Nippon Ronen Igakkai Zasshi. Japanese Journal of Geriatrics
Print ISSN : 0300-9173
Original Articles
Prediction of serum albumin levels by non-invasive factors among elderly female patients
Miki KinoshitaYuko TokudomeKenji TakagiShinji KatoYoshihiro Hotta
Author information

2011 Volume 48 Issue 4 Pages 361-368


AIM: To apply nutrition care management to elderly female patients, we predicted serum albumin (s-Alb) levels by non-invasive factors.
METHODS: After excluding patients with lesions/diseases which were directly related to s-Alb levels, we investigated 147 elderly women aged 75-years or over who were taking meals orally and were hospitalized from April 2008 to April 2009 at a hospital in Toyota. The patients were classified into 2 groups, one of patients with s-Alb levels of 3.5 g/dl or below (n=80), and the other of those with s-Alb levels of over 3.5 g/dl (n=67). Between the 2 groups, we examined differences in age, body mass index (BMI), living arrangements, necessary nursing care level (NNCL), bed confinement level (BCL), OH scale level (OHSL), and dietary intake either by the Student t-test, Mann-Whitney U test or chi-square test. Pearson correlation coefficients were calculated among s-Alb levels and selected variables. Taking into account the correlation coefficients, we conducted multiple regression analysis adopting the s-Alb level as a dependent variable and non-invasive factors as independent variables. For all the performed tests and analyses, a p value of less than 0.05 (on two-tailed analysis) was assumed to represent a statistically significant difference.
RESULTS: S-Alb level was significantly associated with variables, including age, BMI, NNCL, BCL, OHSL, and percentage of protein intake (PPI). Multiple regression analysis revealed 4 significant variables: age, BCL, OHSL, and PPI. The multiple regression equation was y=4.977-(0.098×OHSL)-(0.080×BCL)-(0.016×age)+(0.003×PPI), and the multiple correlation coefficient R2 was 0.398 (p <0.001).
CONCLUSIONS: S-Alb levels among elderly female patients may be predicted by 4 non-invasive variables: age, BCL, OHSL, and PPI.

Information related to the author
© 2011 The Japan Geriatrics Society
Previous article Next article