OBJECTIVES:

Unplanned extubations (UEs) in adult and pediatric populations are associated with poor clinical outcomes and increased costs. In-hospital outcomes and costs of UE in the NICU are not reported. Our objective was to determine the association of UE with clinical outcomes and costs in very-low-birth-weight infants.

METHODS:

We performed a retrospective matched cohort study in our level 4 NICU from 2014 to 2016. Very-low-birth-weight infants without congenital anomalies admitted by 72 hours of age, who received mechanical ventilation (MV), were included. Cases (+UE) were matched 1:1 with controls (−UE) on the basis of having an equivalent MV duration at the time of UE in the case, gestational age, and Clinical Risk Index for Babies score. We compared MV days after UE in cases or the equivalent date in controls (postmatching MV), in-hospital morbidities, and hospital costs between the matched pairs using raw and adjusted analyses.

RESULTS:

Of 345 infants who met inclusion criteria, 58 had ≥1 UE, and 56 out of 58 (97%) were matched with appropriate controls. Postmatching MV was longer in cases than controls (median: 12.5 days; interquartile range [IQR]: 7 to 25.8 vs median 6 days; IQR: 2 to 12.3; adjusted odds ratio: 4.3; 95% confidence interval: 1.9–9.5). Inflation-adjusted total hospital costs were higher in cases (median difference: $49 587; IQR: −15 063 to 119 826; adjusted odds ratio: 3.8; 95% confidence interval: 1.6–8.9).

CONCLUSIONS:

UEs in preterm infants are associated with worse outcomes and increased hospital costs. Improvements in UE rates in NICUs may improve clinical outcomes and lower hospital costs.

What’s Known on This Subject:

Unplanned extubations (UEs) in adult and pediatric populations are associated with increased mortality, prolonged duration of mechanical ventilation and length of stay, and increased hospital costs. In-hospital outcomes and costs of UEs in preterm infants have not been reported.

What This Study Adds:

In a matched cohort of very-low-birth-weight infants, UEs were associated with ∼1-week longer duration of mechanical ventilation, a 10-day increase in length of stay, and $50 000 increased total hospital costs.

Unplanned extubations (UEs) are the most common adverse event during mechanical ventilation (MV) in the NICU.1  Contemporary UE rates vary up to 25-fold among centers, from 0.54 to 16.1 UE per 100 ventilator days,2,3  and 14% to 41% of infants experience a UE during their NICU hospitalization.2,4  Short-term patient harm often follows these events, including reintubation, oxygen desaturation, and hemodynamic instability leading to cardiopulmonary resuscitation.58  In adult and pediatric populations, UEs are associated with poor in-hospital outcomes, including prolonged MV, increased ICU and hospital length of stay (LOS), and increased hospital costs.9,10  In previous studies, it has been indicated that these poor clinical outcomes are primarily driven by patients who require reintubation after UE.912 

Although the short-term risks of UE are well known, and large national quality improvement efforts are underway to reduce the overall incidence,13,14  little is known about the in-hospital consequences of UE in preterm infants. The objective of this study was to determine the effect of UE on clinical outcomes, including duration of MV, LOS, and hospital costs in preterm infants. In addition, we sought to determine if outcomes of infants who were successfully extubated after UE were better than those without UE.

We performed a retrospective observational matched cohort study15  in the 98-bed, level IV (regional), academic NICU of the Vanderbilt University Medical Center from January 1, 2014, through December 31, 2016. All very-low-birth-weight infants (<1500 g birth weight) admitted to the NICU during the study period within 72 hours of birth, without chromosomal or congenital anomalies, and who received MV through an oral endotracheal tube (ETT) for ≥1 hour were eligible for inclusion in the cohort. During the study period, the NICU had no ventilator weaning protocol or sedation guidelines. Consistent with evidence-based recommendations, ventilated infants did not routinely receive sedatives or opioid medications.16,17  The Vanderbilt University Medical Center institutional review board approved the study with a waiver of consent.

