Sage Journals: Discover world-class research

Abstract

The pediatric early warning score (PEWS) is a tool used to predict clinical deterioration. Referenced vital sign parameters are based on expert opinion but heart rate and respiratory rate percentiles in hospitalized children have been published. This retrospective case-control study of unplanned intensive care unit (ICU) transfers compares evidence-based vital signs (EBVS) effect on PEWS sensitivity and specificity, determines the impact of age categories on PEWS deterioration prediction, and evaluates whether EBVS PEWS is associated with need for invasive ICU supports. EBVS PEWS improved sensitivity (43%-71% vs 30%-63%) for unplanned transfers with slightly decreased specificity (88%-98% vs 93%-99%). Logistic regression analysis and odds ratios (ORs) demonstrated EBVS PEWS was associated with increased risk for ICU-specific supports (OR = 1.16, 95% confidence interval [CI] = 1.0-1.34, P = .0498). Evidence-based vital signs can improve PEWS sensitivity to identify unplanned ICU transfers and identify patients requiring ICU-specific interventions.

Keywords

early warning score hospital medicine clinical deterioration intensive care units heart rate respiratory rate

Introduction

Early identification of clinical deterioration is critically important to decrease pediatric morbidity and mortality. Early warning systems (EWS) are tools comprised of objective vital signs and routine nursing assessments designed to predict clinical deterioration and possible need for intensive care unit (ICU) level care.^1,2 Composite scores trigger escalation policies, including thresholds for multidisciplinary bedside huddles and rapid response team (RRT) activation. The PEWS is widely utilized in acute inpatient settings as a predictor of deterioration^2-5 on both general medical and subspecialty services.⁶

Heart rate (HR) and respiratory rate (RR) measurements are essential to clinical assessments in acute illnesses and are key components of composite PEWS. Accurate reference ranges are essential to identifying abnormal values. While there exist widely referenced resources for pediatric vital sign ranges,^7-10 supporting evidence for these data is generally weak¹¹ and sources are not well defined. Moreover, there are discrepancies between these commonly referenced sources and vital sign data derived from patients admitted to acute care units.¹² Even among early studies, there is variability in the vital signs utilized for PEWS scoring, if these values are listed at all.^1,2,5,13

Bonafide et al determined standardized HR and RR percentile curves based on data collected from 116, 000 vital signs of 14, 000 hospitalized children.¹² This study illustrates the discrepancy between EBVS reference ranges for hospitalized children compared with those previously described in healthy pediatric outpatients.¹¹ Although a previous study showed no significant improvement on bedside PEWS predictions of deterioration using EBVS,¹⁴ no study to date has compared the impact of these vital sign parameters on the Brighton PEWS model. Therefore, we aimed to assess whether PEWS calculated using EBVS was better able to identify patients at risk for deterioration compared with vital signs derived from published reference ranges (institution-based vital signs, IBVS) in children. We sought to determine whether more refined age categories, which would better account for physiologic changes across the pediatric age spectrum, would increase PEWS prediction of deterioration independent of vital sign criteria. Finally, we aimed to demonstrate whether there was an association between a need for ICU-specific therapies following transfer and a change in PEWS calculated using EBVS.

Methods

Study Design

We conducted a retrospective case-control study of patients who experienced unplanned ICU transfers between April 2016 and February 2017 at a 323-bed stand-alone, quaternary care urban, academic children’s hospital. This study was reviewed and approved by the Institutional Review Board at our institution.

Study Population and Design

Cases were defined as patients experiencing unplanned escalations of care from acute care medical and surgical units to the ICU via RRT or code team activation. We included patients transferred to the neonatal, pediatric, and cardiovascular ICUs. We selected cases from a patient database compiled by the Pediatric Early Recognition and Resuscitation Committee (PERRC) at our institution. We excluded cases aged >18 years, second ICU transfer during a hospitalization, length of stay >60 days at time of transfer, or planned ICU transfer (eg, expected post-operative care). The highest total PEWS in the 24 hours prior to ICU transfer was rescored. If multiple scores with equivalent values were present one score was selected at random for rescoring. Controls were randomly matched by primary admission service, age, hospital length of stay, and date of transfer (within 2 weeks). Cases that could not be matched exactly to hospital day at ICU transfer were instead matched to controls of equal or greater hospital length of stay to avoid replacement. Controls were selected from a Research Electronic Data Capture (REDCap)^15,16 database compiled by PERRC and hosted by our institution, which stores maximum PEWS values for all hospitalized patients each day.

