A 3-year follow-up study in 245 institutionalised elderly (51 M:194 F; 83.7 ± 8.6 years). Nutritional risk was graded by GNRI (severe, <82; moderate, 82 to <92; mild, 92–98; no risk, >98). Main outcome was overall-cause death.
After the follow-up 99 (26 M:73 F) events occurred. Nutritional risk prevalence was 5.7 % , 24.1 % , 34.7 % and 35.5 % and mortality rates were 71.4 % , 48.6 % 33.7 % and 34.3 % with the GNRI < 82, 82 to <92, 92–98, and >98, respectively. Kaplan–Meier curves were significantly associated to GNRI (p = 0.0068). GNRI < 82 was consistently related to death (odds ratio, OR = 5.29, [95 % CI: 1.43–19.57], p = 0.0127) when compared to GNRI > 98. Similar results were confirmed by Cox regression (hazard ratio, HR = 2.76 [95 % CI: 1.89–4.03], p = 0.0072). Finally, when “severe” and “moderate” risk were analysed as a single class (GNRI < 92) outcome associations were: OR = 2.17, [95 % CI: 1.10–4.28] (p = 0.0245); HR = 1.76 [95 % CI: 1.34–2.23] (p = 0.0315). Survival analysis showed higher mortality rates by GNRI < 92 (p = 0.0188).
Present data support the use of the GNRI in the evaluation of long-term nutrition-related risk of death. We suggest a GNRI < 92 as clinical trigger for nutritional support in institutionalised elderly.