Video Abstract
In 2011, the Neonatal Resuscitation Program (NRP) added consideration of continuous positive airway pressure (CPAP) for spontaneously breathing infants with labored breathing or hypoxia in the delivery room (DR). The objective of this study was to determine if DR-CPAP is associated with symptomatic pneumothorax in infants 35 to 42 weeks’ gestational age.
We included (1) a retrospective birth cohort study of neonates born between 2001 and 2015 and (2) a nested cohort of those born between 2005 and 2015 who had a resuscitation call leading to admission to the NICU and did not receive positive-pressure ventilation.
In the birth cohort (n = 200 381), pneumothorax increased after implementation of the 2011 NRP from 0.4% to 0.6% (P < .05). In the nested cohort (n = 6913), DR-CPAP increased linearly over time (r = 0.71; P = .01). Administration of DR-CPAP was associated with pneumothorax (odds ratio [OR]: 5.5; 95% confidence interval [CI]: 4.4–6.8); the OR was higher (P < .001) in infants receiving 21% oxygen (OR: 8.5; 95% CI: 5.9–12.3; P < .001) than in those receiving oxygen supplementation (OR: 3.5; 95% CI: 2.5–5.0; P < .001). Among those with DR-CPAP, pneumothorax increased with gestational age and decreased with oxygen administration.
The use of DR-CPAP is associated with increased odds of pneumothorax in late-preterm and term infants, especially in those who do not receive oxygen in the DR. These findings could be used to clarify NRP guidelines regarding DR-CPAP in late-preterm and term infants.
Continuous positive airway pressure (CPAP) is a risk factor for pneumothorax. New Neonatal Resuscitation Program guidelines included CPAP as a possible corrective measure for spontaneously breathing infants who have labored breathing or persistent cyanosis.
Among late-preterm and term infants, pneumothorax increased after implementation of the new Neonatal Resuscitation Program guidelines. Delivery room CPAP was associated with pneumothorax. The odds ratio was higher in infants receiving 21% oxygen than in those receiving oxygen supplementation.
The use of delivery room (DR) continuous positive airway pressure (CPAP) has increased in preterm infants in recent years,1 likely because of lower risk of the combined outcome of death or bronchopulmonary dysplasia, as compared with DR endotracheal intubation.2,3 DR intubation decreased in preterm infants at Parkland Health and Hospital System (PHHS) as well,4 coinciding with participation (2005–2009) in the Surfactant, Positive Pressure, and Oxygenation Randomized Trial (SUPPORT), in which researchers compared DR intubation to DR-CPAP.5 New guidelines were published in 2010 and implemented by the Neonatal Resuscitation Program (NRP) in 2011, which included CPAP as a possible corrective measure for spontaneously breathing infants without apnea, gasping, or heart rate <100 beats per minute who have labored breathing or persistent cyanosis, regardless of gestational age (GA).6,7
CPAP improves oxygenation in preterm infants with respiratory distress syndrome (RDS).8 CPAP may also improve functional residual capacity, respiratory work, and gas exchange.9 The effect of positive pressure on lung volume increases with lung elasticity, which is directly related to GA, and decreases with chest wall elasticity, which is inversely related to GA.7,10 The amount of pressure required to cause lung rupture decreases with increasing GA because of increased distensibility and decreased surface tension.11 The association between CPAP and air leak syndrome has been a concern in neonatal literature.12 In the Continuous Positive Airway Pressure or Intubation at Birth (COIN) trial, increased air leak in DR-CPAP versus DR intubation may have resulted from a higher level of CPAP than in similar studies.4,13 CPAP is an appropriate DR therapy for most hypoxia or labored breathing in preterm infants that is caused by RDS or retained lung fluid (both associated with decreased lung compliance). However, in term infants, RDS prevalence is low. Grunting, nasal flaring, or retractions may result from diagnoses for which DR-CPAP administration is not a proven therapy, such as fetal acidemia, perinatal asphyxia, pulmonary hypertension, pneumothorax, congenital diaphragmatic hernia, polycythemia, fever, sepsis, or hypoglycemia.14,15 Therefore, we hypothesized that in late-preterm and term infants, the use of DR-CPAP would be associated with increased frequency of symptomatic pneumothorax.
Methods
Design
This was a retrospective birth cohort study with a nested cohort.
Patients
Entry criteria for the birth cohort included infants 35 to 42 weeks’ GA born at PHHS between 2001 and 2015. To further assess the effects of DR-CPAP, we selected a nested cohort of at-risk infants (1) treated by a resuscitation team (which could provide DR-CPAP), (2) with extensive available information (available in those born between 2005 and 2015 who were admitted to the NICU), and (3) excluding those who received DR positive-pressure ventilation (PPV).