Late rescues were defined as patients who received ICU-specific therapies including intubation, mechanical ventilation, vasopressor initiation, and/or cardiopulmonary resuscitation within 12 hours of ICU transfer. These patients were detected with the use of an automated trigger tool¹⁷ that screens for unplanned ICU transfers. Each unplanned transfer was manually screened for presence of late rescue criteria. However, all cases received ICU-level care following transfer, which may include intubation or initiation of inotropic medications after the 12-hour late rescue window.

Vital Sign and Pediatric Early Warning Score Criteria

PEWS in our institution is based on the Brighton PEWS model¹ and has been in use since 2008. Current IBVS have been in use since 2012. They were determined by an expert panel of experienced nurses and critical care physicians who referenced vital sign parameters from other institutions in addition to Pediatric Advanced Life Support (PALS) vital sign guidelines (Table 1). PEWS is utilized exclusively on acute care medical and surgical units and is based on vital sign and routine nursing assessments in 3 domains: behavior, respiratory, and cardiovascular. Possible scores range from 0 to 3 in each domain based on degree of abnormality. Additional points are added to summative PEWS for presence of tracheostomy, continuous albuterol therapy for ≥12 hours, and/or for persistent post-operative emesis (Supplemental Table 1). Summative PEWS of 0 to 2 results in routine monitoring at 4-hour intervals; scores of 3 require physician notification and re-evaluation within 2 hours; scores of 4 or individually scoring 3 in a single category require prompt bedside evaluation by the primary team, bedside nurse, and charge nurse; and scores of 5 or greater mandate RRT activation. The RRT is composed of a PICU fellow or advanced practitioner, PICU charge nurse, and a PICU-based respiratory therapist. PEWS was calculated automatically by the electronic health record once all components were documented by the bedside nurse. The automated process for calculating PEWS within the electronic medical record has been previously published.¹⁸

Table 1.

Institution-Based Vital Sign Criteria for Hospitalized Children.

Age	HR (beats per minute)	Respirations (per minute)
Newborn-5 mo	100-160	30-60
6-11 mo	110-160	24-38
1-2 y	90-150	22-30
3-4 y	80-125	22-30
5-9 y	70-115	20-24
10-13 y	60-100	16-22
≥14 y	60-100	14-20

Abbreviation: HR, heart rate.

Evidence-based vital signs used in the comparison for this study were based on normal distribution curves for vital signs presented by Bonafide et al¹² (Tables 2 and 3). Pediatric early warning score values for EBVS were assigned as follows: <90th percentile = 0; 90th to 94th percentile = 1; 95th to 98th percentile = 2; and ≥99th percentile = 3. Subjective nursing assessments of mental status, perfusion, and respiratory effort were unchanged in their contribution to revised PEWS. The 5 age groupings used in the current IBVS system were, in addition, compared with the 13 subdivided groups presented by Bonafide in the EBVS system.

Table 2.

Evidence-Based Heart Rates by Age and Percentile for Hospitalized Children.

Age group	90th percentile	95th percentile	99th percentile
0-<3 mo	164	171	186
3-<6 mo	159	167	182
6-<9 mo	156	163	178
9-<12 mo	153	160	176
12-<18 mo	149	157	173
18-<24 mo	146	154	170
2-<3 y	142	150	167
3-<4 y	138	146	164
4-<6 y	134	142	161
6-<8 y	128	137	155
8-<12 y	120	129	147
12-<15 y	112	121	138
15-<18 y	107	115	132

Table 3.

Evidence-Based Respiratory Rates by Age and Percentile for Hospitalized Children.