UE was defined a priori as any removal of an ETT that the medical team did not plan in advance. This included ETTs dislodged by the patient, those removed inadvertently during routine nursing and medical care, and those intentionally removed during an acute resuscitation event or period of patient decompensation if there was not objective evidence that the ETT was in the trachea and patent. Our NICU uses a standardized protocol to assess whether an ETT is in the trachea and patent consisting of auscultation for breath sounds, assessment for expiratory waveforms on the ventilator, and use of an end-tidal CO2 detector. Direct laryngoscopy to determine if the ETT is in the trachea was not routinely used in our NICU during the study period.7,18  We used 4 data sources to prospectively collect UEs: (1) root cause analysis forms completed by the bedside team immediately after a UE, (2) daily medical record review of all intubated infants in our NICU, (3) review of all reports of UEs in our hospital’s voluntary incident reporting system, and (4) an ongoing record kept by the NICU respiratory therapists of all UEs in the NICU. We previously reported the sensitivity of each of these sources to identify UE in our NICU, with daily medical records review being the most reliable method (detected 96% of UEs).7  A study team member (L.D.H.) abstracted potential UEs from the 4 data sources, and these were reviewed monthly by a multidisciplinary group to determine if each event met criteria for UE.

We formed 2 matched cohorts, termed our primary and secondary cohorts. We report the detailed outcomes from our primary cohort in the body of this article and secondary cohort results in the Supplemental Information. To ensure that infants who had a UE were compared with similar infants who were “at risk” for UE, our primary cohort was formed by matching cases (+UE) 1:1 with controls (−UE) on 3 variables in the following order: equal duration of MV at the time of UE, gestational age (GA) ± 14 days, and Clinical Risk Index for Babies (CRIB)19  score at birth. If a case had >1 UE, matching was performed on the basis of the day of ventilation of first UE. For example, if an infant had a UE on MV day 10, we first identified all controls with at least 10 days of MV from the entire cohort of eligible controls. We then identified the control (10 or more MV days) infant with the most similar GA and CRIB score to form the matched pair. The day of UE in the case (MV day 10) and the day of matching in the control (MV day 10) are termed the “day of matching.”

After matching our primary cohort, imbalance in some patient characteristics on the day of matching existed, including postmenstrual age (PMA), postnatal age, and use of high-frequency ventilation at matching. Post hoc, we formed a smaller matched cohort using propensity score methods, termed our secondary cohort, to ensure that findings from our primary cohort could not be attributed to imbalance of baseline covariates.

Controls in both the primary and secondary cohort were retrospectively selected from 287 available infants who did not experience UE at any time. More-detailed explanation of the methods to form both cohorts (Supplemental Methods 1), Stata code for generating our cohorts (Supplemental Methods 2), and a figure (Supplemental Fig 2) describing the matching procedure for our primary cohort are available in the Supplemental Information.

To determine the effects of UE, we focused on outcomes that occurred after UE and could potentially be attributable to the event. The primary outcome was the duration of MV after the day of matching, referred to as postmatching MV. Secondary clinical outcomes included total duration of MV, postmatching LOS, total LOS, total days of supplemental oxygen therapy, oxygen use at 36 weeks’ PMA, oxygen use at discharge, in-hospital death, and ventilator days between matching and successful extubation that we defined as ventilator-free for 3 consecutive days.20  Clinical outcomes and daily ventilation status were obtained from an internal NICU database updated daily by 2 trained research nurses. Total (including direct and indirect), direct, and variable direct hospital costs for each infant were obtained from our local hospital cost accounting database (EPSi, Chesterfield, MO).

Descriptive statistics were calculated for demographics, clinical characteristics at matching, and outcomes by using median and interquartile ranges (IQRs) for all continuous variables (with the exception of costs), median and IQR for differences between cases and their matched controls (continuous variables), and percentages for nominal variables. We assessed covariate imbalance before and after matching using standardized differences for all demographic and clinical characteristics.21  All cost outcomes are adjusted for inflation to 2017 US dollars by using the Consumer Price Index for All Urban Consumers.22  To adjust for potential residual confounding after matching, we calculated adjusted odds ratios (aORs) for all outcomes using either multivariable ordinal cumulative probability (for continuous variables)23,24  or logistic regression (for binary variables) models with a robust covariance estimator (Huber-White) to account for nonindependence of the matched pairs. Final models included covariates with an absolute standardized difference >0.125  after matching as sample size allowed.

Because poor clinical outcomes and increased hospital costs in adults and children with UE are primarily driven by patients who require reintubation after UE,9,10  we performed a matched analysis for the primary outcome and hospital costs stratifying the matched pairs in our primary cohort by need for reintubation within 3 days in the case.

Finally, we performed 4 a priori specified sensitivity analyses to test the robustness of our findings by restricting our primary cohort sample to pairs in which (1) the case had only 1 UE, (2) both infants survived to discharge, (3) neither infant had surgery, and (4) both infants were inborn. We used Wilcoxon matched pairs signed rank test (continuous variables) and Exact McNemar test (nominal variables) to assess for statistical differences in these analyses. Matching procedures were completed in Stata/MP version 15.1. All other analyses were performed in R version 3.5.1. All tests were 2 tailed and used a significance level of 0.05.

