Clinical Signs to Categorize Shock and Target

Vasoactive Medications in Warm Versus Cold

Pediatric Septic Shock*
Sarah B. Walker, MD1; Thomas W. Conlon, MD1; Bingqing Zhang, MPH1; Janell L. Mensinger, PhD2;
Julie C. Fitzgerald, MD, PhD1,3; Adam S. Himebauch, MD1; Christie Glau, MD1;
Akira Nishisaki, MD, MSCE1; Suchitra Ranjit, MD4; Vinay Nadkarni, MD, MSc1;
Scott L. Weiss, MD MSCE1,3,5

Objectives: Determine level of agreement among clinical signs of
shock type, identify which signs clinicians prioritize to determine
shock type and select vasoactive medications, and test the asso-
ciation of shock type-vasoactive mismatch with prolonged organ
dysfunction or death (complicated course).
Design: Retrospective observational study.
Setting: Single large academic PICU.
Patients: Patients less than 18 years treated on a critical care
sepsis pathway between 2012 and 2016.
Interventions: None.
Measurements and Main Results: Agreement among clinical signs
(extremity temperature, capillary refill, pulse strength, pulse pres-
sure, and diastolic blood pressure) was measured using Fleiss
and Cohen’s κ. Association of clinical signs with shock type and
shock type-vasoactive mismatch (e.g., cold shock treated with va-
sopressor rather than inotrope) with complicated course was de-
termined using multivariable logistic regression. Of 469 patients,
clinicians determined 307 (65%) had warm and 162 (35%) had
cold shock. Agreement across all clinical signs was low (κ, 0.25;
95% CI, 0.20–0.30), although agreement between extremity
capillary refill (adjusted odds ratio, 15.7; 95% CI, 7.9–31.3), and
pulse strength (adjusted odds ratio, 21.3; 95% CI, 8.6–52.7)
were associated with clinician-documented shock type. Of the 86
patients initiated on vasoactive medications during the pathway,
shock type was discordant from vasoactive medication (κ, 0.14;
95% CI, –0.03 to 0.31) and shock type-vasoactive mismatch was
not associated with complicated course (adjusted odds ratio, 0.3;
95% CI, 0.1–1.02).
Conclusions: Agreement was low among common clinical signs
used to characterize shock type, with clinicians prioritizing extremity
temperature, capillary refill, and pulse strength. Although clinician-
assigned shock type was often discordant with vasoactive choice,
shock type-vasoactive mismatch was not associated with compli-
cated course. Categorizing shock based on clinical signs should
be done cautiously. (Pediatr Crit Care Med 2020; 21:1051–1058)
Key Words: cold shock; pediatric septic shock; vasoactive therapy;
warm shock

temperature, capillary refill, and pulse strength was better than
with pulse pressure and diastolic blood pressure. Only extremity
temperature (adjusted odds ratio, 26.6; 95% CI, 15.5–45.8),
*See also p. 1085.
1
Department of Anesthesiology and Critical Care, Children’s Hospital of
Philadelphia, University of Pennsylvania, Philadelphia, PA.
2
Department of Epidemiology and Biostatics, Dornsife School of Public
Health, Drexel University, Philadelphia, PA.
3
Pediatric Sepsis Program, Children’s Hospital of Philadelphia, University
of Pennsylvania, Philadelphia, PA.
4
Critical Care, Apollo Children’s Hospital, Chennai, India.
5
Center for Mitochondrial and Epigenomic Medicine, Children’s Hospital
of Philadelphia, University of Pennsylvania, Philadelphia, PA.
Copyright © 2020 by the Society of Critical Care Medicine and the World
Federation of Pediatric Intensive and Critical Care Societies
DOI: 10.1097/PCC.0000000000002481
S
epsis remains a common cause of morbidity and mor-
tality in children (1–3). Guidelines for pediatric sepsis
focus on early recognition and treatment of shock (4, 5),
emphasizing that a shorter time to shock reversal is associated
with improved outcomes (6, 7). For persistent shock despite
fluid resuscitation, the 2014 American College of Critical Care
Medicine guidelines recommended using clinical signs to target
vasoactive medications according to shock type, including ino-
tropes for low cardiac output/high systemic vascular resistance
(i.e., cold shock) and vasopressors for high cardiac output/low
systemic vascular resistance (i.e., warm shock) (4).
However, recent Pediatric Surviving Sepsis Campaign (SSC)
guidelines noted that clinical signs may not reflect underlying
cardiovascular physiology (5). Studies have shown that bed-
side clinical assessment of shock type is often not consistent
with assessment by invasive (8–10) or noninvasive (11–14) he-
modynamic monitoring. In other words, many patients with