Age group	90th percentile	95th percentile	99th percentile
0-<3 mo	56	62	76
3-<6 mo	52	58	71
6-<9 mo	49	54	67
9-<12 mo	46	51	63
12-<18 mo	43	48	60
18-<24 mo	40	45	57
2-<3 y	37	42	54
3-<4 y	35	40	52
4-<6 y	33	37	50
6-<8 y	31	35	46
8-<12 y	28	31	41
12-<15 y	25	28	35
15-<18 y	23	26	32

Statistical Methods

Characteristics data were reported in summary statistics as mean and standard deviation (SD) or median and interquartile range (IQR) for continuous variables, and frequency and proportion for categorical variables. Primary admission service, age, gender, hospital length of stay, and date of transfer were compared between case and control groups using fixed effects regression. Paired t-tests compared the differences between IBVS and EBVS PEWS within each case and control group. Sensitivity and specificity with 95% confidence intervals (CI) for unplanned transfer were calculated for IBVS and EBVS PEWS. To account for matching in this case-control study, conditional logistic regression including matching variables as covariates was used to calculate the receiver operating characteristic (ROC) area under the curves (AUCs) and reported with the corresponding 95% CIs, and differences between AUCs were compared using the chi-square test. Logistic regression was applied to examine the influence of the breakdown of age categories by including the interaction term of age category and PEWS, and was run separately for case and control groups. Late rescue case analysis was carried out by logistic regression for IBVS and for EBVS, respectively. Odds ratio and 95% CI were used to assess the effect of PEWS to predict patients at risk for late rescues. All data analyses were carried out by SAS software version 9.2 (SAS Institute Inc., Cary, NC). A P value of ≤.05 was considered statistically significant.

Results

Study Population Characteristics

During the study period, PEWS of 400 unique ICU transfers and 800 controls was analyzed and rescored. Characteristics of this population are shown in Table 4. There were no statistically significant differences in patient sex (P = .86), mean age (P = .25), or primary service at the time of ICU transfer (P = .58). There was a statistically significant difference in hospital day at transfer as some controls were not able to be matched exactly to length of hospital stay at case transfer while preserving other matching criteria (P = .02). Table 5 shows the distribution of patients by age category in both institutional and data-driven models.

Table 4.

Characteristics of the Study Population.

Characteristic	Cases(n = 400)	Controls (n = 800)	Total patients(n = 1200)	P value
Sex
Male	219 (54.8%)	445 (55.6%)	664 (55.3%)	.8598
Female	181 (45.2%)	355 (44.4%)	536 (44.7%)
Mean age (months)	68.2	64.3	65.6	.2529
Mean hospital day at transfer	4.4	6.0	5.5	.0209
Service
Hospitalist	194 (48.5%)	388 (48.5%)	582	.5839
Neurology/Neurosurgery	37 (9.3%)	74 (9.3%)	111
Hematology/Oncology/BMT	64 (16%)	128 (16%)	192
Pulmonology	31 (7.8%)	62 (7.8%)	93
Cardiology/Cardiovascular surgery	36 (9%)	72 (9%)	108
Gastroenterology	18 (4.5%)	36 (4.5%)	54
Endocrinology	5 (1.3%)	10 (1.3%)	15
Nephrology	5 (1.3%)	10 (1.3%)	15
Surgery	9 (2.3%)	18 (2.3%)	27
Adolescent	1 (0.3%)	2 (0.3%)	3

Abbreviation: BMT, bone marrow transplant.

Table 5.

Study Population Distribution by Age Category.

	IBVS frequency	EBVS frequency
Age category
0-5 months	265 (22.1%)	—
6-11 months	140 (11.7%)	—
1-2 years	162 (13.5%)	—
3-4 years	119 (9.9%)	—
5-9 years	237 (19.8%)	—
10-13 years	130 (10.8%)	—
≥14 years	147 (12.3%)	—
0-3 months	—	183 (15.3%)
3-<6 months	—	82 (6.8%)
6-<9 months	—	73 (6.1%)
9-<12 months	—	67 (5.6%)
12-<18 months	—	67 (5.6%)
18-<24 months	—	51 (4.3%)
2-<3 years	—	45 (3.8%)
3-<4 years	—	75 (6.3%)
4-<6 years	—	80 (6.7%)
6-<8 years	—	105 (8.8%)
8-<12 years	—	157 (13.1%)
12-<15 years	—	97 (8.1%)
15-<18 years	—	118 (9.8%)

Abbreviations: EBVS, evidence-based vital signs; IBVS, institution-based vital signs.

Supplemental Table 2 demonstrates the non-normal distribution of PEWS by IBVS and EBVS among the total study population. In total, 42.8% of cases and 4.6% of controls had increased scores using EBVS (P ≤ .001). Among cases and controls with initial PEWS of 1 or higher, 4.6% of cases and 31.9% of controls had a reduced score (P ≤ .001).

