OBJECTIVES:

Unintended extubations (UEs) lead to significant morbidity in neonates. A quality improvement project was initiated in response to high rates in our level IV NICU. We targeted creating and sustaining UE rates below the published standard of 1 per 100 ventilator days.

METHODS:

This project spanned 4 time periods: baseline, epoch 1 (December 2010–May 2012), sustain, and epoch 2 (May 2015–December 2017) by using standard quality improvement methodology. Epoch 1 interventions included real-time analysis of UE events, standardization of taping, patient positioning and movement, accurate event reporting, and change in nomenclature. Epoch 2 interventions included reduction in daily chest radiographs (CXRs) and development of a high-risk tool. Patient and event characteristics were statistically compared across time points.

RESULTS:

Of the 612 UE events recorded over 10 years, 249 UEs occurred from May 2011 to 2017 involving 184 unique patients. UE rates decreased by 43% (from 1.75 to 0.99 per 100 ventilator days; epoch 1) and were sustained until a notable spike. Epoch 2 interventions led to a further 31% rate reduction. Single CXR use decreased by half. Median corrected gestational age at the time of an event was 35 weeks (interquartile range: 29–41). Seventy percent of infants experiencing an UE required reintubation, 29% had a previous event, and 9% had a code event.

CONCLUSIONS:

A decrease in UE below benchmarks can be achieved and sustained by standardization and mitigation interventions. This decline was also accompanied by a reduction in use of CXRs without increasing UE events.

Initiating and maintaining a culture of safety in the NICU is essential for the treatment of newborns who are critically ill. The use of invasive mechanical ventilation via endotracheal tube (ETT) remains a key component of neonatal care. Unintended extubations (UEs), defined as any incident in which there is a sudden and unexpected dislodgement of an ETT in a patient receiving invasive mechanical ventilation at a time that was not specifically intended or ordered,1,3 is a common problem in all ICUs.4,5 UEs in the NICU are further complicated by limited sedation use, uncuffed ETTs, a humidified environment, a small surface area, and the fragility of an infant’s face.4,8 

UEs are the fourth most common adverse event in the NICU and can lead to airway trauma, intraventricular hemorrhage, and cardiovascular collapse.9,10 Studies have identified risk factors, but there is no standard of care to reduce events and no consensus as to an acceptable benchmark. In 2012, the Vermont Oxford Network suggested keeping the UE median rate below a threshold of 2 events per 100 ventilator days. Additional studies suggest an even lower benchmark of <1 event per 100 ventilator days.2 

In 2007, a notable increase in UE events occurred in the Children’s National Medical Center (CNMC) NICU. Baseline data from 2007 to 2010 revealed rates above the benchmark (mean 1.75 per 100 ventilation days) that reached a maximum of 4.5 events per 100 ventilation days. This rate was unacceptably high, and a quality improvement (QI) project was initiated in December 2010 (epoch 1) with the specific aim of reducing the UE rate to below the suggested benchmark of 1 per 100 ventilator days. Over the next 2 years, vigorous implementation of interventions led to the achievement and sustainment of this goal. In May 2015 (epoch 2), a noticeable elevation led to a renewed effort to return to previous standards and further reduce UE rates by 50% (from 0.99 to 0.50 per 100 ventilator days). In this study, we report on the development of this QI project with which we aimed to reduce UEs in our level IV NICU.

The CNMC NICU is an urban, academic, 54-bed, level IV nondelivery NICU that admits >800 infants per year (24% surgical cases, 6% low birth weight cases, 13% very low birth weight cases, and 1.3% extracorporeal membrane oxygenation). As a QI initiative, this project did not constitute human subjects research and was not under the oversight of the CNMC Institutional Review Board.

All infants who required invasive mechanical ventilation via an ETT were at risk for a UE. A conservative definition of only in-hospital conditions in which an ETT was clearly dislodged or clinically removed from patients who developed an unexpected desaturation or deterioration with no change in the color of a Co2 detector was used.3,11 A code event was defined as cardiovascular instability (heart rate <60 beats per minute) requiring chest compressions, positive pressure ventilation, and/or resuscitative medications. This project evolved over 10 years and consisted of a baseline period (January 2007–December 2010), epoch 1 (December 2010–May 2012), and a sustain period followed by epoch 2 (May 2015–December 2017).