Setting
At PHHS, deliveries for infants 35 to 42 weeks’ GA are typically attended by a provider from the newborn nursery who may use PPV without positive end-expiratory pressure, using a self-inflating bag. The PHHS NICU resuscitation team is called for infants 35 to 42 weeks’ GA at high risk for needing resuscitation (eg, malformations), fetal compromise (ie, nonreassuring fetal heart tones), or poor respiratory effort not resolving with short-term PPV with a self-inflating bag. Only the NICU resuscitation team can provide CPAP or positive end-expiratory pressure to newborns in the DR. Facemask CPAP is provided via NeoPuff T-piece resuscitator. CPAP is typically started at 5 cm H2O and further escalated to a maximum of 8, if needed, in the DR or later in the NICU, on the basis of chest radiograph findings and oxygen requirement. PHHS adopted the NRP guidelines including DR-CPAP in spontaneously breathing infants with respiratory distress or hypoxia in July 2011. Criteria for NICU admission are listed online (Supplemental Information).
Data Sources
Data were obtained from the Parkland neonatal resuscitation and NICU databases. The neonatal resuscitation database is a compilation of data since 2005 pertaining to DR resuscitation of newborns who are admitted to the NICU as well as data at the time of admission to the NICU. The Parkland NICU database provides data during their NICU stay.
Outcome Variables
The primary outcome of the birth cohort was the change in percentage of pneumothorax after implementation of the new NRP guidelines.
The primary outcome of the nested cohort was the odds ratio (OR) of pneumothorax of infants who received DR-CPAP versus those who did not receive DR-CPAP. Secondary outcomes included the use of DR-CPAP and the comparison of OR of pneumothorax with CPAP in infants receiving 21% oxygen in the DR versus OR in those receiving oxygen supplementation.
Approval
This study was approved by the University of Texas Southwestern Medical Center Institutional Review Board and by PHHS.
Statistical Analysis
Continuous variables are presented as mean (SD) or median (interquartile range). Changes over time (with participation in SUPPORT and after implementation of the new NRP guidelines) were analyzed either by statistical process control (P control chart) by using QI Macros for Excel (KnoWare International, Denver, CO) if values were stable or by time series analysis to analyze significance of trends. Other statistical analyses were conducted by using SPSS version 23 (IBM, Inc, Armonk, NY) or SAS version 14.2 (SAS Institute, Inc, Cary, NC) for 2-tailed tests, with P < .05 considered significant. Continuous variables were analyzed by using Student t test, Mann-Whitney test, or time series analysis. Dichotomous outcomes were analyzed by χ2 analysis (exact test followed by pairwise comparisons with Bonferroni correction, as appropriate) and Cochran-Mantel-Haenszel test.
Because this was a retrospective observational study, the association between DR-CPAP and pneumothorax could have resulted from bias because of unequal distribution of prognostic factors between patients exposed (treated) or not exposed (controls) to DR-CPAP. Four approaches were used to adjust for bias.16
First, stepwise logistic regression analysis was conducted to adjust for confounders.
Second, the analysis was adjusted for DR oxygen administration by testing for an additive interaction between DR-CPAP and DR oxygen.
Third, propensity score analysis was conducted to reduce imbalance between treated neonates and controls on many covariates merged into a single variable, the propensity score. As recommended by Austin, the propensity score for pneumothorax included all variables available in the DR (excluding CPAP) and soon after NICU admission that reached significance on forward stepwise regression as well as the change in practice recommended by the NRP.17 Each treated neonate was matched by using a SAS macro18 with 2 controls whose propensity scores differed from the treated infant by <0.01 (ie, caliper matching).19 Imbalance of covariates was assessed by the standardized difference, using a SAS macro.17,20
Fourth, instrumental variable analysis was used to control for unmeasured confounding. A valid instrumental variable is a variable that (1) is independent of the unmeasured confounding, (2) affects the treatment, and (3) affects the outcome only indirectly through its effect on the treatment.21 Epoch (defined by calendar year and change in guidelines) was selected as an instrumental variable because of changes in DR-CPAP use over time. Imbalance of covariates was assessed by the multivariate imbalance coefficient (ranging from 0 to 1, representing perfect balance and imbalance, respectively), using the GI SAS macro.21
To detect a 25% increase in frequency of pneumothorax in the birth cohort from a baseline of 0.4% with a statistical power of 0.9, a total n of 9030 patients in each group was needed, using χ2 analysis. Sample size analysis for the logistic regression in the nested cohort was sufficient to have at least 10 subjects with pneumothorax per variable included in the logistic regression analysis.