During the study period, 453 very-low-birth-weight infants were ventilated in the NICU with 345 out of 453 (76%) infants meeting study inclusion criteria (Fig 1). Of the eligible infants, 58 out of 345 (17%) had ≥1 UEs. We matched 56 out of 58 (97%) cases to a suitable control in our primary cohort and 47 out of 58 (81%) cases in our secondary cohort. Compared with the unmatched sample, matching improved our covariate balance (Supplemental Figs 3 and 4), although differences between the matched pairs remained in both our primary (Table 1) and secondary cohorts (Supplemental Table 3) before adjustment. Study infants were generally born extremely preterm (median: GA <27 weeks in both cohorts) and extremely low birth weight (median birth weight: 765 g in both cohorts).

FIGURE 1

Flow diagram of study subjects in our primary cohort. VLBW, very low birth weight.

FIGURE 1

Flow diagram of study subjects in our primary cohort. VLBW, very low birth weight.

Close modal
TABLE 1

Baseline Variables for Matched Pairs at Birth and Matching in the Primary Cohort (n = 56 Pairs)

Cases (+UE)Controls (−UE)Median Difference Within Pairs (IQR)Standardized Difference
Characteristics at Birth     
 GA, median wk (IQR) 25.5 (25 to 27) 26 (24 to 27) 0 (0 to 0) −0.02 
 Birth wt, median g (IQR) 745 (600 to 910) 785 (640 to 905) 0 (−85 to 87.5) 0.05 
 CRIB score, median (IQR) 7 (4 to 8) 7 (3.5 to 8) 0 (−1 to 0) 0.03 
 Male sex, n (%) 27 (48) 31 (55) — 0.14 
 Antenatal steroids, n (%) 47 (86) 47 (86) — 
 Outborn, n (%) 9 (16) 11 (20) — 0.09 
 Maternal chorioamnionitis, n (%) 6 (11) 7 (13) — 0.06 
Characteristics at UE and matching     
 PMA, median wk (IQR) 28 (26– to 30) 28 (26 to 29) 0 (0 to 1) −0.27 
 Postnatal age, median d (IQR) 19.5 (9.5 to 35.5) 18 (5 to 29.5) 0 (−1 to 5.5) −0.27 
 Ventilator-free d before, median d (IQR) 1 (0 to 10.5) 0 (0 to 3) 0 (−1 to 6) −0.41 
 Duration of MV, median d (IQR) 13.5 (4.5 to 20.5) 13.5 (4.5 to 20.5) 0 (0 to 0) 
 High-frequency ventilation, n (%) 20 (36) 13 (23) — −0.27 
 Wt, median g (IQR) 935 (750 to 1180) 945 (790 to 1105) −25 (−195 to 225) −0.19 
 FiO2, median (IQR) 0.40 (0.30 to 0.55) 0.35 (0.28 to 0.5) 0 (−0.15 to 0.19) −0.08 
 Mean airway pressure, median (IQR) 10 (9 to 11) 9.3 (8.7 to 11) 0.05 (−1.6 to 1.9) 0.001 
 Previous surgery, n (%) 11 (20) 5 (9) — −0.31 
 Receiving antibiotics, n (%) 21 (38) 21 (38) — 
 Receiving sedation or opiate drips, n (%) 6 (11) 7 (13) — 0.06 
 Receiving pressors, n (%) 1 (2) 2 (4) — 0.11 
 NPO, n (%) 19 (34) 21 (38) — 0.07 
Cases (+UE)Controls (−UE)Median Difference Within Pairs (IQR)Standardized Difference
Characteristics at Birth     
 GA, median wk (IQR) 25.5 (25 to 27) 26 (24 to 27) 0 (0 to 0) −0.02 
 Birth wt, median g (IQR) 745 (600 to 910) 785 (640 to 905) 0 (−85 to 87.5) 0.05 
 CRIB score, median (IQR) 7 (4 to 8) 7 (3.5 to 8) 0 (−1 to 0) 0.03 
 Male sex, n (%) 27 (48) 31 (55) — 0.14 
 Antenatal steroids, n (%) 47 (86) 47 (86) — 
 Outborn, n (%) 9 (16) 11 (20) — 0.09 
 Maternal chorioamnionitis, n (%) 6 (11) 7 (13) — 0.06 
Characteristics at UE and matching     
 PMA, median wk (IQR) 28 (26– to 30) 28 (26 to 29) 0 (0 to 1) −0.27 
 Postnatal age, median d (IQR) 19.5 (9.5 to 35.5) 18 (5 to 29.5) 0 (−1 to 5.5) −0.27 
 Ventilator-free d before, median d (IQR) 1 (0 to 10.5) 0 (0 to 3) 0 (−1 to 6) −0.41 
 Duration of MV, median d (IQR) 13.5 (4.5 to 20.5) 13.5 (4.5 to 20.5) 0 (0 to 0) 
 High-frequency ventilation, n (%) 20 (36) 13 (23) — −0.27 
 Wt, median g (IQR) 935 (750 to 1180) 945 (790 to 1105) −25 (−195 to 225) −0.19 
 FiO2, median (IQR) 0.40 (0.30 to 0.55) 0.35 (0.28 to 0.5) 0 (−0.15 to 0.19) −0.08 
 Mean airway pressure, median (IQR) 10 (9 to 11) 9.3 (8.7 to 11) 0.05 (−1.6 to 1.9) 0.001 
 Previous surgery, n (%) 11 (20) 5 (9) — −0.31 
 Receiving antibiotics, n (%) 21 (38) 21 (38) — 
 Receiving sedation or opiate drips, n (%) 6 (11) 7 (13) — 0.06 
 Receiving pressors, n (%) 1 (2) 2 (4) — 0.11 
 NPO, n (%) 19 (34) 21 (38) — 0.07 

