Predictive Monitoring for respiratory decompensation leading to urgent unplanned intubation in the neonatal intensive care unit

Department of Chemical Engineering, University of Virginia, Charlottesville, VA 22904.
Pediatric Research (Impact Factor: 2.31). 11/2012; 73(1). DOI: 10.1038/pr.2012.155
Source: PubMed
ABSTRACT
Background
Infants admitted to the neonatal intensive care unit (NICU), and especially those born with very low birth weight (VLBW; <1500 grams), are at risk for respiratory decompensation requiring endotracheal intubation and mechanical ventilation. Intubation and mechanical ventilation are associated with increased morbidity, particularly in urgent unplanned cases.Methods
We tested the hypothesis that the systemic response associated with respiratory decompensation can be detected from physiological monitoring, and that statistical models of bedside monitoring data can identify infants at increased risk of urgent, unplanned intubation. We studied 287 VLBW infants consecutively admitted to our NICU and found 96 events in 51 patients, excluding intubations occurring within 12 hours of a previous extubation.ResultsIn order of importance in a multivariable statistical model, we found the characteristics of reduced O(2) saturation, especially as heart rate was falling, increased heart rate correlation with respiratory rate, and the amount of apnea all were significant independent predictors. The predictive model, validated internally by bootstrap, had receiver-operating characteristic area of 0.84 + 0.04.Conclusions
We propose that predictive monitoring in the NICU for urgent unplanned intubation may improve outcomes by allowing clinicians to intervene non-invasively before intubation is required.Pediatric Research (2012); doi:10.1038/pr.2012.155.

Full-text

Available from: Douglas E Lake, Oct 07, 2014
104 Pediatric RESEARCH Volume 73 | Number 1 | January 2013
Copyright © 2013 International Pediatric Research Foundation, Inc.
Articles
Clinical Investigation
nature publishing group
BACKGROUND: Infants admitted to the neonatal intensive
care unit (NICU), and especially those born with very low birth
weight (VLBW; <1,500 g), are at risk for respiratory decompensa-
tion requiring endotracheal intubation and mechanical ventila-
tion. Intubation and mechanical ventilation are associated with
increased morbidity, particularly in urgent unplanned cases.
METHODS: We tested the hypothesis that the systemic
response associated with respiratory decompensation can
be detected from physiological monitoring and that statisti-
cal models of bedside monitoring data can identify infants at
increased risk of urgent unplanned intubation. We studied 287
VLBW infants consecutively admitted to our NICU and found 96
events in 51 patients, excluding intubations occurring within
12 h of a previous extubation.
RESULTS: In order of importance in a multivariable statistical
model, we found that the characteristics of reduced O
2
satura-
tion, especially as heart rate was falling; increased heart rate
correlation with respiratory rate; and the amount of apnea were
all significant independent predictors. The predictive model,
validated internally by bootstrap, had a receiver-operating
characteristic area of 0.84 ± 0.04.
CONCLUSION: We propose that predictive monitoring in the
NICU for urgent unplanned intubation may improve outcomes
by allowing clinicians to intervene noninvasively before intu-
bation is required.
I
nfants born prematurely have extended stays in the neona-
tal intensive care unit (NICU). is is particularly true of
infants born with very low birth weight (VLBW, <1,500 g), at
least 65% of whom will require endotracheal intubation for the
administration of mechanical ventilation (data for 2009–2010,
from the Vermont Oxford Network of more than 900 centers,
http://www.vtoxford.org/). ese long ICU stays can be punc-
tuated by clinical deterioration, including frequent apneas
(1,2) or other forms of respiratory decompensation, leading to
urgent unplanned intubation, in which the infant is provided
mechanical ventilation through an endotracheal tube.
In addition to worsening neonatal apnea, urgent unplanned
intubation can result from sepsis, respiratory distress syn-
drome, pneumonia, exacerbation of chronic lung disease, or
critical illness from conditions such as necrotizing enterocolitis.
Although intubation for the purpose of mechanical ventilation
is an eective intervention for respiratory decompensation,
it is also associated with substantial morbidity and mortality,
including pneumonia (3), barotrauma, and volutrauma lead-
ing to pneumothorax or bronchopulmonary dysplasia, and
oxygen toxicity leading to pulmonary and retinal injury. Early
detection of respiratory decompensation may allow for early
and less obtrusive treatment, such as initiation or increase of
noninvasive respiratory support such as continuous positive
airway pressure, administration of bronchodilators, or evalu-
ation and treatment of infection.
We hypothesize that some episodes of apparently sudden
clinical deterioration in the NICU have precursors of altered
control of heart rate and other physiological processes that
require nely adaptive coupling among organs (4). is is the
case with late-onset neonatal sepsis, in which reduced heart
rate variability and transient decelerations can precede clini-
cal signs of illness by 24 h (5–11). In a recent very large ran-
domized clinical trial, we found that display of a multivariable
statistical model that relates these abnormal heart rate char-
acteristics (HRCs) to the fold increase in risk of sepsis in the
next 24 h led to a more than 20% reduction in VLBW NICU
mortality (12).