Results
Birth Cohort
The birth cohort included 200 381 neonates born 2001–2015 (Fig 1). The percentage of symptomatic pneumothorax, initially stable, increased progressively after implementing the new guidelines (Fig 2). The average percentage of pneumothorax increased from 0.38% before implementation (561 out of 148 591) to 0.56% (292 out of 51 790) after implementation of the new NRP guidelines (P < .05).
Study participant flow diagram showing the birth cohort (infants 35–42 weeks’ GA born in 2001–2015) and the nested cohort (infants 35–42 weeks’ GA born in 2005–2015 who had a DR resuscitation team call, did not receive PPV in the DR, and were admitted to the NICU). O2, oxygen.
Study participant flow diagram showing the birth cohort (infants 35–42 weeks’ GA born in 2001–2015) and the nested cohort (infants 35–42 weeks’ GA born in 2005–2015 who had a DR resuscitation team call, did not receive PPV in the DR, and were admitted to the NICU). O2, oxygen.
Evolution of the percentage of neonates with pneumothorax in the birth cohort. The graph shows year on the x-axis and the percentage of neonates 35 to 42 weeks’ GA with pneumothorax on the y-axis. CL, center line; LCL, lower control limit; UCL, upper control limit.
Evolution of the percentage of neonates with pneumothorax in the birth cohort. The graph shows year on the x-axis and the percentage of neonates 35 to 42 weeks’ GA with pneumothorax on the y-axis. CL, center line; LCL, lower control limit; UCL, upper control limit.
Nested Cohort
Among 9304 neonates born at 35 to 42 weeks’ GA in 2005–2015 who had a DR resuscitation team call and were admitted to the NICU, 6913 (nested cohort) did not receive PPV (Fig 1).
In the nested cohort, infants who received DR-CPAP had significantly higher proportions of placental abruption, meconium-stained amniotic fluid, cesarean delivery, and surfactant administration compared with those who did not receive DR-CPAP. Among infants receiving DR-CPAP, there was a lower proportion of girls and Hispanics and lower Apgar scores (Table 1).
Perinatal Characteristics of Infants 35 to 42 Weeks’ GA in the Nested Cohort
| DR-CPAP | No DR-CPAP | P | |
|---|---|---|---|
| n = 1001 | n = 5912 | ||
| Preeclampsia, n (%) | 159 (16) | 815 (14) | .09 |
| Placental abruption, n (%) | 12 (1.2) | 28 (0.5) | .01 |
| Meconium-stained fluid, n (%) | 285 (29) | 1053 (18) | <.001 |
| Cesarean delivery, n (%) | 657 (66) | 3124 (53) | <.001 |
| Female, n (%) | 416 (42) | 2762 (47) | .003 |
| Race and ethnicity, n (%) | .04 | ||
| Hispanic | 725 (72)a | 4526 (77)a | |
| African-American non-Hispanic | 162 (16) | 845 (14) | |
| White non-Hispanic | 78 (8) | 377 (6) | |
| Other | 36 (4) | 164 (3) | |
| GA, mean ± SD, wk | 38 ± 2 | 38 ± 2 | .89 |
| Birth wt, mean ± SD, g | 3159 ± 669 | 3123 ± 729 | .13 |
| Intrauterine growth restriction, n (%) | 21 (2.1) | 264 (4.5) | <.001 |
| Apgar 1 min, median (IQR) | 8 (7–8) | 8 (8–8) | <.001 |
| Apgar 5 min, median (IQR) | 8 (8–9) | 9 (9–9) | <.001 |
| Surfactant administration, n (%) | 18 (1.8) | 19 (0.3) | <.001 |
| DR-CPAP | No DR-CPAP | P | |
|---|---|---|---|
| n = 1001 | n = 5912 | ||
| Preeclampsia, n (%) | 159 (16) | 815 (14) | .09 |
| Placental abruption, n (%) | 12 (1.2) | 28 (0.5) | .01 |
| Meconium-stained fluid, n (%) | 285 (29) | 1053 (18) | <.001 |
| Cesarean delivery, n (%) | 657 (66) | 3124 (53) | <.001 |
| Female, n (%) | 416 (42) | 2762 (47) | .003 |
| Race and ethnicity, n (%) | .04 | ||
| Hispanic | 725 (72)a | 4526 (77)a | |
| African-American non-Hispanic | 162 (16) | 845 (14) | |
| White non-Hispanic | 78 (8) | 377 (6) | |
| Other | 36 (4) | 164 (3) | |
| GA, mean ± SD, wk | 38 ± 2 | 38 ± 2 | .89 |
| Birth wt, mean ± SD, g | 3159 ± 669 | 3123 ± 729 | .13 |
| Intrauterine growth restriction, n (%) | 21 (2.1) | 264 (4.5) | <.001 |
| Apgar 1 min, median (IQR) | 8 (7–8) | 8 (8–8) | <.001 |
| Apgar 5 min, median (IQR) | 8 (8–9) | 9 (9–9) | <.001 |
| Surfactant administration, n (%) | 18 (1.8) | 19 (0.3) | <.001 |
Student t test, Mann-Whitney test, or χ2 analysis (exact test followed by pairwise comparisons with Bonferroni correction, as appropriate. IQR, interquartile range.
