Pediatric Early Warning Systems (PEWS)

Conflict of Interest Declaration

The ILCOR Continuous Evidence Evaluation process is guided by a rigorous ILCOR Conflict of Interest policy. There were no authors with conflicts of interest.

CoSTR Citation

Ong GYK; Acworth J; KC Ng; Chong SL; Goh MSL; Yao SHW, on behalf of the International Liaison Committee on Resuscitation Pediatric Life Support Task Forces.

Pediatric Early Warning Systems (PEWS) With Or Without Rapid Response Teams: Consensus on Science with Treatment Recommendations. International Committee of Resuscitation (ILCOR) Pediatric Life Support Task Force XXX. Available from: http//ilcor.org

Methodological Preamble (and Link to Published Systematic Review if applicable)

The continuous evidence evaluation process for the production of Consensus on Science with Treatment Recommendations (CoSTR) started with a scoping review of Paediatric Early Warning Systems / Scores (Maconochie, 2020, S140) conducted by KC Ng and members of the Pediatric Life Support Task Forces. Evidence for pediatric literature was sought and considered by the Pediatric Task Force. These data were taken into account when formulating the Treatment Recommendations.

This systematic review was registered with PROSPERO (CRD42021269579). All items were reported in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines.

Eligibility Criteria

We included all studies with children < 18 years old performed in the inpatient units and EDs of paediatric hospitals but excluded outpatient clinics. We included studies published between January 2011 and December 2020 on the general paediatric population as well as those that focused on specific populations (e.g. oncology or cardiology units).

We compared patient populations among whom pediatric early warning systems (PEWS) were implemented (intervention) to those among whom PEWS were not implemented (comparator). The implemented PEWS could include any formal standardized set of criteria to alert clinicians to potential deterioration.

We excluded all studies that did not have a comparator group (no PEWS). For outcomes, we included: (1) mortality; (2) cardiopulmonary arrest; (3) unplanned codes; and (4) critical deterioration. We chose the specific outcomes of cardiopulmonary arrests and unplanned codes because these were individually measured and accounted for in the literature. Under critical deterioration, we included the following composite definition of significant clinical deterioration: (a) Unplanned/crash tracheal intubation; (b) Unanticipated fluid resuscitation and inotropic/vasopressor use; (c) Cardiopulmonary resuscitation (CPR) or Extracorporeal Membrane Oxygenation (ECMO); and (d) Death in patients (all-cause mortality) without a Do Not Resuscitate (DNR) order. We also included unplanned or emergency admissions to Paediatric Intensive Care Unit (PICU) as these reflected the utilisation of critical care resources. For study designs, we included randomized controlled trials (RCTs) and non-randomized studies including trials, controlled before- and after-, or with- and without studies. Case series, case control, and systematic reviews, as well as unpublished studies and studies not in English, were excluded.

Information Sources

We included the following electronic databases: Medline, EMBASE, Cochrane Central Register of Controlled Trials, CINAHL (Cumulative Index to Nursing and Allied Health Literature) and Web of Science. To understand if there were concurrent similar systematic reviews being carried out, we searched the following: PROSPERO, ClinicalTrials.gov, International Standard Randomised Controlled Trial Number registry, WHO International Clinical Trials Registry Platform and EU Clinical Trials Register.

Search Strategy (Attached)

We drew up a comprehensive search strategy with inputs from a medical librarian as well as our clinical team. We used and exploded the Medical Subject Headings (MeSH) terms for Medline and Cochrane, and Emtree terms for EMBASE as appropriate for each term’s tree. Their synonyms were included in the title, abstract and keyword searches. We performed the search on 26^th June 2021. We provided the search terms for each electronic database and hand-searched the bibliography of systematic reviews, as well as publications that we included, to ensure that our literature search was comprehensive

PICOST

Do Pediatric Early Warning Systems reduce mortality and significant clinical deterioration? A Systematic Review

