Can We Use Simple Radiographic Measurements to Predict Need for Intervention in Neonatal Pneumothorax?

Background: Pneumothorax (PTX) develops in 1-2% of neonates, leading to significant morbidity and mortality and requiring providers to be comfortable with management. Our objective was to evaluate whether radiographic measurements of PTX size can be used to predict the need for procedural intervention in neonates in order to help guide the need for the availability of specific personnel. Methods: With the help of a data analyst, 62 patients diagnosed with neonatal PTX between March 2016 and October 2024 were identified. Most babies (46) were born in 2023-2024 when our new electronic health record could more easily identify these infants. PTX size was evaluated using radiographs by calculating the ratio of the widest transverse measurement of the PTX on both anteroposterior (AP) and, when available, lateral decubitus (DECUB) divided by the widest transverse measurement of the hemithorax above the diaphragm. Clinical data were collected, and statistical analysis was performed using need for intervention (thoracentesis (TC), chest tube (CT), or both). Results: We found that a larger PTX size ratio, measured in the AP (p < 0.0001) or DECUB view (p < 0.008), was highly associated with need for intervention in this cohort of infants with PTX. Only 33% of PTXs required intervention. Also, 13/14 (93%) cases who underwent TC ultimately required a CT. PTX was more prevalent in males in general, but sex was not associated with needing intervention. The average gestational age (GA) of the cohort was 36 5/7 weeks, with only 12% being < 34 weeks GA. Univariate analysis indicated that lower GA and birth weight were risk factors for intervention. There was a trend (p = 0.075, by Fisher's exact test) suggesting that infants with both respiratory distress syndrome (RDS) and PTX may be more likely (60%) to require intervention (no RDS, 29% intervention). Finally, a receiver operator characteristic curve was derived from the AP ratio based on the yes/no intervention which resulted in an area under the curve statistic of 0.902 and the optimal AP ratio cutoff of 0.184. Conclusions: The ratio of the transverse measurement of the PTX/hemithorax size from radiographs was highly predictive for need for intervention in a cohort of primarily term infants with PTX. Smaller and lower GA infants were at a higher risk for requiring procedural intervention. Nearly all infants who had TC also needed a CT. These findings could inform clinical strategies for managing neonatal PTXs, especially in identifying appropriate needed personnel availability if a TC occurs.