a Indicates pairwise comparisons that reached significance.
DR-CPAP use increased linearly over time (r = 0.71, P = .01) (Fig 3). DR-CPAP in neonates requiring oxygen supplementation in the DR increased during the SUPPORT trial, then transiently decreased and did not increase after implementing the new NRP guidelines beyond the level during SUPPORT (Fig 4). In contrast, DR-CPAP in neonates receiving 21% oxygen increased in 2011 after implementing the new NRP guidelines (Fig 4).
Evolution of the use of CPAP among neonates in the nested cohort. The graph shows year on the x-axis and the percentage of neonates who received CPAP on the y-axis.
Evolution of the use of CPAP among neonates in the nested cohort. The graph shows year on the x-axis and the percentage of neonates who received CPAP on the y-axis.
Control charts of use of CPAP in the DR versus oxygen administration in the DR in the nested cohort. The graphs show year on the x-axis and the percentage of neonates who received CPAP on the y-axis. A, Oxygen administration in the DR. B, Room air in the DR. CL, center line; LCL, lower control limit; UCL, upper control limit.
Control charts of use of CPAP in the DR versus oxygen administration in the DR in the nested cohort. The graphs show year on the x-axis and the percentage of neonates who received CPAP on the y-axis. A, Oxygen administration in the DR. B, Room air in the DR. CL, center line; LCL, lower control limit; UCL, upper control limit.
Association of CPAP With Pneumothorax
Among 6913 neonates, the frequency of pneumothorax was 3.7% among those who did not receive DR-CPAP and 16.9% among those who received DR-CPAP (number needed to harm: 8) (Table 2). Administration of DR-CPAP was significantly associated with pneumothorax (OR: 4.6; 95% confidence interval [CI]: 3.6–6.0) in bivariate analysis. This was confirmed in multivariate analyses (Supplemental Tables 3–6). Both instrumental variable and propensity score analyses substantially reduced imbalance of covariates between treated patients and controls (Supplemental Tables 4 and 5).
Pneumothorax by Epoch Versus Continuous Positive Pressure and Oxygen Administration in the DR in the Nested Cohort
| Epoch | Room Air | Oxygen | Total | |||
|---|---|---|---|---|---|---|
| No DR-CPAP | DR-CPAP | No DR-CPAP | DR-CPAP | No DR-CPAP | DR-CPAP | |
| No. patients with pneumothorax out of total no. patients (%) | ||||||
| 2005 | 12 out of 251 (5) | 0 out of 1 (0) | 9 out of 238 (4) | 6 out of 52 (12) | 21 out of 489 (4) | 6 out of 53 (11) |
| 2006 | 5 out of 280 (2) | — | 10 out of 260 (4) | 8 out of 60 (13) | 15 out of 540 (3) | 8 out of 60 (13) |
| 2007 | 29 out of 495 (6) | 0 out of 1 (0) | 9 out of 50 (18) | 8 out of 49 (16) | 38 out of 545 (7) | 8 out of 50 (16) |
| 2008 | 17 out of 478 (4) | 0 out of 2 (0) | 3 out of 45 (7) | 12 out of 72 (17) | 20 out of 523 (4) | 12 out of 74 (16) |
| 2009 | 24 out of 504 (5) | 0 out of 1 (0) | 3 out of 28 (11) | 5 out of 48 (10) | 27 out of 532 (5) | 5 out of 49 (10) |
| 2010 | 12 out of 423 (3) | 2 out of 16 (13) | 2 out of 67 (3) | 13 out of 90 (14) | 14 out of 490 (3) | 15 out of 107 (14) |