The PICOST (Population, Intervention, Comparator, Outcome, Study Designs and Timeframe)

Population: Infants, children, and adolescents in any inpatient setting

Intervention: Pediatric early warning systems (PEWS) with or without rapid response teams/medical emergency teams (RRTs/METs)

Comparators: No pediatric early warning systems (PEWS) or standard care (without a scoring system)

Outcomes: Significant clinical deterioration event, including but not limited to:

(1) Unplanned/crash tracheal intubation,

(2) Unanticipated fluid resuscitation and inotropic/vasopressor use,

(3) Cardiopulmonary resuscitation (CPR) or Extracorporeal Membrane Oxygenation (ECMO)

(4) Death in patients (all-cause mortality) without a Do Not Resuscitate (DNR) order.

Study Designs: Randomized controlled trials (RCTs) and non-randomized studies (non-randomized controlled trials, interrupted time series, controlled before-and-after studies, cohort studies) are eligible for inclusion.

Timeframe: All years and all languages were included as long as there was an English abstract; unpublished studies (e.g., conference abstracts, trial protocols) were excluded. Literature search updated to 26th June 2021.

NOTE FOR RISK OF BIAS: The methodology used to compare the Risk of Bias (RoB) was assessed per comparison rather than per outcome, since there were no meaningful differences in bias across outcomes. In cases where differences in risk of bias existed between outcomes this was noted; for which there were none noted.

Consensus on Science

We identified 15 studies^1-15, one randomized controlled trial (RCT) and fourteen cohort studies, for inclusion in our systematic review. (See Tables 1 & 2 attached).

We used the Scottish Intercollegiate Guidelines Network (SIGN) to appraise the risk of bias in each study¹⁶. We assessed for risk of bias at both the study and the outcome level. Checklists were used depending on the study design (e.g. cohort versus randomized controlled trial). Overall assessment ranged from unacceptable (0) to high quality (++). Two independent reviewers assessed each article and differences were resolved by consensus. Three cohort studies ^13-15 were not included in the final analysis because of unacceptable bias.

We looked at 3 critical outcomes (mortality, cardiopulmonary arrest events, and “significant clinical deterioration events”), and 1 important outcome (unplanned code events).

“Significant clinical deterioration” referred to a composite measure of late ICU admission and was a composite outcome that occurred among patients without DNR (do not resuscitate) orders with 1 or more of the following: death before ICU admission; provision of cardiopulmonary resuscitation, tracheal intubation, administration of vasoactive medication, or provision of fluid boluses of 60 mL/kg or greater within the 12 hours before ICU admission; tracheal intubation, cardiopulmonary resuscitation, initiation of extracorporeal membrane oxygenation, or death within the first hour of ICU admission.

Summary Measures and Synthesis of Results

Extracted data were combined in a 2-stage meta-analysis approach. In the first step, incidence rate rations (IRR) with their 95% confidence intervals (CI) were estimated from individual studies reporting patient-days such as number of deaths per 1000 patient-days.

Likewise, risk ratios (RR) with 95% CI were estimated from individual studies reporting binary outcome data such as number of cardiopulmonary arrests in a hospital during a specific time period. In the second stage, a restricted Maximum Likelihood (REML) random effects meta-analysis was employed to combine RRs and IRRs from individual studies. Statistical heterogeneity was quantified using the I2 statistic for analyses.¹⁷ Summary statistics were reported as RR with 95% confidence interval. We performed a sensitivity analysis by study design, pooling all studies at first regardless of study design and then only for the observational studies.

Risk of bias across studies

Funnel plots displaying the outcome rate from individual studies were created for the exploration of publication bias. Begg and Mazumdar’s test was used for publication bias for each outcome.¹⁸

Certainty of evidence

We performed the certainty assessment using GRADE. This assessment was based on the study design, risk of bias, inconsistency, indirectness, imprecision and other considerations. ¹⁹

All analyses were conducted using Comprehensive Meta Analysis V5 and SAS 9.4 software.

