The relationship between the baseline geriatric nutritional risk index (GNRI) and neurological function at the convalescence stage in patients with stroke: a cross-sectional study

BACKGROUND: Malnutrition is a common complication after stroke and may worsen neurological outcomes for patients. There are still no uniform tools for screening nutritional status for the patients with stroke. We aimed to explore the relationship between the baseline geriatric nutritional risk index (GNRI) and neurological function at the convalescence stage for patients with stroke and assessed the predictive value of the GNRI for adverse neurological outcomes. METHODS: A total of 311 patients with stroke were enrolled retrospectively. Basic information and laboratory results on admission since onset of stroke were collected. The GNRI on admission was calculated and neurological outcomes evaluated by the Barthel index at 1 month after the onset of stroke. Statistical analyses, including correlation coefficient tests, multivariate regression analyses, and receiver operating characteristic (ROC) analyses, were applied in this study. RESULTS: Compared with the good outcome group, the poor outcome group showed a significantly lower GNRI on admission (P < 0.05). GNRI was associated with Barthel index (r = 0.702, P < 0.01). The GNRI was independently correlated with the Barthel index (Standardization β = 0.721, P < 0.01) and poor outcome 0.885 (95% CIs, 0.855–0.917, P < 0.01) after adjusting for covariates. Compared with no nutritional risk grades (Q4), the OR of GNRI to poor neurological outcome increased across increasing nutritional risk grades of GNRI (OR = 2.803, 95% CIs = 1.330–5.909 in Q3, 7.992, 95% CIs = 3.294–19.387 in Q2 and 14.011, 95% CIs = 3.972–49.426 in Q1, respectively, P for trend < 0.001). The area under ROC curves (AUC) of the GNRI was 0.804, which was larger than that of the NIHSS, BMI, or Albumin (P < 0.01), with an optimal cut-off value of 97.69, sensitivity of 69.51% and specificity of 77.27%. Combined GNRI with NIHSS gained the largest AUC among all the variables (all P < 0.05), with an AUC of 0.855, sensitivity of 84.75 and specificity of 72.73%. CONCLUSIONS: For patients with stroke, higher nutritional risk grades at baseline indicated worse neurological function at the convalescence stage. Compared with NIHSS, BMI, and Albumin, GNRI was a competitive indicator for the risk of poor neurological outcome. The predictive property of GNRI for adverse neurological outcomes might be more powerful when combined with NIHSS.