevents pooling of data.

Dallas 2015

Proposed Consensus on Science statements

For the important outcome of bystander CPR performance (skill acquisition and retention) we identified a moderate level of evidence downgraded for risk of bias from three RCTs (Dracup 1998, Brannon 2009, Blewer 2012) and very low quality of evidence from eleven non-RCTs (Barr 2013, Khan 2010, Kliegal 2000, Komelasky 1990, Komelasky 1993, Moore 1997, Wright 1989, Dracup 1989, Messmer 1993, Sharieff 2001, Sigsbee 1990). The heterogeneous nature of the studies prevents pooling of data. However, these studies consistently showed high scores for CPR performance in those trained versus control and high scores for CPR performance immediately following

• training. Most studies examining retention of skills showed a

decline in CPR performance over time without re-training.

Dallas 2015

Proposed Consensus on Science statements

For the important outcomes (number of people trained) we identified low quality of evidence (downgraded for risk of bias) from two RCTs (Blewer 2012, Greenberg 2011) and very low quality of evidence from four non-RCTs (Barr 2013, Knight 2013, Pane 1989, Kligel 2000). The heterogeneous nature of the studies prevents pooling of data. For the important outcome of willingness to provide CPR we identified a moderate level of evidence (downgraded for risk of bias) from two RCTs (Blewer 2012, Moser 1999) and very low quality of evidence from six non-RCTs (Dracup 1994, Haugk 2006, Komelasky, 1990, Kliegal 2000, Knight 2013, Schneider 2014). The heterogeneous nature of the studies prevents pooling of data, but with a strong signal towards willingness to provide CPR if required.

Dallas 2015

2015 Proposed Treatment Recommendations

Based on moderate-low quality of evidence, we recommend training likely rescuers (e.g. family or care-givers) of high-risk populations in CPR – based on the willingness to be trained and the fact that there is low harm and high potential benefit.

There is low to moderate quality of evidence in 29 studies related

to CPR training in likely rescuers (e.g. family or care-givers) of high-risk OHCA groups, with no strong evidence of improved patients outcomes.

The recommendation for providing training in this group places

higher value on the potential benefits of patients receiving bystander CPR by a family-member or caregiver in the case of cardiac arrest, and the willingness of this group to be trained.

We place lesser value on associated costs, and the potential that

skills may not be retained without on-going CPR training. The cost

• f training are potentially reduced with CPR training self-

instruction kits (e.g. DVD training).

Dallas 2015

Knowledge Gaps

There is a need for higher quality research in the current era. Future research could consider: Separate reporting of patient outcomes and bystander CPR in those witnessed by trained rescuers and controls. Studies examining modified CPR trainings similar to those examined in Blewer 2012 (chest compression only v standard CPR training) versus no training. Adequately powered studies reporting critical clinical outcomes are needed-Use accredited BLS courses and standardized/objective methods of assessment for CPR performance (Real-Time Data recording) Use stronger methods for follow-up e.g. death/hospital records not self- report(particularly for critical outcomes) Report data according to guidelines (e.g. CONSORT/STROBE) Examine different timeframes for recruitment of patients Detail exclusions and compare demographics to included cases

Dallas 2015

Next Steps

This slide will be completed during Task Force Discussion (not EvRev) and should include:

Consideration of interim statement Person responsible Due date