Pediatric Critical Care Medicine www.pccmjournal.org 1051

Copyright © 2020 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
 Walker et al

apparent cold shock using clinical signs were identified to have sign and bedside clinically determined shock type recorded in the
low systemic vascular resistance while those with warm shock window of 2 hours before to 8 hours after pathway start time were
were found to have decreased cardiac output due to myocardial included in the analysis (Fig. 1). For patients with more than one
dysfunction. Additionally, patients frequently have conflicting pathway activation during the study period, only the first episode
clinical signs, as one study found almost no patients (1/239) was included. We excluded patients who had both cold and warm
whose clinical examination findings were consistent with a shock documented in the first hour as this may reflect either an
single shock type (15). When there is a disagreement between inconsistency in charting or rapidly evolving physiology for which
these clinical signs, it is unclear which clinical sign(s) clinicians associations would be prone to misclassification bias. We also
prioritize to categorize shock type and guide vasoactive therapy excluded patients on extracorporeal membrane oxygenation.
selection. Furthermore, the clinical impact of using inotropes
for clinically determined cold shock and vasopressors for clin- Data Collection
ically determined warm shock has not been evaluated. Despite Data were abstracted from the electronic health record (EHR)
these inconsistencies and lack of evidence, clinical signs remain into a standardized case report form within Research Elec-
commonly used as surrogates of underlying physiology to cat- tronic Data Capture (created at Vanderbilt University, hosted
egorize shock type and target treatment. by Children’s Hospital of Philadelphia). Variables included
We sought to determine the level of agreement among the demographics, comorbid conditions, site of infection, vaso-
five clinical signs commonly used to categorize warm and cold active infusion, and Pediatric Index of Mortality-3 (PIM-3)
pediatric septic shock and identify which of these clinical signs risk of mortality (16). Hospital-acquired shock was defined as
clinicians prioritize to categorize shock type and select vaso- sepsis pathway starting greater than or equal to 2 days after
active medications. We also compared clinical outcomes be- hospital admission (17). Data were extracted by a single author
tween patients for whom clinically assigned shock type was not (S.B.W.), with double-data entry performed for a random se-
matched to the initial vasoactive medication (i.e., cold shock lection of 10% of patients by a second author (T.W.C.) to test
treated with a vasopressor or warm shock treated with an ino- for consistent variable extraction. Of the 106 data fields col-
trope) versus those with shock type-vasoactive match. lected per patient, the mean disagreement was only 2% of data
fields (range, 0–5% across patients).
MATERIALS AND
METHODS
Study Design and
Population
We conducted a retrospec-
tive, observational study of
patients treated for suspected
septic shock at a single, large,
academic PICU at the Chil-
dren’s Hospital of Philadelphia
(CHOP) between December 1,
2012, and December 1, 2016.
The CHOP Institutional Re-
view Board approved the
study with a waiver of consent.
All PICU patients less than
18 years old who were treated
using the CHOP Critical Care
Sepsis Pathway during the study
period and were documented
to be in shock were eligible for
inclusion. Diagnosis of sepsis
and decision to treat using the
sepsis pathway were at the dis-
cretion of the treating clinician.
Patients treated on this pathway
had clinical signs of shock re-
corded at regular intervals and
shock type recorded hourly. Figure 1. Patient flow diagram. CHOP = Children’s Hospital of Philadelphia, ECMO = extracorporeal membrane
Patients with at least one clinical oxygenation.