Evidence-Based Vital Signs Improve the Sensitivity of Pediatric Early Warning Score to Predict Intensive Care Unit Transfer

Table 6 compares the IBVS and EBVS PEWS in case group and control groups. The median PEWS for EBVS was statistically significantly higher than IBVS among cases (4.0, IQR: 2.0, 5.0 vs 3.0, IQR: 2.0, 5.0, P < .0001) and similar in control groups (0.0, IQR: 0, 1.0 vs 0.0, IQR: 0, 1.0). The mean time from the highest recorded PEWS to ICU transfer is 4.67 ± 0.23 hours.

Table 6.

Differences in Pediatric Early Warning Score Among Cases and Controls After Evidence-Based Vital Signs Rescoring.

		Median PEWS (IQR)	P value	Maximum PEWS
Case	IBVS	3.0 (2.0, 5.0)	<.0001	9
Case	EBVS	4.0 (2.0, 5.0)	<.0001	9
Control	IBVS	0.0 (0, 1.0)	<.0001	5
Control	EBVS	0.0 (0, 1.0)	<.0001	6

Abbreviations: EBVS, evidence-based vital signs; IBVS, institution-based vital signs; IQR, interquartile range; PEWS, pediatric early warning score.

Table 7 displays the sensitivities, specificities, and AUCs for IBVS and EBVS by the three clinically significant PEWS cutoffs. The ranges for sensitivity were improved across the three PEWS cutoff points for EBVS compared with IBVS (43%-71% vs 30%-63%) with only slightly decreased specificity (88%-98% vs 93%-99%). There was no statistically significant difference in the AUC for EBVS and IBVS across PEWS thresholds.

Table 7.

Sensitivities and Specificity of Pediatric Early Warning Score to Predict Unplanned Intensive Care Unit Transfer Using Evidence-Based Vital Signs versus Institution-Based Vital Signs.

		Sensitivity, % (95% CI)	Specificity, % (95% CI)	AUC (95% CI)	P value
PEWS ≥5	IBVS	30 (26-34)	99 (98-99)	0.71 (0.68-0.74)	.1521
PEWS ≥5	EBVS	43 (38-47)	98 (97-99)	0.74 (0.71-0.77)	.1521
PEWS ≥4	IBVS	47 (42-52)	97 (96-98)	0.78 (0.75-0.81)	.4588
PEWS ≥4	EBVS	56 (51-61)	95 (93-96)	0.79 (0.77-0.83)	.4588
PEWS ≥3	IBVS	63 (58-67)	93 (91-94)	0.83 (0.80-0.85)	.7067
PEWS ≥3	EBVS	71 (67-76)	88 (86-90)	0.84 (0.80-0.86)	.7067

Abbreviations: AUC, area under the curve; CI, confidence interval; EBVS, evidence-based vital signs; IBVS, institution-based vital signs; ICU, intensive care unit; PEWS, pediatric early warning score.

We next evaluated the impact of EBVS PEWS on positive predictive value (PPV) and number of patients needed to evaluate (NNE) for ICU transfer (Table 8). We extrapolated the data to calculate PPV under the EBVS PEWS system by utilizing an unplanned transfer incidence of 0.67% in our institution. As anticipated, there was a slight decrease in the PPV and subsequent increase in the number of patients requiring evaluation for transfer at each PEWS threshold.

Table 8.

Impact of Evidence-Based Vital Signs on Positive Predictive Value and Number Needed to Evaluate.

		PPV	NNE
PEWS ≥5	IBVS	30.1%	3.3
PEWS ≥5	EBVS	25.3%	3.9
PEWS ≥4	IBVS	15.1%	6.6
PEWS ≥4	EBVS	9.6%	10.4
PEWS ≥3	IBVS	7.7%	13
PEWS ≥3	EBVS	5.5%	17.9

Abbreviations: EBVS, evidence-based vital signs; IBVS, institution-based vital signs; NNE, number needed to evaluate; PEWS, pediatric early warning score; PPV, positive predictive value.

