Comments on parenteral nutrition and extracorporeal membrane oxygenation circuit dysfunction

I read with interest the recent article titled “Association between parenteral nutrition and extracorporeal membrane oxygenation circuit dysfunction in critically ill adults requiring extracorporeal life support: A retrospective cohort study,” published in the Journal of Parenteral and Enteral Nutrition.1 I commend the authors for their valuable work and would like to comment on the statistical analysis and lipid emulsion treatment. The study reported that extracorporeal membrane oxygenation (ECMO) circuit dysfunction seemed to occur more frequently in the parenteral nutrition group, although this difference was not statistically significant, and parenteral nutrition was not identified as an independent predictor of ECMO dysfunction.1 However, several observed confounders may have influenced the results. In this retrospective study, the authors compared 38 patients receiving parenteral nutrition with 113 patients not receiving it to examine the effect on ECMO membrane dysfunction. This significant difference in group size raises concerns about statistical power, the influence of outliers, and the ability to adjust for confounding variables. Moreover, the imbalance may reflect nonrandom allocation driven by factors such as disease severity, physician preference, or institutional practices. In this context, using propensity score matching (PSM) could reduce bias by balancing observed covariates and approximating randomization.2 PSM summarizes multiple confounders into a single score, producing more comparable groups and mitigating confounding effects.2 For example, the frequency of distributive shock was significantly higher in the parenteral nutrition group (15.8%) than in the non–parenteral nutrition group (1.8%, P ≈ 0.014), as was the use of venovenous ECMO (71.1% vs 46.0%, P ≈ 0.023).1 Considering that venovenous ECMO has been associated with more frequent oxygenator failure,3 this unequal distribution may confound the observed association between parenteral nutrition and ECMO membrane life span. Although lipid emulsion therapy alleviates cardiovascular collapse due to toxic lipid-soluble drugs,4 high doses (52.5 ml/kg) can rarely contribute to ECMO circuit obstruction following lipid emulsion resuscitation administered for drug-induced toxicity.5 This potential interaction warrants consideration in interpreting ECMO membrane dysfunction in patients receiving parenteral nutrition containing lipid emulsions. Therefore, I suggest that applying PSM in retrospective studies with uneven group sizes and significant baseline differences (P < 0.05) is a reasonable approach to enhance internal validity. Further studies incorporating this method could clarify the relationship between parenteral nutrition, lipid emulsions, and ECMO circuit outcomes. I hope that these suggestions will meaningfully inform future investigations on this subject. None declared.