Results:

• For the critical outcome of mortality, we identified low certainty evidence (downgraded for serious risk of bias and serious imprecision) from 1 RCT {Parshuram 2018 1002} [RR 1.24; 95%CI 0.95 to 1.62 (p=0.110)], and 9 cohort studies {Agulnik 2017 2965; Bonafide 2014 25; Brilli 2007 236; Hanson 2010 314, Kotsakis 2011 72; McKay 2013 48; Sefton 2015 91; Sharek 2007 2267; Tibballs 2009 306} [pooled cohort RR 1.17; 95% CI 1.00 to 1.37 (p=0.087)], that suggested that there was no significant increased mortality when no PEWS was compared to PEWS.

Pooled analysis of all studies demonstrated an overall increased trend for mortality when no PEWS was compared to PEWS; which was limited by heterogeneity [pooled RR 1.18; 95% CI 1.01 to 1.38 (p=0.036)].

• For the critical outcome of cardiopulmonary arrest, we identified very low certainty evidence (downgraded for very serious risk of bias and very serious imprecision) from 6 cohort studies {Bonafide 2014 25; Brilli 2007 236; Hanson 2010 314; Hunt 2008 117; Kotsakis 2011 72; Tibballs 2009 306}, that showed no difference between the use of no PEWS versus PEWS.

There was an overall trend for increased cardiopulmonary arrest events when no PEWS was compared to PEWS, but this was not statistically significant [pooled IRR/RR 1.22; 95% CI 0.93 to 1.59 (p=0.153)].

• For the critical outcome of significant clinical deterioration events (a composite outcome defined by Parshuram {Parshuram 2018 1002}), we identified low certainty evidence (downgraded for serious risk of bias and serious imprecision) from 1 RCT {Parshuram 2018 1002} [RR 1.67; 95% CI 1.34 to 2.08 (p=<0.001)], and 5 cohort studies {Agulnik 2017 2965; Bonafide 2014 25; McKay 2013 48; Parshuram 2011 18; Sefton 2015 91}, [pooled cohort RR 1.09; 95% CI 0.84 to 1.42 (p=0.517)], that was equipoised on the use of no PEWS versus PEWS.

Pooled analysis of all studies demonstrated an overall non-statistically significant trend of increased significant clinical deterioration events when no PEWS was compared to PEWS; which was limited by heterogeneity [pooled RR 1.21; 95% CI 0.91 to 1.62 (p=0.199)].

• For the important outcome of unplanned code events, we identified very low certainty evidence (downgraded for very serious risk of bias and very serious imprecision) from 4 cohort studies {Brilli 2007 236; Kotsakis 2011 72; McKay 2013 48; Sharek 2007 2267}, that showed a statistically significant increase in unplanned code events when no PEWS was compared to PEWS [pooled IRR/RR 1.73; 95% CI 1.01 to 2.96 (p=0.046)].

Treatment Recommendations

We suggest using pediatric early warning systems to monitor hospitalized children with the aim of identifying those who may be deteriorating (weak recommendation, low quality evidence).

Justification and Evidence to Decision Framework Highlights

The Pediatric Life Support (PLS) Task Force concluded that the implementation of pediatric early warning systems should be part of an overall clinical response system, with the task force placing a higher value on improving healthcare provider ability to recognize and intervene for patients with deteriorating illness over the expense incurred by a healthcare system committing significant resources to implement these systems. The task force also noted that the complex process of optimizing patient care is likely to include both the implementation of pediatric early warning systems and ongoing healthcare provider education. The PLS Task Force agreed that the decision to use pediatric early warning systems should be balanced between use of existing resources and capabilities of the healthcare setting to adapt to its use and the consequences of its use.

In making these recommendations, the PLS Task Force considered the following:

Values, Preferences, and Task Force Insights

The evidence is equipoised to justify the use of pediatric early warning systems to significantly decrease in-hospital pediatric mortality, significant clinical deterioration, and cardiopulmonary arrest events. However, in systems with available resources that prioritize and value the potential to decrease the incidence of code events for inpatient pediatric patients, there was very weak evidence to support the use of pediatric early warning systems in this context.

