BACKGROUND AND OBJECTIVES:

The use of Pediatric Early Warning Scores is becoming widespread to identify and rapidly respond to patients with deteriorating conditions. The ability of Pediatric Early Warning Scores to identify children at high risk of deterioration or death has not, however, been established in resource-limited settings.

METHODS:

We developed the Pediatric Early Warning Score for Resource-Limited Settings (PEWS-RL) on the basis of expert opinion and existing scores. The PEWS-RL was derived from 6 equally weighted variables, producing a cumulative score of 0 to 6. We then conducted a case-control study of admissions to the pediatrics department of the main public referral hospital in Kigali, Rwanda between November 2016 and March 2017. We defined case patients as children fulfilling the criteria for clinical deterioration, who were then matched with controls of the same age and hospital ward.

RESULTS:

During the study period, 627 children were admitted, from whom we selected 79 case patients and 79 controls. For a PEWS-RL of ≥3, sensitivity was 96.2%, and specificity was 87.3% for identifying patients at risk for clinical deterioration. A total PEWS-RL of ≥3 was associated with a substantially increased risk of clinical deterioration (odds ratio 129.3; 95% confidence interval 38.8–431.6; P <.005).

CONCLUSIONS:

This study reveals that the PEWS-RL, a simple score based on vital signs, mental status, and presence of respiratory distress, was feasible to implement in a resource-limited setting and was able to identify children at risk for clinical deterioration.

What’s Known on This Subject:

Pediatric Early Warning Scores in Western health systems can be used to identify children at risk for deterioration, allowing earlier intervention. When used with a rapid-response team, they can reduce clinical deterioration events, emergency resuscitations, and PICU transfers.

What This Study Adds:

This study reveals that in a single center in Rwanda, a resource-limited setting, we were able to implement a simplified Pediatric Early Warning Score with good predictive ability to identify those children at risk for clinical deterioration.

With inpatient pediatric mortality rates of 5% to 15% at many hospitals in resource-limited settings, there is a great deal of room for improvement in survival rates.1 One potential intervention is the early identification and triage of patients by using a Pediatric Early Warning Score (PEWS). PEWSs are tools designed to detect the early deterioration of pediatric inpatients with the goal of reducing in-hospital morbidity and mortality. The use of such PEWSs is becoming widespread among Western health systems.2,8 Studies have shown that PEWSs in these settings can be used to identify >80% of children at risk for deterioration, allowing for earlier intervention.1,6 They are often used in conjunction with a rapid-response team and have been found to be associated with fewer clinical deterioration events,9 emergency resuscitations,10 and PICU transfers.11 

These PEWSs, however, have been minimally studied in low-resource settings, where the staffing ratios and level of training of nurses and other providers are often different. It is unclear whether these same factors will prove to be significant in a scoring system and whether they can be reliably and consistently collected. Given these differences, there have been previous attempts to adapt the PEWS concept to these settings. The inpatient triage assessment and treatment was the first PEWS developed and validated in a low-resource setting in Malawi.12 However, this score, unlike PEWS tools in better-resourced settings, uses solely vital signs, and any indicators of behavior or respiratory effort are not examined because of lack of trained personnel and available data. The use of solely vital signs leads to an easily administered yet less sensitive score (sensitivity 44%) that misses a significant proportion of children at risk for deterioration.12 

In middle-income countries, modifications to existing PEWS systems have generated more sensitive and specific performance. For instance, Agulnik et al11,13,14 modified the Boston Children’s Hospital Children’s Hospital Early Warning Score for use in a pediatric oncology hospital in Guatemala and found that it was highly correlated with the need for PICU transfer (area under receiver operating characteristic: 0.96 and 0.94) and, when combined with an escalation of care algorithm, resulted in lower rates of inpatient clinical deterioration events and PICU admissions, but no sensitivity or specificity of the score was calculated. Similarly, the Brighton PEWS, a score that relies on clinical evaluation without the need for measurement of vital signs, was modified, translated, and studied in Brazil by Miranda et al,15 and it was found to have a sensitivity of 90.2% and a specificity of 74.4%. However, both the modified Brighton PEWS–Brazil and the Children’s Hospital Early Warning Score require a significant amount of clinical judgment by nurses that may be absent in lower-resource settings.