1052 www.pccmjournal.org December 2020 • Volume 21 • Number 12

Copyright © 2020 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited

Sepsis pathway documentation within the EHR included
five common clinical signs of shock (extremity temperature,
capillary refill, pulse strength, pulse pressure, and diastolic
blood pressure) and shock type (warm, cold, or none) every
15 minutes for the first hour and then at least hourly there-
after. Although the sepsis pathway was initiated by a physician
order, documentation of these variables was completed by
bedside nurses after routine unit-based educational curric-
ulum. Clinical signs were recorded from both arms and legs
for extremity temperature, capillary refill, and pulse strength.
Because clinical signs assessed in both arms and legs were
highly correlated (eTable 1, Supplemental Digital Content 1,
http://links.lww.com/PCC/B415), we limited analyses to lower
extremity values. Invasive blood pressure values were preferred,
but noninvasive oscillometric measurements were collected
if unavailable. The values for each clinical sign were a priori
categorized as indicating low cardiac output/high systemic
vascular resistance (i.e., cold shock), high cardiac output/low
systemic vascular resistance (i.e., warm shock), or indetermi-
nate (Table 1).
All eligible patients were included in the analysis of the level
of agreement across clinical signs and to shock type. For the
subset who had initiation of vasoactive medications after bed-
side shock type categorization, we categorized each patient ei-
ther as a shock type-vasoactive match or mismatch based on
common clinical signs referenced by 2014 ACCM pediatric
sepsis guidelines. We allowed for the vasoactive to precede
shock type documentation by up to 30 minutes to account
for delays in documentation while limiting the time expected
for the vasoactive medication to itself influence shock type.
Additional data collection on this subset of patients included
organ dysfunction on day 7 following sepsis pathway initiation
and vital status at 28 days. Organ dysfunction was determined
using established criteria for pediatric sepsis (18) with minor
modifications for practicality: cardiovascular dysfunction was
defined as hypotension or need for vasoactive medication, res-
piratory dysfunction as noninvasive or invasive mechanical
ventilation at increased levels from baseline, neurologic dys-
function as greater than or equal to 3 point decline in Glasgow
Coma Scale from baseline, and hematologic dysfunction
excluded patients with chronic or medication-induced throm-
bocytopenia or coagulopathy.
TABLE 1. Classification of Clinical Signs
Clinical Sign Warm Shock
Extremity temperature “Warm”
Capillary refill <2s
Pulse strength “Bounding”
Pulse pressure a
DBP < 1/2 SBP
DBP b
Low
DBP = diastolic blood pressure, SBP = systolic blood pressure.
Pulse pressure defined by (18).
a
Low DBP defined by: < 1 yr: < 30 mm Hg; 1 to < 2 yr: < 35 mm Hg; 2 to < 6 yr: < 40 mm Hg; 6 to < 13 yr: < 45 mm Hg; and ≥ 13 yr: < 50 mm Hg.
b
Dashes indicate no values exist for these criteria.
Pediatric Critical Care Medicine www.pccmjournal.org 1053
Copyright © 2020 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
Feature Articles
Vasoactive medications were a priori dichotomized as pri-
mary inotropes or vasopressors. Inotropes were defined as
dopamine less than 10 µg/kg/min, any dose of epinephrine,
dobutamine, or milrinone (8). Vasopressors were defined as
dopamine greater than or equal to 10 µg/kg/min, any dose of
norepinephrine, phenylephrine, or vasopressin (8). In order
to account for the change in our pathway recommendation
of preferred initial vasoactive medication from dopamine to
epinephrine for indeterminate shock type due to 2014 ACCM
guidelines, we defined a variable called “time cohort” with
patients treated between December 1, 2012, and January 31,
2016, as belonging to the “dopamine-first” period and between
February 1, 2016, and December 1, 2016, as “epinephrine-
first.” Clinical signs and shock type documented immediately
prior initiation of vasoactive were used to assess for a shock
type-vasoactive match if available and differed from the first
documented values of the pathway. The main clinical outcome
was complicated course, defined as dysfunction of two or more
organ systems on day 7 or death by day 28 (19, 20).
Statistical Analysis
Analyses were performed using STATA (version 15.1; Stata-
Corp, College Station, TX), SAS (Version 9.4; SAS Institute,
Cary, NC), and R software (version 3.5.1; R Core Team, Vi-
enna, Austria). Data were reported as medians with interquar-
tile range (IQR) or frequencies with proportions. Continuous
variables were analyzed with the Wilcoxon rank-sum test and
categorical variables using chi-square or Fisher exact tests.
Pair-wise agreement between clinical signs was calculated
using Cohen’s κ, and agreement across greater than or equal
to 3 clinical signs was calculated using Fleiss κ. We considered
κ less than 0.4 as poor, 0.4–0.59 as fair, 0.6–0.74 as good, and
0.75–1.0 as excellent agreement (21). Clinical signs indicative
of indeterminate shock type were excluded from analyses. We
used multivariable logistic regression to determine the asso-
ciation between each clinical sign and clinician-documented
shock type. Potential confounders that differed between
patients with cold and warm shock with p value of less than
0.1 were included in final multivariable models after assessing
for collinearity. We included race/ethnicity and age as a priori
confounders regardless of statistical significance in the bivar-
iate analysis due to their potential impact on the assessment of
Indeterminate Cold Shock
“Other” “Cold”
2–3 s >3s
“Other” “Weak” or “absent”
— DBP ≥ 1/2 SBP
— Normal or elevated
 Walker et al
capillary refill. Cancer was not included in the final model de-
spite being significant in bivariable analyses due to collinearity
with PIM-3 risk of mortality.