More Refined Age Categories Do Not Improve Pediatric Early Warning Score Prediction of Deterioration

Table 9 presents the Wald tests for the interaction terms of PEWS with age categories for each scenario including IBVS PEWS versus IBVS age, IBVS PEWS versus EBVS age, EBVS PEWS versus IBVS age, and EBVS PEWS versus EBVS age. In each scenario, none of the p values for PEWS ≥5, PEWS ≥4, and PEWS ≥3 were statistically significant, indicating that more granular age categories do not influence the tool’s prediction of deterioration.

Table 9.

Association Between Vital Sign Criteria and Age Categories on Pediatric Early Warning Score Prediction of Deterioration.

		Wald test statistic	P value
Total PEWS	IBVS PEWS × IBVS Age	5.33	.50
	IBVS PEWS × EBVS Age	18.82	.09
	EBVS PEWS × IBVS Age	8.05	.23
	EBVS PEWS × EBVS Age	18.91	.09
PEWS ≥5	IBVS PEWS × IBVS Age	1.18	.98
	IBVS PEWS × EBVS Age	1.27	1.00
	EBVS PEWS × IBVS Age	3.41	.76
	EBVS PEWS × EBVS Age	10.19	.60
PEWS ≥4	IBVS PEWS × IBVS Age	5.00	.55
	IBVS PEWS × EBVS Age	5.54	.94
	EBVS PEWS × IBVS Age	3.35	.76
	EBVS PEWS × EBVS Age	5.92	.92
PEWS ≥	IBVS PEWS × IBVS Age	2.42	.88
	IBVS PEWS × EBVS Age	5.45	.94
	EBVS PEWS × IBVS Age	8.57	.20
	EBVS PEWS × EBVS Age	12.64	.40

Abbreviations: EBVS, evidence-based vital signs; IBVS, institution-based vital signs; PEWS, pediatric early warning score.

Late Rescue Intensive Care Unit Transfers and Patients Requiring Intensive Care Unit -Specific Therapies Have Higher Pediatric Early Warning Score Using Evidence-Based Vital Signs

Table 10 describes the ICU-specific therapies cases received following transfer and the influence of EBVS on the PEWS of these patients. The median PEWS was increased across all groups when comparing EBVS and IBVS PEWS, and all changes were statistically significant. There was a small proportion of patients whose PEWS was decreased with EBVS ranging from 1% to 11%. In contrast, a substantial proportion of each ICU-specific intervention group had increased PEWS with EBVS, ranging from 49% to 88%. “Other” therapies included continuous insulin infusions, continuous electrolyte infusions, exchange transfusion, and therapies requiring intensive monitoring such as anticoagulation infusions, antimalarial agents, or specialized respiratory supports (eg, Heliox). Patients who did not receive one of the listed ICU-specific therapies in the first 12 hours still received critical care support and monitoring following their transfer.

Table 10.

Change in Pediatric Early Warning Score by Intensive Care Unit -Specific Therapies.

	Frequency (%)	Median PEWS (IBVS)	Median PEWS (EBVS)	P value	Case frequency with increased EBVS PEWS	Case frequency with decreased EBVS PEWS	Mean change in EBVS PEWS
HFNC	114 (28.5)	4.0	5.0	<.001	67 (58.8%)	2 (1.8%)	0.7
NIPPV	93 (23.3)	4.0	5.0	<.001	46 (49.5%)	6 (6.5%)	0.7
Mechanical ventilation	31 (7.8)	4.0	5.0	.0003	16 (51.6%)	2 (6.5%)	0.8
Vasopressor initiation	27 (6.8)	2.0	5.0	<.0001	20 (74.1%)	2 (7.4%)	1.1
CPR	1 (0.3)	2.0	2.0	—	—	—	—
Other	9 (2.3)	2.0	4.0	.013	8 (88.9%)	1 (11.1%)	0.9
No ICU-specific Intervention	172 (43)	3.0	3.0	<.0001	57 (33.1%)	12 (7.0%)	0.4

Abbreviations: CPR, cardiopulmonary resuscitation; EBVS, evidence-based vital signs; HFNC, high flow nasal cannula; IBVS, institution-based vital signs; ICU, intensive care unit; NIPPV, non-invasive positive pressure ventilation; PEWS, pediatric early warning score.