Extubation data collected from 2007 to 2008 only included the date and time of the event. An improved apparent cause analysis (ACA) form was developed in 2008 and continually modified (Supplemental Fig 6). Complications secondary to UEs, specifically cardiac instability requiring cardiac compressions and/or code medications, were added. Weekly and collective monthly review of these ACAs revealed areas for improvement. The Ishikawa cause-and-effect diagram revealed a complex etiology for UEs (Fig 1). Major causes in epoch 1 were equipment failures (loose tape, unsecure ventilator tubing, and variable taping techniques), processes (excessive motion or handling during procedures), measurement errors (inaccurate event reporting), and inadequate sedation. Apparent causes in epoch 2 additionally included processes (inadequate positioning at rest) and measurement errors (inadequate identification of high-risk patients). The key driver diagram (Fig 2) was used to map out interventions to be tested.

Multiple interventions were implemented. Those with a measure of reliability level 1 were used to create a standardized workflow, level 2 measures were used to error-proof the system and prevent workarounds, and level 3 interventions were used to redesign the system and change performance.12 

Epoch 1 Interventions

  1. Intervention 1: Airway alert card (level 2). A card placed on the ventilator was used to document the depth of ETT insertion and date of the last chest radiograph (CXR) performed. As a quick reference guide, the card allowed discrepancies to be corrected during routine checks and emergencies before dislodgement occurred by comparing ETT markings with the ideal documented depth.

  2. Intervention 2:

    • Real-time ACA huddle (level 1). All personnel involved in an event met within 72 hours to determine contributing factors from eyewitness accounts. Weekly review of ACAs furthered problem identification and intervention development.

    • Airway safety protection team (level 1). In September 2010, a multidisciplinary team of respiratory therapists (RTs), nurses, physicians, and physician extenders was assembled to review events and develop solutions from frontline provider perspectives. Data were entered into an expanded database used to generate baseline data for this period.

  3. Intervention 3: UE bundle implementation. In May 2011, the following bundle of interventions was implemented and refined over several months:

    • Communication during morning rounds (level 1): The presence of RTs on rounds was mandated for any intubated patient to report ETT position and voice concerns regarding its maintenance. Physicians were mandated to place electronic orders denoting size and depth and document radiograph confirmation.

    • Standardization of ETT securement and placement (level 1): The ETT affixation was standardized to a double “Y” taping method at the corner of the mouth, taught to all staff members via in-services and peer teaching, and audited by senior providers. Insertion depth was determined from the gum by using a weight-based formula, and radiograph confirmation occurred immediately after placement. If ETT tape was found loose, staff would reinforce the securement; however, any ETT tape found “off the cheek” would require retaping.

    • Standardization of infant positioning during morning radiographs (level 1) and 2-person handling (level 2): The infant’s position for radiograph confirmation of ETT position was standardized to a vertical neutral position to decrease frequent untaping and repositioning attempts. The presence of 2 staff members at times of major anticipated motion was also enforced.

    • Accurate event reporting and documentation (level 1): Several staff educational sessions regarding what constituted a UE and how to complete the ACA form were performed.

    • Event name change (level 1): Change from self-extubation and/or accidental extubation to a UE was made to alter the perception of a UE from an unpreventable occurrence to a preventable or “never” event.

Epoch 2 Interventions

Interventions implemented in epoch 1 were continued throughout the sustain period. In July 2015, a clinically notable increase in events was detected. The following interventions were enforced:

  1. Intervention 4: Multidisciplinary team assembly and data review, reinforcement of epoch 1 initiatives (level 1), and standardization of infant positioning at rest (level 2). The multidisciplinary airway safety protection team was reassembled as the Stop Unintended Extubation team. The cause-and-effect diagram was reviewed (Fig 1), and previous standards from epoch 1 were reinforced. Standardization of infant position at rest was determined to be an additional factor. Interventions included positioning the ventilator tubing coming in from the head of the bed, flipping the ETT across the face and not side hanging, and supporting infants with rolls to prevent downward sliding within the isolette and placing tension on the ETT (level 2).

  2. Intervention 5: CXR reduction protocol (level 3). A reduction in the use of CXRs for all stable intubated neonates from daily to twice weekly decreased unnecessary motion and frequency of ETT manipulation. An additional process measure (number of CXRs per patient day) was added.

