Original Research

Journal of Intensive Care Medicine

2020, Vol. 35(1) 74-81
Factors Associated With Unplanned ª The Author(s) 2017
Article reuse guidelines:
sagepub.com/journals-permissions
Extubation in Children: A Case–Control Study DOI: 10.1177/0885066617731274
journals.sagepub.com/home/jic

Paulo Sérgio Lucas da Silva, MSc, MD1,

and Marcelo Cunio Machado Fonseca, MD, PhD2

Abstract
Purpose: Although several studies assess unplanned extubation (UE) in children, few have addressed determinants of UE and
factors associated with reintubation in a case-controlled manner. We aimed to identify the risk factors and outcomes
associated with UE in a pediatric intensive care unit. Methods: Cases of UE were randomly matched with control patients at a
ratio of 1:4 for age, severity of illness, and admission diagnosis. For cases and controls, we also collected data associated with UE
events, reintubation, and outcomes. Results: We analyzed 94 UE patients (0.75 UE per 100 intubation days) and found no
differences in demographics between the 2 groups. Logistic regression revealed that patient agitation (odds ratio [OR]: 2.44;
95% confidence interval [CI]: 1.28-4.65), continuous sedation infusion (OR: 3.27; 95% CI: 1.70-6.29), night shifts (OR: 9.16;
95% CI: 4.25-19.72), in-charge nurse experience <2 years (OR: 2.38; 95% CI: 1.13-4.99), and oxygenation index (OI) >5
(OR: 76.9; 95% CI: 16.79-352.47) were associated with UE. Risk factors for reintubation after UE included prior level of
sedation (COMFORT score < 27; OR: 7.93; 95% CI: 2.30-27.29), copious secretion (OR: 11.88; 95% CI: 2.20-64.05), and
OI > 5 (OR: 9.32; 95% CI: 2.45-35.48). Conclusions: This case–control study showed that both patient- and nurse-associated
risk factors were related to UE. Risk factors associated with reintubation included lower levels of consciousness, copious
secretions, and higher OI. Further evidence-based studies, including a larger sample size, are warranted to identify predisposing
factors in UEs.

Keywords
intensive care, mechanical ventilation, pediatrics, unplanned extubation, reintubation

Introduction costs associated with UEs. Of the 3 pediatric case–control

studies assessing risk factors for UE,3,6,7 only 1 matched the
Health-care organizations focus on reducing the number of
control patients.3
preventable safety events by implementing programs and We could ease the task of identifying research evidence
process improvement initiatives. Unplanned extubation (UE)
relevant to clinical practice with a higher number of case–con-
is an important complication of mechanical ventilation (MV)
trol studies on UEs.8 Hence, we sought to study the determi-
and the ninth most common adverse event in North American
nants of UE and secondarily to determine the factors associated
pediatric intensive care units (PICUs). Seventy percent of them
with reintubation.
are deemed to be preventable.1
The literature review shows several available studies asses-
sing the clinical circumstances preceding or concurrent with
UE events, as well as quality improvement initiatives to reduce
their occurrence.2 However, other studies on the determinants 1
Pediatric Intensive Care Unit, Department of Pediatrics, Hospital do Servidor
of UE revealed inconsistent findings and methodological weak- Público Municipal, São Paulo, Brazil
2
nesses, such as use of retrospective designs and lack of a com- Health Technologies Assessment Center, Federal University of São Paulo,
parison group. The case–control design enables researchers to São Paulo, Brazil
study the relationship of multiple factors for one outcome and Received April 18, 2017. Received revised August 18, 2017. Accepted
is therefore appropriate for studying infrequent outcomes, such August 23, 2017.
as UE. To date, 5 studies assess UEs in a case-controlled
manner.3-7 Only 3 studies use multivariate analyses to assess Corresponding Author:
Paulo Sérgio Lucas da Silva, Pediatric Intensive Care Unit, Department
independent risk factors for UE.3,6,7 Ream et al4 studied the of Pediatrics, Hospital do Servidor Público Municipal, Rua Castro Alves, 60,
influence of staffing and workload on different shifts for UE São Paulo 01532-900, Brazil.
occurrence, while Roddy et al5 focused on attributable hospital Email: psls.nat@terra.com.br
da Silva and Fonseca 75

