Volume 63, Issue 11 p. 1209-1216
Free Access

Unplanned extubation in a paediatric intensive care unit: impact of a quality improvement programme

P. S. L. Da Silva

P. S. L. Da Silva

Consultant, Paediatric Intensive Care Unit

Search for more papers by this author
V. E. De Aguiar

V. E. De Aguiar

Statistician, Hospital Estadual de Diadema (UNFESP), Diadema

Search for more papers by this author
H. M. Neto

H. M. Neto

Consultant, Paediatric Intensive Care Unit

Search for more papers by this author
W. B. De Carvalho

W. B. De Carvalho

Professor, Department of Paediatrics, Paediatric Intensive Care Unit, Universidade Federal de São Paulo (UNIFESP), São Paulo, Brazil

Search for more papers by this author
First published: 09 October 2008
Citations: 40
Dr Paulo Sergio Lucas da Silva
[email protected]


Unplanned tracheal extubation is an important quality issue in current medical practice as it is a common occurrence in paediatric intensive care units. We have assessed the effectiveness of a continuous quality improvement programme in reducing the incidence of unplanned extubation over a 5-year period. After a 2-year baseline period, we developed action plans to address the issues identified. Following implementation of the programme, the overall incidence of unplanned extubation decreased from 2.9 unplanned extubations per 100 intubated patient days in the first year to 0.6 in the last year (p = 0.0001). This reduction was the result of a decrease in unplanned extubation in children younger than 2 years of age. Although mortality was similar to that of children who did not experience an unplanned extubation, those with an unplanned extubation had a significantly longer duration of mechanical ventilation, longer stay in the intensive care unit, and longer hospital stay. We found that the implementation of a continuous quality improvement programme is effective in reducing the overall incidence of unplanned extubations.

Besides cost-controlling efforts and continuing advances in science and technology, quality of care has emerged as an important issue [1]. Because of the acuity of care and specialised systems of health care required, the intensive care unit (ICU) offers a unique area of focus for quality improvement [2]. Thus, the implementation of quality improvements in the ICU represents a challenging activity, the results of which can sometimes be discouraging [2].

In any quality system, routines for reporting undesirable events are key elements, while screening for adverse occurrences is a well-established method of quality control in hospitals [3].

Unplanned tracheal extubation in ICU adult patients has been extensively investigated over the past decade because they are associated with a greater risk of mortality, morbidity, and increased resource utilisation [4]. The reported incidence of unplanned extubation in adults ranges from 2.8% to 20.6% [5]. Because of its frequent occurrence and importance, unplanned extubation is a worthwhile target for quality improvement programmes [6–8]. There are a number of quality improvement studies reporting a reduction in unplanned extubation in adults after implementation of a monitoring and provider education programme [6, 9–12]. Few reports have been conducted assessing the impact of unplanned extubation on outcome [4, 13–15]. These have shown that patients with unplanned extubation had prolonged mechanical ventilation, ICU stay and hospital stay. However, there were conflicting results with regard to mortality – unplanned extubation was not associated with increased mortality [4, 13], or resulted in reduced mortality [14, 15]. Only one study in children demonstrated an association between prolonged mechanical ventilation and paediatric intensive care unit (PICU) stay and unplanned extubation [8], and no studies have yet assessed mortality. The primary objective of this study was to assess the effectiveness of a continuous quality improvement programme in reducing the incidence of unplanned extubation in a PICU.


A prospective cohort study was carried out in a five-bed PICU within a university-affiliated hospital. The Ethics Committee of the hospital approved the study and waived the need for written informed consent. From 1 January 2002 to 1 May 2007 all mechanically ventilated children were monitored for incidents of unplanned extubation. The study population included all consecutive patients admitted to the PICU who met the following conditions: age between 1 month and 12 years, and need for invasive mechanical ventilation. Patients with a tracheostomy were excluded. All patients were cared for by attending physicians who were board certified in paediatric critical care medicine, along with registered nurses experienced in critical care. The nurse to patient ratio was 1 : 1.