We have tested the hypothesis that respiratory decompensa-
tion leading to urgent unplanned intubations can also be pre-
ceded by changes apparent from bedside physiological moni-
toring. Similar to the development of the HRC index, or HeRO
score, we have developed logistic regression models on the
basis of physiological waveforms conventionally recorded in
the NICU, including cardiac, respiratory, and pulse oximetry
vital signs. Unlike the HeRO score, however, the new predic-
tive model includes information from the respiratory as well as
the cardiac system, and the interactions between the two.
Received 8 May 2012; accepted 3 October 2012; advance online publication 5 December 2012. doi:10.1038/pr.2012.155
Predictive monitoring for respiratory decompensation
leading to urgent unplanned intubation in the neonatal
intensive care unit
Matthew T. Clark
1
, Brooke D. Vergales
2
, Alix O. Paget-Brown
2
, Terri J. Smoot
3
, Douglas E. Lake
3,4
, John L. Hudson
1
, John B. Delos
5
,
John Kattwinkel
2
and J. Randall Moorman
3
International Pediatric Research Foundation, Inc.
2013
10.1038/pr.2012.155
8 May 2012
3 October 2012
Clinical Investigation
Articles
5 December 2012
1
Department of Chemical Engineering, University of Virginia, Charlottesville, Virginia;
2
Division of Neonatology, Department of Pediatrics, University of Virginia School of
Medicine, Charlottesville, Virginia;
3
Division of Cardiovascular Medicine, Department of Medicine, University of Virginia School of Medicine, Charlottesville, Virginia;
4
Department
of Statistics, University of Virginia, Charlottesville, Virginia;
5
Department of Physics, College of William and Mary, Williamsburg, Virginia. Correspondence:
Matthew T. Clark (mtc2h@virginia.edu)
Page 1
Volume 73 | Number 1 | January 2013 Pediatric RESEARCH 105
Copyright © 2013 International Pediatric Research Foundation, Inc.
Articles
Predicting urgent intubation in the NICU
RESULTS
Patient Population
Times when no vital signs were recorded because of technical
problems were excluded, leaving a population of 309 VLBW
infants who had monitoring data available. Of the 309 VLBW
infants with available data, data from 22 patients could be
collected only when they were mechanically ventilated, and
these patients were therefore excluded from the study. A total
of 287 patients were included in the study. e demographic
data of the patients in the study, and those of the subset of
patients who had an urgent unplanned intubation, are shown
in Tabl e 1. In the population of 287 VLBW infants for whom
data were recorded for at least 12 h before intubation, we found
96 unplanned intubation events in 51 patients.
An Example of the Analysis
Figure 1 shows the time series of cardiorespiratory parameters
for an infant born with a birth weight of 1,460 g at 29 wk esti-
mated gestational age. is infant had an urgent unplanned
intubation for respiratory acidosis 9 d aer birth, and param-
eters are shown relative to the time of this event. e clini-
cal goal is to identify patterns that are predictive of urgent
unplanned intubations in VLBW infants.
e le column in Figure 1 shows the means and SDs of
conventionally monitored vital signs, including heart rate,
respiratory rate, and pulse oximetry level. During the time
leading to intubation, this infants physiological measurements
present conicting information. For example, the heart rate
variability and arterial oxygen saturation are rising, consistent
with improving status. Concurrently, the respiratory rate is
falling and pulse oximetry variability is rising, consistent with
deteriorating status.
e right column of Figure 1 shows correlations between
the vital signs, as well as three measures of physiological sta-
bility: the level of cardiorespiratory coupling (13), duration of
time spent in apnea with associated bradycardia and desatura-
tion (or apnea burden) (14), and the output of a model for pre-
dicting neonatal sepsis on the basis of HRCs, the HeRO score
(12). Correlations between heart and respiratory rates, and
between heart rate and pulse oximetry, rise in the 2 d before
urgent unplanned intubation. At the same time, the correlation
between respiratory rate and pulse oximetry falls. e level of
cardiorespiratory coupling 1 d before intubation is 25% of the
value taken 2 d before intubation and falls nearly to zero by the
Heart rate
(bpm)
<HR RR>
Cross-corr
coefficent
<HR O
2
>
Cross-corr
coefficient
<RR O
2
>
Cross-corr
coefficient
Cardio-
respiratory
coupling
(proportion)
Apnea
burden
(s/12 h)
HeRO
score
Heart rate
SD
(bpm)
Resp.