Methods vasopressor or inotropic medication other than low-dose

dopamine or dobutamine (5 mg/kg/min), (6) a fraction of
Study Design inspired oxygen (FIO2) 40%; (7) pH of 7.32 to 7.47 in the
We carried out a nested case–control study from January 1, most recent blood gas analysis, (8) positive end-expiratory
2007, to December 31, 2015, and performed it in an pressure 8 cmH2O, and (9) plateau pressure 25 cmH2O.
8-bedded PICU of a tertiary hospital. The local internal review Once we made the decision, we considered patients ready for
board approved the study and waived the need for informed extubation if they tolerated 30 minutes of minimal ventilator
consent from the children’s parents. support, defined as (1) continuous positive airway pressure
The study population included consecutive patients admit- of 5 cmH2O or (2) pressure support ventilation adjusted for
ted to the PICU between 1 month and 16 years old and needing endotracheal tube size.11,12 As part of our weaning protocol,
invasive MV for at least 12 hours. Patients with a tracheostomy we obtained an arterial blood gas sample before extubation;
were excluded. Our population is predominantly composed of the attending physician decided on weaning management and
children with medical conditions; however, we also admit final extubation. In case of UE, the attending physician
patients to the PICU who are undergoing planned and emer- decided whether or not to reintubate. In those UE patients
gency surgical thoracosurgical and neurosurgical procedures. who did not require emergency intubation, noninvasive
Nursing staff works in 3 shifts: 8:00 AM to 2:00 PM, 2:00 PM to positive pressure ventilation or supplemental oxygen was
8:00 PM, and 8:00 PM to 8:00 AM. initially attempted.

Patient Management
Each shift the nursing staff is composed of 1 nurse in-charge
Data Collection
and other licensed practical nurses with a minimum of We extracted data from a prospective database designed to
6 months experience in PICU setting. All patients who register information for planned and UEs in the PICU. This
undergo orotracheal intubation have their endotracheal tubes electronic database is part of our continuous quality improve-
secured in a standardized way. We used the same type of ment program to monitor and reduce UE events.13 We also
adhesive tape (Tensoplast sport elastic adhesive bandage; assessed potential risk factors for UEs with a data collection
BSM Medical, Charlotte, North Carolina) as well as an tool of patient characteristics, circumstances of the extubation,
“H”-shaped piece of elastic tape to affix the tube to the face, and outcomes. In order to avoid missing data, we interviewed
with 2 arms of the H placed on the upper lip and the other the nurses who witnessed or discovered the UE, as the inter-
2 arms wrapped around the tube, completing at least 2 com- views tend to include a root cause analysis. We based our
plete wraps. Both nursing and respiratory therapy personnel variables on a comprehensive literature review of previous
retaped or repositioned the endotracheal tubes in a joint UE studies2-4,6,7 and practical insights from the medical PICU
effort. In line with the institute’s policy, we used physical staff. In addition, we collected data for cause analysis, which
restraints when it was deemed necessary or whenever phar- include date and time of the event, use of patient restraints,
macological and nonpharmacological interventions were circumstances of UE, presence of patient agitation, level of
unsuccessful to ensure patient safety; we also used mittens sedation, time and doses of the last infused medications, and
or 2-point soft wrist restraining therapies. Based on the rec- nursing activity at the time of the event.
ommendations of the American College of Radiology,9 we In this study, we prospectively collected demographic and
did not perform daily routine chest radiographs. Respiratory laboratory data, amount of secretions, oxygenation measures,
therapists monitored the position of the endotracheal tube at outcomes, and other data, which we reviewed each day, while
the lip at least once per shift to detect short displacements the primary investigator (da Silva, PS) transcribed them with a
and adjust its position as required. double entry into a specifically designed database.
The nurses assessed the sedation level every 4 hours with the Demographic and clinical variables included age, gender,
COMFORT scale, which ranges from 8 to 40.10 We kept weight, PICU admission diagnosis, and severity of illness at
patients at the desired sedation level (COMFORT score 17 admission as measured by the Pediatric Risk of Mortality
and 26) while on MV. A COMFORT score of <17 implied (PRISM) II14 and the Pediatric Logistic Organ Dysfunction
oversedation and a score of >26 implied undersedation.10 The score.15 We also registered the level of sedation prior to
vast majority of our patients on invasive MV received contin- extubation (COMFORT score),10 use of physical restraints,
uous intravenous sedation and analgesia. The sedative agents duration of MV, length of PICU/hospital stay, ventilator-
included fentanyl and midazolam. For this study, we used the associated pneumonia (VAP) rate, and need for tracheostomy.
last recorded COMFORT score in our analysis. The following laboratory data were collected based on the last
We considered patients to be at the weaning phase if they recorded data point before extubation: arterial blood gas values
fulfilled the following criteria: (1) improvement or resolution (pH, partial pressure of carbon dioxide, and partial pressure of
of the underlying disease, (2) possessing spontaneous arterial oxygen), arterial oxygen tension (PaO2)/FIO2 ratio, and
respiratory effort, (3) gag or cough with suctioning, (4) oxygenation index (OI ¼ [mean airway pressure/PaO2/FIO2
acceptable level of consciousness for extubation, (5) no ratio]  100).
76 Journal of Intensive Care Medicine 35(1)

