Comparing the Effect of Beractant (Beraksurf™) with That of Poractant Alfa (Curosurf®) on the Need for Intermittent Positive Pressure Ventilation in Neonatal Respiratory Distress Syndrome by Adopting a Semi-parametric Approach: Re-Analyzing Data of a Randomized Controlled Trial.

Abstract

Introduction: Randomized controlled trial (RCT) data are analyzed by two challengeable adjusted and non-adjusted approaches. Performing appropriate adjusted analysis leads to a more interpretable and efficient estimation of treatment effects. Semiparametric adjustment approach modifies the estimating equations solved by the marginal treatment effect estimator by adding an augmentation function, which makes use of the baseline covariates and estimate the unbiased marginal treatment effect with improved precision. The effect of the intervention obtained using the semi-parametric adjustment method, similar to the unadjusted method and contrary to the adjusted parametric method, is marginal, resulting in better interpretability. Moreover, due to leveraging baseline covariates, it is more efficient compared to the unadjusted models. This study aimed to estimate the effect of beractant (Beraksurf™, Tekzima Company), compared with the Poractant alfa (Curosurf®, Chiesi Pharmaceuticals), as surfactant replacement therapy, on the need for Intermittent Positive Pressure Ventilation (IPPV) in Neonatal Respiratory Distress Syndrome (NRDS) more precisely by fitting a semi-parametric efficient model adjusted for appropriate covariates.

Method: This study is secondary and we re-analyzing data of a published RCT. This RCT was conducted in the NICU of Alzahra Hospital in Tabriz, Iran for eight months, and 200 infants were assigned to two groups receiving either 100 mg/kg BeraksurfTM (n = 99) or 200 mg/kg Curosurf® (n = 101). The effect of the treatments was evaluated regarding the need for IPPV by fitting semi-parametric logistic regression models, adjusted for the best subset of covariates selected by the forward variable selection algorithm and confounders identified by the expert panel. IPPV in our study was administered via an endotracheal tube, as per the protocol followed in the primary trial. The need for IPPV was determined based on the clinical judgment of neonatologists, considering the infants' respiratory distress levels, oxygen saturation, and arterial blood gas measurements. The decision was made within the first 72 h after surfactant administration.

Results: The proportions of the required IPPV outcome were 29.3% and 59.4% in the BeraksurfTM group and Curosurf®, respectively. While the unadjusted comparison between the two treatments was significant (OR = .283, 95% CI: (.157, .509), P-value < .001), adjusted OR in semi-parametric logistic regression by adjusting for the best subset of covariates selected by forward method including steroids, Apgar score at min1, and initial Fio2 wasn't significant (OR = .751, 95% CI: (.510, 1.111), P-value = .151). The efficacy of this semi-parametric model over the unadjusted model was 1.81. The results of the expert-based adjusted model, adjusting for the gestational age, birth weight, and initial FiO2, were consistent with those derived from the R²-based selection, supporting the robustness of our findings (OR = .685, 95% CI: (.449, 1.529), P-value = .078).

Conclusion: Fitting a semi-parametric model adjusted for the baseline covariates resulted in a marginal, unbiased, more efficient and interpretable estimation of BeraksurfTM versus Curosurf® effects. Although the crude model showed that BeraksurfTM was more effective than Curosurf®, the results of the efficient semi-parametric model with adjustment for the best subset of covariates revealed no statistically significant difference between the two drugs regarding their effects. We hope that the use of this method and its findings will contribute to a better understanding of covariate adjustment.