| 2011aa | 3 out of 186 (2) | 1 out of 15 (7) | 0 out of 30 (0) | 2 out of 20 (10) | 3 out of 216 (1) | 3 out of 35 (9) |
| 2011ba | 12 out of 258 (5) | 3 out of 18 (17) | 3 out of 50 (6) | 10 out of 36 (28) | 15 out of 308 (5) | 13 out of 54 (24) |
| 2012 | 15 out of 538 (3) | 8 out of 39 (21) | 4 out of 81 (5) | 2 out of 66 (3) | 19 out of 619 (3) | 10 out of 105 (10) |
| 2013 | 17 out of 482 (4) | 11 out of 45 (24) | 4 out of 82 (5) | 22 out of 110 (20) | 21 out of 564 (4) | 33 out of 155 (21) |
| 2014 | 10 out of 432 (2) | 9 out of 34 (27) | 3 out of 86 (4) | 13 out of 97 (13) | 13 out of 518 (3) | 22 out of 131 (17) |
| 2015 | 9 out of 510 (2) | 19 out of 59 (32) | 3 out of 58 (5) | 16 out of 69 (23) | 12 out of 568 (2) | 35 out of 128 (27) |
| 2005–2015 | 165 out of 4837 (3) | 53 out of 231 (23) | 53 out of 1075 (5) | 117 out of 770 (15) | 218 out of 5912 (4) | 170 out of 1001 (17) |
| No. patients needing thoracocentesis or thoracostomy out of total no. with pneumothorax (%) | ||||||
| 2005–2015 | 8 out of 165 (5) | 7 out of 53 (13) | 7 out of 53 (13) | 9 out of 117 (8) | 15 out of 218 (7) | 16 out of 170 (9) |
| Epoch | Room Air | Oxygen | Total | |||
|---|---|---|---|---|---|---|
| No DR-CPAP | DR-CPAP | No DR-CPAP | DR-CPAP | No DR-CPAP | DR-CPAP | |
| No. patients with pneumothorax out of total no. patients (%) | ||||||
| 2005 | 12 out of 251 (5) | 0 out of 1 (0) | 9 out of 238 (4) | 6 out of 52 (12) | 21 out of 489 (4) | 6 out of 53 (11) |
| 2006 | 5 out of 280 (2) | — | 10 out of 260 (4) | 8 out of 60 (13) | 15 out of 540 (3) | 8 out of 60 (13) |
| 2007 | 29 out of 495 (6) | 0 out of 1 (0) | 9 out of 50 (18) | 8 out of 49 (16) | 38 out of 545 (7) | 8 out of 50 (16) |
| 2008 | 17 out of 478 (4) | 0 out of 2 (0) | 3 out of 45 (7) | 12 out of 72 (17) | 20 out of 523 (4) | 12 out of 74 (16) |
| 2009 | 24 out of 504 (5) | 0 out of 1 (0) | 3 out of 28 (11) | 5 out of 48 (10) | 27 out of 532 (5) | 5 out of 49 (10) |
| 2010 | 12 out of 423 (3) | 2 out of 16 (13) | 2 out of 67 (3) | 13 out of 90 (14) | 14 out of 490 (3) | 15 out of 107 (14) |
| 2011aa | 3 out of 186 (2) | 1 out of 15 (7) | 0 out of 30 (0) | 2 out of 20 (10) | 3 out of 216 (1) | 3 out of 35 (9) |
| 2011ba | 12 out of 258 (5) | 3 out of 18 (17) | 3 out of 50 (6) | 10 out of 36 (28) | 15 out of 308 (5) | 13 out of 54 (24) |
| 2012 | 15 out of 538 (3) | 8 out of 39 (21) | 4 out of 81 (5) | 2 out of 66 (3) | 19 out of 619 (3) | 10 out of 105 (10) |
| 2013 | 17 out of 482 (4) | 11 out of 45 (24) | 4 out of 82 (5) | 22 out of 110 (20) | 21 out of 564 (4) | 33 out of 155 (21) |
| 2014 | 10 out of 432 (2) | 9 out of 34 (27) | 3 out of 86 (4) | 13 out of 97 (13) | 13 out of 518 (3) | 22 out of 131 (17) |
| 2015 | 9 out of 510 (2) | 19 out of 59 (32) | 3 out of 58 (5) | 16 out of 69 (23) | 12 out of 568 (2) | 35 out of 128 (27) |
| 2005–2015 | 165 out of 4837 (3) | 53 out of 231 (23) | 53 out of 1075 (5) | 117 out of 770 (15) | 218 out of 5912 (4) | 170 out of 1001 (17) |
| No. patients needing thoracocentesis or thoracostomy out of total no. with pneumothorax (%) | ||||||
| 2005–2015 | 8 out of 165 (5) | 7 out of 53 (13) | 7 out of 53 (13) | 9 out of 117 (8) | 15 out of 218 (7) | 16 out of 170 (9) |
—, not applicable.