Among the 400 cases, 47 cases were identified as late rescues (Table 11). Mean PEWS for late rescues was 2.0 with SD of 3.0 for IBVS and 5.0 with SD of 3.0 for EBVS. Compared with non-late rescue cases, EBVS PEWS was statistically associated with the risk of late rescues with OR of 1.16 (95% CI = 1.0-1.34, P = .0498). There was no significant association between IBVS PEWS and the risk of late rescues with the OR of 1.05 (95% CI = 0.91-1.22, P = .502). There was no statistically significant difference when adjusting for age and gender covariates among late rescue and non-late rescue groups.

Table 11.

Association Between Pediatric Early Warning Score and Late Rescue Intensive Care Unit Transfers.

	Mean PEWS (SD)	OR (95% CI)	P value
IBVS	3.36 (2.2)	1.05 (0.91-1.22)	.502
EBVS	4.30 (2.1)	1.16 (1.0-1.34)	.0498

Abbreviations: CI, confidence interval; EBVS, evidence-based vital signs; IBVS, institution-based vital signs; ICU, intensive care unit; OR, odds ratio; PEWS, pediatric early warning score.

Discussion

To our knowledge, this is the first study to investigate the effect of EBVS on the Brighton PEWS model. Our retrospective analysis demonstrated that EBVS resulted in a small but non-statistically significant improvement in the sensitivity of PEWS to identify risk for clinical deterioration in patients admitted to acute care units who experienced unplanned ICU transfer. EBVS PEWS did not significantly improve the PPV of the tool. Patients meeting late rescue criteria and requiring ICU-specific therapies were more likely to have had elevated PEWS using EBVS. This suggests that the degree of vital sign abnormality may have been underestimated by current parameters in patients who received escalations of care and that EBVS may have a greater impact to detect more severe deterioration events. In addition, more granular age categories do not significantly improve the sensitivity of PEWS prediction of deterioration.

Previous research has similarly demonstrated no significant improvement of EBVS using the bedside PEWS model.¹⁴ Both studies share similarities in study methodology, although Ross et al converted historical data for subjective scoring of the respiratory data set to generalize to their population. In contrast, we preserved the patient-level data collected for each member of the study population, which included an assessment of respiratory effort at the time of data collection. In both cases, application of EBVS did not significantly alter the overall predictive utility of either PEWS model. In addition, the case population studied by Ross et al excluded patients requiring ICU transfer exclusively for intensive nursing care. While this allows for analysis of a larger population requiring ICU-specific medical therapies, it is not entirely representative of the population that receives care in many critical care settings, particularly in those larger academic settings that provide care for a high proportion of children with medical complexity and intensive nursing needs. Together, these studies reinforce the limitations of existing EWS tools to reliably predict deterioration in the pediatric population.

There is significant variability among the vital sign parameters currently implemented in PEWS scoring, with previously little evidence to support the use of these criteria. Textbooks and reference range resources^7-9 including PALS guidelines¹⁰ that currently inform vital sign parameters for PEWS are largely based on consensus opinion. A meta-analysis¹¹ of vital signs from well children in outpatient settings highlights that these reference ranges correspond poorly to data-driven assessments of percentile curves in the well pediatric population. Bonafide et al have demonstrated even more substantial discrepancies among hospitalized non-ICU patients, citing that 38% of vital sign values measured in inpatient settings would be assigned points in PEWS using parameters developed by expert opinion. The original Brighton PEWS publication intentionally omitted specific vital sign parameters for ease of manual score calculations; however, the introduction of automated PEWS calculations in electronic medical records has removed this labor-intensive barrier. While there may be limited effectiveness of EBVS on PEWS systems, there is still value in utilizing evidence-based parameters for monitoring hospitalized children during routine care and in periods of clinical deterioration.

The physiologic variability of HR and RR by age in children adds additional layers of complexity to PEWS. While EBVS percentile curves demonstrate subtle variations with age on HR and RR, we have demonstrated that this variable is not independently associated with deterioration and does not significantly alter PEWS sensitivity. The reason for this lack of effect is not clearly understood, however may be related to broad definitions of “normal” vital signs for age group in previous PEWS models as a means to simplify calculations prior to automation. This data may have other applications outside of PEWS, however, including in assigning physiologic alarm parameters that may impact alarm fatigue. The cognitive burden on physicians and nurses to recognize abnormal vital signs should be minimized as safely as possible without sacrificing the sensitivity of the system to detect abnormalities.