FiO2, fraction of inspired oxygen; NPO, nil per os; —, not applicable.

During a total of 3158 ventilator days, 84 UE events occurred in our primary cohort, yielding a UE rate of 2.65 out of 100 ventilator days. Thirty-four infants (61%) had 1 UE, 17 (30%) had 2 UEs, 4 (7%) had 3 UEs, and 1 infant had 4 UEs. Most infants were given a trial of noninvasive support after UE with immediate reintubation in only 16 out of 84 events (19%). Ultimately, however, most UEs (64 out of 84, 76%) resulted in reintubation by 24 hours. Nearly all cases (49 out of 56, 88%) were reintubated within 3 days after their primary UE. Most infants who were reintubated within 3 days of their primary UE (32 out of 49, 65%) had increased respiratory support after reintubation as evidenced by a higher mean airway pressure or inspired oxygen concentration in the 24 hours after reintubation compared with those values on the day of UE. Cardiopulmonary resuscitation, either chest compressions (8 out of 84, 10%) or administration of bolus epinephrine (2 out of 84, 2%), occurred after 8 UE events in 7 (13%) infants.

Primary Cohort

Compared with their matched controls, cases had significantly longer postmatching MV (median: 12.5 [IQR: 7 to 26.5] vs 6 [IQR: 2 to 12.5] days; aOR: 4.3 [95% confidence interval (CI): 1.9–9.5]). Cases also had longer total MV (median: 30 [IQR: 17 to 44] vs 22.5 [IQR: 8.5 to 33] days; aOR: 4.5 [95% CI: 2–10]) and longer ventilator time from matching to successful extubation (median: 9 [IQR: 1 to 20] vs 4 [IQR: 1– to 9] days; aOR: 3.1 [95% CI: 1.4–6.7]) than controls. Cases also had significantly higher postmatching LOS (aOR: 3.3 [95% CI: 1.5–7.1]), total days of in-hospital supplemental oxygen use (aOR: 2.4 [95% CI: 1.1–5.1]), and supplemental oxygen use at 36 weeks’ PMA (aOR: 4.6 [95% CI: 1.5–14.1]) (Table 2). There was no difference in mortality between cases and controls after adjusting for patient characteristics (aOR: 0.3 [95% CI: 0.04–2]).

TABLE 2

Clinical Outcomes and Hospital Costs for Matched Pairs in the Primary Cohort (n = 56 Pairs)