Definitions continuous variables and the w2 test for categorical variables or

Fisher exact test if we expected that the cell value was less than
Unplanned extubation is defined as any displacement of the
5 (categorical variables).
endotracheal tube from the trachea when a provider does not
Moreover, we performed a logistic regression analysis to
deliberately remove it. Alternatively, a planned extubation is
assess risk factors for both UE events and reintubation after
defined as removal of the endotracheal tube after a physician’s
UE: All variables in the univariate analysis with a P value less
order, according to hospital policies and procedures. Reintuba-
than 0.05 were included as independent variables, and addi-
tion is defined as replacement of the endotracheal tube within
tional variables were considered clinically relevant. We did not
24 hours of UE. For the purpose of this study, in patients with
simultaneously include collinear variables in the model if the
multiple episodes of UEs, we only assessed the first episode;
variance inflation factor was not larger than 1.0. The Kaplan-
thus, cases are composed of consecutive patients with an UE in
Meier survival analysis assessed mortality in mechanically
the study period.
ventilated patients, while log-rank tests determined differences
Control patients had planned extubations and were matched
between curves. Significance was assessed at a P value of less
according to 3 criteria3: (1) diagnosis category, (2) age, and (3)
than .05. We analyzed data with the Statistical Program for
illness severity as defined by PRISM II. Age is matched +3
Social Sciences (Chicago, Illinois), version 16.0.
months if the patient’s age is 12 months and +12 months if
the patient’s age is >12 months. For illness severity, patients
were matched using PRISM II score +3. We screened the
PICU database to randomly find 4 control matches and identify
Results
2 groups of patients: those for whom reintubation is required During the study period, the hospital admitted 2023 patients to
and those in whom it was not. the PICU. Of these, 1055 (52%) patients required more than
Agitation was defined as excessive restlessness/excitement 12 hours of MV, resulting in 14 087 ventilation days (median:
and increased motor activity.13 This information was obtained 6 days; range: 3-11 days). Within the study period, there were
from postevent interview of the caregivers at bedside. For this 106 UEs (0.75 UE per 100 intubation days) in 94 patients, with
study, we included a subjective evaluation, as the sedation level 376 controls included.
at the time of UE could not be the same as previously recorded We report on demographic and clinical characteristics in
with our sedation assessment tool. Table 1. As stated, the patients were well matched for age,
We assessed the time until the UE as the period from the gender, illness severity scores, and admitting diagnosis.
initial intubation until the unplanned procedure, whereas the
time of MV is the number of days in which patients are main- Determinants of UEs
tained on invasive MV.
Respiratory therapists documented the endotracheal secre- We provide determinants associated with UE in the univariate
tions according to a semi-quantitative scale: no secretions/ analysis, as seen in Table 2. The logistic regression assessed the
minimal secretions (suctioning required every 2-4 hours), relationship between specific independent variables and UE
moderate secretions (suctioning required every 1-2 hours), (Table 3). Patient agitation, use of continuous sedation infu-
or copious secretions (suctioning required several times per sion, night shifts, in-charge nurse experience <2 years, and OI >
hour).16 Ventilator-associated pneumonia is defined by the 5 were also associated with UE.
Center for Disease Control and Prevention and National
Healthcare Safety Network.17 Follow-Up After UEs
The outcomes for UE patients and their matched controls are
End Points shown in Table 4. The Kaplan-Meier curves demonstrates
that both days on a ventilator (hazard ratio: 0.47; 95% CI:
Primarily, we evaluated the possible determinants associated 0.39-0.58) and time until PICU discharge (hazard ratio: 0.52;
with UE. Secondarily, we compared duration of MV, PICU and 95% CI: 0.42-0.63) were significantly shorter in the control
hospital stay, and mortality between patients who experienced group compared to the case group (Figure 1). The PICU
UE and those who did not. We also assessed potential risk mortality of UE patients was significantly higher than that
factors for reintubation after UE. of matched control patients. However, there was no death
directly ascribed to UE.
Unplanned extubation patients who required reintubation
Statistical Analysis had a similar median duration of MV prior to the UE, compared
We performed descriptive statistics for all variables, with to those who did not require reintubation (72 vs 108 hours,
results expressed as numerical values and percentages for cate- P = .152). On the other hand, UE patients requiring reintuba-
gorical variables as well as medians and quartiles (25th–75th tion had a markedly longer length of PICU stay (20.5 vs 10.5
percentile) for continuous variables. We calculated the median days, P ¼ 0.008) and prolonged MV time (13.5 vs 5 days,
difference and 95% confidence interval (CI) when appropriate P = .001) than did UE patients without reintubation. Logistic
and based our comparisons on the Mann-Whitney U test for regression analysis showed that risk factors associated with
da Silva and Fonseca 77