Many nurses and clinical assistants in low-resource settings have far less training than those in developed settings, including some with only high school level education or 1 or 2 years of nursing education,16,19 making reliable administration and calculation of these scores difficult in these settings. Furthermore, many of these scores contain a large number of elements2,6,8,20 and are time consuming in hospitals with an overwhelming burden of disease and low nurse-to-patient ratios.

Therefore, we developed and assessed the validity and test characteristics of a novel version of a PEWS, the Pediatric Early Warning Score for Resource-Limited Settings (PEWS-RL), for use in resource-limited settings.

Centre Hospitalier Universitaire de Kigali (CHUK) is a large tertiary hospital in Kigali, Rwanda, that receives ∼70% of the referred case patients from across the country. The pediatric department of CHUK consists of 66 beds and is divided into 4 units: the pediatric emergency department, pediatric wards (including PICU), the pediatric outpatient department, and neonatology. In 2015, the pediatric department admitted 2242 patients, with a mortality rate of 10.4%, and the PICU admitted 108 patients.21 

We developed the PEWS-RL on the basis of expert opinion and existing PEWS scoring systems. A preliminary chart review was conducted to assess nursing documentation and frequency of assessments. A pilot study was then done to attempt to assess utility of elements taken from a wide variety of existing PEWS (eg, doctor, nurse, and family concern; vital signs; respiratory distress severity; receiving oxygen; alert, verbal, painful, unresponsive mental status; capillary refill; color; behavior; and presence of sclerema). We were able to determine that certain measures, such as parental or nursing concern, were consistently not used by our nurses, likely because of cultural differences. Similarly, measures such as assessing the level of mental status on a Glasgow Coma Scale or an alert, verbal, painful, unresponsive scale or assessing the severity of respiratory distress, capillary refill, or color were beyond the skill level or time availability of nurses. We also discovered that nurses were confused by having to calculate a score that was based on a separate tool. Our goal was to create a tool that could be easily administered without a great deal of training in the subjective assessment of patients. We found that the South Children’s Observation and Severity Tool Children’s Early Warning Score22 simplified score calculation dramatically by merely requiring nurses to add up the number of marks that fell in a highlighted area of the vital signs chart. We therefore adapted their format to include our target elements of assessment.

Our PEWS-RL initially consisted of 7 variables, with a possible score on each variable of 0 or 1, producing a cumulative score between 0 and 7, with higher scores corresponding to a greater acuity. These variables included the following: respiratory rate, respiratory distress (defined as any increased work of breathing), heart rate, temperature, blood pressure (BP), oxygen use, and mental status (normal versus abnormal). Nurse training was done on signs of respiratory distress, including retractions, nasal flaring, fast breathing, inability to speak in full sentences, etc, although no specific study definition was laid out. In earlier iterations, we were unsuccessful in having nurses grade the degree of respiratory distress according to more specific definitions. Normal vital signs ranges by age were taken from the South Children’s Observation and Severity Tool Children’s Early Warning Score.22 Although we originally included BP in our score, during analysis we discovered that few BPs were actually recorded, likely secondary to lack of availability of appropriate cuff sizes. Therefore, this measure was removed before analysis, and the PEWS-RL was modified to consist of 6 variables, producing a cumulative score between 0 and 6. However, BP remained in the vital signs charts, and PEWS calculations remained out of a total of 7 in the patient files.

The PEWS-RL was incorporated into standard practice at the hospital and became the official vital signs collection form for admitted patients23 (Supplemental Figs 4–13). To reflect changes in normal vital signs ranges by age, we used 5 different collection sheets that were inserted into the files on the basis of the patient’s age (0–1 and 1–11 months and 1–4, 5–12, and ≥13 years). Training sessions were conducted with all pediatric nurses on vital sign assessment skills and PEWS-RL calculation and documentation.

This was a case-control study. We drew a convenience sample of case patients and controls from all children from 0 to 18 years of age admitted to the pediatric department for >24 hours between November 2016 and March 2017. A case patient had to meet at least 1 of the criteria below (which were selected to capture events that would trigger an admission to a PICU in higher-resource settings) to constitute a clinical deterioration:

  • use of bag-mask ventilation (BMV);

  • cardiopulmonary resuscitation (CPR);

  • use of noninvasive respiratory support (continuous positive airway pressure);

  • use of mechanical ventilation;

  • transfer to the PICU;

  • use of inotropes; and

  • death.