OutcomesaCases (+UE)Controls (−UE)Median Difference within Pairs (IQR)aOR (95% CI)b
Primary outcome     
 Postmatching duration of MV, d 12.5 (7 to 26.5) 6 (2 to 12.5) 6 (0 to 17) 4.3 (1.9–9.5) 
Secondary outcomes     
 Total duration of MV, d 30 (17 to 44) 22.5 (8.5 to 33) 6 (0 to 17) 4.5 (2–10) 
 Postmatching LOS, d 85.5 (70 to 107) 75 (56 to 90) 8.5 (−11 to 32) 3.3 (1.5–7.1) 
 Total LOS, d 113.5 (95 to 134) 93 (75 to 112) 14.5 (−7.5 to 40) 3 (1.4–6.6) 
 Days of in-hospital supplemental oxygen use 88 (64 to 117) 68 (39 to 97) 12 (−18 to 58) 2.4 (1.1–5.1) 
 Supplemental oxygen use at 36 wk PMA, n (%) 48 (86) 37 (66) — 4.6 (1.5–14.1) 
 Supplemental oxygen use at discharge, n (%) 33 (59) 35 (63) — 0.8 (0.3–2.1) 
 Postmatching MV days to successful extubationc 9 (1 to 20) 4 (1 to 9) 4 (−1 to 12) 3.1 (1.4–6.7) 
 In-hospital death, n (%) 3 (5) 7 (13) — 0.3 (0.04–2) 
Hospital cost typed     
 Total hospital cost, $ — — 49 587 (−15 063 to 119 826) 3.8 (1.6–8.9) 
 Total direct cost, $ — — 29 996 (−8805 to 75 492) 3.9 (1.7–8.9) 
 Total variable direct cost, $ — — 28 828 (−7741 to 69 176) 3.8 (1.6–8.7) 
OutcomesaCases (+UE)Controls (−UE)Median Difference within Pairs (IQR)aOR (95% CI)b
Primary outcome     
 Postmatching duration of MV, d 12.5 (7 to 26.5) 6 (2 to 12.5) 6 (0 to 17) 4.3 (1.9–9.5) 
Secondary outcomes     
 Total duration of MV, d 30 (17 to 44) 22.5 (8.5 to 33) 6 (0 to 17) 4.5 (2–10) 
 Postmatching LOS, d 85.5 (70 to 107) 75 (56 to 90) 8.5 (−11 to 32) 3.3 (1.5–7.1) 
 Total LOS, d 113.5 (95 to 134) 93 (75 to 112) 14.5 (−7.5 to 40) 3 (1.4–6.6) 
 Days of in-hospital supplemental oxygen use 88 (64 to 117) 68 (39 to 97) 12 (−18 to 58) 2.4 (1.1–5.1) 
 Supplemental oxygen use at 36 wk PMA, n (%) 48 (86) 37 (66) — 4.6 (1.5–14.1) 
 Supplemental oxygen use at discharge, n (%) 33 (59) 35 (63) — 0.8 (0.3–2.1) 
 Postmatching MV days to successful extubationc 9 (1 to 20) 4 (1 to 9) 4 (−1 to 12) 3.1 (1.4–6.7) 
 In-hospital death, n (%) 3 (5) 7 (13) — 0.3 (0.04–2) 
Hospital cost typed     
 Total hospital cost, $ — — 49 587 (−15 063 to 119 826) 3.8 (1.6–8.9) 
 Total direct cost, $ — — 29 996 (−8805 to 75 492) 3.9 (1.7–8.9) 
 Total variable direct cost, $ — — 28 828 (−7741 to 69 176) 3.8 (1.6–8.7) 

—, not applicable.

a

Median (IQR) unless noted otherwise.

b

aORs for continuous variables have been calculated adjusting for PMA at matching, postnatal age at matching, ventilator-free days before matching, high-frequency ventilation use at matching, sex, weight at matching, and surgery of any type before matching. aORs for binary outcomes are only adjusted for the first 4 variables in list.

c

Successful extubation defined as ventilator-free for 3 d; because of deaths, this was available for 47 pairs.

d

All costs adjusted to 2017 US dollars using the Consumer Price Index for All Urban Consumers.

In 36 out of 56 (64%) matched pairs, total hospital costs were higher in the case than in the control. Median costs were $49 587 (total), $29 996 (direct), and $28 828 (variable direct) higher in cases than in controls. All cost differences were significant.

Secondary Cohort

Results of the propensity score–matched cohort were similar to the primary cohort (Supplemental Table 4), although the difference in postmatching MV between cases and controls was larger (median: 14 [IQR: 7 to 29] vs 6 [IQR: 2 to 16] days; aOR: 3 [95% CI: 1.4–6.7]) than in the primary cohort.

Only 7 cases were successfully extubated immediately after their primary UE. There were no significant differences in postmatching MV time (median: 0 days [IQR: 0 to 7] in cases versus 5 days [IQR: 1 to 12] in controls, P = .26) between cases who remained successfully extubated after UE and their matched controls, although total hospital costs (median total cost differential: $55 283, P = .03) were higher in cases. Postmatching MV time was significantly longer for cases that were reintubated than their matched controls (median: 14 days versus 6 days, P < .001) as were total hospital costs (median total cost differential: $34 610, P = .007).

