OBJECTIVE: We evaluated the Pediatric Early Warning Score (PEWS) sensitivity as an early indicator of patients deterioration leading to a Rapid Response Team (RRT)/code event. We hypothesized that at least 80% of patients had a critical PEWS preceding the event. We determined staff awareness of deterioration in patient status prior to the event as evidenced by consults, addition of monitoring equipment or increased frequency of assessment. The timing of these events was compared to critical PEWS times.
METHODS: One hundred and seventy non-ICU RRT and 16 code events were identified between October 2006 and February 2008. We completed retrospective PEWS at four-hour intervals or less for twenty-four hours preceding the event. The PEWS algorithm, guiding staff to consult at a critical score ≥4 or a single domain score equal to 3, was applied.
RESULTS: For 85.5% of patients the earliest indicator of deterioration, evidenced by a critical PEWS, was a median of 11 hours 36 minutes and the earliest preceding the event was 30 minutes. For 97.1% of patients the earliest median time to a consult was 80 minutes. Oximetry was added 6.9 hours for 43.5% of patients. 7% of patients had increased nursing assessment. A sub-group of patients had 1) critical PEWS, 2) consult and 3) addition of a monitor. The median time for earliest critical PEWS for these was significant (P < 0.001).
CONCLUSION: PEWS can potentially provide a forewarning time >11 hours, alerting the team to adapt the care plan and possibly averting an RRT or code.
- Pediatric Early Warning Score
- screening tool
- risk assessment
- clinical deterioration
- Rapid Response Team
- code blue
- early recognition
WHAT'S KNOWN ON THIS SUBJECT:
Two previous studies addressed PEWS sensitivities to unplanned transfers to a higher level of care.
WHAT THIS STUDY ADDS:
No study has focused exclusively on patients who experience and rapid response team or code call. Neither has any addressed the lead time of recognition by using PEWS.
Hospitalized pediatric patients are at risk for sudden deterioration that results in respiratory and/or cardiac arrest and death. Rapid response teams (RRTs) have been put in place at many hospitals to respond to these emergencies.1,–,3 This departure from usual chain-of-command for medical care is intended to ensure that clinically deteriorating patients receive assistance before their conditions are an emergency. Although there has been evidence of significant decreases in cardiopulmonary code (code blue) rates outside the ICU and deaths as a result of RRT implementation, they have not been eliminated.2,4 Efforts have shifted to earlier recognition of change in physiologic status by using objective clinical indicators and risk tools. Earlier response and interventions provide the opportunity to decrease the need for calling the RRT.
Several adult and subsequent pediatric early warning scores have been developed and refined.5,–,13 Haines et al10 used a retrospective record review and nurse interview to evaluate trigger criteria for patient identification of high-dependence nursing care needs. Their 14-item tool was based on criteria for airway, breathing, circulation, disability, and other selected diagnostic elements. Duncan and colleagues8,9 reported on the development of a 20-item early warning score. Their validation study suggested that three-quarters of patients who requiring a code blue call might have had at least a 1-hour warning on the basis of their score. The Pediatric Early Warning Score (PEWS), designed by Monaghan12 and his team at the Children's Hospital in Brighton, England, is a more easily scored tool that is based on only 5 domains: behavior, cardiovascular status, respiratory status, nebulizer use, and persistent postsurgical vomiting. The tool was further supported by an algorithmic response that is based on the score. Quist-Therson11 and his team at the Hertfordshire Partnership NHS Trust Children's Services in England adapted the Monaghan model by using colors as indicators of deterioration. The simplicity, efficiency, and color-coded score hierarchy made it a desirable choice for trial at Children's Hospitals and Clinics of Minnesota (Children's), a 325-bed facility located at 2 urban sites in the cities of Minneapolis and St Paul. Tume14 examined clinical observations of patients who were transferred to a higher level of care and found that PEWS would have identified 87% of these patients who were at risk for deterioration. Tucker et al15 in a similar prospective study described the sensitivity and specificity of PEWS for detecting clinical deterioration that results in unplanned transfer to the PICU. They reported sensitivity of 84.2% at a score of ≥4 and concluded that PEWS could identify children who require transfer.
We began our implementation and evaluation of the Monaghan and Quist-Therson PEWS tool in January 2008. The primary objective of this retrospective study was to evaluate the sensitivity of PEWS for a group of patients who had a documented RRT or code blue event as well as the lead time for the earliest and latest critical PEWS before the event. We hypothesized that at least 80% of patients had a critical PEWS, defined as a score of ≥4 or a domain score of 3, before the event. If true, then this would provide evidence of good sensitivity and support its use in the pediatric inpatient setting to identify early physiologic deterioration. A secondary objective was to examine staff awareness of deterioration in patient status before the event and to determine whether use of PEWS would have provided significantly earlier recognition. We measured staff awareness by key indicators of (1) consultation with another nurse, physician, or respiratory therapist; (2) the addition of monitoring equipment; and/or (3) increased frequency of patient assessment.