  3. Intervention 6: Identification of high-risk patients (level 2). In January of 2016, the Zell UE Risk Tool was developed by drawing on known risk factors in an attempt to mitigate occurrence of events (Supplemental Figs 7 and 8). The tool was revised and edited throughout epoch 2 but has not been validated. Each patient was scored and assigned a point value, with a higher value (>6) denoting high risk. Scores were completed by the outgoing nurse on each shift and discussed by the incoming nurse with the rounding team. An elevated score triggered a series of actionable items whereby a “UE RISK” sign was placed at the bedside, staff did not offer patients unreasonable motion, and nurses checked securement of the ETT and intravenous access. Staff members were also empowered to notify physicians if the patient required increased sedation. Pictures of high-risk positions were added to the back of the tool as reminders of proper positioning (Supplemental Fig 8).

Continued reinforcement and monitoring occurred for the above interventions.

The primary outcome measure was the rate of UEs calculated as the number of UEs per 100 patient ventilator days minus the number of tracheostomy days. Extubation data were collected at the time of the event by the bedside team on ACA forms and reviewed within 72 hours by the airway safety protection team that included physician and nurse leaders. Ventilator days were derived from electronic charts. The number of days between events was calculated as a secondary outcome measure. Adverse events stemming from UE were measured by the number of UE-related code events per month recorded on the ACA form. An additional process measure, the number of single CXRs per patient day for all patients, was added in epoch 2 when the CNMC NICU transitioned to a CXR reduction strategy. Although this strategy aligned with best practices,13 the resultant UE rate was monitored closely (UE rate constituted a balancing measure relative to CXR reduction). The number of CXRs and net charges per CXR were tracked by using Trendstar billing data.

Statistical process control charts were developed by using QI Macros software (KnowWare International, Denver, CO). The UE rate and CXRs per patient days were displayed on u charts. A t chart was used to depict days between events. Established rules for differentiating special versus common-cause variation were applied.14 Statistical analysis of demographic data and risk factors using χ2, analysis of variance, Kruskal-Wallis, and Tukey’s studentized range were performed where appropriate by using SAS software (SAS Institute, Inc, Cary, NC).

Between 2007 and 2010 (baseline), the UE rate was 1.75 per 100 ventilator days. In 2010, the first 2 interventions were initiated but did not result in a decreased UE rate. They improved the documentation and aided in the elucidation of causes and iterative plan-do-study-act cycle design. In May 2011, the bundle of interventions was initiated and augmented by a 1-on-1 bedside discussion with each team member after each UE resulting in a 43% reduction (0.99 events per 100 ventilator days) (Fig 3), which was sustained with ongoing efforts and monitored closely for 3 years (sustain). The start of epoch 2 (July 2015) was heralded by a significant elevation in UE rate (2.67 per 100 ventilator days) (ie, ∼10–12 events per month compared with 3–4 per month previously). Epoch 2 interventions were initiated, leading to rapid decrease within 2 months to previous rates followed by an added push to decrease rates further by focusing on reduction of CXRs as part of a best practices initiative. Staff were reeducated to proper ETT positioning via bedside visual aids, and the Zell Risk Tool was introduced, yielding a reduction of 31% (0.68 events per 100 ventilator days) (Fig 3). Overall, a 61% rate reduction occurred over 10 years. The raw annual number of UEs decreased by 51% (from 74 [2007–2011] to 36 UEs per year [2012–2017]).

Within 10 months of the initiation of epoch 1 interventions, the number of days between UEs increased from 3 to 7 days. Epoch 2 interventions increased the number of days between UEs from 6 to 9 days (longest interval 75 days) (Fig 4). Analysis of events for adverse outcomes of UE (May 2011–Dec 2017) revealed that only 23 UEs (9%) resulted in a code event (Table 1).

After the initiation of the CXR reduction initiative in epoch 2, the number of single CXRs declined from 0.45 to 0.27 single CXRs per patient day (mean) (Fig 5). Total charges for single CXRs decreased by 45% (from $289 520 per month to $159 424 per month), saving a projected $1.5 million per year. The CXR reduction strategy was not associated with increased rates of UE.

