Importance of nutrition adequacy by enteral nutrition in the acute phase of critical illness

Résumé fourni par la source

We appreciate the opportunity to answer questions about our manuscript. We note that most of the questions are related to a study whose design would be a clinical trial. Our research was a cohort study, so the methods used were measures of association. In this letter, we answer the questions of the letter sent to the editor: It is plausible that patients who received less feeding did so because their condition was already deteriorating, rather than worsening because of reduced feeding. This reverse causality is a concern that should be addressed in future clinical studies. Our research was a cohort study, the objective of which was to identify associations and not cause and effect. Critical illness involves complex physiological processes; factors such as organ failure, infection, or the severity of the underlying disease are likely to have contributed significantly to mortality. It is indeed more reasonable that nonsurvivors received less nutrition because of limitations imposed by their medical condition. Factors such as gastrointestinal intolerance, hemodynamic instability, and the need for life-sustaining interventions (eg, mechanical ventilation and vasopressors) often lead to reduced feeding in intensive care unit settings. Thus, reduced feeding observed in nonsurvivors may be a marker of disease severity rather than a direct cause of mortality. This distinction is crucial to understand the relationship between nutrition and outcomes in patients who are critically ill. Future studies, ideally with randomized designs, could help clarify the true impact of feeding on mortality and better account for the confounding effects of illness severity. The sample size was calculated to detect differences in the average protein adequacy between the survivor and nonsurvivor groups with a difference of 6.91% being relevant for the study.1 Considering a power of 80%, a significance level of 5%, and a standard deviation of 12.53% we calculated a total sample size of 106 patients, 58 survivors, and 48 nonsurvivors. Adding 10% for possible losses and refusals, the sample size should be 119 (65 in survivors and 54 in nonsurvivors). For this purpose, the PSS Health online version tool was used. Directed acyclic graphs (DAGs) are a helpful and robust tool commonly used in clinical and epidemiologic research to guide investigators to understand the interaction among variables and consider those beyond the exposure and their related outcomes. This strategy is particularly desirable to guide researchers’ decisions, not only about which variables need to be controlled in statistical analyses to minimize bias in estimating effects, but also to presume which variables could introduce bias, if controlled in the analysis. Specifically, DAGs can identify variables that, if controlled for in the design or analysis phase, are sufficient to eliminate confounding and some forms of selection bias. DAGs also help recognize variables that, if controlled for, bias the analysis (eg, mediators or factors influenced by both exposure and outcome). For these reasons, we chose the graphical criteria for selecting adjusted covariates to define the minimum set of covariates to adjust for confounding and reduce variable selection bias. DAGs are created based on current knowledge concerning biological and behavioral clusters linked to specific causal research questions. The covariates identified by the backdoor criterion to adjust for confounding were (1) age, body mass index (BMI), use of mechanical ventilation (MV), use of vasopressor drugs, sequential organ failure assessment (SOFA), and modified nutrition risk in the critically (m-NUTRIC) for energy adequacy and (2) age, BMI, use of MV, use of vasopressor drugs, SOFA, acute physiology and chronic health evaluation II, and m-NUTRIC for protein adequacy (the DAG figures and their explanations are described in the manuscript). Time-dependent effects: complex phenomena, especially in survival analyses, may involve time-dependent effects where certain variables have different impacts over time (eg, feeding adequacy may have a different impact on patient survival at different stages of critical illness). Simple models might not capture these effects well, whereas more sophisticated models could better represent the data. Scientific progress requires complexity. Historically, scientific advancements have often come from developing more complex models that explain phenomena better than simple, earlier theories. Although Occam's razor helps avoid unnecessary complexity, it should not prevent the development of more advanced, accurate models when warranted by the data. The authors declare no conflict of interest.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.