Integration of vital sign criteria that are more specific to inpatient settings should lead to earlier recognition of decompensation and more timely initiation of goal-directed therapies. This may result in differences in clinical management that may mitigate or promote earlier ICU team involvement. In addition, a more sensitive EWS system may facilitate clinical interventions in patients at high risk of clinical decompensation and ultimately may decrease the rate of preventable ICU transfers or the intensity of therapies to rescue patients transferred to the ICU. Given the retrospective application of EBVS in this study, it is difficult to fully quantify implications of our results on clinical management in real time. Follow-up studies will be necessary to assess outcome measures related to rates of ICU transfers and patient morbidity.

This study has several limitations. First, while PEWS models across institutions share similarities, there are variations in system design and specific vital sign criteria that may impact the generalizability of this study. The magnitude of the impact of EBVS PEWS in a given institution may vary compared with our results. However even among the initial Brighton PEWS validation studies, there is variation among the vital sign criteria listed,^2,5 if listed at all.^1,13 The lack of validated and universally utilized vital signs is a major limitation to PEWS research on the whole and is one which this study seeks to address. Second, our model for PEWS only takes into account vital sign abnormalities ≥90th percentile when calculating total score and does not incorporate criteria for severe bradycardia or bradypnea, which may signal impending cardiac or respiratory arrest. Additional studies are needed to determine the effect that incorporating these parameters would have on PEWS sensitivity and specificity and clinician notification fatigue. Third, transfers meeting criteria for late rescues were rare, and therefore, we may lack statistical power to detect significant differences in the IVBS regression analysis for severe deterioration. Dedicated studies in this population will be beneficial moving forward and may include analysis of other metrics of deterioration.^19,20 Finally, this retrospective study analyzes a single timepoint in a patient’s hospital course prior to ICU transfer. The highest IBVS PEWS does not necessarily correlate with the highest possible EBVS PEWS within the given time period, and these data do not inform time from a clinician’s first recognition of decompensation in patients transferred to the ICU. Moreover, a single PEWS value does not, in isolation, hold any specific predictive value for ICU transfer in the absence of clinical context. Additional research will provide valuable data to characterize the effect of RRT utilization and post-transfer morbidity following EBVS implementation.

Conclusion

In summary, we have demonstrated that EBVS resulted in small though non-statistically significant improvements in Brighton-based PEWS sensitivity to identify unplanned ICU transfers from acute care units. We demonstrated that EBVS were better able to identify patients requiring ICU-specific interventions. However, EBVS did not significantly impact PEWS sensitivity by age. EBVS PEWS may be beneficial to facilitate directed interventions in early clinical decompensation to reduce unplanned ICU transfers, and may be used in conjunction with other track and trigger tools to minimize preventable morbidity.

Author Contributions

Dr LR and Dr ND: Conceptualized and designed this study. Dr LR: Collected and scored the data. Dr AZ: Performed statistical analysis of the data. Dr LR: Drafted and edited the manuscript with assistance from Dr AZ. Dr ND: Provided mentorship, oversight, and critical review of the data and manuscript. All authors revised and reviewed the manuscript and approved the final work as submitted.

Supplemental Material

sj-pdf-1-cpj-10.1177_00099228231166264 – Supplemental material for Use of Evidence-Based Vital Signs in Pediatric Early Warning Score to Predict Clinical Deterioration on Acute Care Units

Supplemental material, sj-pdf-1-cpj-10.1177_00099228231166264 for Use of Evidence-Based Vital Signs in Pediatric Early Warning Score to Predict Clinical Deterioration on Acute Care Units by Lisa Rickey, Anqing Zhang and Nathan Dean in Clinical Pediatrics

Supplemental Material

sj-pdf-2-cpj-10.1177_00099228231166264 – Supplemental material for Use of Evidence-Based Vital Signs in Pediatric Early Warning Score to Predict Clinical Deterioration on Acute Care Units

Supplemental material, sj-pdf-2-cpj-10.1177_00099228231166264 for Use of Evidence-Based Vital Signs in Pediatric Early Warning Score to Predict Clinical Deterioration on Acute Care Units by Lisa Rickey, Anqing Zhang and Nathan Dean in Clinical Pediatrics

Footnotes

Acknowledgements

We acknowledge the assistance of Ms Susan Keller in a review of the literature for this study and thank Ms Parissa Safari who maintains the PERRC patient database.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Lisa Rickey

Supplemental Material

Supplemental material for this article is available online.