Table 1. Comparison of Unplanned Extubation Cases Versus “Matched” Controls.a

Variables All (N ¼ 470) Controls (n ¼ 376) UE Cases (n ¼ 94) P Value

Age, months 6 (2-32) 6 (2-32) 7 (2.75-32) .695b

Age distribution, n (%) .713c
<6 months 218 (46.38) 177 (47.07) 41 (43.61)
6-12 months 94 (20) 74 (19.68) 20 (21.27)
13-72 months 105 (22.34) 82 (21.80) 23 (24.46)
>73 months 53 (11.27) 43 (11.43) 10 (10.63)
Weight, kg 7 (4.3-13) 7 (4.5-12.35) 6.2 (3.87-13.6) .271b
Male sex, n (%) 271 (57.65) 211 (56.11) 60 (63.82) .167c
Admission diagnosis, n (%) 1.000c
Respiratory 265 (56.38) 212 (56.38) 53 (56.38)
Sepsis 70 (14.89) 56 (14.89) 14 (14.89)
Postsurgery 30 (6.38) 24 (6.38) 6 (6.38)
Trauma 35 (7.44) 28 (7.44) 7 (7.44)
Others 70 (14.89) 56 (14.89) 14 (14.89)
PRISM score 12 (10-17) 10 (12-17) 13 (9-16) .936b
PELOD score 2 (1-11) 2 (1-11) 2 (1-11) .274b
Abbreviations: PELOD, pediatric logistic organ dysfunction; PRISM, pediatric risk of mortality; UE, unplanned extubation.
a
Data are expressed as median (percentile 25, percentile 75) or number (%).
b
Mann-Whitney U test.
c 2
w test.

reintubation after UE were at the prior level of sedation (COM- of intubation or amount of secretions with UE. Marcin et al3
FORT score < 27; odds ratio [OR]: 7.9; 95% CI: 2.3-27.29, P ¼ and Fitzgerald et al6 found a strong association of agitation
.001), copious secretion (OR: 11.8; 95% CI: 2.20-64.05, P ¼ with UE (OR: 2.9-9.5), but not with the sedation administration
.004), and OI > 5 (OR: 9.3; 95% CI: 2.45-35.48, P ¼ .001). Of method (continuous vs intermittent dosing). From our view,
94 UE patients, 64 required reintubation, while 59 (62.7%) of study design and sedation management may be the key factors
94 UE patients received full ventilator support. Of these, 46 of such differences. We should note that our patients receive
patients required reintubation, whereas 13 patients did not continuous sedation infusion, so it is possible that these patients
(71.8% vs 43.3%, P ¼ .007), yielding an unadjusted OR of develop tolerance during treatment. Whether an approach with
reintubation of 3.34 (95% CI: 1.35-8.25). Fifty-eight (90%) careful attention to escalating needs or narcotic rotation could
reintubations occurred within 30 minutes of UE. The reasons reduce UE raises questions for future quality improvement
for reintubation after UE were hypoxia (n ¼ 36, 56.2%), studies focusing on UE.
increased work of breathing (n ¼ 10, 15.6%), hemodynamic Patients with an OI > 5 were more likely to have an UE (OR:
impairment (n ¼ 9, 14%), hypoventilation (n ¼ 5, 7.8%), and 76.9). This patient subgroup may identify more unstable chil-
upper airway obstruction (n ¼ 4, 6.2%). dren in need of intensive airway management and therefore
more exposed to UE events. This finding is corroborated by
other investigators, showing that a worse clinical status, which
Discussion includes patients with respiratory conditions, shows an
This case–control study of UE demonstrates that UE events increased incidence of UE.18-21
have both patient- and nurse-associated risk factors. The case We found that UE was strongly associated with nurse–
group had a longer length of PICU stay and MV with an patient assignment ratio >1:2 and nurse experience <2 years.
increased rate of reintubation compared to the control group. Studies assessing nurse-related factors in UE events3,4 reveal
Secondary analysis showed that patients who experienced UE conflicting results. In fact, while our findings are in line with
and needed reintubation had a longer PICU stay and duration of previous studies,4,22 other investigators show no relationship
MV than those who did not require reintubation. In addition, a between experience of nurse years with UE3,6 or any associ-
lower level of consciousness, copious secretions, and higher OI ation of nursing ratio with UEs.6 While the largest group in
were risk factors for reintubation. the qualified nursing workforce in our study comprised
nurses with <5 years of experience, those with experience
<2 years of experience carried an OR of UE at 2.38. This
Risk Factors for UEs finding is expected, as junior staff nurses may not have the
We found that agitation (OR: 2.4), a requirement for continu- anticipatory or management skills for unexpected events, as
ous sedation infusion (OR: 3.2), and OI > 5 were the most they usually operate between an advanced beginner and a
important related factors associated with UEs. On the other more competent stage. 23 Quality improvement programs
hand, we did not find an association with a longer duration would provide education and training to facilitate
78 Journal of Intensive Care Medicine 35(1)