All patients had undergone orotracheal intubation and the method of securing the tracheal tubes was standardised using the same type of adhesive tape. The tube was affixed to the face with an ‘H’ shaped piece of elastic tape with the two arms of the H placed across the upper lip and the other two arms wrapped in a spiral fashion around the tube for at least two complete wraps. In line with the institute’s policy, physical restraints were used when deemed necessary and whenever both pharmacological and non-pharmacological interventions had proved unsuccessful in ensuring patient safety. Mittens or two-point soft wrist restraining therapies were used.

We developed our improvement model based on the ‘Plan-Do-Check-Act’ cycle. During phase I (‘plan’), running from 1 December 2001 to 30 December 2001 a multi-disciplinary task force (or continuous quality improvement (CQI) team) was set up consisting of a consultant intensivist (the primary investigator), respiratory therapists and nursing staff. The CQI team devised a one-page event report to collect and analyse information on patient status at the point of extubation. Data were collected during phase II (‘do’), spanning from 1 January 2002 to 30 April 2004. The first 4 months of this phase were considered a pilot phase for testing and improving the event report. Thus, only data collected for each year (1 May–30 April), beginning on 1 May 2002 were considered in the analysis. Risk factors for unplanned extubation were assessed based on a previously published event report form [5, 9] consisting of three parts: (i) patient characteristics, (ii) description of the extubation and (iii) outcome. The event report form was filled out by the nurse caring for the patient and double-checked by the primary investigator to ensure accuracy. Prior to study commencement, PICU nurses were trained to fill out these forms. In case of doubt, nurses could consult the CQI team or the primary investigator who regularly visited the unit to collect data and to ensure compliance with the study. For children without unplanned extubations (controls), the form excluded questions about the extubation. These data were used to compare features of children with and without unplanned extubations. In phase III (‘check’), which ran from 1 May 2004 to 1 June 2004, the team quantified and benchmarked the incidence of unplanned extubation and conducted a root-cause analysis of associated factors as well as evaluating the previously collected data. In phase IV (‘act’), from 2 June 2004, to 1 June 2005, we developed action plans to address the issues identified. After this the incidence of unplanned extubation was again evaluated and results of the study were made known to clinical providers. At the outset of our programme, we established a target of one unplanned extubation per 100 intubated patient days, based on the study by Kuracheck et al. [17]. This incidence was adopted as our internal benchmark against which to compare our future performance.

The following major interventions were implemented: (i) educational courses with literature review for PICU staff focusing on the principles and practices of continuous quality improvement as well as the importance of preventing unplanned extubation, (ii) standardising selected procedures, including tracheal tube fixation, tube suctioning, patient hygiene and transport, (iii) identifying younger patients (< 2 years of age) as high-risk for unplanned extubation and (iv) standardising sedation practices by implementing a protocol to direct the titration of dose and frequency of boluses to mechanically ventilated patients according to the COMFORT scale [16]. The COMFORT scale utilises eight parameters – five behavioural and three physiologic – and scores each with a value of 1–5 to give a total score range between 8 and 40 after a 2-min period of observation. The eight parameters measured are alertness, calmness/agitation, physical movement, muscle tone, facial tension, mean arterial blood pressure, heart rate and respiratory response. A score of 8 represents ‘deeply sedated’ and a score of 40 represents ‘alert and distressed’. The majority of patients received continuous infusions of opioids and benzodiazepines, and fentanyl and midazolam were the most frequently prescribed. Infusion rates for these drugs started at 2 μg.kg−1.h−1 (range: 2–10 μg.kg−1.h−1) for fentanyl and 2 μg.kg−1.min−1 (range: 2–10 μg.kg−1.min−1) for midazolam, and were titrated by the intensivist on the basis of clinical assessment and the COMFORT scale (aiming for a score between 17 and 26). For periods of agitation, boluses equivalent to the hourly drug infusion amounts were used on an ‘as-needed’ basis. Rocuronium was used occasionally – at the discretion of the attending physician, mainly in children who had failed to achieve synchronous ventilation with adequate sedation. They received a continuous intravenous infusion at an initial rate of 10 μg.kg−1.min−1 (range: 5–20 μg.kg−1.min−1).