rate
(bpm)
Resp. rate
SD
(bpm)
O
2
saturation
(%)
O
2
Sat
SD
(%)
170
155
13
6
50
30
25
15
100
90
14
0
2.0 1.5 1.0
Days relative to intubation
0.5
0.0
2.0 1.5 1.0
Days relative to intubation
0.5
0.0
a
g
h
i
j
k
l
b
c
d
e
f
0.3
0.0
0.3
0.4
0.4
0.0
0.1
0.1
0.2
0.0
35
0
2
0
0.0
Figure 1. Time series of physiological measures for one patient before unplanned intubation. Unplanned intubation occurs at zero, on the right edge of
the plots. During this time period, (a) mean heart rate dips, (b) heart rate SD increases, (c) mean respiratory rate decreases, (d) respiratory rate SD remains
unchanged, (e) oxygen saturation remains unchanged, and (f) oxygen saturation SD increases. Moreover, during this time, (g) correlation between heart
rate and respiratory rate increases, (h) correlation between heart rate and oxygen saturation increases, (i) correlation between respiratory rate and oxy-
gen saturation decreases, (j) cardiorespiratory coupling decreases, (k) apnea burden increases, and (l) the HeRO score increases. bpm, beats per minute;
cross-corr, cross-correlation; HR, heart rate; O
2
sat, oxygen saturation; Resp. rate, respiratory rate; RR, respiratory rate.
Table 1. Demographic characteristics of the study population
All infants in the study (N = 287)
EGA (wk) 27 (25, 29)
Males 147
Birth weight (g) 1,010 (783, 1,268)
Length of stay (d) 61 (35, 95)
Ventilator days 15 (2, 37)
PMA at discharge (wk) 37 (36, 39)
Infants with events (N = 51)
Events 96
Events due to sepsis 11
Males 32
EGA (wk) 26 (25, 28)
Birth weight (g) 810 (708, 1,060)
Length of stay (d) 97 (67, 107)
Ventilator days 27 (8, 43)
PMA at discharge (wk) 39 (37, 41)
PMA at urgent unplanned intubation 29 (26, 31)
Data are presented as median (25th, 75th percentile).
EGA, estimated gestational age; PMA, postmenstrual age.
Page 2
106 Pediatric RESEARCH Volume 73 | Number 1 | January 2013
Copyright © 2013 International Pediatric Research Foundation, Inc.
Articles
Clark et al.
time of intubation. e patients apnea burden is high through-
out, and the HeRO score increases by fourfold over the day
leading up to intubation.
us, the clinician has multiple streams of physiological
data, all varying with time, interrelated to various degrees, and
oen with inconsistent trajectories. is justies an approach
using multivariate time series methods.
Univariate Analyses
Examination of patient records indicated that patient physiol-
ogy undergoes changes before clinically relevant incidents (15)
and urgent unplanned intubation, in particular. We exploit
this fact by developing logistic regression models for intuba-
tion on the basis of physiological parameters. Figure 2 shows
the relative risk of unplanned intubation in the next 24 h on the
ordinate and the percentile of each physiological parameter on
the abscissa. For example, the lowest and highest heart rates
observed in our sample of infants are represented by the 0 and
100 percentiles, respectively. e nomenclatures μ
i
and σ
i
indi-
cate the mean and SD of vital sign i, respectively, and <i j> indi-
cates the cross-correlation coecient between vital signs i and
j at zero lag. High respiratory and heart rates, and high respira-
tory and oxygenation variability are associated with increased
risk of intubation, as is low oxygen saturation. Risk of intuba-
tion has a nonlinear relation with heart rate variability.
e curves in Figure 2 indicate the importance of each phys-
iological parameter in predicting intubation. Parameters that
provide a large dynamic range between risks at the 0 and 100
percentiles are good candidates for a model. Mean and vari-
ability of the pulse oximetry level, correlation between heart
rate and pulse oximetry, cardiorespiratory coupling, and the
HeRO score all have high dynamic ranges. e association of
c
d
5a
4
3
Relative risk
2
1
0
02550
Percentile
75 100
5
4
3
Relative risk
2
1
0
02550
Percentile
75 100
5
4
3
Relative risk
2
1
0
02550
Percentile
75 100
b
5
4
3
Relative risk
2
1
0
02550
Percentile
75 100
Figure 2. Relative risk of unplanned intubation in the next 24 h as a func-
tion of variable percentile. Percentiles are based on all values observed for
a given variable, and variables are calculated over half-hour windows. (a)
Relative risk vs. percentile of mean heart rate (solid line), respiratory rate
(dashed line), and pulse oximetry (dashed-dotted line). (b) Relative risk
vs. percentile of SD of heart rate (solid line), respiratory rate (dashed line),
and pulse oximetry (dashed-dotted line). (c) Relative risk vs. percentile of
correlation between heart rate and respiratory rate (solid line), heart rate
and pulse oximetry (dashed line), and respiratory rate and pulse oximetry
(dashed-dotted line). (d) Relative risk vs. percentile of the HeRO score
(solid line), coupling (dashed line), fraction of beats during inhalation
(dashed-dotted line), and apnea burden (gray dotted line).