Of the 612 UE events recorded over 10 years, 249 events representing 184 individual patients (May 2011–December 2017) had detailed patient and event characteristics additionally recorded and analyzed to determine contributing factors (Table 1). The median gestational age (GA) at birth was 25 weeks (interquartile range [IQR]: 24–32), median corrected GA at the time of event was 35 weeks (IQR: 29–41), and mean event weight was 2108 g (±1275 g). During epoch 2, neonates had significantly lower event weights, GAs, and postmenstrual ages compared with the sustain period. Twenty-nine percent of neonates had a previous event (1.4 ± 0.89 events per infant). The mean number of ventilated days was 35 days (IQR: 14.75–66), 14% were on a sedation wean, and 70% required reintubation within 24 hours, with less reintubations occurring during the sustain period compared with other epochs. Fewer patients were sedated during epoch 1 compared with the other periods. No statistically significant differences were noted between total events occurring during the day as compared with the night.

This is a dual-epoch QI study during which we successfully sustained UE rates below both the national benchmark (2 events per 100 ventilation days) and the suggested rate of 1 per 100 ventilation days2 and an overall reduction of 61% over 10 years.

Strengths of this study were the presence of 2 study periods, use of consistent definitions, and consistent auditors during both epochs (J.C. and J.P.G.). Success may be attributed to the strong culture of safety around airway protection in our NICU (presence of a dedicated airway protection team, real-time analysis, standard taping, positioning, and ETT monitoring, and implementation of a risk tool to identify high-risk patients). These techniques combined with consistent reeducation of staff and continued improvements in the communication between the physicians, nurses, and RTs have driven the UE rate down. With the implementation of a limited CXR strategy, reductions in the manipulation of the ETT and infant handling have furthered reductions while allowing potential cost savings of >$1.5 million per year and adhering to best practices. Continued surveillance by our team also provided an early warning and rapid response to sudden spikes in events that could have been easily missed. The use of multiple level 1 and 2 interventions enabled us to obtain a 43% reduction in UEs. However, level 2 and 3 interventions allowed us to modify the system with a higher reliability and reduced our failure rate to <1%.

Standardization of ETT taping was previously noted to be more important than the device or method of securement used.2,15 Despite only using the double “Y” method of taping, our rates continued to decline. This is supported by Fontanez-Neives et al,16 who observed a decrease in UE after standardizing the ETT taping method, but differed from Merkel et al,2 who found no significant change in the rate of UE after standardization of taping. Most likely, a consistent, uniform approach combined with education and reevaluation is the key to improvement rather than any commercially available device, adhesives, or soft restraints.

Communication between the physician, nurse, and RT has significantly improved through the usage of the risk tool. This tool gave nurses and RTs an unbiased document with which to express their opinion of the patient’s condition and empowered them to initiate plans. Moreover, positioning pictures increased awareness among nurses of the ideal ETT, ventilator tubing, and patient position. This tool also created a common clear language. Bedside staff ACA reports now contain sufficient detail describing the event to ascribe cause. This is a clear indication of their in-depth understanding of the process and prevention strategies.

UE events occurred commonly in very premature infants, and UE events occurred more frequently at a corrected GA of 35 weeks (IQR: 29–41). Our study differed from that of Veldman et al,17 in which UE occurred at ≤31 weeks corrected GA. Our study correlated well with other studies, illustrating that UEs occur predominantly in infants >34 weeks corrected GA.18 Contrary to previous studies,16,19 UE rates occurred at an average event weight of >1500 g. Infants had an average total period of ventilation of 35 days (IQR: 14.75–66). Similarly, a study by Fontánez-Nieves et al16 revealed that UE occurred at a median of 14 days (3–25). This may indicate that a length of ventilation ≥2 weeks could be a risk factor for UEs.19 

Our reintubation rate within 24 hours after UEs was 70%, which is much higher than that reported by Veldman et al17 (28%) but similar to Loughead et al1 (66%) and Fontánez-Nieves et al16 (75%). This would suggest that infants who had a UE still required invasive mechanical ventilation for stability.