The taskforce recognized the significant limitations of available evidence in its treatment recommendations, but also the importance and the potential value of improving healthcare providers’ ability to recognize and intervene for patients with deteriorating illness. The use of pediatric early warning systems should be balanced with the expense incurred by a healthcare system committing significant resources to implement these systems. This complex process of optimizing patient care is likely to include both the implementation of pediatric early warning systems and ongoing healthcare provider education.

The PLS Task Force agreed that the decision to use a pediatric early warning system should be balanced between use of existing resources and capabilities of the healthcare setting to adapt to its use, and the consequences of its use.

For existing settings using a pediatric early warning system, local validation, site-specific adaptation of its use, and longitudinal evaluation of its effectiveness are important.

Knowledge Gaps & Recommendations

• The amount and quality of evidence in children compared with adults for the role of early warning systems in the inpatient setting is very low. A major limitation to evaluation of these systems is the low rate of pediatric cardiopulmonary arrest and mortality (especially outside the intensive care unit setting), including within the hospitals from which the data in this analysis originate. As such, demonstrating a statistically significant effect after a new implementation is difficult.

This is apparent in our findings that found trends of improving cardiopulmonary arrest rate or mortality, and significant clinical deterioration events, although not to statistically significant levels. However, we did find a significant decrease in code events. These findings support implementation of Pediatric early warning systems in the pediatric inpatient setting.

We recommend that a workgroup should be set up to recommend & standardize important clinical outcomes that should be tracked and measured following implementation of Pediatric early warning systems in hospitals and healthcare systems.

• The other major limitation in our analysis is the use of before-and-after studies, with the inherent limitations of unaccounted or confounding variables and inability to develop a comparable control group associated with the problems of confounding variables and contemporaneous trends. Studies should not be limited to RCTs but include comparative study approaches as well as Quality Improvement (QI) and longitudinal studies. Quality Improvement methodology could be used to regulate the impact of a series of changes that include educational processes, documentation review with feedback systems, and modification of other factors thought to improve the delivery of care.

• Further studies for pediatric early warning systems should focus on controlled trials evaluating rapid response teams (RRT) compared to without, and composition of efferent arms; look into specific pediatric subgroups including pediatric patients in the emergency department setting and specific subgroups of pediatric disease populations – e.g. pediatric oncology; pediatric patients in the out-of-hospital setting as well as pediatric patients in resource-rich countries and patients from resource-limited countries; prospectively evaluate different Pediatric early warning systems for predicting, identifying, and provide early intervention for patients at risk for different forms of decompensation, including primary respiratory, circulatory, and neurologic etiologies.

These further studies should also look not only at pediatric early warning systems in resource-rich healthcare institutions but also in healthcare institutions from resource-limited countries and these studies should be powered with more analyzable data and these should be stratified by resource-availability e.g. Gross National Income or Sociodemographic Index status of the country.

• With regards to pediatric early warning systems implementation considerations, studies should look into staff training/education methodology for PEWS implementation, resourcing; feasibility; cost-effectiveness; equity and acceptability of PEWS into the existing healthcare systems.

Attachments:PEWS Search Strategy, PEWS Table 1, PEWS Table 2, PEWS Co STR GRADE Summary of Evidence Et D