Table 2. Performance of univariate logistic regression models for
unplanned intubation
Variable ROC P value Sign χ
2
Vital signs
μ
HR
0.53 0.18 + 0.2
σ
HR
a
0.61 0.007 + 7.2
μ
RR
0.60 0.05 + 7.8
σ
RR
0.61 * + 13.2
μ
SpO2
0.70 * 37.1
σ
SpO2
0.62 * + 13.0
Correlations
<HR RR> 0.65 * + 15.6
<HR SpO
2
> 0.74 * + 47.5
<RR SpO
2
> 0.57 0.06 3.7
Physiological stability
Coupling 0.62 0.03 4.8
CVC 0.50 0.95 <0.1
Beat fraction 0.58 0.02 + 5.0
Apnea burden 0.70 * + 31.5
HeRO score 0.81 * + 49.5
CVC, cardioventilatory coupling; HR, heart rate; ROC, receiver-operating characteristic;
RR, respiratory rate; SpO
2
, peripheral oxygen saturation; , mean; σ, SD.
a
A nonlinear transform was applied to heart rate variability before use in a logistic
regression model. Specifically, the absolute difference between the measured heart
rate variability and the median of all heart rate variability values were used. *P < 0.001.
Page 3
Volume 73 | Number 1 | January 2013 Pediatric RESEARCH 107
Copyright © 2013 International Pediatric Research Foundation, Inc.
Articles
Predicting urgent intubation in the NICU
HeRO is, in part, due to the 11 of 96 urgent unplanned intuba-
tion events in response to sepsis.
We deployed candidate predictor variables in a univariate
logistic regression model to predict urgent unplanned intubation
in the next 24 h. Table 2 shows the performance of these models,
including the receiver-operating characteristic, signicance (P
value) and sign of the coecient, and the goodness of t (χ
2
). To
account for the nonlinear relation for heart rate variability, we
recast it as the absolute dierence between the variability and the
median variability for all patients at all times. Univariate models
were based only on the times when the parameter was available.
Heart rate itself had little predictive information. Low and,
counterintuitively, high heart rate variability were both asso-
ciated with upcoming unplanned intubation (16). e HeRO
score had the best association with upcoming intubation,
with receiver-operating characteristic area of 0.81 and χ
2
value
of 49, followed by the cross-correlation of heart rate and O
2
saturation (0.74 and 48, respectively). e former reects the
reduced heart rate variability and transient decelerations that
can accompany early phases of neonatal sepsis, and the latter
reects the coordinated bradycardia and O
2
desaturation that
accompany neonatal apneas.
Multivariable Analysis
We used the parameters in which univariate coecients were
signicant (Table 2, P 0.05) as inputs to a multivariate logistic
regression model, 11 in total for 96 events. We determined this
to be acceptable, as meaningful multivariate models are known
to require 6–10 events per predictor (17). During periods when
a parameter was not available, the median value of that param-
eter (for all patients at all times) was used. Parameters in which
multivariate regression coecients did not reach signicance
were eliminated from the model, and a new model was created.
We note that the HeRO score, the best-performing individual
predictor, did not make the nal model. e HeRO score does
not add information to models that include the correlation
between heart rate and pulse oximetry: the two measures have
a moderate correlation (r = 0.45). Table 3 shows the coe-
cients and SEs for each parameter used in the nal model. e
area under the receiver-operating curve for the nal model is
0.84 ± 0.04 as determined by bootstrapping (5,18).
e output of the multivariate logistic regression model is
the probability of urgent unplanned intubation in the next
24 h. Figure 3 (le) shows the model output for 96 events in
51 patients, normalized to the relative risk by dividing by the
average rate of intubation in the next 24 h (0.5%). e median
increases by two-thirds during the day before intubation, and
the median output 12 h before intubation is signicantly higher
than that 36 h before intubation (gray dots, P = 0.001).
Figure 3 (right) shows the probability (solid) that the null
hypothesis of the Wilcoxon signed-rank test is true, i.e., that the
median model output t days before intubation is equal to the
corresponding median model output t − 1.5 d before intubation.
e cuto for rejecting the null hypothesis (P = 0.05) is shown as
the dashed horizontal line. Model outputs throughout the 24-h
period before urgent unplanned intubation are signicantly
higher than model outputs 36–60 h before intubation.
Internal Validation
Bootstrapping showed the 95% condence interval to be ±0.04
(18).
Implementation of the Predictive Model for an Individual Patient
Figure 4 shows the relative risk for the patient whose records
are shown in Figure 1 on the basis of our multivariate logistic
Table 3. Performance of multivariate model for unplanned
intubation
Variable
Coefficient
(normalized)
a
Coefficient SE P value χ
2
μ
SpO2
−0.03 −0.13 0.02 * 32.1
<HR RR> 24.98 3.51 0.70 0.017 24.9
<HR SpO
2
> 23.24 3.48 0.64 * 30.1
Apnea
burden
0.009 0.04 0.02 0.02 5.1
HR, heart rate; RR, respiratory rate; SpO
2
, peripheral oxygen saturation; , mean.
a
Coefficients normalized by SD of the variable. *P < 0.001.
a
b
5
10
0
P (Y
t
= Y
tn
)
10
1
10
2
10
3
10
4
4
3
Relative risk
2
1
0
4 3 2
Days relative to intubation
10
3 2
Days relative to intubation
10
Figure 3. Model output for the patient sample relative to time of intubation. (a) Median (solid) and 25%, 75% (dashed) model output for 96 urgent
unplanned intubation events in 51 patients. The median model output 12 h before intubation is signicantly higher than the output 36 h before intuba-
tion (gray dots, P = 0.001 based on a signed-rank test). (b) The probability that the null hypothesis of the Wilcoxon signed-rank test is true. Paired data are
separated by 36 h (n = 36). The dashed line shows the cuto for rejecting the null hypothesis (P = 0.05). Model outputs in the day before intubation are
signicantly higher than the values 36 h before intubation.