The primary outcome variable was the incidence of unplanned extubation which was calculated as the number of unplanned extubations per 100 intubated patient days [18]. The demographic and clinical variables collected included age, gender, weight, PICU admission diagnosis and the severity of illness at PICU admission and throughout PICU stay as measured by the Paediatric Risk of Mortality (PRISM) II [19] and the Paediatric Logistic Organ Dysfunction (PELOD) scores [20]. The secondary outcome measures were as follows: duration of mechanical ventilation, duration of PICU stay, duration of hospital stay, the need for re-intubation after unplanned extubation and PICU mortality. All data were transcribed with double entry by the primary investigator to a specifically designed computer database. An unplanned extubation was defined as an unscheduled event in which an airway was observed, directly or radiographically, to be completely dislodged from the trachea, or as a sudden loss of bilateral breathing sounds that could not be re-established by suctioning or advancement of the airway. Airway dislodgement was then confirmed by direct observation or by the absence of exhaled carbon dioxide assessed via capnometry [8, 21–23]. Replacement of the tracheal tube within 24 h was considered as re-intubation [8]. The decision to re-intubate was based entirely on the judgment of the attending physician. Patients were considered to have received sedation if a sedative had been administered within 2 h of the extubation [8, 18]. The level of consciousness of the patient at extubation time was classified as calm or agitated. Agitation was defined as excessive restlessness resulting in increased physical activity [16]. Major complications following an unplanned extubation were defined as occurrences that resulted in increased morbidity such as pneumothorax, pneumomediastinum, haemorrhage during intubation, atelectasis, cardiopulmonary arrest and asphyxia with temporary or permanent brain injury and death [21].

Data are expressed as numerical values and percentages for categorical variables, and as medians and quartiles (25th–75th percentile) for continuous variables. The median difference and 95% confidence interval (CI) was calculated where appropriate. Comparisons were based on chi-square test or chi-square for trends for categorical data, and Mann–Whitney U-test or Kruskal–Wallis tests for continuous data where appropriate. A p value < 0.05 was considered significant. Data were analysed with the statistical programme for social sciences (SPSS, Chicago, IL, USA) version 11.5 software.


Of 964 patients admitted during the five-year period, 283 (29.3%) required invasive mechanical ventilation. After the exclusion of five patients with long-term tracheostomies, 278 orally intubated patients presenting a total of 323 extubation events, were eligible for analysis. These patients were ventilated for a total duration of 3464 days. There were 60 unplanned extubation events occurring in 52 (18.7%) patients. Of these 52 children, six experienced two unplanned extubations each and one patient had three unplanned extubations. Demographic and clinical data on unplanned extubation patients and controls are shown in Table 1. The groups were similar in terms of gender, severity of illness scores and admission diagnosis. Patients experiencing unplanned extubations were significantly younger (7.5 vs 14 months) than patients from the control group. The distribution by age of children experiencing an unplanned extubation is shown in Table 2.

Table 1. Demographic data and admission diagnosis in PICU patients with unplanned extubation and controls. Data are median [interquartile range] or number (%).
Unplanned extubation
(n = 52)
Control patients
(n = 226)
Median difference
(95% CI)
p value
Age; months 7.5 [3–23.5] 14 [4–60] (−0.0, 9.0) 0.0478
Weight; kg 6.2 [3.6–11.6] 9.3 [5.3–18] (1.101, 4.73) 0.0008
Male gender 33 (63.46%) 128 (56.64%) (−0.214, 0.077) NS
PRISM score on admission 12.5 [9–15] 12.5 [9–17] (−1.00, 2.00) NS
PELOD score during PICU stay 6.0 [1–11] 2.0 [1–12] (0.00, 0.00) NS
Medical 48 (92%) 209 (92%) NS
 Respiratory failure 25 (50%) 106 (47%) NS
 Cardiovascular 2 (4%) 10 (4%) NS
 Sepsis 10 (19%) 49 (22%) NS
 CNS 7 (13%) 19 (8%) NS
 Trauma 2 (4%) 22 (10%) NS
 Other 1 (2%) 3 (1%) NS
Surgical 4 (8%) 17 (8%) NS
  • PRISM, paediatric risk of mortality; PELOD, paediatric logistic organ dysfunction; CI, confidence interval; PICU, paediatric intensive care unit; CNS, central nervous system; NS, not significant.
Table 2. Age distribution of children experiencing an unplanned extubation. Data are number (%).
Age; years Unplanned extubation
(n = 52)
(n = 226)
≤1 32 (61.5%) 108 (47.8%)
1–2 8 (15.5%) 35 (15.5%)
2–5 7 (13.4%) 32 (14.2%)
> 5 5 (9.6%) 51 (22.5%)