Sensitivity analyses (Supplemental Tables 5 through 8) in subgroups of matched pairs confirmed our main findings. Median postmatching MV differential was significantly longer in cases (3–6 days) in 3 of the 4 analyses. Similarly, the total, direct, and variable direct hospital costs were significantly higher in cases in 3 of the 4 analyses.

We have shown that UEs in preterm infants are associated with significantly poorer in-hospital outcomes and increased financial costs, in addition to their well-known acute harmful effects. In our primary cohort, exposure to ≥1 UEs was associated with a nearly 1-week increase in the duration of postmatching MV, a 10-day increase in LOS, and a nearly $50 000 increase in total hospital costs. These effects remained significant even after adjusting for differences in clinical characteristics between the groups.

Our estimates of the effect size of UE in preterm infants are higher than those reported for adult and PICU patients. In one report, children in a PICU who had a UE had an increased ICU and hospital LOS of 5.5 and 6.5 days, respectively, and increased hospital costs of $36 692 compared with age- and diagnosis-matched controls without a UE.10  In adults, UEs have been associated with increased MV duration (4 days) and increased ICU and hospital LOS (5 and 6 days, respectively).9  Several reasons may explain the increased effect size seen in our cohort. First, although the majority of cases in our study had a trial of noninvasive ventilation after UE, almost all (49 out of 56, 88%) infants were reintubated within 72 hours of their primary UE. This rate of reintubation is higher than in nearly all previous studies of UE and may be partially explained by the small size and physical immaturity of our cohort. Reports consistently show that patients who require reintubation after UE drive many of the adverse clinical and financial outcomes.912  Studies in adults have even shown a survival advantage for patients who do not require reintubation after UE.11,12  Because nearly all of our cases required reintubation within 72 hours, we would expect our clinical outcomes to be worse than other populations. Another potential explanation for the increased effect size seen in our study is that 39% of our cases had >1 UE. If the adverse effects of UE are additive, such that each UE results in an incremental worsening of outcomes, we would expect these infants to have the worst clinical outcomes, as our sensitivity analysis showed. Additionally, the longer duration of MV that often follows UE increases the at-risk time for repeated UE, and thus one UE increases the chances of having a subsequent UE.

In addition to prolonged duration of MV, UEs were associated with worsened respiratory outcomes, including supplemental oxygen use at 36 weeks’ PMA and total days of in-hospital supplemental oxygen use. Previous work by Jensen et al26  showed that cumulative duration of MV and not failed extubation attempts are associated with increased risk of bronchopulmonary dysplasia. Although the observational nature of our study does not allow us to determine causality, we hypothesize that the effect of increasing duration of MV and not the UE event itself may be responsible for the differences in respiratory outcomes we observed. We were unable to determine the exact mechanism for the increased duration of MV seen after UE. However, on the basis of our data, we hypothesize 2 potential reasons. First, the majority of infants (65%) who were reintubated after UE required escalation of their ventilator settings compared with settings immediately before UE, consistent with previous reports.8  This acute worsening of respiratory status after UE may lead to prolongation of MV to stabilize respiratory status. Second, it is possible that clinicians may delay planned extubation of infants who were recently reintubated after a UE, especially those who required cardiopulmonary resuscitation. Our observation that controls achieved successful extubation at a median of 4 vs 9 days postmatching in cases supports this possibility.

Our findings in preterm infants have significant public health implications. In 2017 in the United States, ∼25 000 infants were born at 22 to 28 weeks’ GA and weighed <1500 g.27  Up to 85% of these fragile infants are supported with MV during their NICU hospitalization.28  Assuming conservatively that 10% of these infants have a UE in the NICU, far less than the 17% we observed, our findings suggest that UE could result in an annual increase of ∼14 000 ventilator days, 22 000 hospital days, and >60 million dollars in direct hospital costs in this population of infants alone.