For the subset who had initiation of vasoactive medications
after bedside shock type categorization, we used Cohen’s κ to
determine the level of agreement of clinical signs and shock
type with the initial vasoactive medication. We then used mul-
tivariable logistic regression to determine the adjusted odds
ratio (aOR) of complicated course in patients exposed to
shock type-vasoactive mismatch. Potential confounding vari-
ables that differed between patients with and without com-
plicated course with p value of less than 0.1 were included in
the final multivariable model after assessing for collinearity. In
addition, time cohort was forced into the final model because
we anticipated more cold shock-vasoactive mismatches in the
“dopamine-first” period and more warm shock-vasoactive
mismatches in the “epinephrine-first” period. PIM-3 score was
forced into the model given known associations with mortality.
Statistical significance was defined as p value of less than 0.05.
As some children treated for suspected septic shock were ul-
timately determined not to have an infectious diagnosis by the
treating clinician, we performed a sensitivity analysis including
only those patients with confirmed septic shock. Finally, we
performed sensitivity analyses to determine whether clinician-
documented shock type was different than randomly assigned
shock type. These analyses were performed through computer-
generated simulations in which the clinician-documented
shock type was replaced with a randomly assigned shock type.
Simulations of 1,000 iterations of the study’s sample size and
prevalence were generated, resulting in a sampling distribution
of statistics. Results from the study’s observed data were then
compared with the distribution of point estimates obtained
from the simulation studies. If the study’s observed data fell
outside of the 95% bounds of the simulated data, then it is
likely that clinician-documented shock type was significantly
different than randomly assigned shock type.
RESULTS
Over the 4-year study period, there were 991 activations of the
Sepsis Pathway, of which 469 were eligible for analysis (Fig. 1),
including 307 categorized as warm shock and 162 as cold shock.
Overall, median age was 4.5 years (IQR, 1.4–9.8 yr) and PIM-3
risk of mortality was 1.6% (IQR, 0.6–6.2%). Patient character-
istics by shock type are shown in Table 2. Patients identified as
having cold shock had a higher PIM-3 risk of mortality than
warm shock (2.6% vs 1.1%; p < 0.001), had a lower proportion
of cancer (7% vs 21%; p < 0.001), and were more likely to be
treated with a vasoactive medication (65% vs 45%; p < 0.001).
There were similar rates of hospital-acquired sepsis between
warm and cold shock (34% vs 30%; p = 0.32).
Agreement Across Clinical Signs
Overall agreement across all five clinical signs was poor (Fleiss
κ, 0.25; 95% CI, 0.20–0.30), although agreement among ex-
tremity temperature, capillary refill, and pulse strength alone
was good (Fleiss κ, 0.63; 95% CI, 0.54–0.72). When clinical
1054 www.pccmjournal.org December 2020 • Volume 21 • Number 12
Copyright © 2020 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
signs underwent pair-wise comparison, extremity tempera-
ture, capillary refill, and pulse strength exhibited fair to good
agreement with one other, while pulse pressure and diastolic
blood pressure exhibited poor agreement with other clinical
signs (Table 3).
Association of Clinical Signs With Shock Type
After adjusting for age, race, and PIM-3, extremity tempera-
ture (aOR, 26.6; 95% CI, 15.5–45.8), capillary refill (aOR, 15.7;
95% CI, 7.9–31.3), and pulse strength (aOR, 21.3; 95% CI, 8.6–
52.7) were associated with shock type. Pulse pressure (aOR,
1.4; 95% CI, 0.9–2.2) and diastolic blood pressure (aOR, 0.6;
95% CI, 0.4–1.1) were not associated with shock type (Fig. 2;
and eTable 2, Supplemental Digital Content 1, http://links.
lww.com/PCC/B415).
Agreement Between Clinical Signs, Shock Type, and
Vasoactive Selection
Of the 469 eligible patients, 243 (52%) were treated with vas-
oactive medications. Of these, 86 had initiation of vasoactive
medications after shock type categorization (18%) and were
included in the analyses comparing clinical signs/shock type
to vasoactive selection and shock type-vasoactive mismatch
(Fig. 1). The most frequently used initial vasoactive medica-
tions were epinephrine (34%) and dopamine less than 10 µg/
kg/min (33%), followed by norepinephrine (17%) and dopa-
mine greater than 10 µg/kg/min (12%). Patients with warm
shock were more likely to be treated with a mismatched vas-
oactive (33/52 or 63%) than patients with cold shock (7/34 or
21%; p < 0.001). The level of agreement between each of the
clinical signs and type of vasoactive medication was low (all κ
< 0.21; eTable 3, Supplemental Digital Content 1, http://links.
lww.com/PCC/B415). Shock type also exhibited a poor level of
agreement with class of vasoactive medication (κ, 0.14; 95%
CI, –0.03 to 0.31).
Association of Shock Type-Vasoactive Mismatch With
Outcome
Patients with shock type-vasoactive mismatch trended toward
a lower rate of complicated course than patients with shock
type-vasoactive match (20% vs 33%; p = 0.23), but this dif-
ference was not significant after adjusting for age, PIM-3, and
time cohort in multivariable analysis (aOR, 0.3; 95% CI, 0.1–
1.02). Mismatch between each of the five clinical signs and vas-
oactive medication was also not associated with complicated
course (eTable 4, Supplemental Digital Content 1, http://links.
lww.com/PCC/B415).
Sensitivity Analyses
Ninety-one patients (19%) were ultimately determined to have
a noninfectious source of shock. Sensitivity analyses excluding
these patients yielded similar results as the overall analyses
(eTables 5–8, Supplemental Digital Content 1, http://links.
lww.com/PCC/B415). Simulation studies in which clinician-
documented shock type was replaced by randomly assigned
shock type confirmed the associations observed in the study.
 Feature Articles