Page 4
108 Pediatric RESEARCH Volume 73 | Number 1 | January 2013
Copyright © 2013 International Pediatric Research Foundation, Inc.
Articles
Clark et al.
regression model (Table 3). e relative risk increases vefold
from 24 to 12 h before the event.
DISCUSSION
We studied changes in bedside physiological monitoring param-
eters in premature infants at risk for respiratory decompensa-
tion. We used conventional and cross-correlation measures
based on vital signs, novel variables based on cardiorespiratory
waveforms, and multivariate logistic regression to predict epi-
sodes of urgent unplanned intubation. Our predictive statistical
model has good performance, with a receiver-operating charac-
teristic area of 0.84, and allowed the identication of character-
istics that added independent information to one another aer
taking them all into account. e clinical characteristics of the
decompensating infant include, in order of decreasing predic-
tive importance on the basis of goodness of t, low O
2
satura-
tion, coincident uctuations in heart rate and O
2
saturation,
correlated heart rate and respiratory rate, and increasing apneas.
e heart rate–based HeRO score, which had the highest uni-
variate predictive performance with receiver-operating charac-
teristic area of 0.81, was displaced in the nal model by these
other parameters, all of which incorporate information about
respiration.
e value of these ndings is in the possibility of bedside
predictive monitoring for neonatal respiratory decompensa-
tion. e strengths of this analysis are that we used data that are
conventionally available in the NICU and require no new sen-
sors or contact with the infant. A limitation is that the model is
not yet externally validated. In addition, further studies should
investigate the impact that changing respiratory support and
medication administration may have on the model output.
Inputs to the multivariate model listed in Table 3 were selected
to optimize model performance and t to the data. e signi-
cance of physiological parameters, including those not in the
multivariate model, provides insight into mechanisms under-
lying clinical decompensation leading to the need for urgent
unplanned intubation. e importance of mean and variability
of pulse oximetry, and correlation between pulse oximetry and
heart rate as well as heart rate and respiratory rate, indicates a
role for hypoxemia. e association of increased HRC index
with intubation indicates a decline in cardiac control through
extracellular signaling (19) and the autonomic nervous system
(20,21). Decreased cardiorespiratory coupling is an indicator
of critical deterioration, in agreement with the concept of sys-
temic inammatory response syndrome in adult patients (15).
e modern age of high-speed data analysis allows great
opportunities for synthesizing the large number of data streams
available to the intensive care clinician. Although the insights
into the clinical picture of the decompensating infant from this
study are not surprising, there is potentially great value in bed-
side predictive monitoring that is constantly available, requires
no new contact with the infant, and optimally leverages data
that are already present. Such monitoring could never replace
the clinical judgment of experienced doctors and nurses; how-
ever, when considering an apparently stable infant in a busy
NICU, a rising risk score might place the clinician at the right
bedside at the right time.
METHODS
Patient Population
We collected cardiorespiratory waveforms and vital signs from 1,438
consecutive admissions to the University of Virginia NICU from
January 2009 through June 2011. For the 320 VLBW infants, we also
collected demographic data including admission and discharge dates,
types and times of respiratory support, nursing documentation of
apnea and bradycardia, and disposition and status at discharge. Times
when no vital signs were recorded because of technical problems were
excluded. e University of Virginia Institutional Review Board gave
permission for this study with waiver of consent status.
Denition of Urgent Unplanned Intubation
Urgent unplanned intubations were dened as nonelective initiation
of mechanical ventilation. Accepted causes included worsening respi-
ratory status from primary lung disease, increasing apnea, respiratory
acidosis, and increasing requirement for inspired oxygen. ere is no
protocol in the University of Virginia NICU that denes when to intu-
bate for these causes. Decisions are made on a case-by-case basis when
less-invasive treatment (e.g., continuous positive airway pressure)
proves ineective. Oen, intubations occur overnight on the basis of
need as perceived by the NICU sta, and by their nature are considered
urgent and unplanned.
We excluded planned intubations before surgery or other elective
procedures, protocol-driven surfactant administration requiring <12 h
of intubation, and 19 instances of reintubation within 12 h of a prior
extubation. ese clinically important events were excluded because
they do not provide 12 h of nonventilated data on which to develop a
model. Two clinical experts independently investigated patient records
for each intubation, and only events deemed by both reviewers to meet
our criteria for urgent unplanned intubation were included.