We then selected controls among children aged 0 to 18 years admitted to the pediatric department for >24 hours between November 2016 and March 2017 who did not exhibit any of the clinical deterioration criteria listed above. Controls were matched to case patients by calendar day, hospitalization site (ward), and age group (<5 and ≥5 years).

The following groups of patients were excluded:

  • Patients whose parent or caretaker refused consent;

  • patients admitted to the NICU;

  • patients already admitted to the PICU before study enrollment; and

  • patients whose conditions deteriorated within the first 24 hours of admission.

For each enrollee, we recorded the following data: age, sex, HIV status, nutrition status, duration of hospitalization, and final diagnosis. Malnutrition in children <5 years of age was defined as a weight-for-age z score < −2 by using World Health Organization 2006 standards.24 Malnutrition for children aged ≥5 years was defined as a weight-for-age less than the third percentile by using Centers for Disease Control and Prevention 2000 growth charts.25,26 We also considered bilateral edema from nutritional origin as malnutrition, as recommended by the World Health Organization.24 

Nurses recorded the PEWS-RL concurrent with each vital sign collection. Nurses were instructed by hospital administration that physician notification was required for a first score of ≥3 or for an escalation of score by >3 points within 24 hours for future scores. These instructions were incorporated into the PEWS-RL form, with a section to record the time of physician notification and response. We abstracted all PEWS-RLs for the 24 hours before clinical deterioration for case patients. For each matched control, we used the PEWS-RLs collected during the same calendar day that the corresponding case patient’s scores were collected. We used the highest PEWS-RL within the 24-hour surveillance period for analysis. We did not collect data on physician notification or physician response, and no quality improvement data or education was done regarding these notifications because we were first attempting to validate the predictive ability of the score.

This study was approved by the Rwanda National Research Ethics Committee through the Institutional Review Boards at CHUK and the University of Rwanda College of Medicine.

Univariate statistics (χ2 test and Wilcoxon rank sum test for categorical and continuous variables, respectively) were used to compare case patients and controls on demographic and baseline clinical factors. To test the association between clinical deterioration and the modified PEWS-RL, we estimated a logistic regression model with case-patient status as the dependent variable and the PEWS-RL as the independent variable. To further assess the discriminatory ability of the modified PEWS-RL, we generated a receiver operating characteristic (ROC) curve with case-patient status as the outcome variable and the modified PEWS-RL as the predictor variable. As an overall measure of classification accuracy, we calculated the area under the curve and the corresponding 95% confidence interval (CI). We also calculated the sensitivity and specificity, derived from using each PEWS-RL value (from 1 to 6) as the cut point to define case-patient status.

Because of the unforeseen loss of our paper data collection forms, we were unable to maintain the linkage data that matched each case patient to his or her control. However, all other individual data were recorded for analysis after input and data checking.

To address any bias this may have introduced because of the paired nature of our data, a sensitivity analysis was conducted, in which we randomly assigned a control to each case patient within an age-group stratum and estimated a conditional logistic regression model using case-patient status as the dependent variable and the modified PEWS as the independent variable. We performed 1000 iterations of this procedure, each time randomly creating case-control pairs. We calculated the mean odds ratio (OR) and 95% CI bounds resulting from these iterations.

During the study period, there were 627 patients admitted to the pediatrics department. Among them, 104 case patients were eligible for the study. Eight case patients were excluded because their conditions deteriorated within 24 hours of admission, and 12 were excluded because they were admitted to the PICU directly from the operating room. Sixteen patients were excluded secondary to incomplete study data. In total, 68 patients fulfilled the inclusion criteria and were enrolled in the study, and 70 controls were also enrolled (Fig 1). All families that were approached consented to participation.

FIGURE 1

Patient flow diagram.

FIGURE 1

Patient flow diagram.

Close modal

The demographic characteristics of the sample are presented in Table 1. As shown, no statistically significant differences between case patients and controls were detected in age, sex, malnutrition status, or HIV status.