TABLE 2. Characteristics of Patients With Warm Versus Cold Shock

Warm Shock Cold Shock
Characteristic (n = 307)a (n = 162)a p

Age, yr 5.1 (1.5–10.2) 3.7 (1.3–8.9) 0.20

Male sex 152 (50) 79 (49) 0.88
Race 0.77
 White 134 (44) 68 (42)
 Black 88 (29) 44 (27)
 Other 85 (28) 50 (31)
Pediatric Index of Mortality-3 risk of mortality 1.1 (0.5–4.4) 2.6 (0.8–13.6) < 0.001
Previously healthy 68 (22) 31 (19) 0.45
Cardiac disease 45 (15) 24 (15) 0.96
Cancer 63 (21) 11 (7) < 0.001
Source of shock 0.23
 Respiratory 100 (32) 61 (38)
 Blood 30 (10) 10 (6)
 Abdominal 26 (8) 14 (8)
 Other/unknown 97 (32) 40 (25)
 Noninfectious 54 (18) 37 (23)
Positive blood culture 48 (16) 17 (11) 0.13
Hospital acquired 105 (34) 48 (30) 0.32
Time cohortb 0.48
 “Dopamine-first” 231 (75) 117 (72)
 “Epinephrine-first” 76 (25) 45 (28)
Vasoactivec 137 (45) 106 (65) < 0.001
Data presented as median (interquartile range) or n (%).
a

Time cohort defines patients treated between December 1, 2012, and January 31, 2016, as belonging to the “dopamine-first” period and patients treated
b

between February 1, 2016, and December 1, 2016, as belonging to the “epinephrine-first” period, corresponding to the publication date for the 2014 American
College of Critical Care Medicine update and change in recommendations within our pathway.
Vasoactive started any time during hospital encounter, including pre-and post-study period.
c

TABLE 3. Cohen’s κ Across Clinical Signs

Extremity Capillary Pulse Pulse Diastolic
Clinical Sign Temperature Refill Strength Pressure BP

Extremity temperature
Capillary refill 0.64 (0.55–0.74)
Pulse strength 0.55 (0.45–0.65) 0.59 (0.47–0.72)
Pulse pressure 0.11 (0.04–0.19) 0.02 (–0.08 to 0.12) 0.01 (–0.12 to 0.14)
Diastolic BP 0.01 (–0.04 to 0.07) –0.08 (–0.16 to –0.01) 0.02 (–0.10 to 0.13) 0.36 (0.28–0.45)
BP = blood pressure.
Data reported as κ (95% CI).

Most notably, the observed association of extremity tem- contributed to clinician-documented shock type (eFig. 1 A–C,
perature, capillary refill, and pulse strength with clinician- Supplemental Digital Content 1, http://links.lww.com/PCC/
documented shock type were outside of the 95% bounds of B415). In contrast, CIs of the association of pulse pressure and
the simulated data, supporting that these clinical signs likely diastolic blood pressure with clinician-documented shock type