References

1. Agulnik A, Mora Robles LN, Forbes PW, et al. Improved outcomes after successful implementation of a pediatric early warning system (PEWS) in a resource-limited pediatric oncology hospital. Cancer. 2017 Aug 1;123(15):2965-2974. doi: 10.1002/cncr.30664
2. Bonafide CP, Localio AR, Roberts KE, et al. Impact of rapid response system implementation on critical deterioration events in children. JAMA Pediatr. 2014 Jan;168(1):25-33. doi:10.1001/jamapediatrics.2013.3266.
3. Brilli RJ, Gibson R, Luria JW, et al. Implementation of a medical emergency team in a large pediatric teaching hospital prevents respiratory and cardiopulmonary arrests outside the intensive care unit. Pediatr Crit Care Med. 2007 May;8(3):236-46; quiz 247. doi: 10.1097/01.PCC.0000262947.72442.EA.
4. Hanson CC, Randolph GD, Erickson JA, et al. A reduction in cardiac arrests and duration of clinical instability after implementation of a paediatric rapid response system. Postgrad Med J. 2010 May;86(1015):314-8. doi: 10.1136/qshc.2007.026054.
5. Hunt EA, Zimmer KP, Rinke ML, Shilkofski NA, Matlin C, Garger C, Dickson C, Miller MR. Transition from a traditional code team to a medical emergency team and categorization of cardiopulmonary arrests in a children's center. Arch Pediatr Adolesc Med. 2008 Feb;162(2):117-22. doi: 10.1001/archpediatrics.2007.33.
6. Kotsakis A, Lobos AT, Parshuram C, et al. Implementation of a multicenter rapid response system in pediatric academic hospitals is effective. Pediatrics. 2011 Jul;128(1):72-8.
7. McKay H, Mitchell IA, Sinn K, et al. Effect of a multifaceted intervention on documentation of vital signs and staff communication regarding deteriorating paediatric patients. J Paediatr Child Health. 2013 Jan;49(1):48-56. doi: 10.1111/jpc.12019.
8. Parshuram CS, Bayliss A, Reimer J, et al. Implementing the Bedside Paediatric Early Warning System in a community hospital: A prospective observational study. Paediatr Child Health. 2011 Mar;16(3):e18-22. doi: 10.1093/pch/16.3.e18.
9. Parshuram CS, Dryden-Palmer K, Farrell C et al. Effect of a Pediatric Early Warning System on All-Cause Mortality in Hospitalized Pediatric Patients. The EPOCH Randomized Clinical Trial. JAMA. 2018;319(10):1002-1012. doi:10.1001/jama.2018.0948
10. Sefton G, McGrath C, Tume L, et al. What impact did a Paediatric Early Warning system have on emergency admissions to the paediatric intensive care unit? An observational cohort study. Intensive Crit Care Nurs. 2015 Apr;31(2):91-9. doi: 10.1016/j.iccn.2014.01.001.
11. Sharek PJ, Parast LM, Leong K, et al. Effect of a rapid response team on hospital-wide mortality and code rates outside the ICU in a Children's Hospital. JAMA. 2007 Nov 21;298(19):2267-74. doi:10.1001/jama.298.19.2267.
12. Tibballs J, Kinney S. Reduction of hospital mortality and of preventable cardiac arrest and death on introduction of a pediatric medical emergency team. Pediatr Crit Care Med. 2009 May;10(3):306-12. doi: 10.1097/PCC.0b013e318198b02c.
13. Anwar-ul-Haque, Saleem AF, Zaidi S, Haider SR. Experience of pediatric rapid response team in a tertiary care hospital in Pakistan. Indian J Pediatr. 2010 Mar;77(3):273-6. doi: 10.1007/s12098-010-0032-2.
14. Douglas K, Collado JC, Keller S. Implementation of a Pediatric Early Warning Scoring System at an Academic Medical Center. Crit Care Nurs Q. 2016 Oct-Dec;39(4):363-70. i:10.1097/CNQ.0000000000000130
15. Kroeger AR, Morrison J, Smith AH. Predicting unplanned readmissions to a pediatric cardiac intensive care unit using predischarge Pediatric Early Warning Scores. Congenit Heart Dis. 2018 Jan;13(1):98-104. doi: 10.1111/chd.12525.
16. Scottish Intercollegiate Guidelines Network. SIGN Checklists. https://www.sign.ac.uk/what-we-do/methodology/checklists/. Accessed November 1, 2021.