TABLE 1

Demographic and Clinical Characteristics

CharacteristicControls (n = 70)Case Patients (n = 68)Pa
Age, mo, median (IQR) 17 (1–132) 11 (3–72) .65 
Male sex, n (%) 44 (63) 39 (57) .51 
Malnutrition status, n (%) 31 (47) 27 (40) .40 
Positive HIV status, n (%) 0 (0) 1 (2) .99 
Hospital length of stay, median (IQR), d 23 (17–32) 15 (5–27) <.01 
CharacteristicControls (n = 70)Case Patients (n = 68)Pa
Age, mo, median (IQR) 17 (1–132) 11 (3–72) .65 
Male sex, n (%) 44 (63) 39 (57) .51 
Malnutrition status, n (%) 31 (47) 27 (40) .40 
Positive HIV status, n (%) 0 (0) 1 (2) .99 
Hospital length of stay, median (IQR), d 23 (17–32) 15 (5–27) <.01 

IQR, interquartile range.

a

Case-control comparison by Wilcoxon rank sum test for continuous variables; χ2 test or Fisher’s exact test used for categorical variables.

The types of clinical deterioration events among case patients are displayed in Table 2. The most common deterioration event was death, followed by BMV and CPR. Eighty-one percent of case patients had ≥2 causes documented.

TABLE 2

Types of Clinical Deterioration Events Among Case Patients (n = 68)

Clinical Deterioration Indicatorn (%)
Death 57 (84) 
Mechanical support 11 (16) 
Noninvasive respiratory support 9 (13) 
BMV 54 (79) 
PICU admission 12 (18) 
Use of inotropes 13 (19) 
CPR 49 (72) 
Clinical Deterioration Indicatorn (%)
Death 57 (84) 
Mechanical support 11 (16) 
Noninvasive respiratory support 9 (13) 
BMV 54 (79) 
PICU admission 12 (18) 
Use of inotropes 13 (19) 
CPR 49 (72) 

Categories are not mutually exclusive.

The risk factors captured by the modified PEWS-RL were more common among case patients than among controls, as depicted in Fig 2. In a logistic regression model, the PEWS-RL was a significant predictor of case-patient status (OR 4.93; 95% CI 2.98–8.13). The ROC curve of the PEWS-RL used to predict case-patient status is shown in Fig 3. The overall discriminatory power of the PEWS-RL to detect case-patient status, as measured by the area under the curve, was 0.96 (95% CI 0.93–0.99). The test characteristics resulting from the use of each value along the range of PEWS-RLs as a cut point to define case-patient status are shown in Table 3. For a PEWS-RL of ≥3, sensitivity was 96.2%, and specificity was 87.3% for identifying patients at risk for clinical deterioration. A total PEWS-RL of ≥3 was associated with a substantially increased risk of clinical deterioration (OR 129.3; 95% CI 38.8–431.6; P < .005).

FIGURE 2

Highest PEWS.

FIGURE 3

ROC curve.

TABLE 3

Test Characteristics of PEWS

CutoffSensitivity, %Specificity, %
≥0 100.0 0.0 
≥1 98.5 42.9 
≥2 98.5 75.7 
≥3 94.1 85.7 
≥4 88.2 95.7 
≥5 63.2 98.6 
≥6 23.5 100.0 
CutoffSensitivity, %Specificity, %
≥0 100.0 0.0 
≥1 98.5 42.9 
≥2 98.5 75.7 
≥3 94.1 85.7 
≥4 88.2 95.7 
≥5 63.2 98.6 
≥6 23.5 100.0 

In sensitivity analyses, we randomly assigned a control to each case patient within an age-group stratum and estimated a conditional logistic regression model. Across 1000 replications of this procedure, the mean OR was virtually identical to the standard logistic regression at 4.94 ± 1.72, with a range of 2.26 to 11.36. The mean lower bound of the 95% CI was 1.66 ± 0.15, with a range from 1.23 to 2.01. Thus, in every replication, the PEWS remained a significant predictor of case-patient status.

The mortality rate of pediatric inpatients in low-income countries remains unacceptably high. In this study, we found that an easily administered clinical scoring algorithm, the PEWS-RL, could be implemented and used to detect risk of subsequent clinical deterioration. With higher inpatient mortality rates and lower rates of availability of ICU services in resource-limited settings, interventions that are based on such scores may make more of an impact than in developed settings. The earlier identification of those who are at risk for deterioration allows earlier intervention before ICU-level services are required (as ICU services are frequently unavailable).