Data Collection
Vital signs and waveforms were collected from all bedside monitors
in our 45-bed NICU by a centralized server (BedmasterEx, Excel
Medical, Jupiter, FL) behind the clinical rewall. Vital signs (heart
rate, respiratory rate, and pulse oximetry) were calculated by the
monitor by averaging over the previous 10 s and were collected every
2 s. Waveforms included signals from three electrocardiogram leads
digitized at 240 Hz, chest impedance pneumograph digitized at 60
Hz, and oximetry plethysmography digitized at 120 Hz. Data were
transferred to our parallel computing and storage cluster. All infants
had continuous HeRO monitoring (Medical Predictive Science,
Charlottesville, VA) and HRC indexes (12) were collected hourly.
Data Analysis
Calculations were made on 30-min blocks of data collected dur-
ing periods of spontaneous ventilation, i.e., when the infant was not
Relative risk
8
6
4
2
0
4 3 2
Days relative to event
10
Figure 4. Relative risk of urgent unplanned intubation for the patient
shown in Figure 1 based on the multivariate logistic regression model
dened in Table 3. From 1 d before unplanned intubation to the time of
intubation, the estimated risk increases eightfold.
Page 5
Volume 73 | Number 1 | January 2013 Pediatric RESEARCH 109
Copyright © 2013 International Pediatric Research Foundation, Inc.
Articles
Predicting urgent intubation in the NICU
receiving mechanical ventilation. As candidate predictors, we calcu-
lated the mean and SD of each vital sign—heart rate, respiratory rate,
and O
2
saturation—along with the cross-correlation of each vital sign
with the others. From the continuous waveforms—electrocardiogram,
chest impedance, and oximetry plethysmography—we calculated more
complex physiological and statistical measures: cardiorespiratory cou-
pling, cardioventilatory coupling, fraction of heartbeats during inhala-
tion, apnea burden, and the HeRO score.
Cardiorespiratory coupling (hereaer referred to as “coupling”) and
fraction of heartbeats during inhalation were calculated every 30 s over
the previous 4 min when data were of sucient quality for analysis (13).
Coupling is preferential alignment of heartbeats within the respiratory
cycle, and this phenomenon was dened as epochs for which the dis-
tribution of heartbeats within the respiratory cycle had <0.1% chance
of occurring from random numbers (13). Each half hour, the fraction
of measures that exhibited coupling was calculated and averaged over
the previous 12 h. e fraction of beats during inhalation was dened
as the mean over the previous 12 h. Cardioventilatory coupling is the
preferential alignment of inhalation to the heartbeat and was calculated
every 30 s over the previous 10 min. We dened cardioventilatory cou-
pling as epochs for which the relationship between each inhalation and
the previous R-wave had less than a 5% chance of occurring from noise
given the number of intervals (22,23). Each half hour, the fraction of
measures that exhibited cardioventilatory coupling was calculated and
averaged over the previous 12 h.
Cardiorespiratory waveforms were automatically analyzed to detect
central apnea using the methods of Lee et al. (14). Briey, breathing
cessations were detected as low-variance epochs in the chest imped-
ance pneumograph aer notch ltering in heart-clock time to eliminate
cardiac artifact and high-pass ltering to remove movement artifact.
Heartbeats were detected using a threshold-based algorithm (24) as
implemented by Cliord and co-workers (25,26). Apneas were dened
as breathing cessation of at least 10 s with associated bradycardia (heart
rate < 100 beats per minute) and desaturation (peripheral oxygen satu-
ration < 80%) (14,27). Apnea burden was dened as the number of sec-
onds that the infant was apneic during the previous 12 h.
HRC indexes were collected from monitors in the NICU each hour,
and values were carried over to the subsequent half hour. e HRC
index is an output of a logistic regression model based on the RR inter-
val SD, sample asymmetry, and sample entropy that detects reduced
variability and transient decelerations and reports the fold increase in
the risk of neonatal sepsis in the next 24 h. It has been externally vali-
dated and has been shown to add information to the laboratory tests
and clinical signs, and to reduce mortality when displayed (4–12).
Development and Internal Validation of Logistic Regression
Models
We used these conventional and novel physiological variables as inputs
to logistic regression models. Measurements within the 24 h before an
urgent unplanned intubation event were labeled as an outcome of 1 and
used as events to be predicted. All other measurements (excluding data
before failed extubation) were labeled as outcome 0. Standard maximum
likelihood estimation was used to determine the coecients for the
logistic regression model (28). is approach corrects both for unequal
variances and correlated responses from individual patients. More spe-
cically, estimates of regression coecients and other parameters of the
model are obtained in a standard fashion, but the P values are corrected
using the sandwichestimator of SEs (29). For internal validation, we
used a cluster bootstrap technique whereby 1,000 new samples of the
same size were obtained by resampling the infants with replacement
(30). e 2.5 and 97.5 percentiles of the sample of risks are used as lower
and upper limits for a 95% condence interval, respectively.
e multivariate predictive model for urgent unplanned intubation
was developed by rst creating univariate models for each conventional
and novel physiological variable on the basis of all available data for
that variable. Variables that yielded statistically signicant models (as
dened by coecients with P < 0.05) were incorporated into a multi-
variate model. Variables whose coecients were not signicant in the
multivariate model were removed. Not all parameters could be mea-
sured at all times, and we replaced missing data with the median value
for all measurements of that variable from all patients when creating the
multivariate model. Values were imputed in this way to allow the model
to be calculated as oen as possible, with the trade-o of decreasing the
accuracy of the nal model.