A retrospective chart review was completed on 170 RRT calls and 16 code blue events that occurred for 186 unique patients between October 2006 and February 2008. All events occurred on medical surgical units excluding ICU and ICU step-down units. Study approval was obtained from Children's institutional review board.
Code blue and RRT events were included because they represent unexpected and/or critical deterioration characteristics in some hospitalized pediatric patients. In addition, assessment and communication processes are similar. At Children's, employees, patients, parents, and families may unconditionally request urgent medical assistance (RRT) for a patient who is perceived to be in distress without advance consultation with housestaff or the patient's attending physicians. We tested the sensitivity of PEWS in this population as an objective tool to predict critical deterioration and to support structured communication of patient status between team members. RRT and code blue events provide measurable outcomes that we hope to reduce with implementation of PEWS.
This was a retrospective study design of cases before implementation of PEWS. It is the only design that is not contaminated by interventions in response to PEWS. Once PEWS is implemented, it becomes unclear as to whether the event would have occurred as a result of interventions that are responses to critical PEWS.
Assignment of PEWS
PEWS are assigned in 3 domains: behavior, respiratory, and cardiovascular. Scores in each domain can range from 0 to 3 points (Fig 1). In addition, 2 points are added for nebulizations that are continuous or every 15 minutes and 2 points for persistent postoperative vomiting. The total score can range from 0 to 13.
PEWS were calculated for 24 hours before the event at intervals of at least 4 hours for each of the identified patients. This is consistent with assessment frequencies for nurses on these units. When assessment charting was more frequent than 4 hours, additional PEWS were calculated. Behavioral scoring required the actual use of the words shown in Fig 1, such as playing, sleeping, and/or irritable. During waking shifts, this would not be customary documentation. When these words were not documented, the item was scored as missing. When data were not available for a specific domain, it was marked as missing. To ensure consistency, a single “expert” performed all scoring.
Training the Expert in PEWS
A licensed registered nurse with extensive intensive care, rapid response, and code blue experience reviewed patient charts and determined PEWS. Working with a hospital advanced practice nurse, they reviewed charts, discussed and refined the scoring rules, and then independently scored patients until interrater reliability was achieved on a sample of 10 patient records. Our electronic medical record included many more descriptors of patient status than are included in the tool, which required agreement on which terms would be recommended to staff to facilitate scoring PEWS. One example is the option for patient color. PEWS has 3 choices, and the electronic medical record has 6. By comparing their results data, interpretation was refined until agreement was achieved. This agreement of terms for scoring was also used to train staff.
PEWS Action Algorithm: Critical PEWS
We agreed to use the Monaghan12 criteria that a total score of ≥4 or a score of 3 in any of the PEWS domains reflected a critical value that required a consultative action. We refer to these criteria as critical PEWS.
Additional Data Collected
During the 24-hour pre-event period, we collected data on increased frequency of nursing assessment and the addition of equipment, including pulse oximetry and cardiac monitoring. Consultations by another nurse, physician, or respiratory therapist were also collected. The purpose was to evaluate these as indicators of staff awareness of deteriorating patient status. An exhaustive review of multiple documentation elements such as nursing flow sheets, nursing narrative notes, physician progress notes, and physician orders was used to retrieve data elements. Demographic data included age, gender, race, diagnosis, and All Patient Refined Diagnosis Related Groups (APRDRG)16 service line and whether the patient had had surgery. APRDRG is a recognized patient classification scheme available from 3M that relates patient cost to his or her diagnosis and comorbidities. APRDRG service line classification is 1 component of this method.
Sensitivity was defined as percentage of patients who had a critical PEWS before the RRT or code event. Median and range were used to describe continuous variables such as the time from a critical PEWS, first consultation, and first monitor added, to the time of the RRT or code blue event.
For a subgroup of patients who had (1) a critical PEWS, (2) a consultation, and (3) addition of a monitor, median time from each of these to the RRT or code event were compared by using the Friedman nonparametric test with a 2-sided P < .05 required for significance. The nonparametric Wilcoxon Signed Ranks test was used for paired comparisons of these median times where a Bonferroni correction was applied requiring a 2-sided P < .015 for significance. Because of the small number of code blue events it was not possible to compare outcomes for these events to RRT events. SPSS 15.0 (SPSS Inc, Chicago, IL) was used for statistical analysis.