The PEWS-RL is based solely on vital signs and the presence or absence of an abnormal mental status and/or of respiratory distress. It was easily integrated into our patient files in the same format as the previous vital signs form, eliminating the need for duplicate documentation and making the transition as simple as possible for nurses. It does not require any calculation of the score but rather just adding up the number of responses (up to 6) that fall in a red zone. This avoids a problem with existing PEWSs that include a substantial number of items requiring subjective measures that rely on nursing education and additional calculations.2,4,7,9,27,29 

Despite this simplicity, we were able to demonstrate excellent sensitivity and specificity with this score, which was able to match or prove superior to most previously identified PEWSs in developed settings.6,9,30,33 By using a cutoff score of 3, the sensitivity and specificity were 94.1% and 85.7%, respectively. At a cutoff score of 2, we gain sensitivity as expected, but our specificity decreases to 75.7%, leading to more children being inaccurately identified as at risk for deterioration and more physicians notified unnecessarily. This can be problematic in a setting with limited physician and nursing resources.

This study has several limitations worth noting. It was only conducted at a single center (CHUK), and therefore, it is unknown whether the predictive ability of the score will translate to other hospitals with different patient populations and health care workers. If the level of training of the physicians, nurses, and clinical assistants differs substantially, they may be more or less able to complete the score accurately. It clearly needs to be replicated to ensure that it holds its test characteristics across a wide variety of low-resource settings. The decision to use the highest score within a 24-hour period was made because of the infrequency of vital signs within this setting, ensuring that we captured all patients with elevated scores. However, this method could have artificially inflated the score performance. Given the effect size, we believe the influence of this potential error to be small, but replication by using the score within a certain number of hours would more accurately reflect score performance if we had better data collection abilities within our setting. Furthermore, our relatively small sample size is a limitation, but given the high mortality rate, we were still able to demonstrate a substantial difference between groups with good sensitivity and specificity. Our analysis was also limited by our loss of data pairing because of the early destruction of our paper data-entry forms before pairing entry into our database. Unfortunately, data collection in resource-limited settings is often paper-based because of funding. Also, research assistant training and compliance with protocols can be more challenging in these settings. Although this is a significant limitation to our analysis that was unplanned, we believe that we were able to demonstrate with our sensitivity analysis that, regardless of the pairing, our results were consistent.

This study demonstrates that the PEWS-RL can be easily integrated into clinical care and has excellent test characteristics for identifying those children at risk for clinical deterioration at a tertiary hospital in Rwanda. A PEWS of ≥3 was associated with an increased risk of clinical deterioration. Further study is needed to assess the performance of the PEWS-RL across a variety of different low-resource settings and to evaluate its impact on clinical care and outcomes. We are currently undertaking a study looking at the impact of PEWS-RL nursing and physician education and a rapid-response team system on PEWS-RL implementation measures and inpatient pediatric mortality.

Dr Rosman conceptualized and designed the Pediatric Early Warning Score for Resource-Limited Settings, conceptualized and designed the study, assisted with the design of the data collection instrument and oversaw data collection, drafted the initial manuscript, and reviewed and revised the manuscript; Dr Karangwa conceptualized and designed the study, designed the data collection instruments, collected data, drafted the initial manuscript, and reviewed and revised the manuscript; Dr McCall conceptualized and designed the study, assisted with the design of the data collection instrument and oversaw data collection, and reviewed and revised the manuscript; Dr Law consulted on the study design, contributed to data analysis, drafted the initial manuscript, and reviewed and revised the manuscript; Dr Monuteaux consulted on the study design, conducted data analysis, drafted the initial manuscript, and reviewed and revised the manuscript; Ms Briscoe conceptualized and designed the Pediatric Early Warning Score for Resource-Limited Settings, conceptualized and designed the study, and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

FUNDING: Dr Law received salary support through a Canada Research Chair and a Michael Smith Foundation for Health Research Scholar award.

COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2018-4058.

