10.3390/polym8030084

Vollständiger Abstract

Worum geht es in dieser Arbeit?

<b>Background/Objectives:</b> Institutionalized older adults often present overlapping functional, cognitive, nutritional, intake-related, anthropometric, and biochemical abnormalities. We developed a prespecified weighted multidomain nutritional priority score (VIVA-NIPI) to organize the urgency of comprehensive nutritional review and assessed its internal construct concordance and ranking stability. <b>Methods:</b> Routine-care data from 124 residents in 10 long-term care facilities in Valencia, Spain, collected from June 2022 to April 2024, were combined across functional-cognitive, nutritional-intake/anthropometric, and selected biochemical-indicator domains. Weights were defined by the research team from clinical and methodological judgement rather than derived through a formal expert pairwise-comparison exercise; algebraically generated consistency ratios were not interpreted as validation evidence. The reference constructs overlapped with score components and therefore assessed internal concordance rather than independent clinical validity. <b>Results:</b> The continuous score averaged 24.4 ± 12.2. The partly overlapping k-means phenotype yielded an AUC of 0.918 (95% bootstrap CI 0.870-0.958); AUCs were 0.769 for MNA-SF ≤ 11, 0.602 for intake < 75%, and 0.782 for albumin < 3.5 g/dL. These values do not establish diagnostic or prognostic performance. Sensitivity analyses showed Spearman correlations of 0.991 after removing age, 0.960 after removing MNA-SF, 0.957 for an unweighted score, 0.983 when retaining only albumin and total proteins in the biochemical domain, and 0.901 after removing that do-main. <b>Conclusions:</b> VIVA-NIPI is a preliminary weighted multidomain research score, not a clinically validated triage tool. Independent clinical adjudication or prospective outcomes, external centre-aware validation, formal stakeholder weight elicitation, missing-data modelling, and calibration of actionable thresholds are required before implementation.

Abstract: PubMed · Datensatz