References

Monaghan

. Detecting and managing deterioration in children. Paediatr Nurs. 2005;17(1):32-35.

Skaletzky

Raszynski

Totapally

. Validation of a modified pediatric early warning system score: a retrospective case-control study. Clin Pediatr. 2012;51(5):431-435. doi:10.1177/0009922811430342.

Subbe

Kruger

Rutherford

Gemmel

. Validation of a modified Early Warning Score in medical admissions. QJM. 2001;94(10):521-526.

Haines

Perrott

Weir

. Promoting care for acutely ill children—development and evaluation of a paediatric early warning tool. Intensive Crit Care Nurs. 2006;22(2):73-81. doi:10.1016/j.iccn.2005.09.003.

Akre

Finkelstein

Erickson

Liu

Vanderbilt

Billman

. Sensitivity of the pediatric early warning score to identify patient deterioration. Pediatrics. 2010;125(4):e763-e769. doi:10.1542/peds.2009-0338.

Dean

Fenix

Spaeder

Levin

. Evaluation of a pediatric early warning score across different subspecialty patients. Pediatr Crit Care Med. 2017;18(7):655-660. doi:10.1097/PCC.0000000000001176.

Kleinman

McDaniel

Molloy

. The Harriet Lane Handbook: The Johns Hopkins Hospital. 22nd ed. Philadelphia, PA: Elsevier; 2020.

Kliegman

St. Geme

. Nelson Textbook of Pediatrics. 21st ed. Philadelphia, PA: Elsevier; 2020.

Hazinski

. Manual of Pediatric Critical Care. Vol. 974. Philadelphia, PA: Elsevier Mosby; 1999.

10.

Chameides

Samson

Schexnayder

Hazinski

. Pediatric assessment. In: Ralston

Hazinski

Zaritsky

Arno L.

Schexnayder

Stephen M.

(eds.) Pediatric Advanced Life Support: Provider Manual. Dallas, TX: American Heart Association; 2006:9-16.

11.

Fleming

Thompson

Stevens

, et al. Normal ranges of heart rate and respiratory rate in children from birth to 18 years of age: a systematic review of observational studies. Lancet. 2011;377(9770):1011-1018. doi:10.1016/S0140.

12.

Bonafide

Brady

Keren

Conway

Marsolo

Daymont

. Development of heart and respiratory rate percentile curves for hospitalized children. Pediatrics. 2013;131(4):e1150-e1157. www.aappublications.org/news.

13.

Tucker

Brewer

Baker

Demeritt

Vossmeyer

. Prospective evaluation of a pediatric inpatient early warning scoring system. J Spec Pediatr Nurs. 2009;14(2):79-85.

14.

Ross

Harrysson

Goel

, et al. Integration of single-center data-driven vital sign parameters into a modified pediatric early warning system. Pediatr Crit Care Med. 2017;18(5):469-476. doi:10.1097/PCC.0000000000001150.

15.

Harris

Taylor

Thielke

Payne

Gonzalez

Conde

. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42(2):377-381. doi:10.1016/j.jbi.2008.08.010.

16.

Harris

Taylor

Minor

, et al. The REDCap consortium: building an international community of software platform partners. J Biomed Inform. 2019;95:103208. doi:10.1016/j.jbi.2019.103208.

17.

Lemon

Stockwell

. Automated detection of adverse events in children. Pediatr Clin North Am. 2012;59(6):1269-1278. doi:10.1016/j.pcl.2012.08.007.

18.

Dean

Cheng

Crumbley

DuVal

Maldonado

Ghebremariam

. Improving accuracy and timeliness of nursing documentation of pediatric early warning scores. Pediatr Qual Saf. 2020;5(2):e278. doi:10.1097/pq9.0000000000000278.

19.

Hussain

Sosa

Ambroggio

Gallagher

Brady

. Emergency transfers: an important predictor of adverse outcomes in hospitalized children. J Hosp Med. 2019;14(8):482-485. doi:10.12788/jhm.3219.

20.

Bonafide

Localio

Roberts

Nadkarni

Weirich

Keren

. Impact of rapid response system implementation on critical deterioration events in children. JAMA Pediatr. 2014;168(1):25-33. doi:10.1001/jamapediatrics.2013.3266.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.04 MB

0.02 MB