We thank the CHUK hospital administration for their recognition of the importance of incorporating tools for the early recognition of clinical deterioration into their patient care systems. We also thank the pediatric nurses and pediatric residents for their willingness to learn and adapt to a new system in an effort to improve clinical care for their patients. Finally, we thank Drs Lisine Tuyisenge and Emmanuel Rusingiza for their support in conducting this study.

BMV

bag-mask ventilation

BP

blood pressure

CHUK

Centre Hospitalier Universitaire de Kigali

CI

confidence interval

CPR

cardiopulmonary resuscitation

OR

odds ratio

PEWS

Pediatric Early Warning Score

PEWS-RL

Pediatric Early Warning Score for Resource-Limited Settings

ROC

receiver operating characteristic

1
Obermeyer
Z
,
Abujaber
S
,
Makar
M
, et al;
Acute Care Development Consortium
.
Emergency care in 59 low- and middle-income countries: a systematic review.
Bull World Health Organ
.
2015
;
93
(
8
):
577
586G
[PubMed]
2
Monaghan
A
.
Detecting and managing deterioration in children.
Paediatr Nurs
.
2005
;
17
(
1
):
32
35
[PubMed]
3
Tibballs
J
.
Systems to prevent in-hospital cardiac arrest.
Paediatr Child Health (Oxford)
.
2011
;
21
(
7
):
322
328
4
Duncan
H
,
Hutchison
J
,
Parshuram
CS
.
The Pediatric Early Warning System score: a severity of illness score to predict urgent medical need in hospitalized children.
J Crit Care
.
2006
;
21
(
3
):
271
278
[PubMed]
5
Haines
C
,
Perrott
M
,
Weir
P
.
Promoting care for acutely ill children-development and evaluation of a paediatric early warning tool.
Intensive Crit Care Nurs
.
2006
;
22
(
2
):
73
81
[PubMed]
6
Edwards
ED
,
Powell
CVE
,
Mason
BW
,
Oliver
A
.
Prospective cohort study to test the predictability of the Cardiff and Vale paediatric early warning system.
Arch Dis Child
.
2009
;
94
(
8
):
602
606
[PubMed]
7
Parshuram
CS
,
Hutchison
J
,
Middaugh
K
.
Development and initial validation of the Bedside Paediatric Early Warning System score.
Crit Care
.
2009
;
13
(
4
):
R135
[PubMed]
8
Tucker
KM
,
Brewer
TL
,
Baker
RB
,
Demeritt
B
,
Vossmeyer
MT
.
Prospective evaluation of a pediatric inpatient early warning scoring system.
J Spec Pediatr Nurs
.
2009
;
14
(
2
):
79
85
[PubMed]
9
Parshuram
CS
,
Duncan
HP
,
Joffe
AR
, et al
.
Multicentre validation of the bedside paediatric early warning system score: a severity of illness score to detect evolving critical illness in hospitalised children.
Crit Care
.
2011
;
15
(
4
):
R184
[PubMed]
10
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
;
8
(
3
):
236
246; quiz 247
[PubMed]
11
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
;
123
(
15
):
2965
2974
[PubMed]
12
Olson
D
,
Davis
NL
,
Milazi
R
, et al
.
Development of a severity of illness scoring system (inpatient triage, assessment and treatment) for resource-constrained hospitals in developing countries.
Trop Med Int Health
.
2013
;
18
(
7
):
871
878
[PubMed]
13
Agulnik
A
,
Méndez Aceituno
A
,
Mora Robles
LN
, et al
.
Validation of a pediatric early warning system for hospitalized pediatric oncology patients in a resource-limited setting.
Cancer
.
2017
;
123
(
24
):
4903
4913
[PubMed]
14
Agulnik
A
,
Forbes
PW
,
Stenquist
N
,
Rodriguez-Galindo
C
,
Kleinman
M
.
Validation of a pediatric early warning score in hospitalized pediatric oncology and hematopoietic stem cell transplant patients.
Pediatr Crit Care Med
.
2016
;
17
(
4
):
e146
e153
[PubMed]
15
Miranda
JO
,
Camargo
CL
,
Nascimento Sobrinho
CL
, et al
.
Translation and adaptation of a pediatric early warning score [in Portuguese].
