Table 3.

Hazard risks of GNRI < 92 for all-cause mortality and cardiovascular events.

Variables	HR (95% CI)	p-Value
All-cause death
Unadjusted	3.87 (1.62–8.72)	0.003
Age-adjusted	3.91 (1.62–8.94)	0.003
Multivariable-adjusted a	5.95 (1.94–18.04)	0.002
Cardiovascular disease event
Unadjusted	8.26 (2.73–25.84)	<0.001
Age-adjusted	13.44 (3.99–50.51)	0.001
Multivariable-adjusted b	9.48 (2.47–44.31)	<0.001

Abbreviations: GNRI, geriatric nutritional risk index; HR, hazard ratio; CI, confidence interval. ^a Adjusted for age, diabetes mellitus, C-reactive protein, urinary protein-to-creatinine ratio, ratio of non-protein calorie to nitrogen, hemoglobin concentration, and baseline estimated glomerular filtration rate (A graphic approach (log–log plots) was used to assess the proportionality of HRs. Continuous variables were divided into two groups by their median values. All variables in which HR was estimated satisfied this assumption. If serum albumin concentration or phosphorus levels were used as the dependent variable, results remained similar (data not shown)); ^b Adjusted for age, diabetes mellitus, resistant high blood pressure, C-reactive protein, urinary protein-to-creatinine ratio, ratio of non-protein calorie to nitrogen, hemoglobin concentrations, and baseline estimated glomerular filtration rate (A graphic approach (log–log plots) were used to assess the proportionality of HRs. Continuous variables were divided into two groups by their median values. All variables in which HR was estimated satisfied this assumption. If serum albumin concentration or phosphorus levels were used as the dependent variable, results remained similar (data not shown)). The time from first visit to the dietary clinic to the day of dietary recording was not associated with the investigated outcomes on univariate analyses. If this interval was used as the dependent variable, the results remained similar (data not shown).