STATEMENT OF FINANCIAL SUPPORT
This work was funded by National Institutes of Health grant no.
1RC2HD064488.
Disclosure: D.E.L. and J.R.M. wish to disclose nancial interest in Medical Pre-
dictive Science Corporation (Charlottesville, VA).
REFERENCES
1. Darnall RA, Ariagno RL, Kinney HC. e late preterm infant and the con-
trol of breathing, sleep, and brainstem development: a review. Clin Perina-
tol 2006;33:883–914; abstract x.
2. Barrington K, Finer N. e natural history of the appearance of apnea of
prematurity. Pediatr Res 1991;29:372–5.
3. Apisarnthanarak A, Holzmann-Pazgal G, Hamvas A, Olsen MA, Fraser VJ.
Ventilator-associated pneumonia in extremely preterm neonates in a neo-
natal intensive care unit: characteristics, risk factors, and outcomes. Pedi-
atrics 2003;112:1283–9.
4. Fairchild KD, O’Shea TM. Heart rate characteristics: physiomarkers for
detection of late-onset neonatal sepsis. Clin Perinatol 2010;37:581–98.
5. Grin MP, O’Shea TM, Bissonette EA, et al. Abnormal heart rate char-
acteristics preceding neonatal sepsis and sepsis-like illness. Pediatr Res
2003;53:920–6.
6. Grin MP, Lake DE, Bissonette EA, et al. Heart rate characteristics:
novel physiomarkers to predict neonatal infection and death. Pediatrics
2005;116:1070–4.
7. Grin MP, O’Shea TM, Bissonette EA, et al. Abnormal heart rate character-
istics are associated with neonatal mortality. Pediatr Res 2004;55:782–8.
8. Grin MP, Lake DE, Moorman JR. Heart rate characteristics and labora-
tory tests in neonatal sepsis. Pediatrics 2005;115:937–41.
9. Grin MP, Lake DE, O’Shea TM, Moorman JR. Heart rate characteristics
and clinical signs in neonatal sepsis. Pediatr Res 2007;61:222–7.
10. Kovatchev BP, Farhy LS, Cao H, et al. Sample asymmetry analysis of heart
rate characteristics with application to neonatal sepsis and systemic inam-
matory response syndrome. Pediatr Res 2003;54:892–8.
11. Moorman JR, Delos JB, Flower AA, et al. Cardiovascular oscillations at the
bedside: early diagnosis of neonatal sepsis using heart rate characteristics
monitoring. Physiol Meas 2011;32:1821–32.
12. Moorman JR, Carlo WA, Kattwinkel J, et al. Mortality reduction by heart
rate characteristic monitoring in very low birth weight neonates: a ran-
domized trial. J Pediatr 2011;159:900–6.e1.
13. Clark MT, Rusin CG, Hudson JL, et al. Breath-by-breath analysis of cardio-
respiratory interaction for quantifying developmental maturity in prema-
ture infants. J Appl Physiol 2012;112:859–67.
14. Lee H, Rusin CG, Lake DE, et al. A new algorithm for detecting central
apnea in neonates. Physiol Meas 2012;33:1–17.
15. Godin PJ, Buchman TG. Uncoupling of biological oscillators: a comple-
mentary hypothesis concerning the pathogenesis of multiple organ dys-
function syndrome. Crit Care Med 1996;24:1107–16.
16. Grin MP, Scollan DF, Moorman JR. e dynamic range of neonatal heart
rate variability. J Cardiovasc Electrophysiol 1994;5:112–24.
17. Grin MP, Lake DE, Bissonette EA, et al. Heart rate characteristics:
novel physiomarkers to predict neonatal infection and death. Pediatrics
2005;116:1070–4.
18. Feng Z, McLerran D, Grizzle J. A comparison of statistical methods for
clustered data analysis with Gaussian error. Stat Med 1996;15:1793–1806.
19. Küster H, Weiss M, Willeitner AE, et al. Interleukin-1 receptor antagonist
and interleukin-6 for early diagnosis of neonatal sepsis 2 days before clini-
cal manifestation. Lancet 1998;352:1271–7.
20. Ellenby MS, McNames J, Lai S, et al. Uncoupling and recoupling of autonomic
regulation of the heart beat in pediatric septic shock. Shock 2001;16:274–7.
21. Buchman TG, Stein PK, Goldstein B. Heart rate variability in critical illness
and critical care. Curr Opin Crit Care 2002;8:311–5.
22. Larsen PD, Galletly DC. Cardioventilatory coupling in heart rate variability:
the value of standard analytical techniques. Br J Anaesth 2001;87:819–26.
Page 6
110 Pediatric RESEARCH Volume 73 | Number 1 | January 2013
Copyright © 2013 International Pediatric Research Foundation, Inc.
Articles
Clark et al.