Pediatric Critical Care Medicine www.pccmjournal.org 1055

Copyright © 2020 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
 Walker et al
Figure 2. Association of clinical signs with shock type. Data are pre-
sented as the adjusted odds ratio of individual clinical sign to predict
shock type. A positive adjusted odds ratio indicates higher agreement
between the clinical sign and documented shock type (i.e., the clinical sign
was indicative of “cold shock” when shock type was documented as cold,
or the clinical sign was indicative of “warm shock” when shock type was
documented as warm). A negative adjusted odds ratio indicates disagree-
ment between the clinical sign and shock type.
fell inside of the 95% bounds of the simulated data, suggesting
that the association of these signs with clinician-documented
shock type may not be different than random variation (eFig.
1, D and E, Supplemental Digital Content 1, http://links.lww.
com/PCC/B415). Compared to randomly assigned shock type,
clinician-documented shock type was not a better predictor
of vasoactive choice (eFig. 2, Supplemental Digital Content 1,
http://links.lww.com/PCC/B415) and shock type vasoactive
mismatch was not a better predictor of complicated course
(eFig. 3, Supplemental Digital Content 1, http://links.lww.
com/PCC/B415), with both results having CIs falling inside the
95% bounds of the simulated data.
DISCUSSION
Despite common use of bedside clinical signs to categorize
type of pediatric septic shock, we found inconsistent agree-
ment between these signs. Extremity temperature, capillary
refill, and pulse strength exhibited fair to good agreement with
each other, while diastolic blood pressure and pulse pressure
exhibited poor agreement. Extremity temperature, capillary
refill, and pulse strength were strongly associated with shock
type, suggesting that clinicians prioritized these signs even
when diastolic blood pressure and pulse pressure supported
the alternative shock type. For patients treated with vasoac-
tive medications as part of the sepsis pathway, neither shock
type nor any of the clinical signs of shock type were strongly
matched with vasoactive selection, but the resulting shock
type-vasoactive mismatch was not associated with compli-
cated course. These data support SSC recommendations not
to make the determination of shock type based on clinical
signs alone (5).
Prior studies have demonstrated that direct measurement
of cardiac index and systemic vascular resistance can be used to
categorize shock type and hone vasoactive therapy in children
with septic shock (8–14). In contrast, this study looked solely
1056 www.pccmjournal.org December 2020 • Volume 21 • Number 12
Copyright © 2020 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
at the clinical signs used to differentiate warm versus cold pe-
diatric septic shock and clinician’s determination of shock
type rather than direct measurements of underling cardiovas-
cular physiology. As such, this study elucidates the real-world
practice of how clinicians initially diagnose and treat pediatric
septic shock, rather than a comparison to a measured gold
standard of physiologic dysfunction.
Despite common use in clinical practice, there are few data
to validate the use of clinical signs to differentiate warm versus
cold pediatric septic shock in the absence of additional inva-
sive or noninvasive measurements of underling cardiovascular
physiology. The classification of septic shock as warm versus
cold was put forth prior to the development of the flow-directed
pulmonary artery catheter in the 1960s (22). With introduc-
tion of the pulmonary artery catheter and human studies using
radionuclide cineangiography, it became evident that cardiac
output could be maintained or even increased despite myocar-
dial dysfunction and a reduction in ejection fraction (22, 23).
These observations suggested that cardiovascular manifestations
of septic shock were more complex than what could be ascer-
tained solely from the physical examination. Since that time,
other pediatric studies have specifically noted this discrepancy
between clinical signs and shock type determined either nonin-
vasively (8–10) or noninvasively (11–14). Our findings similarly
demonstrate that clinical signs are often discordant, and thus
may imply conflicting information about patient physiology.
Clinicians appeared to prioritize clinical signs assessed through
direct physical examination (i.e., extremity temperature, capil-
lary refill, and pulse strength) even when electronically displayed
or calculated indices (i.e., blood pressure and pulse pressure)
supported the alternative shock type.
Despite the challenges of using bedside clinical signs to ac-
curately characterize shock physiology, there is evidence that
these clinical assessments are useful. Scott et al (15) reported
that abnormalities in clinical signs in the emergency depart-
ment were associated with organ dysfunction in pediatric
patients with high specificity. Furthermore, time to normali-
zation of blood pressure and capillary refill are frequently used
markers in assessing response to resuscitation in the early stages
of shock (6, 24, 25). In a recent randomized trial in adults, a
strategy of peripheral perfusion-targeted resuscitation during
early septic shock was associated with reduced organ dysfunc-
tion and a trend toward reduced mortality (34.9% vs 43.4%; p
= 0.06) compared with lactate level-targeted resuscitation (26).
Our study, even with low agreement across all signs, did not
find clear evidence that a mismatch between clinical signs/shock
type and initial vasoactive medication was harmful. Because
this analysis was limited to a small subset of patients started
on vasoactive medications after shock type was documented in
the PICU, a conclusive interpretation is not possible. However,
one explanation for the observed trend of shock type-vasoac-
tive mismatch with a lower rate of complicated course is that
clinical signs, on their own, are an insufficient measure of he-
modynamic physiology on which to precisely target therapy.
Alternatively, our findings also raise the hypothesis that early
recognition and treatment of shock may be more important