Table 2. Univariate Analysis of Factors Associated With Unplanned Extubation.a

Variables All (N ¼ 470) Controls (n ¼ 376) UE Cases (n ¼ 94) Odds Ratio (95% CI) P Value

COMFORT score 21 (18-26) 21 (18-26) 22 (16-28) .710b

COMFORT score level distribution, n (%) .387c
16 71 (15.10) 46 (12.23) 25 (26.59)
17-26 290 (61.70) 249 (66.22) 41 (43.61)
27 109 (23.19) 81 (21.54) 28 (29.78)
Patient agitation, n (%) 200 (42.55) 147 (39.07) 53 (56.38) 2.01 (1.27-3.18) .002c
Use of restraints, n (%) 263 (55.95) 212 (56.38) 51 (54.25) 0.91 (0.58-1.44) .711c
Continuous sedation infusion, n (%) 211 (44.89) 140 (37.23) 71 (75.53) 5.20 (3.11-8.70) <.001c
Time of MV before UE, hours 72 (120-225) 72 (120-237) 90 (48-218) .004b
Time of MV before UE >5 days 217 (46.17) 183 (48.67) 34 (36.17) 0.59 (0.37- 0.95) .025c
Time of day, n (%) <.001c
Day shift 230 (48.93) 206 (54.78) 24 (25.53)
Evening shift 174 (37.02) 144 (38.29) 30 (31.91)
Night shift 67 (14.04) 26 (6.91) 41 (43.61)
Night shift, n (%) 67 (14.25) 26 (6.91) 41 (43.61) 10.41 (5.88-18.41) <.001c
UE during change of shift, n (%) 71 (15.10) 57 (15.15) 14 (14.89) 0.97 (0.51-1.84) .948c
Nurse-to-patient ratio 1:1.8 (1.7-2.0) 1:1.8 (1.7-2) 1:1.8 (1.6-2) 1:1.8 (1.6-2) .585b
Nurse-to-patient ratio distribution, n (%) <.001c
1:1 19 (4.04) 10 (2.65) 9 (9.57)
1:2 424 (90.21) 350 (93.08) 74 (78.72)
>1:2 26 (5.53) 15 (3.98) 11 (11.70)
Nurse-to-patient ratio
>1:2, n (%) 26 (5.53) 15 (3.98) 11 (11.70) 3.18 (1.41-7.19) .026c
In-charge nurse—years of experience in 4 (3-5) 4 (3-5) 3 (3-4) <.001b
PICU
In-charge nurse—years of experience <.001c
distribution, n (%)
0-2 78 (16.59) 51 (13.56) 27 (28.72)
3-5 307 (65.31) 247 (65.69) 60 (63.82)
>5 85 (18.08) 78 (20.74) 7 (7.44)
In-charge nurse—years of experience 78 (16.59) 51 (13.56) 27 (28.72) 2.56 (1.50-4.38) <.001c
<2 years
Copious secretion, n (%) 77 (16.38) 50 (13.29) 27 (28.72) 2.62 (1.53-4.49) <.001c
pH 7.41 (7.37-7.46) 7.41 (7.37-7.46) 7.40 (7.34-7.45) .054b
PCO2, mm Hg 37 (31-43) 37 (31-43) 37 (30.23-45.25) .683b
PaO2/FIO2, mm Hg 281.75 (220.23-380.23) 293.30 (233.35-387.28) 240 (151.2-361) <.001b
Oxygenation index 2.1 (1.5-3.0) 2 (1.3-2.5) 3.8 (2.3-6.6) <.001b
Oxygenation index 5, n (%) 430 (91.5) 374 (99.5) 56 (59.6) 0.00 (0.00-0.03) <.001c
FIO2, % 0.30 (0.30-0.40) 0.30 (0.30-0.40) 0.40 (0.40-0.50) <.001b
Abbreviations: CI, confidence interval; FIO2, fractional inspired oxygen concentration; MV, mechanical ventilation; PICU, pediatric intensive care unit; PaO2, partial
pressure of arterial oxygen; PCO2, partial pressure of carbon dioxide; UE, unplanned extubation.
a
Data are expressed as median (percentile 25, percentile 75) or number (%).
b
Mann-Whitney U test.
c 2
w test.