Figure 1 shows the annual rate of unplanned extubations for the overall population and for children≤2 years and > 2 years. The overall unplanned extubation rate over the study period was 1.7 unplanned extubation per 100 intubated patient days. The first period (baseline) corresponds to the first 2 years of the programme (2002–2004) (phases I and II) during which unplanned extubation rates were stable (2.9 vs 2.6 unplanned extubation/100 intubated patient days). The second period (2004–2005) represents the transition year during which the intervention was developed, adopted and instituted as a requirement of every intubated PICU patient. The third period (2005–2007) represents the two subsequent years (phase IV), with unplanned extubation rates ranging from 0.5–0.6 unplanned extubation/100 intubated patient days. Children younger than 2 years presented greater unplanned extubation rates than children over 2 years (2.5 vs 0.4 unplanned extubation/100 intubated patient days).

Details are in the caption following the image

Unplanned extubations in PICU. The vertical dotted line indicates the point at which the new interventions were started (June 2004). *p = 0.0001, χ2 for trend.

Of the 60 unplanned extubation events, 45 occurred in children younger < 2 years and 47 in children who had received sedation within 2 h of the unplanned extubation. At the time of their unplanned extubation, 34 (57%) were using physical restraints. No difference was observed in unplanned extubation events in relation to the time of day, use of physical restraints and level of consciousness. With regard to the originating activity for occurrence of these events, 28/60 were patient-related (extubation due to patient agitation or deliberate self-extubation) and 32/60 caregiving-related (extubation during nursing care or during manipulation of the patient by other health-care providers). As shown in Table 3, planned extubation patients were frequently extubated during the morning shift (p = 0.03) and less commonly received sedation within 2 h of the event (p = 0.005) compared with unplanned extubation patients.

Table 3. Characteristics of patients and circumstances of unplanned extubation. Data are number (%)
Unplanned extubations
(n = 60)
Planned extubations
(n = 263)
p value
Age; years
 ≤2 45 (75%) 174 (66%) NS
 > 2 15 (25%) 89 (34%) NS
Time of day
 Morning shift 19 (32%) 158 (60%) 0.03
 Evening shift 19 (32%) 92 (35%) NS
 Night shift 22 (36%) 13 (5%) NS
Sedation use
 Yes 47 (78%) 141 (54%) 0.005
 No 13 (22%) 122 (46%) NS
Level of consciousness
 Calm 25 (42%) 144 (55%) NS
 Agitated 35 (58%) 119 (45%) NS
Restraints in use
 Yes 34 (57%) 150 (57%) NS
 No 26 (43%) 113 (43%) NS
Admission diagnosis
 Surgical 5 (8%) 20 (8%) NS
 Medical 55 (92%) 243 (92%) NS
  • NS, not significant.

Fifty-one (85%) of the 60 unplanned extubations occurred during full ventilatory support while 39 (65%) required re-intubation within 24 h and 38 required re-intubation immediately (within 15 min). Re-intubation was not required in 6 out of 9 patients who were weaning at the time of the unplanned extubation. There were no unplanned extubations while patients were receiving paralysing agents.

Table 4 shows the outcome for patients with unplanned extubations compared with controls. There was no difference between the groups in mortality. In contrast, patients with unplanned extubation spent a median of 2.5 times longer on mechanical ventilation (14 vs 6 days) and spent twice as long within the PICU (26.5 vs 12 days) and in hospital (40 vs 23 days). The median time to occurrence of unplanned extubation after the onset of mechanical ventilation was 5 days [2–15] (95% CI: 3–9). Major complications occurred in 10 out of 60 unplanned extubation events: bradycardia (4), tracheal intubation difficulties (3), cardiorespiratory arrest (2) and pulmonary haemorrhage (1). After the CQI programme implementation the incidence of major complications reduced from 9 out of 43 to 1 out of 17. No deaths were ascribed directly to unplanned extubation. There was no difference in the median duration of mechanical ventilation during the study period (to determine whether the use of the sedation protocol led to over-sedation of children).