than the precision of hemodynamic support, at least in the in-
itial stages of resuscitation. Indeed, early reversal of shock has
been shown to improve survival in children with sepsis (25, 27).
Additionally, the frequent use of dopamine and epinephrine in
our study—the vasoactive medications most readily available
in our institution—suggests that clinicians already prioritize
rapid blood pressure normalization over a targeted vasoactive.
Although two pediatric trials have demonstrated improved sur-
vival with initial use of epinephrine rather than dopamine in
pediatric septic shock (28, 29), additional study is warranted to
determine if universal epinephrine initiation targeted to hemo-
dynamic improvement is as or more effective than attempting to
match initial vasoactive support to a shock type.
There are several limitations to this study. First, data on
clinical signs and shock type were abstracted from nursing
documentation. Despite a unit-based educational curriculum,
nursing documentation may deviate from physician interpre-
tation and treatment decisions. Although we considered using
physician notes to determine clinical signs and shock type,
such documentation was more variably recorded, could not be
accurately timed to when the assessment took place, and gen-
erally did not provide serial assessments over the initial hours
of therapy. In addition, we were not able to assess if factors
beyond clinical signs influenced shock type determination.
Second, we were only able to evaluate the agreement between
clinical signs and shock type, not the relationship of either to
patients’ actual physiology. Third, inter-rater reliability of clin-
ical examination findings is potentially problematic in children
(30–32). As this is a single-center study, the results could reflect
institutional practices of these assessments. Fourth, in contrast
to prior studies (8, 11, 14, 33), warm shock was more common
in our study cohort overall and was not associated with hospi-
tal-acquired sepsis. This may reflect reliance on clinical docu-
mentation of shock type in our study rather than invasive or
noninvasive hemodynamic assessments, as in previous studies
(8, 11, 14, 33), or our inclusion of a more heterogeneous group
of patients. For example, unlike Ceneviva et al (8) and Ranjit
et al (11), our study included patients who did not require vas-
oactive medications. Although the prevalence of shock type
should not affect the concordance of clinical signs or their as-
sociation with shock type, it could signify institutional prac-
tices that might affect generalizability of our results (e.g., in
selection or availability of vasoactive infusions or propensity
to prioritize some clinical signs over others). Fifth, we excluded
many patients who were treated with vasoactive medications
because this therapy was started before documentation of clin-
ical signs and shock type. The small subset of patients included
in the vasoactive analyses tempers the strength of our conclu-
sions about the association of shock type-vasoactive mismatch
and clinical outcomes. Therefore, these findings should be
considered exploratory and hypothesis-generating.
CONCLUSIONS
In this retrospective study from a single, large, academic PICU,
there was good agreement between only three (extremity tem-
perature, capillary refill, and pulse strength) of the five clinical
Pediatric Critical Care Medicine www.pccmjournal.org 1057
Copyright © 2020 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
Feature Articles
signs commonly used to assign pediatric septic shock type.
These three clinical signs also demonstrated the strongest as-
sociation with assigned shock type, suggesting that clinicians
prioritized extremity temperature, capillary refill, and pulse
strength over pulse pressure and diastolic blood pressure. Al-
though the clinically assigned shock type did not appear to
drive selection of initial vasoactive for most patients (especially
those with warm shock), shock type-vasoactive mismatch was
not associated with worse clinical outcomes. Our data sup-
port recent SSC campaign recommendations that clinical signs
should not be used in isolation to categorize shock type. Ad-
ditional research is needed to identify the optimal approach
to early hemodynamic assessment that is most likely to drive
good outcomes in pediatric sepsis.
Supplemental digital content is available for this article. Direct URL cita-
tions appear in the printed text and are provided in the HTML and PDF
versions of this article on the journal’s website (http:/journals.lww.com/
pccmjournal).
Supported, in part, by grant from the Department of Anesthesiology and
Critical Care at the Children’s Hospital of Philadelphia.
Dr. Mensinger’s institution received funding from Children’s Hospital of
Philadelphia. Drs. Himebauch, Glau, and Nishisaki received funding from
the Society of Critical Care Medicine. Dr. Weiss’s institution received fund-
ing from National Institute of General Medical Sciences K23GM110496.
The remaining authors have disclosed that they do not have any potential
conflicts of interest.
This study was conducted at the Children’s Hospital of Philadelphia.
For information regarding this article, E-mail: sbwalker@luriechildrens.org
REFERENCES
1. Balamuth F, Weiss SL, Neuman MI, et al: Pediatric severe sepsis in
U.S. children’s hospitals. Pediatr Crit Care Med 2014; 15:798–805
2. Hartman ME, Linde-Zwirble WT, Angus DC, et al: Trends in the epi-
demiology of pediatric severe sepsis*. Pediatr Crit Care Med 2013;
14:686–693
3. Weiss SL, Fitzgerald JC, Pappachan J, et al; Sepsis Prevalence,
Outcomes, and Therapies (SPROUT) Study Investigators and
Pediatric Acute Lung Injury and Sepsis Investigators (PALISI)
Network: Global epidemiology of pediatric severe sepsis: The sepsis
prevalence, outcomes, and therapies study. Am J Respir Crit Care
Med 2015; 191:1147–1157
4. Davis AL, Carcillo JA, Aneja RK, et al: The American College of