development of expert knowledge and intuition and

Outcomes Related to UE Events
appropriate airway management. We found that UE patients had a longer PICU stay and length
With regard to daily event distribution, we found that of MV compared to controls, similar to other case–control
night shifts increased the risk of UE and speculate that this studies5,7 and also found a higher mortality rate among UE
finding is a result of the combination of decreased nursing patients, in contrast to a previous study.7 Although these
staff and less patient surveillance. Other studies4,6 show no findings are largely attributable to the higher proportion of
association between UE events for night/day shifts. A direct patients who fail to tolerate the UE (68%), a higher number
comparison between these studies may not be helpful, as the of patients are required to confirm these results. In fact, rein-
practice environment varies among PICUs; in general, expe- tubation may be considered a marker of illness severity,25 as
rienced nurses work day shifts, while new or inexperienced it may result from the patient’s inability to maintain sponta-
staff work night shifts, resulting in inconsistent experience neous ventilation, including a worse clinical condition, lead-
and skill mix.24 ing to a negative outcome.25 The development of a new
da Silva and Fonseca 79

Table 3. Factors Associated With Unplanned Extubation.a

Variables Coefficient Standard Error Odds Ratio (95% CI) P Value

Nursing assignment ratio >1:2 0.544913 0.597361 1.72 (0.53-5.56) .362

In-charge nurse of experience <2 years 0.865743 0.378568 2.38 (1.13-4.99) .022
Patient agitation 0.892522 0.328482 2.44 (1.28-4.25) .007
Continuous sedation infusing 1.18543 0.333323 3.27 (1.70-6.29) <.001
Night shift 2.21456 0.391319 9.16 (4.25-19.72) <.001
Time of MV before UE >5 days 0.132272 0.337531 0.88 (0.45-1.70) .695
Copious secretion 0.725591 0.382566 2.07 (0.98-4.37) .058
Oxygenation index >5 4.34296 0.776533 76.94 (16.79-352.47) <.001
Abbreviations: CI, confidence interval; MV, mechanical ventilation; UE, unplanned extubation.
a
Hosmer-Lemeshow test: w2 ¼ 5.942, df ¼ 7, P ¼ .547. Area under the Receiver Operating Charactyeristic curve. (AUC): 0.88 (95% CI, 0.85-0.91).

Table 4. Outcomes of the Patients Experiencing UE Compared to the Matched Control Participants.a

Variables All (N ¼ 470) Controls (n ¼ 376) UE Cases (n ¼ 94) Odds Ratio (95% CI) P Value

Length of PICU stay, days 12 (8-21) 11 (7-18) 16.5 (10-34.5) <.001b

Duration of MV, days 7 (4-12) 6 (4-10) 11 (5-28.5) <.001b
Length of hospital stay, days 22 (15-37) 22 (15-36) 30 (18-58.5) <.001b
Tracheostomy, n (%) 20 (4.25) 13 (3.45) 7 (7.44) .164c
VAP, n (%) 55 (11.10) 47 (12.50) 8 (8.51) 0.65 (0.29-1.42) .233d
Reintubation, n (%) 103 (21.91) 39 (10.37) 64 (68.08) 18.43 (10.67-31.81) <.001d
Mortality, n (%) 11 (2.34) 4 (1.06) 7 (7.44) 7.48 (2.14-26.12) .002c
Abbreviations: CI, confidence interval; MV, mechanical ventilation; PICU, pediatric intensive care unit; UE, unplanned extubation; VAP, ventilator-associated
pneumonia.
a
Data are expressed as median (percentile 25, percentile 75) or number (%).
b
Mann-Whitney U test.
c
Fisher exact test.
d 2
w test.

Figure 1. Kaplan-Meier curves showing the length of mechanical ventilation (A) and PICU stay (B) for both unplanned extubation patients
(cases) and planned extubation patients (controls). Data have been censored for patients who died. MV indicates mechanical ventilation; PICU,
pediatric intensive-care unit.

medical condition in the interval between extubation and tracheostomy rate is higher in the UE group, while the VAP
reintubation or a new problem requiring MV may have con- rate is not.
tributed to higher mortality in this group.26 Although our Remarkably, 72% of the UEs that required reintubation
results show no statistical differences, we observe that the occurred in patients receiving full mechanical ventilatory
80 Journal of Intensive Care Medicine 35(1)

support, although more than half of those occurring in weaning Thus, if patients are identified as having a high risk of UE,
patients did not require reintubation. In line with previous temporarily intensified surveillance may be needed. Identifica-
reports in adults, this finding may imply that some patients can tion of factors associated with UE must be used in future stud-
be on MV longer than necessary.27,28 ies to assess care focused on quality initiatives and resource
We found that many factors significantly correlate with allocation, which would provide improved health care even
reintubation, for example, a lower level of consciousness, with limited resources.
higher OI, and an increased secretion burden predisposed to In conclusion, the reasons for UEs are multifactorial, but
reintubation after UE. Other investigators corroborate these contributing factors associated with this adverse event include
findings.29-32 From a clinical view, pediatric intensivists who patient agitation, use of continuous sedation infusion, and OI
are aware of these risk factors should incorporate them into >5. In PICUs with a high nurse–patient ratio, greater attention
their decision-making and not postpone reintubation in patients may be warranted regarding nonnursing workforce factors such
presenting these factors. as patient acuity and patient numbers.