The median age of subjects was 25.5 months ranging from 0–252 months. Gender was 60% male and 40% female. Races included 55.9% white, 17.2% Black/African American, 7.5% Asian, and 7% Hispanic/Latino. Only 23.1% of patients were surgical. Forty-six percent of patients received care from the APRDRG respiratory service line followed by infectious disease (9.6%), cancer care (4.8%), cardiac care (4.2%) and digestive disease (3.2%). The remainder of patients received their care from a variety of APRDRG service lines.
Pediatric Early Warning Score
The sensitivity of PEWS was 85.5%, defined as the patient's having had a critical score within 24 hours before the event. As shown in Table 1, the median time from the first critical PEWS to an RRT or code event was 696 minutes (11 hours, 36 minutes) and the latest critical score was 30 minutes for 159 (85.5%) of 186 patients in this study. A total of 73.1% of patients had a critical PEWS just before the RRT or code event (not shown in Table 1). The median time from a critical PEWS just before the event was 30 minutes. Consistent with pediatric patterns of compensation, critical respiratory domain scores (score of 3) were the earliest to precede the critical event at 21 hours, 8 minutes for 51.6% of patients.
Consultations and Added Monitoring
A total of 181 (97.3%) of 186 patients received at least 1 consultation from a variety of providers, including physicians, nurses, and respiratory therapists. For first consultations, 159 (87.8%) included a medical doctor. The median time from the first consultation before the event was 80 minutes. As shown in Table 2, >98% of study patients had pulse oximetry monitoring initiated before the RRT or code blue event. This is used instead of cardiac monitoring for patients who are outside the ICU, and nurses can independently initiate it. A total of 81 (43.5%) patients had monitoring added during the 24-hour pre-event period. The median time of the first addition of a monitor before the event was 6 hours, 54 minutes. This additional monitoring was interpreted as evidence of some staff awareness of change in patient condition. Evidence of documented increased frequency of nursing assessment was rare at 7%.
We evaluated a subgroup of 72 patients who shared 3 common findings: (1) critical PEWS; (2) clinician consultation; and (3) addition of a monitor. When all median times to the event were compared, they were significantly different (P < .001). The median time to first consultation was 73 minutes, which was significantly less than first critical PEWS at 602 minutes (10 hours, 2 minutes; P < .001). The median time for addition of a monitor was 406 minutes (6 hours, 46 minutes), similar to critical PEWS (P = .42).
RRT Interventions and Subsequent Placement in the PICU
Nearly 91% of patients who experienced an RRT event received a significant medical intervention: 37.1% received oxygen, 27.1% received nebulization, 21.1% received oral/nasal/pharynx suctioning, 17.6% received cardiac monitoring, and 21.1% received arterial blood gas/venous blood gas. A total of 40 (23.5%) patients who experienced an RRT event were moved to the PICU. These outcomes are similar to those reported for Children's by Zenker et al.3
As shown in Table 3, for 9 of 16 patients who experienced a code blue event and also had a critical PEWS, the median time from the earliest critical PEWS to the critical event was 6 hours, 45 minutes and the latest was 55 minutes. The remaining 7 patients did not have a critical PEWS. Eight patients had addition of an oximeter at a median time of 7 hours, 20 minutes, and 3 patients had no interruption of oximetry monitoring during the 24 hours before the critical event.
For all patients who experienced a code blue event, significant comorbidity existed. Six of the 16 experienced prolonged seizure with hypoxia, 2 had cardiopulmonary arrests, and 2 had equipment failure related to trach-ventilator dependence. The remaining patients had sudden respiratory failure as a result of an acute event that included apnea, aspiration, and airway occlusion with cough. Twelve of the 16 patients who experienced a code blue event were transferred to a higher level of care, 3 remained on their unit, and 1 was unresponsive to resuscitation efforts and died.
We have shown that PEWS are likely to be critical at points in time during the 24-hour period before an RRT or code blue event, giving valuable forewarning. To our knowledge, this is the first report of the evaluation of the sensitivity of the PEWS as an indicator of risk for an RRT or code blue event during the 24 hours before these events. Tucker et al15 as well as Tume14 used only unplanned transfer to a higher level of care as a measure of deterioration. Also and as important, this is the only study that documented the lead time for which PEWS would have signaled deterioration earlier than standard clinical observation. Hospitalized pediatric patients are known to exhibit prolonged compensatory stabilization before decompensation. Earliest recognition and intervention on the basis of any single critical PEWS could support these children during their compensatory phase and improve outcomes and reduce cost of care. During the compensatory phase, PEWS would be expected to vary, depending on the patient's underlying diagnosis and comorbidities and ability to compensate. This variation in comorbidities and the ability to compensate may account for the intermittent consultations on the patients before the events and also account for <100% critical scores for the study patients. In addition, the severity of comorbidity for patients who experienced a code blue event contributed to sudden changes in physiologic status that were not always signaled by a critical PEWS, as shown in Table 3.