Table 4. Outcome of PICU patients suffering an unplanned extubation. Data are median [interquartile range] or number (%).
Unplanned extubation (n = 52) Controls (n = 226) Median difference (95% CI) p value
Duration of mechanical ventilation; days 14 [5–30] 6 [3–11] (−10, −3) 0.0001
Duration of PICU stay; days 26.5 [12.7–44] 12 [6–18.2] (−18, −7) 0.0001
Duration of hospital stay; days 40 [22.7–82] 23 [14–37] (−26.9, −10.00) 0.0001
PICU mortality 11 (21.1%) 35 (15.5%) (−0.06, −0.17) NS
  • CI, confidence interval; PICU, paediatric intensive care unit; NS, not significant.


Although we did not design this prospective study such that the effectiveness of specific interventions might be evaluated, we nevertheless clearly demonstrated the overall effectiveness of a CQI programme in reducing the incidence of unplanned extubation. During the programme, it was evident that children younger than 2 years represented a high-risk group for unplanned extubation and benefited the most from the programme.

The majority of studies available in the paediatric population are limited to reporting the incidence and identifying risk factors for unplanned extubation [21, 22, 24–27] with few reports having assessed effectiveness of the strategies implemented to reduce unplanned extubation incidence [8, 28]. Sadowski et al. [8] observed a significant reduction in unplanned extubations through continuous monitoring on a quality improvement programme which included the adoption of protocols for weaning patients from mechanical ventilation and sedation, along with an education programme for PICU staff. Popernack et al. [28] reported a reduced number of unplanned extubation using a sedation algorithm which, according to the authors, requires further validation. Studies conducted over the past 20 years have indicated that unplanned extubation occurs in 0.6–13.3% of intubated paediatric patients, with a rate of 0.11–1.26 events for every 100 ventilation days [27]. No consensus has been reached establishing an acceptable unplanned extubation rate. If the desired goal is an unplanned extubation rate of zero, this may, in turn lead to overuse of sedatives and neuromuscular blockade or excessively early extubation (when patients have not met extubation criteria) and result in worse outcome.

Sadowiski et al. [8] reported lower unplanned extubation rates prior to implementation of a continuous quality improvement programme, a lower proportion of patients receiving full ventilatory support at unplanned extubation time, and a lower re-intubation rate for unplanned extubation patients, than those observed in our study. These differences may be the result of dissimilar patient populations. The entry criteria in their study included tracheotomised patients. Given that tracheotomised children have a longer duration of mechanical ventilation and a safer airway which is less likely to become dislodged [21], it may lead to an underestimation of the unplanned extubation rate. Furthermore, in the Sadowski study population with a lower risk of unplanned extubation than our primarily medical patients was assessed.

An important strength of this study is its methodology of assessing intervention effectiveness. In the first 2 years of the programme, a consistently high annual rate of unplanned extubation was observed along with opportunities for improvement. The stability of the rate of unplanned extubations during this period suggests that there was no ‘Hawthorne effect’ resulting in a change in practice patterns and performance over time resulting in a change in PICU performance prior to the implementation of the interventions. (Awareness of being observed can alter the way in which a person acts or performs – the ‘Hawthorne effect’ [29]). Stability during the baseline period is required to rule out extraneous trends and to document convincingly that intervention is superior to the process responsible for performance [30]. Following implementation of the intervention it was, therefore, possible to determine the effectiveness of the intervention by applying immediacy, magnitude and stability criteria to the results obtained (Figure 1) [30]. Despite the fact that implementation projects can be very disappointing and tend to induce a short-term effect that disappears within a certain period [31], we established a positive and strong effect over a period of time.

