Age and sex adjusted adiposity estimators predict all cause and pneumonia related mortality in hospitalized older patients with severe dysphagia receiving artificial nutrition.

BACKGROUND

Older adults with severe dysphagia who require percutaneous endoscopic gastrostomy (PEG) feeding and/or total parenteral nutrition (TPN) have substantial mortality risk, yet practical tools for prognostic stratification are limited. Whether age- and sex-adjusted adiposity estimation formulas (CUN-BAE, ECORE-BF, and the Deurenberg formula) can improve risk prediction for all-cause and pneumonia-related mortality in this setting remains unclear.

METHODS

This study is a secondary analysis of a previously established single-center Japanese retrospective cohort of 247 patients aged ≥50 years with severe dysphagia receiving percutaneous endoscopic gastrostomy (PEG) and/or total parenteral nutrition (TPN). Associations of adiposity estimators with all-cause and pneumonia-related mortality were evaluated using Kaplan-Meier analysis, Cox regression, restricted cubic splines (RCS), time-dependent ROC analysis, and the C-index. Incremental predictive value beyond the baseline model was assessed using net reclassification improvement (NRI) and integrated discrimination improvement (IDI), with sensitivity analyses including multiple imputation, exclusion of deaths within 30 days, and additional adjustment for the Clinical Frailty Scale.

RESULTS

Across tertiles, Kaplan-Meier curves separated significantly for both all-cause and pneumonia-related mortality. In fully adjusted Cox models, the highest tertile was associated with higher all-cause mortality (HR 2.02-2.33) and markedly higher pneumonia-related mortality (HR 3.78-5.09) compared with the lowest tertile, with evidence of monotonic trends. Restricted cubic spline analyses supported largely linear dose-response relationships. Predictive discrimination improved over time; at 3 years, CUN-BAE and Deurenberg showed higher AUCs than ECORE-BF for both endpoints. Incremental analyses indicated added value for all-cause mortality with CUN-BAE and Deurenberg.

CONCLUSION

Adiposity estimation formulas, particularly CUN-BAE and Deurenberg, provide clinically useful mortality risk stratification in severe dysphagia receiving PEG/TPN.