This study has inherent limitations. Prospective data regarding patients who did not experience a UE event were not collected, and the type and amount of sedation was not captured. Our standard sedation practices involved the use of a combination of morphine and/or midazolam intravenously and varied according to the patient’s underlying morbidity. We continue to strive to standardize sedation and weaning protocols and explore methods of determining extubation readiness.17,20 The effect of the limitation of kangaroo care on parent bonding and satisfaction was not captured. Kangaroo care was limited only in infants with high-risk scores, of which there were few. Finally, although there has been adequate awareness surrounding the risk tool, inconsistent discussion between nurses and the team still occurs. This tool is currently being validated and is only used as a communication tool.

Future directions include comparing patients with and without UEs, validating the risk tool, measuring the amount of sedation and duration of kangaroo care as balancing measures, improving data collection for code events, and revising nursing/patient ratios to determine their effects on UE.

Reducing UE can be successfully accomplished in a NICU setting to achieve levels below national benchmarks through standardization and mitigation interventions. Continued vigilance is imperative to sustain reduction, identify special causes, and reinforce measures with staff.

Dr Galiote and Ms Ridoré contributed to the study’s inception, collected data, performed statistical analysis, and drafted, reviewed, and revised the manuscript; Ms Carman participated in the study design and design of the data collection instruments, reviewed events in real time, implemented interventions, and critically reviewed the manuscript; Ms Zell participated in the study design, developed the Zell unintended extubation tool, supervised the implementation of interventions, and critically reviewed the manuscript; Ms Brant and Ms Gayle served as the primary respiratory therapy liaisons for all aspects of this work, participated in the study design, supervised the implementation of interventions, and reviewed the manuscript; Dr Short initiated this quality improvement project more than a decade ago in an attempt to gain insight into accurate reporting of data, supervised the development of interventions, reviewed data, and critically reviewed the manuscript; Dr Klugman played an integral role in the inception of the study during epoch 1 and data collection, served as a content expert in unintended extubation events, and reviewed the manuscript; Dr Soghier conceptualized and designed the study in epoch 2, developed the cause-and-effect diagram, modified the key driver diagram, collected data, supervised development and implementation of interventions, drafted the initial manuscript, and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

FUNDING: No external funding.

We acknowledge our nurse educators (Carmen Blake, Dawn Brittingham, and Ann-Marie Brown) and all our nurses who tirelessly watch over our infants and their tubes and ensure that they receive the best care possible. We thank the following nurses: Jessica Eitel, Ivy Prothro, Rey Emmanuel, Jessica Mcauliffe, Jennifer Kentros, Morgan Lee, Katelyn Schmidt, Kizzie Reed, and Miranda Ross, who participated in this project. We also thank Dr Mohamed El-Dib, Amy Garman, May-Britt Sten, Giovanni Jordan, Kathy Mills, Katherine Pham, William Powell, and Sanya Tyler for their support and contribution to the project over the years.

