Forecasting and validating fat mass ratio models through anthropometric measurements and health-related factors among people with HIV: a cross-sectional investigation.

Background: There is a limited research on predictive models of fat mass ratio (FMR) in people living with human immunodeficiency virus (HIV) (PWH). This study aimed to develop models considering anthropometric and health-related factors to predict and validate FMR in PWH regardless of sex.
Methods: One hundred and six Brazilian PWH (46.4±9.8 years) were evaluated for body composition using dual-energy X-ray absorptiometry (DXA), body circumference (BC), and skinfold thicknesses (SKs). FMR predictive models were developed using stepwise linear regression, and their agreement with DXA was assessed using Bland-Altman plots. Cross-validation was performed using the predicted residual error sum of squares (PRESS) method.
Results: Six FMR estimation models were developed for PWH, with adjusted R ² ranging from 0.43 to 0.72, standard error of the estimate (SEE) from 0.16% to 0.22%, and 95% confidence interval (CI) from 1.03 to 1.15. Model 6, including thigh SK, waist BC, therapy duration, subscapular SK, education years, and abdominal SK, exhibited the highest determination power (R ² adjusted 0.72, SEE 0.16%, and 95% CI: 1.06-1.15). The agreement between DXA-based FMR and predictive models showed minimal bias (-0.03 to +0.04) and narrower limits of agreement, particularly for the top-performing model (-0.33 to +0.30). Model 6 exhibited a high adjusted Q ² PRESS (0.70) and low SPRESS (0.17).
Conclusions: Our predictive models advance the study of body composition in PWH by consolidating the use of anthropometry for diagnosing and monitoring lipodystrophy regardless of sex.
Competing Interests: Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://atm.amegroups.com/article/view/10.21037/atm-23-1946/coif). A.P.d.S. received a scholarship from the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - (Finance Code 001, grant No. 88882.317622/2019-01), and from the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) (grant No. 201126/2022-2). Additionally, the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) will pay the article processing charges for this manuscript. The other authors have no conflicts of interest to declare.
(2024 Annals of Translational Medicine. All rights reserved.)