Infants and young children are at particular risk for unplanned extubation due to their short tracheal length, the use of uncuffed, oral tracheal tubes and the patients’ developmental immaturity, which may limit co-operation with ICU personnel [27]. We observed that children younger than 2 years old, particularly those under 1 year, were consistently more likely to experience unplanned extubation than older patients. This finding differs to those of previous studies [8, 24] which have identified patients aged younger than 5 years as a high-risk group. Similar to previous reports [23, 32] unplanned extubations were equally distributed across all nursing shifts. Differences in timing of unplanned extubation have been attributed to increased ICU activity during day shifts [9, 14] or a low level of staff vigilance at night [33]. In regard to the use of restraints, our data show that a large number of events (57%) occurred despite restraint use, a finding consistent with previous studies [13, 14, 23, 34, 35] which have reported that this measure did not prevent patients from self-extubating.

Although the study groups were similar in relation to their severity of illness scores (PRISM II and PELOD) and diagnoses on admission, patients experiencing an unplanned extubation had average periods on ventilation and duration of PICU stay and hospital stay twice as long as the control group. Earlier studies in adult ICU patients demonstrated that these effects were due to the outcome in patients who failed to tolerate unplanned extubation [4, 13–15]. Moreover, other investigators have also demonstrated that patients requiring re-intubation after unplanned extubation have a large impact on resource utilization and costs [15, 36]. In our series, we observed that 65% of patients required re-intubation after unplanned extubation, which may in turn translate into more complications and increased costs. Even though our findings are in line with other studies [4, 8, 13–15], question remains as to whether the increased ventilation time in patients with unplanned extubation are a cause or consequence of this event, since a longer period of mechanical ventilation exposes the patient to greater risk of unplanned extubation. In our study, the median time duration of intubation before unplanned extubation was 5 days, similar to previous studies [6, 10, 37, 38]. Although a sedation protocol constitutes one of the components of intervention in reducing unplanned extubation rate, the longer length of mechanical ventilation cannot be linked to sedative overuse since there was no difference in ventilation support time during the study period.

An important finding in our study was the high proportion of patients with unplanned extubation receiving full ventilatory support (85%). This may be associated with a greater overall re-intubation rate (65%) compared with other studies in children (38–52%) [8, 24, 26, 28]. The different re-intubation rates among studies may be due to different patient populations. Re-intubation is more common among medical ICU patients in whom the rate ranges from 40% to 78% [5, 14, 33] than surgical ICU patients. This rate is similar to that observed in our primarily medical population.

The impact of unplanned extubations can be significant, with an abrupt loss of airway control and mechanical ventilatory support resulting in cardiopulmonary complication and even death [37]. In our series, major complications following unplanned extubations were reported in 16% of patients. This finding is similar to those of other reports (5–28%) [14, 37] and underscores the importance of preventing unplanned extubation. We observed that after the quality improvement intervention the number of major complications was reduced by 88%. There were no differences in mortality between patients with or without an unplanned extubation. Reports in adults linking unplanned extubations with increased mortality are inconclusive. Studies which associated mortality with unplanned extubations identified patients who failed to tolerate unplanned extubation as a high-risk group for increased mortality [4, 13, 15].

There are limitations to the present study. Firstly, we did not randomise the programme to create a control group. Quality improvement is typically conducted in settings where random allocation to groups is not feasible for ethical reasons or where the environmental conditions prevent experimental manipulation [30]. Secondly, our study represents the experience in a single PICU, and comparison of our outcomes to other PICUs or studies must be made with caution. Finally, we did not account for potential changes in patient characteristics, severity of illness or staffing issues that may have changed over time and influenced the rate of unplanned extubation. As demonstrated in the first 2 years of the study, it is unlikely that this limitation affected the data, while monitoring of these factors was beyond the scope of our investigation.

In conclusion, our quality improvement intervention over a 5-year period was associated with a significant reduction in the rate of unplanned extubation. Children younger than 2 years old were identified as being at the greatest risk of unplanned extubation. As unplanned extubations are associated with increased morbidity, a longer duration of mechanical ventilation, and a longer stay in the PICU and hospital stay, this intervention should be helpful in enhancing quality of care and patient safety. By employing a concerted CQI programme and identifying risk factors for unplanned extubation, providers should be able to decrease the risk of unplanned extubation for patients in PICUs.


The authors would like to thank the team involved with the development of this programme, including Aline Torturete, RRT, Laerte Fernandes, RRT, Suely Pereira, RN, Patricia Bezerra, RN, and the nursing staff.