Clinical significance of small traumatic pneumothoraces

Advances in computed tomography (CT) have enabled detection of small or trace traumatic pneumothoraces. However, the Abbreviated Injury Scale (AIS) assigns all pneumothoraces a minimum severity of AIS 2, which can contribute to Injury Severity Score (ISS) calculation when pneumothorax is the highest severity injury in the chest region or to the New Injury Severity Score (NISS) if it is among the three most severe injuries. The objectives of this study were to evaluate the relationship between pneumothorax size and clinically relevant adverse clinical outcomes and to assess the impact of pneumothorax coding within AIS on the predictive performance of ISS.

We conducted a retrospective cohort study of adult occupants enrolled in the Crash Injury Research and Engineering Network from 2017 through 2023 who sustained pneumothorax or hemothorax and had analyzable chest CT imaging. Pneumothorax size was quantified using standardized volumetric segmentation and expressed as the percentage of hemithorax involvement. The primary outcome was a composite of endotracheal intubation, in-hospital mortality, or hospital length of stay greater than 7 days. Multivariable logistic regression evaluated age, pneumothorax size, bilateral pneumothorax, and Injury Severity Score excluding pneumothorax codes (ISS*) as predictors of the primary outcome. Predictive discrimination of a base model including age and ISS was compared with models substituting ISS* and ISS** (excluding pneumothoraces <1% of the hemithorax) using leave-one-out cross-validated area under the receiver operating characteristic curve.

There were 94 occupants who met inclusion criteria, of whom 57 (61%) experienced adverse clinical outcomes. Eighty patients sustained pneumothorax, including 15 with bilateral involvement. Pneumothoraces were typically small: 54 measured <1% of the hemithorax and only 5 exceeded 20%. Median pneumothorax size was larger among patients with the primary outcome, though not statistically significant in unadjusted analysis (p = 0.187). In multivariable analysis, ISS* was the strongest predictor (OR 1.18, 95% CI 1.10-1.29), and pneumothorax size remained independently associated with the primary outcome (OR 1.15 per percent increase, 95% CI 1.04-1.32). Age and bilateral pneumothorax were not statistically significant. A base model including age and ISS demonstrated an AUC of 0.70 (95% CI 0.59-0.81). Removing pneumothorax codes (ISS*) improved discrimination to 0.76 (95% CI 0.66-0.86; p = 0.03), while removing only pneumothoraces <1% (ISS**) improved discrimination to 0.77 (95% CI 0.67-0.87; p < 0.01). There was no significant difference between ISS* and ISS** (p = 0.36).

Very small traumatic pneumothoraces are common in modern CT-based trauma evaluation and appear to contribute limited incremental clinical risk. Inclusion of these injuries in ISS calculations may reduce predictive performance. Refinement of pneumothorax severity classification in future AIS revisions, including downgrading trace pneumothoraces and introducing greater size-based granularity, may improve alignment between anatomic injury coding and observed clinical outcomes.