Our study has several limitations. First, even after matching and adjustment for covariate imbalance, unmeasured differences may exist between cases and controls and could affect outcomes. Some patient factors remained different between the cases and controls even after matching. We would expect some of these factors, including older PMA and less-severe respiratory disease (using ventilator-free days as a surrogate) at the time of matching, to bias our results toward the null hypothesis that UE had no effect on outcomes. Other factors, including more high frequency ventilation at matching in cases, could explain some of our findings. To address this limitation, we included all factors with imbalance in our adjusted models. Second, 3 of the 4 data streams we used to identify UEs depended on self-report by clinicians, introducing the possibility of misclassification of exposure. To avoid this, we reviewed medical records of all intubated infants to ensure complete capture of UEs. We would expect misclassification of exposure, meaning uncaptured UE in controls, to bias our study results toward the null hypothesis that UEs have no effect on outcomes. Finally, our study was conducted in a single center, and although our findings are likely generalizable to critically ill preterm infants, they may not be generalizable to all infants cared for in an NICU, especially in populations who may have higher rates of successful extubation after UE.

UEs in preterm infants are associated with poorer in-hospital outcomes and increased costs. Evidence-based strategies, such as standardization of ETT fixation,3,18  2- or 3-person ETT retaping and patient transfers,3  and prospective identification of infants at high risk of UE,29,30  have been reported to decrease UE rates and can be implemented by NICUs. Modest investments by hospital systems in quality improvement programs to decrease UEs in the NICU may yield significant cost savings while improving overall quality of care.

Dr Hatch conceptualized and designed the study, coordinated and supervised data collection, performed the matching procedure, and drafted the initial manuscript; Mrs Scott performed data collection, cleaning, and transformation and assisted in designing the statistical analyses; Dr Slaughter and Ms Xu assisted in study design and designed and completed all statistical analyses; Dr Smith assisted in study design and performed data collection for financial analyses; Drs Stark, Patrick, and Ely assisted in study design; and all authors reviewed and revised the final manuscript, approved the final manuscript as submitted, and agree to be accountable for all aspects of the work.

FUNDING: Dr Hatch was supported by the Vanderbilt Department of Pediatrics Katherine Dodd Faculty Scholars Award. Use of the Research Electronic Data Capture program was supported by UL1 TR000445 from the National Center for Advancing Translational Sciences and National Institutes of Health. Funded by the National Institutes of Health (NIH).

