This was a single-institution prospective cohort study. A total of 386 patients aged ≥65 years admitted to a home care support hospital between September 2014 and October 2015 and May to December 2019 were enrolled. After excluding 18 patients who died during initial hospitalization, 22 who were lost to follow-up after 1 year, 40 who had missing data on admission, and 1 who discontinued follow-up, only 318 were included in the final analysis. The primary outcome was death within 1 year after discharge; based on the MNAⓇ-SF score, the patients were classified as well nourished, at risk, and malnourished. Based on the GNRI and GLIM criteria, the patients were classified as well nourished, mildly malnourished, moderately malnourished, and severely malnourished (Fig 1).
The patients received oral explanations and signed a consent form to participate in the study. If an individual had difficulty responding, a proxy consenter was requested. The study was approved by the Ethics committee of the Okubo Hospital (no.16).
Outcome
The primary outcome was the presence or absence of death within 1 year from the date of discharge at the time of initial admission. The patients were divided into a survival group and mortality group. Patients who did not die during the study period were censored.
Assessment
Within 1 week of admission, the registered dietitian transcribed the following items from the medical record or obtained information on them by interviewing the patient or family members: age, sex, level of need for long-term care, type of residence (home [alone or otherwise], residential care), and primary person in charge of cooking. Diseases were identified from the medical records and scored using the Charlson Comorbidity Index (CCI) [16]: 0 as “low,” 1–2 as “medium,” 3–4 as “high,” and ≥5 as “very high.” Basic ADL was assessed using the Barthel index [17].
A registered dietitian assessed the following anthropometric parameters on the day of admission. Height, weight, and triceps skinfold thickness (TSF) were measured by the same person using an insert tape and adipometer (Abbott Japan Co., Ltd.). Calf circumference (CC) was measured using a calf circumference measure (Nestle Japan Co., Ltd. Tokyo). The mean values of AC, TSF, and CC were measured thrice.
The levels of the following laboratory parameters measured within 2 weeks of admission were obtained from the medical records: albumin (Alb), blood urea nitrogen (BUN), creatinine (Cre), red blood cell (RBC), hemoglobin (Hb), and C-reactive protein (CRP).
Nutritional assessment
1)MNAⓇ-SF
Within 1 week of hospital admission, a registered dietitian interviewed the patient or a family member for evaluation. Patients who scored 12–14 points were classified as “well nourished, “8–11” points as “at risk,” and 0–7 as “malnourished.”
2)GNRI
The GNRI was calculated using the following formula:
GNRI=[1.489×Alb(g/dl)]+[41.7×actual weight/ideal body weight]
The calculated values were divided into the following categories: good nutrition (≥98), mild malnutrition (92 to <98), moderate malnutrition (82 to <92), and severe malnutrition (<82).
3)GLIM criteria
The MNAⓇ-SF was used for screening the risk of malnutrition. Patients with an MNAⓇ-SF score of ≥12 points were classified into the good nutrition group, while those with a score of ≤11 points with no actual disease and/or etiology were classified into the mild malnutrition group. The phenotype was assessed using a BMI <20 kg/m2 (≥70 years <22 kg/m2) and CC (<34 cm for men and <33 cm for women, to diagnose sarcopenia). A CRP level of ≥0.5 g/dL was used to determine the presence of absence of inflammation owing to an underlying disease. Patients who met the BMI and/or CC criteria and the CRP level criteria were classified into the moderate malnutrition group. Patients with a BMI <18.5 kg/m2 (≥70 years and <20 kg/m2) and/or CC <31 cm (used in the MNAⓇ-SF assessment) also met the CRP level criteria and were classified into the severe malnutrition group.
Statistical analysis
Statistical analyses were performed using IBM SPSS version 27 for Windows (IBM Japan, Ltd., Tokyo, Japan) and SAS ver. 9.4 (SAS Institute Japan Co., Ltd., Tokyo, Japan). Data are expressed as mean (standard deviation [SD]), median (interquartile range), or frequency (%). The t-test or Mann–Whitney U test was used for pair-wise comparisons of continuous variables between the survival group and death group. Categorical data were analyzed using the chi-square test and Fisher’s exact test.
The survival data of groups classified according to the MNAⓇ-SF, GNRI, and GLIM criteria were analyzed using the Kaplan–Meier method. The log-rank method was used for evaluating the significant difference between the two groups.
Multivariate Cox proportional hazard analysis with mortality as an outcome was performed to adjust for the effects of confounding factors and identify the most useful nutritional indicators for predicting patients’ prognosis. Multivariate analysis was conducted to assess the accuracy of prognostic nutritional assessment tools, including sex, age (>85 years), care level, residence type (residence [alone or other] or facility occupancy), CCI, MNAⓇ-SF score, GNRI, and GLIM criteria. The GNRI and GLIM criteria were divided into three categories: well and mild malnutrition, moderate malnutrition, and severe malnutrition. Considering the multicollinearity, the MNAⓇ-SF, GNRI, and GLIM criteria were analyzed using separate models. In all analyses, a two-sided test had a significance level set at < 0.05.