Strengths and Limitations of the Study Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to
An important strength of our study is the large number of
the research, authorship, and/or publication of this article.
patients who present with UE and are then evaluated. We
closely matched each UE patient with 4 corresponding control
patients chosen randomly, creating statistically equivalent Funding
groups for age, severity of illness score, and diagnosis. We The author(s) received no financial support for the research, author-
prospectively collected data related to UE events as part of our ship, and/or publication of this article.
quality improvement program, in contrast to current detection
systems using self-report and administrative coding systems; References
these may underdetect serious adverse events, given that data 1. Agarwal S, Classen D, Larsen G, et al. Prevalence of adverse
quality is not often validated.33 Although the majority of stud- events in pediatric intensive care units in the United States.
ies assessing UEs does not establish definitions of contributing Pediatr Crit Care Med. 2010;11(5):568-578.
factors, such as level of sedation, amount of secretions, and 2. Lucas da Silva PS, de Carvalho WB. Unplanned extubation in
weaning status, we attempted to control these measures by pediatric critically ill patients: a systematic review and best prac-
using objective definitions. tice recommendations. Pediatr Crit Care Med. 2010;11(2):
This study has also several limitations. First, we used a 287-294.
matched case–control design to assess our data. We selected 3. Marcin JP, Rutan E, Rapetti PM, Brown JP, Rahnamayi R,
a control group from a random sample of all patients at risk of Pretzlaff RK. Nurse staffing and unplanned extubation in the
experiencing the studied outcome. Although both groups are pediatric intensive care unit. Pediatr Crit Care Med. 2005;6(3):
similar in age, gender, admission diagnosis, and severity of 254-257.
illness, there is no perfect matching technique. While selecting 4. Ream RS, Mackey K, Leet T, et al. Association of nursing work-
controls during the same shift as cases of UE is attractive for load and unplanned extubations in a pediatric intensive care unit.
identifying patient-specific factors, it may limit the identifica- Pediatr Crit Care Med. 2007;8(4):366-371.
tion of factors affected by unit staffing.6 Second, we collected 5. Roddy DJ, Spaeder MC, Pastor W, Stockwell DC, Klugman D.
our data at a single institution, which may limit the general- Unplanned extubations in children: impact on hospital cost and
izability of our study. Other investigators may use different length of stay. Pediatr Crit Care Med. 2015;16(6):572-575.
criteria for ventilator weaning, which in turn also limit the 6. Fitzgerald RK, Davis AT, Hanson SJ. Multicenter analysis of the
generalization of our results. Third, we did not assess the vari- factors associated with unplanned extubation in the PICU. Pediatr
able proportion of administrative duties and patient care Crit Care Med. 2015;16(7):e217-e223.
assumed by each charge nurse in this study, which might 7. Kanthimathinathan HK, Durward A, Nyman A, Murdoch IA,
impact the workload. Finally, although we explored several Tibby SM. Unplanned extubation in a paediatric intensive care
risk factors for UE, other variables not included in this study unit: prospective cohort study. Intensive Care Med. 2015;41(7):
may influenced the UE occurrence, such as delirium, presence 1299-1306.
of caregivers at the bedside when the UE episode occurs, float- 8. Polit D, Beck CT. Essentials of Nursing Research: Appraising
ing nurses from other units, and the nursing workload in both Evidence for Nursing Practice. 7th ed. Philadelphia, PA: Wolters
groups. Further studies incorporating these variables may yield Kluwer Health, Lippincott Williams & Wilkins; 2010.
additional information. 9. Amorosa JK, Bramwit MP, Mohammed TL, et al. ACR appropri-
ateness criteria routine chest radiographs in intensive care unit
patients. J Am Coll Radiol. 2013;10(3):170-174.
Next Steps and Conclusion 10. Ambuel B, Hamlett KW, Marx CM, Blumer JL. Assessing dis-
Development of a stratification tool to identify patients at risk tress in pediatric intensive care environments: the COMFORT
of UE based on clinical factors would be extremely helpful. scale. J Pediatr Psychol. 1992;17(1):95-109.
da Silva and Fonseca 81