     
  • aOR

    adjusted odds ratio

  •  
  • CI

    confidence interval

  •  
  • CRIB

    Clinical Risk Index for Babies

  •  
  • ETT

    endotracheal tubem

  •  
  • GA

    gestational age

  •  
  • IQR

    interquartile range

  •  
  • LOS

    length of stay

  •  
  • MV

    mechanical ventilation

  •  
  • PMA

    postmenstrual age

  •  
  • UE

    unplanned extubation

1
Sharek
PJ
,
Horbar
JD
,
Mason
W
, et al
.
Adverse events in the neonatal intensive care unit: development, testing, and findings of an NICU-focused trigger tool to identify harm in North American NICUs
.
Pediatrics
.
2006
;
118
(
4
):
1332
1340
2
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
3
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
4
Ferraz
P
,
Barros
M
,
Miyoshi
M
,
Davidson
J
,
Guinsburg
R
.
Bundle to reduce unplanned extubation in a neonatal intensive care unit [published online ahead of print January 28, 2019]
.
J Matern Fetal Neonatal Med
.
doi:10.1080/14767058.2019.1568981
5
Hatch
LD
,
Grubb
PH
,
Lea
AS
, et al
.
Endotracheal intubation in neonates: a prospective study of adverse safety events in 162 infants
.
J Pediatr
.
2016
;
168
:
62
66.e6
6
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
7
Hatch
LD
,
Grubb
PH
,
Markham
MH
, et al
.
Effect of anatomical and developmental factors on the risk of unplanned extubation in critically ill newborns
.
Am J Perinatol
.
2017
;
34
(
12
):
1234
1240
8
Kambestad
KK
,
Huack
A
,
Nair
S
, et al
.
The adverse impact of unplanned extubation in a cohort of critically ill neonates
.
Respir Care
.
2019
;
64
(
12
):
1500
1507
9
Epstein
SK
,
Nevins
ML
,
Chung
J
.
Effect of unplanned extubation on outcome of mechanical ventilation
.
Am J Respir Crit Care Med
.
2000
;
161
(
6
):
1912
1916
10
Roddy
DJ
,
Spaeder
MC
,
Pastor
W
,
Stockwell
DC
,
Klugman
D
.
Unplanned extubations in children: impact on hospital cost and length of stay
.
Pediatr Crit Care Med
.
2015
;
16
(
6
):
572
575
11
Krinsley
JS
,
Barone
JE
.
The drive to survive: unplanned extubation in the ICU
.
Chest
.
2005
;
128
(
2
):
560
566
12
de Groot
RI
,
Dekkers
OM
,
Herold
IH
,
de Jonge
E
,
Arbous
MS
.
Risk factors and outcomes after unplanned extubations on the ICU: a case-control study
.
Crit Care
.
2011
;
15
(
1
):
R19
13
Solutions for Patient Safety
.
Our results
. Available at: www.solutionsforpatientsafety.org/our-results/. Accessed June 11, 2018
14
Meyers
JM
,
Pinheiro
J
,
Nelson
MU
.
Unplanned extubation in NICU patients: are we speaking the same language?
J Perinatol
.
2015
;
35
(
9
):
676
677
15
Greenland
S
,
Morgenstern
H
.
Matching and efficiency in cohort studies
.
Am J Epidemiol
.
1990
;
131
(
1
):
151
159
16
Ng
E
,
Taddio
A
,
Ohlsson
A
.
Intravenous midazolam infusion for sedation of infants in the neonatal intensive care unit
.
Cochrane Database Syst Rev
.
2017
;
1
:
CD002052
17
Bellù
R
,
de Waal
K
,
Zanini
R
.
Opioids for neonates receiving mechanical ventilation: a systematic review and meta-analysis
.
Arch Dis Child Fetal Neonatal Ed
.
2010
;
95
(
4
):
F241
F251
18
Hatch
LD
,
Rivard
M
,
Bolton
J
, et al
.
Implementing strategies to identify and mitigate adverse safety events: a case study with unplanned extubations
.
Jt Comm J Qual Patient Saf
.
2019
;
45
(
4
):
295
303
19
The CRIB (clinical risk index for babies) score: a tool for assessing initial neonatal risk and comparing performance of neonatal intensive care units. The International Neonatal Network [published correction appears in Lancet 1993;342(8871):626]
.
Lancet
.
1993
;
342
(
8865
):
193
198
20
Giaccone
A
,
Jensen
E
,
Davis
P
,
Schmidt
B
.
Definitions of extubation success in very premature infants: a systematic review
.
Arch Dis Child Fetal Neonatal Ed
.
2014
;
99
(
2
):
F124
F127
21
Austin
PC
.
Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples
.
Stat Med
.
2009
;
28
(
25
):
3083
3107
22
US Bureau of Labor Statistics
.
2017
Consumer price index
. Available at: https://www.bls.gov/cpi/. Accessed September 4, 2018
23
Liu
Q
,
Shepherd
BE
,
Li
C
,
Harrell
FE
 Jr.
.
Modeling continuous response variables using ordinal regression
.
Stat Med
.
2017
;
36
(
27
):
4316
4335
24
Harrell
FE
.
Regression Modeling Strategies with Applications to Linear Models, Logistic and Ordinal Regression, and Survival Analysis
.
Cham, Switzerland
:
Springer-Verlag
;
2015
25
Normand
ST
,
Landrum
MB
,
Guadagnoli
E
, et al
.
Validating recommendations for coronary angiography following acute myocardial infarction in the elderly: a matched analysis using propensity scores
.
J Clin Epidemiol
.
2001
;
54
(
4
):
387
398
26
Jensen
EA
,
DeMauro
SB
,
Kornhauser
M
,
Aghai
ZH
,
Greenspan
JS
,
Dysart
KC
.
Effects of multiple ventilation courses and duration of mechanical ventilation on respiratory outcomes in extremely low-birth-weight infants
.
JAMA Pediatr
.
2015
;
169
(
11
):
1011
1017
27
Centers for Disease Control and Prevention
.
Natality public-use data, 2007–2018. 2018
. Available at: http://wonder.cdc.gov/natality-current.html. Accessed Jun 24, 2019
28
Stoll
BJ
,
Hansen
NI
,
Bell
EF
, et al;
Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network
.
Trends in care practices, morbidity, and mortality of extremely preterm neonates, 1993-2012
.
JAMA
.
2015
;
314
(
10
):
1039
1051
29
Hatch
LD
,
Scott
TA
,
Rivard
M
, et al
.
Building the driver diagram: a mixed-methods approach to identify causes of unplanned extubations in a large neonatal ICU
.
Jt Comm J Qual Patient Saf
.
2019
;
45
(
1
):
40
46
30
Galiote
JP
,
Ridoré
M
,
Carman
J
, et al
.
Reduction in unintended extubations in a level IV neonatal intensive care unit
.
Pediatrics
.
2019
;
143
(
5
):
e20180897

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