Critical Care Medicine clinical practice parameters for hemodynamic
support of pediatric and neonatal septic shock: Executive summary.
Pediatr Crit Care Med 2017; 18:884–890
5. Weiss SL, Peters MJ, Alhazzani W, et al: Surviving sepsis campaign
international guidelines for the management of septic shock and sep-
sis-associated organ dysfunction in children. Pediatr Crit Care Med
2020; 21:e52–e106
6. Carcillo JA, Kuch BA, Han YY, et al: Mortality and functional morbidity
after use of PALS/APLS by community physicians. Pediatrics 2009;
124:500–508
7. Oliveira CF, Nogueira de Sá FR, Oliveira DS, et al: Time- and fluid-
sensitive resuscitation for hemodynamic support of children in septic
shock: Barriers to the implementation of the American College of
Critical Care Medicine/Pediatric Advanced Life Support Guidelines
in a pediatric intensive care unit in a developing world. Pediatr Emerg
Care 2008; 24:810–815
8. Ceneviva G, Paschall JA, Maffei F, et al: Hemodynamic support in
fluid-refractory pediatric septic shock. Pediatrics 1998; 102:e19
9. Egan JR, Festa M, Cole AD, et al: Clinical assessment of cardiac per-
formance in infants and children following cardiac surgery. Intensive
Care Med 2005; 31:568–573
 Walker et al
10. Tibby SM, Hatherill M, Marsh MJ, et al: Clinicians’ abilities to estimate
cardiac index in ventilated children and infants. Arch Dis Child 1997;
77:516–518
11. Ranjit S, Aram G, Kissoon N, et al: Multimodal monitoring for hemo-
dynamic categorization and management of pediatric septic shock: A
pilot observational study. Pediatr Crit Care Med 2014; 15:1–10
12. Razavi A, Newth CJL, Khemani RG, et al: Cardiac output and systemic
vascular resistance: Clinical assessment compared with a noninva-
sive objective measurement in children with shock. J Crit Care 2017;
39:6–10
13. Ranjit S, Kissoon N: Bedside echocardiography is useful in assessing
children with fluid and inotrope resistant septic shock. Indian J Crit
Care Med 2013; 17:224–230
14. Brierley J, Peters MJ: Distinct hemodynamic patterns of septic
shock at presentation to pediatric intensive care. Pediatrics 2008;
122:752–759
15. Scott HF, Donoghue AJ, Gaieski DF, et al: Effectiveness of physical
exam signs for early detection of critical illness in pediatric systemic
inflammatory response syndrome. BMC Emerg Med 2014; 14:24
16. Straney L, Clements A, Parslow RC, et al; ANZICS Paediatric Study
Group and the Paediatric Intensive Care Audit Network: Paediatric
index of mortality 3: An updated model for predicting mortality in pe-
diatric intensive care*. Pediatr Crit Care Med 2013; 14:673–681
17. Haque M, Sartelli M, McKimm J, et al: Health care-associated infec-
tions - an overview. Infect Drug Resist 2018; 11:2321–2333
18. Goldstein B, Giroir B, Randolph A; International Consensus
Conference on Pediatric Sepsis: International pediatric sepsis con-
sensus conference: Definitions for sepsis and organ dysfunction in
pediatrics. Pediatr Crit Care Med 2005; 6:2–8
19. Wong HR, Cvijanovich NZ, Anas N, et al: Developing a clinically fea-
sible personalized medicine approach to pediatric septic shock. Am J
Respir Crit Care Med 2015; 191:309–315
20. Signoff JK, Fitzgerald JC, Teachey DT, et al: Hypofibrinogenemia is
associated with poor outcome and secondary hemophagocytic lym-
phohistiocytosis/macrophage activation syndrome in pediatric severe
sepsis. Pediatr Crit Care Med 2018; 19:397–405
21. Cicchetti DV: Guidelines, criteria, and rules of thumb for evaluating
normed and standardized assessment instruments in psychology.
Psychol Assess 1994; 6:284–290
1058 www.pccmjournal.org December 2020 • Volume 21 • Number 12
Copyright © 2020 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.
Unauthorized reproduction of this article is prohibited
22. Funk DJ, Parrillo JE, Kumar A: Sepsis and septic shock: A history. Crit
Care Clin 2009; 25:83–101, viii
23. Parker MM: Profound but reversible myocardial depression in patients
with septic shock. Ann Intern Med 1984; 100:483
24. de Oliveira CF: Early goal-directed therapy in treatment of pediatric
septic shock. Shock 2010; 34(Suppl 1):44–47
25. Han YY, Carcillo JA, Dragotta MA, et al: Early reversal of pediatric-
neonatal septic shock by community physicians is associated with
improved outcome. Pediatrics 2003; 112:793–799
26. Hernández G, Ospina-Tascón GA, Damiani LP, et al; The
ANDROMEDA SHOCK Investigators and the Latin America Intensive
Care Network (LIVEN): Effect of a resuscitation strategy targeting pe-
ripheral perfusion status vs serum lactate levels on 28-day mortality
among patients with septic shock: The ANDROMEDA-SHOCK ran-
domized clinical trial. JAMA 2019; 321:654–664
27. Shime N, Kawasaki T, Saito O, et al: Incidence and risk factors for
mortality in paediatric severe sepsis: Results from the national pae-
diatric intensive care registry in Japan. Intensive Care Med 2012;
38:1191–1197
28. Ventura AM, Shieh HH, Bousso A, et al: Double-blind prospective
randomized controlled trial of dopamine versus epinephrine as first-
line vasoactive drugs in pediatric septic shock. Crit Care Med 2015;
43:2292–2302
29. Ramaswamy KN, Singhi S, Jayashree M, et al: Double-blind random-
ized clinical trial comparing dopamine and epinephrine in pediatric
fluid-refractory hypotensive septic shock. Pediatr Crit Care Med
2016; 17:e502–e512
30. Lobos AT, Menon K: A multidisciplinary survey on capillary refill time:
Inconsistent performance and interpretation of a common clinical test.
Pediatr Crit Care Med 2008; 9:386–391
31. Lobos AT, Lee S, Menon K: Capillary refill time and cardiac output in
children undergoing cardiac catheterization. Pediatr Crit Care Med
2012; 13:136–140
32. Fleming S, Gill P, Jones C, et al: Validity and reliability of measurement
of capillary refill time in children: A systematic review. Arch Dis Child
2015; 100:239–249
33. Deep A, Goonasekera CD, Wang Y, et al: Evolution of haemodynam-
ics and outcome of fluid-refractory septic shock in children. Intensive
Care Med 2013; 39:1602–1609