2011a and 2011b are before and after implementation of the new NRP guidelines, respectively.
The OR of pneumothorax associated with CPAP was higher (P < .001) in infants receiving 21% oxygen (OR: 8.5; CI: 5.9–12.3; P < .001) than in those receiving oxygen supplementation (OR: 3.5; CI: 2.5–5.0; P < .001) in bivariate analysis (Fig 1). Similar results were obtained in multivariate analyses (Supplemental Tables 5, 7, and 8). The frequency of pneumothorax was not significantly associated with the level of CPAP (P = .68).
Other Outcomes
Pulmonary interstitial emphysema occurred in 8 out of 6913 patients: 3 out of 1001 on CPAP versus 5 out of 5912 without CPAP (P = .01).
Among 388 patients with pneumothorax, 7.9% received either thoracocentesis or thoracostomy (Table 2). The percentage of treatment was similar in those with versus those without DR-CPAP. The frequency of treatment was 7 out of 210 (3%) in those without respiratory support at the time of diagnosis versus 24 out of 178 (13%) in those on positive airway pressure, P < .001.
Among infants given DR-CPAP, pneumothorax increased with increasing GA and was higher in boys and lower in infants who received oxygen in the DR (Supplemental Table 9).
Discussion
We present a large birth cohort study in which an increase in pneumothorax was shown after implementation of the new NRP guidelines. In a nested cohort, we found a progressive increase in DR-CPAP in all infants, regardless of whether they needed oxygen supplementation or not. DR-CPAP was associated with increased odds of pneumothorax. The OR of pneumothorax with CPAP in neonates receiving DR-CPAP and 21% oxygen was 3 times as high as in those receiving DR-CPAP with supplemental oxygen. Similar results were obtained by instrumental variable analysis and propensity score analysis. The odds of DR-CPAP–associated pneumothorax increased with increasing GA and decreased with oxygen requirement, suggesting that the risk of CPAP-induced pneumothorax is higher in more-mature lungs with less underlying disease.
The interaction between oxygen administration and DR-CPAP, as well as the increasing odds of pneumothorax with increasing GA among neonates receiving DR-CPAP, is consistent with Adler’s data in which an increased risk of pneumothorax in neonates with higher lung compliance and lower chest wall compliance was shown.10 Moreover, our data are consistent with previous studies in which an association between DR-CPAP and pneumothorax in late-preterm and term neonates was shown.
In a large Canadian retrospective cohort study of 71 237 infants, it was found that CPAP was protective from developing pneumothorax in early-preterm neonates but was a risk factor for pneumothorax in moderate-to-late-preterm and term neonates.22
In 2015, Hishikawa et al23 found that implementation of new Japanese resuscitation guidelines including DR-CPAP was associated with increased pneumothorax in early-term infants but not in those born beyond 38 weeks. The association between guideline implementation and pulmonary air leak disappeared by adjusting for DR mask CPAP.23
In 2017, Clevenger and Britton24 reported increased odds of pneumothorax with DR-CPAP (adjusting for GA) both in a case-control study and in a later cohort study in term neonates. However, because controls of the cohort study were healthy, the possibility of selection bias or confounders was not taken into account in the analysis, suggesting a possible overestimation of the association between DR-CPAP and pneumothorax.24
In at least 3 randomized trials, researchers have shown the benefit of CPAP in late-preterm infants and term infants with respiratory distress and oxygen requirement in the NICU. In one study in neonates with transient tachypnea of the newborn, DR-CPAP using a T-piece–based infant resuscitator versus free-flow oxygen reduced the duration and severity of respiratory distress.25 In another study among infants with meconium aspiration syndrome, CPAP reduced the need for mechanical ventilation.26 In the third study among infants born at >30 weeks’ GA with respiratory distress, CPAP versus headbox oxygen administration reduced the need for up-transfer of infants with respiratory distress in nontertiary centers. There was a clinically relevant but not statistically significant increase in pneumothorax.27
Our study has several strengths that distinguish it from previous studies in which researchers have evaluated DR-CPAP and pneumothorax. First, both the birth cohort and the nested cohort had a large sample size. Second, analysis of the association of pneumothorax with CPAP was done in a nested cohort of high-risk neonates, excluding those who received PPV in the DR, thereby avoiding attribution of a pneumothorax to CPAP in a patient who also had received PPV. Third, we evaluated the interaction between oxygen requirement and DR-CPAP in regards to probability of pneumothorax. This is, to our knowledge, the only study in which this interaction has been examined.
