Cord blood biomarkers predict neonatal respiratory dysfunction after prenatal smoke exposure: A decision tree model

Comprehensive Summary

The aim of this study was to quantify the effects of prenatal smoke exposure on neonatal respiratory function. The prospective cohort study analyzed the relationship between cord blood cotinine levels, oxidative stress biomarkers, and tidal breath analysis which allowed researchers to develop a decision tree model that can predict whether a newborn may have respiratory dysfunction from smoke exposure. The study used a sample size of 91 mother and newborn pairs of which 50 are smoke exposed, and 41 are unexposed. Of the 50 exposed, it was broken down further into active smokers, passive exposure, and a combination of both. The study used a ratio of time to peak tidal expiratory flow to total expiratory time (TPTEF/TE) of less than 35% to characterize low respiratory function and results indicated that cord blood cotinine levels had a high diagnostic performance (AUC=0.744, p < 0.001) and cord blood total oxidant status had good predictive ability (AUC 0.690, p = 0.003). The Classification and Regression Tree (CART) decision tree model identified both those variables as primary and secondary predictors at greater than 26.1 ng/mL and 9.755 umol/L respectively to predict groups at increased risk for respiratory dysfunction. After bootstrap correction, the model had an AUC of 0.724, sensitivity of 75.8%, and specificity of 68.9% showing promise as a screening and/or risk-stratification tool.

Outcomes and Implications

The study was limited by a modest sample size, short term follow-up, and being completed in a single center, thus it lacks generalizability. Furthermore, the CART model’s moderate specificity may cause unnecessary attention to otherwise healthy newborns. However, it is important to better predict and understand how smoking affects the respiratory system in newborns and this study accomplishes this by highlighting markers that physicians may use to better treat their patients and using those markers to predict which newborns are most at risk. With further testing using larger samples, longer follow-ups, and external validation, this model will have potential to screen newborns with smoking exposures and determine which will need closer respiratory attention in the future.