ACA

apparent cause analysis

CNMC

Children’s National Medical Center

CXR

chest radiograph

ETT

endotracheal tube

GA

gestational age

IQR

interquartile range

QI

quality improvement

RT

respiratory therapist

UE

unintended extubation

1
Loughead
JL
,
Brennan
RA
,
DeJuilio
P
,
Camposeo
V
,
Wengert
J
,
Cooke
D
.
Reducing accidental extubation in neonates.
Jt Comm J Qual Patient Saf
.
2008
;
34
(
3
):
164
170, 125
2
Merkel
L
,
Beers
K
,
Lewis
MM
,
Stauffer
J
,
Mujsce
DJ
,
Kresch
MJ
.
Reducing unplanned extubations in the NICU.
Pediatrics
.
2014
;
133
(
5
). Available at: www.pediatrics.org/cgi/content/full/133/5/e1367
[PubMed]
3
Meyers
JM
,
Pinheiro
J
,
Nelson
MU
.
Unplanned extubation in NICU patients: are we speaking the same language?
J Perinatol
.
2015
;
35
(
9
):
676
677
4
Sadowski
R
,
Dechert
RE
,
Bandy
KP
, et al
.
Continuous quality improvement: reducing unplanned extubations in a pediatric intensive care unit.
Pediatrics
.
2004
;
114
(
3
):
628
632
[PubMed]
5
Rachman
BR
,
Watson
R
,
Woods
N
,
Mink
RB
.
Reducing unplanned extubations in a pediatric intensive care unit: a systematic approach.
Int J Pediatr
.
2009
;
2009
:
820495
[PubMed]
6
Fitzgerald
RK
,
Davis
AT
,
Hanson
SJ
;
National Association of Children’s Hospitals and Related Institution PICU Focus Group Investigators
.
Multicenter analysis of the factors associated with unplanned extubation in the PICU.
Pediatr Crit Care Med
.
2015
;
16
(
7
):
e217
e223
7
Barber
JA
.
Unplanned extubation in the NICU.
J Obstet Gynecol Neonatal Nurs
.
2013
;
42
(
2
):
233
238
[PubMed]
8
Kiekkas
P
,
Aretha
D
,
Panteli
E
,
Baltopoulos
GI
,
Filos
KS
.
Unplanned extubation in critically ill adults: clinical review.
Nurs Crit Care
.
2013
;
18
(
3
):
123
134
[PubMed]
9
Klugman
D
,
Berger
JT
,
Spaeder
MC
,
Wright
A
,
Pastor
W
,
Stockwell
DC
.
Acute harm: unplanned extubations and cardiopulmonary resuscitation in children and neonates.
Intensive Care Med
.
2013
;
39
(
7
):
1333
1334
[PubMed]
10
Silva
PS
,
Reis
ME
,
Aguiar
VE
,
Fonseca
MC
.
Unplanned extubation in the neonatal ICU: a systematic review, critical appraisal, and evidence-based recommendations.
Respir Care
.
2013
;
58
(
7
):
1237
1245
[PubMed]
11
Mbi Ndakor
S
,
Nelson
MU
,
Pinheiro
JMB
.
Counting unplanned extubations: marked variation among neonatologists.
J Perinatol
.
2017
;
37
(
6
):
698
701
12
Nolan
T
,
Resar
R
,
Haraden
C
,
Griffin
FA
.
Improving the Reliability of Health Care
.
Boston, MA
:
Institute of Healthcare Improvement
;
2004
13
Ho
T
,
Dukhovny
D
,
Zupancic
JA
,
Goldmann
DA
,
Horbar
JD
,
Pursley
DM
.
Choosing wisely in newborn medicine: five opportunities to increase value.
Pediatrics
.
2015
;
136
(
2
). Available at: www.pediatrics.org/cgi/content/full/136/2/e482
[PubMed]
14
Montgomery
DC
.
Introduction to Statistical Quality Control
. 4th ed.
New York, NY
:
John Wiley
;
2001
15
Crezeé
KL
,
DiGeronimo
RJ
,
Rigby
MJ
,
Carter
RC
,
Patel
S
.
Reducing unplanned extubations in the NICU following implementation of a standardized approach.
Respir Care
.
2017
;
62
(
8
):
1030
1035
[PubMed]
16
Fontánez-Nieves
TD
,
Frost
M
,
Anday
E
,
Davis
D
,
Cooperberg
D
,
Carey
AJ
.
Prevention of unplanned extubations in neonates through process standardization.
J Perinatol
.
2016
;
36
(
6
):
469
473
17
Veldman
A
,
Trautschold
T
,
Weiss
K
,
Fischer
D
,
Bauer
K
.
Characteristics and outcome of unplanned extubation in ventilated preterm and term newborns on a neonatal intensive care unit.
Paediatr Anaesth
.
2006
;
16
(
9
):
968
973
[PubMed]
18
da Silva
PS
,
de Aguiar
VE
,
Neto
HM
,
de Carvalho
WB
.
Unplanned extubation in a paediatric intensive care unit: impact of a quality improvement programme.
Anaesthesia
.
2008
;
63
(
11
):
1209
1216
[PubMed]
19
Carvalho
FL
,
Mezzacappa
MA
,
Calil
R
,
Machado Hda
C
.
Incidence and risk factors of accidental extubation in a neonatal intensive care unit.
J Pediatr (Rio J)
.
2010
;
86
(
3
):
189
195
[PubMed]
20
Jarachovic
M
,
Mason
M
,
Kerber
K
,
McNett
M
.
The role of standardized protocols in unplanned extubations in a medical intensive care unit.
Am J Crit Care
.
2011
;
20
(
4
):
304
311; quiz 312

Competing Interests

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

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

Supplementary data