Propriedades psicométricas de dois instrumentos utilizados na avaliação nutricional de pessoas hospitalizadas com doença hepática crônica

Abstract

Introduction: Complications from chronic liver disease (CLD) influence the complexity of malnutrition diagnosis and its high prevalence. Thus, a careful nutritional assessment is essential in these situations, requiring nutrition-focused physical examination and a global nutritional assessment to minimize estimation bias. In this context, two new tools, Specific Nutritional Assessment for Advanced Chronic Liver Disease (SNE-ACLD) and Systematized Nutritional Semiology (SNS), address the specificities of this population. Objective: To evaluate the psychometric properties of SNE-ACLD and SNS tools and establish respective cut-off points for evaluating people with CLD. Method: An observational, multicenter study that included cross-sectional and longitudinal data was conducted in hospitalized people with CLD over 18 years of age, who were evaluated by SNE-ACLD and SNS. Data on sociodemographic, clinical, and nutritional status characteristics were collected. Psychometric properties verified the internal and external validity of these instruments, namely: structural construct validity, based on factor analysis; reliability, through internal consistency analysis and interobserver reliability; criterion validity, with concurrent and predictive mortality validation; interpretability analysis, considering the severity of CLD; and floor and ceiling effect. In addition, cut-off points were established for both instruments through cluster analysis. Results: Two hundred participants were evaluated, primarily male and older adults. The most prevalent etiology was alcohol, with the highest frequency of decompensated CLD. Malnutrition was identified by 79.0% of the sample by the ANE-DHCA and 80.0% by the SNS. The SNE-ACLD achieved construct validity through exploratory factor analysis by presenting components consistent with the malnutrition construct. Concurrent validity with AUC: 0.713 (95%CI: 0.570 – 0.843) and predictive validity with HR: 4.87 (95%CI: 2.22-10.71) in the model adjusted for sex, age, and disease severity were also confirmed. The instrument showed internal consistency (α-Conbrach:0.70) and interpretability for both severity markers for CLD, the Child-Pugh (p-value<0.001) and the MELD-Na (p-value: 0.030). SNE-ACLD did not show a floor and ceiling effect, indicating a good distribution of the total score and the ability to distinguish between those who score in the lowest and highest values. The SNS presented structural construct validity, addressing the domain of nutritionfocused physical examination in the construct of malnutrition. It also showed reliability for internal consistency (α-Conbrach: 0.91) and interobserver reliability (ICC: 0.85; 95%CI: 0.78 – 0.92). Interpretability was achieved when severity was considered through the Child-Pugh disease marker (p-value: 0.009). The tool presented a total score distribution of less than 15% in extreme values without floor and ceiling effect. The cut-off points established for SNE-ALCD and SNS were 20 and 15 points, indicating the presence of malnutrition. Conclusion: The new nutritional assessment instruments had internal and/or external validity, contributing to their application in clinical practice in a reliable and assertive way.