23. Tzeng YC, Larsen PD, Galletly DC. Mechanism of cardioventilatory cou-
pling: insights from cardiac pacing, vagotomy, and sinoaortic denervation in
the anesthetized rat. Am J Physiol Heart Circ Physiol 2007;292:H1967–77.
24. Pan J, Tompkins WJ. A real-time QRS detection algorithm. IEEE Trans
Biomed Eng 1985;32:230–6.
25. Li Q, Mark RG, Cliord GD. Robust heart rate estimation from multiple
asynchronous noisy sources using signal quality indices and a Kalman l-
ter. Physiol Meas 2008;29:15–32.
26. Tarassenko L, Cliord G, Townsend N. Detection of ectopic beats in the
electrocardiogram using an auto-associative neural network. Neural Proc
Lett 2001;14:15–25.
27. Finer NN, Higgins R, Kattwinkel J, Martin RJ. Summary proceed-
ings from the apnea-of-prematurity group. Pediatrics 2006;117:
S47–51.
28. White H. Maximum-likelihood estimation of mis-specied models.
Econometrica 1982;50:1–25.
29. Wei LJ, Lin DY, Weissfeld L. Regression-analysis of multivariate incom-
plete failure time data by modeling marginal distributions. J Am Stat Assoc
1989;84:1065–73.
30. Steyerberg EW, Harrell FE Jr, Borsboom GJ, et al. Internal validation of
predictive models: eciency of some procedures for logistic regression
analysis. J Clin Epidemiol 2001;54:774–81.
Page 7
  • [Show abstract] [Hide abstract] ABSTRACT: Purpose of review: Predictive monitoring is an exciting new field involving analysis of physiologic data to detect abnormal patterns associated with critical illness. The first example of predictive monitoring being taken from inception (proof of concept) to reality (demonstration of improved outcomes) is the use of heart rate characteristics (HRC) monitoring to detect sepsis in infants in the neonatal ICU. The commercially available 'HeRO' monitor analyzes electrocardiogram data from existing bedside monitors for decreased HR variability and transient decelerations associated with sepsis, and converts these changes into a score (the HRC index or HeRO score). This score is the fold increase in probability that a patient will have a clinical deterioration from sepsis within 24 h. This review focuses on HRC monitoring and discusses future directions in predictive monitoring of ICU patients. Recent findings: In a randomized trial of 3003 very low birthweight infants, display of the HeRO score reduced mortality more than 20%. Ongoing research aims to combine respiratory and HR analysis to optimize care of ICU patients. Summary: Predictive monitoring has recently been shown to save lives. Harnessing and analyzing the vast amounts of physiologic data constantly displayed in ICU patients will lead to improved algorithms for early detection, prognosis, and therapy of critical illnesses.
    No preview · Article · Feb 2013 · Current opinion in pediatrics
  • [Show abstract] [Hide abstract] ABSTRACT: Bronchopulmonary dysplasia (BPD), or chronic lung disease of prematurity, occurs in ∼30% of preterm infants (15,000 per year) and is associated with a clinical history of mechanical ventilation and/or high inspired oxygen at birth. Here, we describe changes in ventilatory control that exist in patients with BPD, including alterations in chemoreceptor function, respiratory muscle function, and suprapontine control. Because dysfunction in ventilatory control frequently revealed when O2 supply and CO2 elimination are challenged, we provide this information in the context of four important metabolic stressors: stresses: exercise, sleep, hypoxia, and lung disease, with a primary focus on studies of human infants, children, and adults. As a secondary goal, we also identify three key areas of future research and describe the benefits and challenges of longitudinal human studies using well-defined patient cohorts.
    No preview · Article · Jul 2013 · Respiratory Physiology & Neurobiology
  • [Show abstract] [Hide abstract] ABSTRACT: To identify clinical conditions associated with a large increase (spike) in the heart rate characteristics index in very low birth weight (VLBW) infants. Retrospective medical record review within a day of all large heart rate characteristics index spikes (increase of ≥3 from the previous 5-day average) in VLBW infants at a single center enrolled from 2007 to 2010 in a multicenter trial of heart rate characteristics monitoring. In the trial, infants were randomized to having their heart rate characteristics index displayed to clinicians or not displayed. Of 274 eligible infants, 224 large heart rate characteristics spikes occurred in 105 infants. Thirty-three spikes were associated with surgery or procedures requiring anesthetic or anticholinergic medications, and infection-related conditions were the most common clinical association with the other spikes. Of the first spikes in 47 infants randomized to conventional monitoring (heart rate characteristics index not displayed to clinicians), 53% were associated with suspected or proven infection. Respiratory deterioration without suspected infection occurred with 34%, and no association was identified in 13%. Infants randomized to having their heart rate characteristics index displayed were more likely to have antibiotics initiated around the time of a large heart rate characteristics index spike. Sepsis, other infectious or systemic inflammatory conditions, respiratory deterioration, and surgical procedures are the most common clinical associations with a large increase in the heart rate characteristics index in VLBW infants. This information may improve use of heart rate characteristics monitors in patients in the neonatal intensive care unit.
    No preview · Article · Jan 2014 · The Journal of pediatrics
Show more