Rev Bras Enferm
.
2016
;
69
(
5
):
888
896
[PubMed]
16
Gitembagara
A
,
Relf
MV
,
Pyburn
R
.
Optimizing nursing and midwifery practice in Rwanda.
Rwanda Journal
.
2015
;
2
(
2
):
26
30
17
McHugh
MD
,
Lake
ET
.
Understanding clinical expertise: nurse education, experience, and the hospital context.
Res Nurs Health
.
2010
;
33
(
4
):
276
287
[PubMed]
18
Munyiginya
P
,
Brysiewicz
P
,
Mill
J
.
Critical care nursing practice and education in Rwanda.
Southern African Journal of Critical Care
.
2016
;
32
(
2
):
55
57
19
World Health Organization
. Working together for health: the world report 2006. Policy briefs. Available at: https://www.who.int/hrh/documents/policy_brief_en.pdf?ua=1, Accessed March 4, 2019
20
Parshuram
CS
,
Dryden-Palmer
K
,
Farrell
C
, et al;
Canadian Critical Care Trials Group
.
Evaluating processes of care and outcomes of children in hospital (EPOCH): study protocol for a randomized controlled trial.
Trials
.
2015
;
16
:
245
[PubMed]
21
Centre Hospitalier Universitaire de Kigali
. CHUK annual report 2015.
2015
22
Sandell
J
. South COAST Children’s Early Warning Score: children’s unit. Available at: https://www.england.nhs.uk/wp-content/uploads/2015/03/pews-infant-coast-score-jan15.pdf. Accessed February 10, 2018
23
Engorn
B
,
Flerlage
J
;
Johns Hopkins Hospital
.
The Harriet Lane Handbook
. 20th ed.
Philadelphia, PA
:
Elsevier Inc
;
2015
24
de Onis
M
.
The new WHO child growth standards.
Paediatria Croatica
.
2008
;
52
(
suppl 1
):
13
17
25
Centers for Disease Control and Prevention
.
Birth to 36 Months: Girls Length-for-Age and Weight-for-Age Percentiles
.
Hyattsville, MD
:
National Center for Health Statistics
;
2001
26
Centers for Disease Control and Prevention
.
Birth to 36 Months: Boys Length-for-Age and Weight-for-Age Percentiles
.
Hyattsville, MD
:
National Center for Health Statistics
;
2001
27
Sefton
G
,
Lane
S
,
Killen
R
, et al
.
Accuracy and efficiency of recording pediatric early warning scores using an electronic physiological surveillance system compared with traditional paper-based documentation.
Comput Inform Nurs
.
2017
;
35
(
5
):
228
236
[PubMed]
28
Ahmed
M
,
Sobithadevi
D
,
Lall
R
,
Ghose
A
,
Boswell
S
,
Reynolds
T
.
1483 Burton paediatric early warning system score [abstract].
Arch Dis Child
.
2012
;
97
(
suppl 2
):
A421
29
Opio
MO
,
Nansubuga
G
,
Kellett
J
.
In-hospital mortality of acutely ill medical patients admitted to a resource poor hospital in sub-Saharan Africa and to a Canadian regional hospital compared using the abbreviated VitalPAC Early Warning Score.
Eur J Intern Med
.
2014
;
25
(
2
):
142
146
[PubMed]
30
Akre
M
,
Finkelstein
M
,
Erickson
M
,
Liu
M
,
Vanderbilt
L
,
Billman
G
.
Sensitivity of the pediatric early warning score to identify patient deterioration.
Pediatrics
.
2010
;
125
(
4
). Available at: www.pediatrics.org/cgi/content/full/125/4/e763
[PubMed]
31
Seiger
N
,
Maconochie
I
,
Oostenbrink
R
,
Moll
HA
.
Validity of different pediatric early warning scores in the emergency department.
Pediatrics
.
2013
;
132
(
4
). Available at: www.pediatrics.org/cgi/content/full/132/4/e841
[PubMed]
32
McLellan
MC
,
Gauvreau
K
,
Connor
JA
.
Validation of the children’s hospital early warning system for critical deterioration recognition.
J Pediatr Nurs
.
2017
;
32
:
52
58
[PubMed]
33
Skaletzky
SM
,
Raszynski
A
,
Totapally
BR
.
Validation of a modified pediatric early warning system score: a retrospective case-control study.
Clin Pediatr (Phila)
.
2012
;
51
(
5
):
431
435
[PubMed]

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data