There were several limitations to our study. First, this was a retrospective analysis, and randomization was not possible. This weakness was in part mitigated by using 4 methods: logistic regression, analysis adjusted for DR oxygen administration, analysis stratified for epoch as instrumental variable, and propensity score analysis.17 Secondly, it was impossible to distinguish whether some of the pneumothoraxes noted on chest radiograph were in fact spontaneous pneumothoraxes present before application of DR-CPAP. Third, data on infants who had a resuscitation team call but were not admitted to the NICU are not available because these infants are not included in the resuscitation database; therefore, the frequency of pneumothorax among all infants who had a resuscitation team call could not be assessed. Fourth, the true overall incidence of pneumothorax in this cohort is not known because chest radiograph was obtained only in symptomatic neonates.
With the findings from our study, 2 clinical implications are suggested. First, caution should be applied regarding providing DR-CPAP, especially in the term population. Second, the use of CPAP should be re-evaluated as therapy for grunting, nasal flaring, or tachypnea in infants 35 to 42 weeks’ GA without an oxygen requirement in the DR. These findings could be used to clarify NRP guidelines regarding the use of DR-CPAP. Further studies may be necessary to clarify the role DR-CPAP plays in the treatment of term infants with oxygen requirement in the DR.
Acknowledgments
We thank Patti Jeannette Burchfield, RN, for her work in collecting and extracting data for the NICU database. We thank Lucy Christie, RN, and Anita Thomas, RN, for their work in collecting and extracting data for the resuscitation database. Their expertise made this project possible. We also thank Robert Haley, MD, for his help with propensity score analysis.
Dr Smithhart conceptualized and designed the study, merged the spreadsheets of the 2 databases, and wrote the first draft of the manuscript; Drs Wyckoff, Jaleel, Kapadia, Nelson, and Kakkilaya conceptualized and designed the study; Mr Brown conducted statistical analyses; Dr Brion conceptualized and designed the study and conducted statistical analyses; and all authors participated in the interpretation of the data, critically reviewed the revisions, approved the final manuscript as submitted, and agree to be accountable for all aspects of the work.
COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2019-1720.
FUNDING: No external funding.
Comments
RE: Delivery Room Continuous Positive Airway Pressure and Pneumothorax
We thank Dr. Fatemizadeh and colleagues for their comment. Evidence should be obtained to assess risks and benefits of new therapies such as CPAP in the delivery room (DR) for term and late-preterm neonates. In 2010 evidence for CPAP in preterm was accumulating. ILCOR and NRP stated: “There is no evidence to support or refute the use of CPAP in the term baby.” Nevertheless, the 2010 ILCOR, NRP and Japanese Resuscitation Council guidelines recommended or considered CPAP for neonates with labored breathing or persistent cyanosis, regardless of gestational age. These guideline changes were followed by increased use of DR-CPAP and frequency of pneumothorax in term and late-preterm neonates in two cohort studies.1,2
All limitations highlighted in Dr. Fatemizadeh’s comment are listed in the manuscript.1 In this observational study we showed association, not causation, of pneumothorax with DR-CPAP after adjustment for selection and treatment bias by gold standard methods. The adjusted odds for pneumothorax in association with CPAP was higher in neonates without DR-O2 and thus less likely to have lung disease.
Dr. Fatemizadeh incorrectly states that the rate of thoracostomy/thoracentesis did not change over time in our cohort. In the nested cohort, the proportion of neonates who received either procedure doubled from 13/3750 (0.3%) to 23/3150 (0.7%) after implementing the new NRP guidelines, P=0.03.
Dr. Fatemizadeh questions whether more chest X-rays may have been obtained in asymptomatic neonates after NRP implementation. Chest X-rays were obtained in neonates with respiratory distress upon NICU admission, regardless of treatment in the DR. In 2012, a protocol limited chest X-rays in term and late-preterm neonates with respiratory distress to those needing >40% O2 or positive airway pressure, or with respiratory distress that was worsening or persistent for >6 hours.