For the subgroup of patients who had a combination of factors indicating risk—(1) PEWS ≥4, (2) monitoring added, and (3) at least 1 medical consultation—the timing of these factors was noteworthy. The time for added monitoring was not significantly different from the earliest critical PEWS. This close proximity of timing supports the premise that PEWS is an objective tool that validates nurses' clinical appraisal of their patient's changing status, yet PEWS ≥4 indicated a risk for deterioration 6 hours earlier than the first documented medical consultation. This is consistent with the discussion by Andrews and Waterman7 of the need to have “a definitive way to get doctors' attention and convince them to review patients … by presenting quantifiable evidence of deterioration” because “quantifiable changes are used by doctors to prioritize their workload.” The actions specified in the PEWS algorithm supports nurses to consult with the medical provider at the earliest time that deterioration in patient status is identified.
Nursing autonomy permits supportive interventions to assist the compensating patient before consultation with another clinician. In the absence of a standardized tool, nurses rely on their clinical attentiveness, intuition, and subjective evaluation to interpret clinical findings. Nurses do small interventions in repeated cycles to maintain a patient's vital signs without recognizing that they are compensating for the patient. Scoring from the normal pediatric baseline, as with PEWS, could facilitate greater situational awareness of the narrow range of patient changes to be tolerated without consultation. The results of this study clearly demonstrate a prolonged period of patient change, affording the opportunity for consultation and intervention before a level of decompensation that requires an RRT or a code blue event. Evaluation of tool sensitivity was important to demonstrate the benefit of an objective assessment tool to support concise, prompt team communication of patient status changes over time. One physician stated, “The nice thing about PEWS is that a score of 4 is a 4 regardless of the nurse, their experience, the unit, or the patient's underlying condition.”
There were several limitations to this study. We depended on the accuracy of patient charting and nursing notes to generate retrospective PEWS. Nurses' normalization to their patient population influences their charting by exception and may have contributed to the missing data elements in this retrospective review. Usually the missing domains were behavior state at 25.4% because staff members do not always document the alert state during the day and Cardiovascular color at 26.1% because staff members do not tend to document normal status. Even under these conditions, we have shown that PEWS was highly sensitive in identifying patients' deterioration and was indicative of risk for an RRT or code blue event. The percentage of patients with increased frequency of assessment may have been higher than the reported 7%; however, there was not documentation to back up this belief. The PEWS tool is relatively new. As such, much of the data about its use and results are not yet published, and comparison of results is not possible.
With 85.5% of patients having at least 1 critical PEWS before the RRT or code blue event at a median time of >11 hours, we believe this to be a strong indicator of its potential to alert the care provider of clinical deterioration. Our results are evidence that the PEWS can potentially play a critical role in (1) quantifying patient status, (2) supporting early recognition of clinical deterioration, and (3) promoting concise communication among care providers to alter plans of care in response to changing patient status. Additional studies are imperative to confirm these findings and to describe potential impact of PEWS on clinical outcomes when there is earlier recognition of clinical deterioration.
We acknowledge our colleagues at Cincinnati Children's for sharing their early experience with clinical implementations of PEWS and their ongoing support in our efforts to implement the tool.
- Accepted January 7, 2010.
- Address correspondence to Mari Akre, PhD, MS, MSN, RN, NEA-BC, Children's Hospitals and Clinics of Minnesota, Clinical Nursing Administration, 910 East 26th St, Suite 40-420, Minneapolis, MN 55403. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- RRT =
- rapid response team •
- PEWS =
- Pediatric Early Warning Score •
- APRDRG =
- All Patient Refined Diagnosis Related Groups
- Vandenberg SD,
- Hutchinson JS,
- Parshuram CS
- Tarassenko L,
- Hann A,
- Young D
- Duncan HP
- 11.↵Quist-Therson E and the Hertfordshire Partnership NHS Trust. Acute Children's Services: Guidelines Services: Guidelines for using paediatric early warning scoring tool. National Health System, Hertfordshire Partnership, England. 82006
- Edwards ED,
- Powell CV,
- Mason BW,
- Oliver A
- 16.↵3M3M. APR DRG Software Versions 20 and 24; 2007.All Patient Refined Diagnosis Related Groups. Available at: http://multimedia.3m.com/mws/mediawebserver?66666UuZjcFSLXTtlxfEMxM2EVuQEcuZgVs6EVs6E666666. Accessed February 15, 2010
- Copyright © 2010 by the American Academy of Pediatrics