Weight estimation in people aged 65 years and over admitted to hospitalisations units

TL;DR

This study develops a new accurate method to estimate body weight in hospitalized older adults, improving nutritional assessments and clinical decisions.

Contribution

A new predictive model using mid-arm and mid-calf circumferences for weight estimation in older hospitalized adults is developed and validated.

Findings

The new two-variable model (BC + CC) showed good predictive accuracy with minimal bias.

Published equations underestimated BMI, while the new model preserved BMI classification.

The new model demonstrated excellent reliability in validation samples with no systematic bias.

Abstract

Malnutrition remains highly prevalent among hospitalised older adults, yet obtaining an accurate body weight, an essential parameter for nutritional assessment, medication dosing, and broader clinical decision-making, often poses substantial challenges in routine care. In patients with reduced mobility, direct weight measurement is frequently unfeasible due to limitations in available equipment and clinical resources. This study aimed to evaluate the validity and accuracy of three commonly used body-weight estimation equations and to develop and internally validate a new predictive model specifically designed for hospitalised adults aged ≥65 years.

A multicentre, cross-sectional study was conducted in nine Spanish hospitals using data from the NUTRIFAG research project, which investigates the nutritional status of hospitalized older adults and was carried out between October 2022 and February 2024. Measured weight, mid-arm circumference (BC), mid-calf circumference (CC), and heel–knee length (KH) were obtained following standardized procedures. Agreement between measured and estimated weight from three published equations was analyzed using intraclass correlation coefficients (ICC), linear regression, and Bland–Altman plots. A new predictive equation was developed through multiple linear regression in a derivation sample (60%) and validated in an independent subsample (40%). The impact of weight estimation on BMI and BMI-based nutritional according to GLIM criteria was also examined.

A total of 1,196 patients were included (54.7% men, 45.3% women; mean age 78.27 ± 7.95 years). All three reference equations systematically underestimated measured weight, particularly in men. Equation 1 yielded the highest consistency ICC (men = 0.769; women = 0.834) but lower absolute agreement. The new two-variable model (BC + CC) demonstrated good predictive capacity (R2 = 0.639 in men; R2 = 0.704 in women) and minimal mean bias (< 1 kg). In the validation sample, ICC was 0.764 (95% CI 0.710–0.810) for men and 0.824 (95% CI 0.777–0.862) for women, confirming excellent reliability without systematic bias. Previously published equations markedly underestimated BMI, resulting in a substantial overestimation of low BMI prevalence according to GLIM criteria, whereas the new equation showed minimal bias and preserved BMI classification comparable to measured weight.

The proposed equation based on mid-arm and mid-calf circumferences offers a simple, accurate, and clinically applicable method for estimating body weight in hospitalized older adults with reduced mobility. By reducing BMI classification errors, its use may contribute to more appropriate nutritional assessment and clinical decision-making. External validation in other settings is recommended.

Figures