In a local unpublished quality improvement (QI) project started in April 2017, a recommendation was made to avoid DR-CPAP for term or late-preterm neonates who did not need DR-O2. Among 4653 neonates without DR-O2 or DR-PPV and admitted to the NICU, significant decreases were observed in DR-CPAP use (6.8% vs. 1.4%, P<0.001), DR-CPAP-associated pneumothorax (1st chest x-ray and 1st day postnatal, 1.4% vs. 0.1%, P<0.001), but not in spontaneous pneumothorax (no positive pressure in DR or NICU, 1.6% vs. 1.3%, P=0.44). Thoracentesis was done in 1 of 44 neonates with CPAP-associated pneumothorax. Among 78965 term or late-preterm neonates, there was no change in NICU admissions for respiratory distress (without DR-O2 or DR-PPV), 0.8% vs. 0.8%, P=0.86 or duration of hospitalization (median 6 days, interquartile range 4-8 days in both epochs, P=0.46). In summary, CPAP-associated pneumothorax in neonates not needing DR-O2 or DR-PPV decreased after limiting DR-CPAP. We are now working at obtaining the number of neonates who received CPAP but were not admitted to the NICU.
These new results in addition to those in our manuscript could be used to clarify NRP guidelines regarding the use of DR-CPAP. We also confirm, as stated in our manuscript, that additional studies are needed to assess risks and benefits of DR-CPAP in term and late-preterm neonates.
Delivery Room CPAP in Late-Preterm and Term Infants: Jury’s Still Out
Delivery Room CPAP in Late-Preterm and Term Infants: Jury’s Still Out
Authors: Roxana Fatemizadeh, Pablo Lohmann, Caraciolo Fernandes
We read with interest the article by Smithhart and colleagues that found an increase in CPAP use and pneumothorax incidence in their nested cohort after implementing the 2011 NRP guideline changes.1 We applaud the authors for investigating the potential consequences of CPAP in a population in which it has not been fully studied; nonetheless, we feel compelled to highlight the major limitations of their study to ensure their results are interpreted appropriately.
The first major limitation is the inherent selection bias that occurs with studies of retrospective, non-randomized cohorts. Table 1 documents the CPAP group was statistically more likely to be delivered via cesarean delivery, have meconium-stained fluid, be growth restricted, and have surfactant administered1—these factors are associated with an increased risk of lung disease and consequently pneumothoraces.2–4 This point reminds us association is not causation - CPAP use and pneumothoraces may both be harbingers of the underlying lung disease.
Another limitation is that only neonates admitted to the NICU were included in the nested cohort. Without data on neonates exposed to DR-CPAP who did not require NICU admission, it is impossible to evaluate the true effects of the NRP guideline change. The omitted population may have shown benefit from the CPAP exposure (and guideline change) through prevention of a NICU admission. To assess the true exposure, the authors could include data on the number of neonates in their birth cohort with DR-CPAP who were never admitted to the NICU. With this total number, one might better determine the true number needed to harm (or treat) via DR-CPAP. By excluding this population that did well with DR-CPAP, the current number needed to harm is misleading.
Finally, the protocol for obtaining chest radiographs over the study period is unclear, and detection bias may be present. Did the increase in patients admitted on CPAP rather than on oxygen/room air lead to an increase in radiographs done? More radiographs will detect more pneumothoraces, and since the rate of thoracostomy/thoracentesis did not change, it is plausible that the increase in pneumothoraces is due to asymptomatic spontaneous pneumothoraces that were previously undetected.
Although the apparent increase in pneumothoraces is concerning, in order to determine the true effect of DR-CPAP, the authors should consider a prospective study wherein late preterm and term neonates with respiratory distress are initially randomized to CPAP or supplemental oxygen without positive pressure. Study endpoints, gathered on all patients, should include both beneficial and potential adverse effects of DR-CPAP, such as NICU admission rate, length of stay in NICU, and need for thoracostomy/thoracentesis.
Overall, the authors have tackled an important subject, and their findings raise questions on whether the risks of delivery room CPAP to the late preterm and early term population are acceptable. However, there isn’t enough data to assess how many patients benefitted from the 2011 NRP changes, and the rate of pneumothoraces requiring intervention did not increase in their population. We caution against implementing guideline changes based on this retrospective study given the listed limitations cited above.
References:
1. Smithhart, W. et al. Delivery Room Continuous Positive Airway Pressure and Pneumothorax. Pediatrics 144, e20190756 (2019).
2. Smith, J., Schumacher, R. E., Donn, S. M. & Sarkar, S. Clinical course of symptomatic spontaneous pneumothorax in term and late preterm newborns: Report from a large cohort. Am. J. Perinatol. 28, 163–167 (2011).
3. Katar, S., Devecioǧlu, C., Kervancioǧlu, M. & Ülkü, R. Symptomatic spontaneous pneumothorax in term newborns. Pediatr. Surg. Int. 22, 755–758 (2006).
4. Duong, H. et al. Pneumothorax in neonates: Trends, predictors and outcomes. J. Neonatal. Perinatal. Med. 7, 29–38 (2014).