11. Bock KR, Silver P, Rom M, Sagy M. Reduction in tracheal lumen workload, skill mix and staff supervision. Intensive Care Med.
due to endotracheal intubation and its calculated clinical signifi- 2004;30(6):1160-1166.
cance. Chest. 2000;118(2):468-472. 23. Benner P. From Novice to Expert: Excellence and Power in
12. Randolph AG, Wypij D, Venkataraman ST, et al. Effect of Clinical Nursing Practice. Commemorative Edition. Vol. 3. New
mechanical ventilator weaning protocols on respiratory outcomes Jersey: Prentice Hall Health; 2001:13-36.
in infants and children: a randomized controlled trial. JAMA. 24. Hamilton P, Eschiti VS, Hernandez K, Neill D. Differences
2002;288(20):2561-2568. between weekend and weekday nurse work environments and
13. da Silva PS, de Aguiar VE, Neto HM, de Carvalho WB. patient outcomes: a focus group approach to model testing.
Unplanned extubation in a paediatric intensive care unit: impact J Perinat Neonatal Nurs. 2007;21(4):331-341.
of a quality improvement programme. Anaesthesia. 2008;63(11): 25. Kiekkas P, Aretha D, Panteli E, Baltopoulos GI, Filos KS.
1209-1216. Unplanned extubation in critically ill adults: clinical review. Nurs
14. Pollack MM, Ruttimann UE, Getson PR. Pediatric Risk of Mor- Crit Care. 2013;18(3):123-134.
tality (PRISM) score. Crit Care Med. 1988;16(11):1110-1116. 26. Frutos-Vivar F, Esteban A, Apezteguia C, et al. Outcome of rein-
15. Leteurtre S, Martinot A, Duhamel A, et al. Validation of the tubated patients after scheduled extubation. J Crit Care. 2011;
paediatric logistic organ dysfunction (PELOD) score: prospec- 26(5):502-509.
tive, observational, multicentre study. Lancet. 2003;362(9379): 27. Krinsley JS, Barone JE. The drive to survive: unplanned extuba-
192-197. tion in the ICU. Chest. 2005;128(2):560-566.
16. Mokhlesi B, Tulaimat A, Gluckman TJ, Wang Y, Evans AT, 28. Bouza C, Garcia E, Diaz M, Segovia E, Rodriguez I. Unplanned
Corbridge TC. Predicting extubation failure after successful com- extubation in orally intubated medical patients in the intensive
pletion of a spontaneous breathing trial. Respir Care. 2007; care unit: a prospective cohort study. Heart Lung. 2007;36(4):
52(12):1710-1717. 270-276.
17. Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance 29. da Silva PS, Fonseca MC. Unplanned endotracheal extubations in
definition of health care-associated infection and criteria for spe- the intensive care unit: systematic review, critical appraisal, and
cific types of infections in the acute care setting. Am J Infect evidence-based recommendations. Anesth Analg. 2012;114(5):
Control. 2008;36(5):309-332. 1003-1014.
18. Chang LC, Liu PF, Huang YL, Yang SS, Chang WY. Risk factors 30. Faustino EV, Gedeit R, Schwarz AJ, Asaro LA, Wypij D, Curley
associated with unplanned endotracheal self-extubation of hospi- MA. Accuracy of an extubation readiness test in predicting suc-
talized intubated patients: a 3-year retrospective case-control cessful extubation in children with acute respiratory failure from
study. Appl Nurs Res. 2011;24(3):188-192. lower respiratory tract disease. Crit Care Med. 2017;45(1):
19. Chao YF. A conceptual framework for predicting selected com- 94-102.
plications of ICU patients. Crit Care Nurse. 1993;13(5):48-54. 31. Fontela PS, Piva JP, Garcia PC, Bered PL, Zilles K. Risk factors
20. Vassal T, Anh NG, Gabillet JM, Guidet B, Staikowsky F, Offen- for extubation failure in mechanically ventilated pediatric
stadt G. Prospective evaluation of self-extubations in a medical patients. Pediatr Crit Care Med. 2005;6(2):166-170.
intensive care unit. Intensive Care Med. 1993;19(6):340-342. 32. Miu T, Joffe AM, Yanez ND, et al. Predictors of reintubation in
21. Boulain T. Unplanned extubations in the adult intensive care unit: critically ill patients. Respir Care. 2014;59(2):178-185.
a prospective multicenter study. Association des reanimateurs du 33. Tsang C, Palmer W, Aylin P. National Institute for Health
centre-ouest. Am J Respir Crit Care Med. 1998;157(4 pt 1): Research. Patient safety indicators: a systematic review of the
1131-1137. literature. 2008. https://www1.imperial.ac.uk/resources/
22. Tibby SM, Correa-West J, Durward A, Ferguson L, Murdoch IA. 0D60DED9-022F-4F8D-BBE5-48D35857CF94/. Accessed
Adverse events in a paediatric intensive care unit: relationship to March 9, 2017.