Predicting early nonelective hospital readmission in nutritionally compromised older adults.

This study determined predictors of early nonelective hospital readmission in 92 (49 women and 43 men) nutritionally compromised Medicare patients. Subjects ranged in age from 65 to 92 y and represented patients hospitalized previously for medical or surgical services. The study used a repeated-measures design of multiple variables representing demographics, anthropometric and clinical values, and functional status. Data were collected during hospitalization and during home visits at 1 and 3 mo postdischarge. There were 26 readmissions, making the 4-mo nonelective readmission rate 26%. Subjects who were readmitted nonelectively were compared with those not readmitted. Univariate analyses suggested strong relations between readmission outcome and serum albumin, total lymphocyte count, change in weight, and change in white blood cell count. Sociodemographic variables were less useful in predicting readmission than were measurements of patients' clinical status. Measurements of change in clinical variables were generally more predictive of readmission than was any one single measurement. Multivariate-logistic-regression analyses suggested a model consisting of change in weight and change in serum albumin from hospitalization to 1 mo after discharge as being highly predictive of early nonelective readmission. Individuals with any amount of weight loss and no improvement in albumin concentrations during the first month after hospitalization were at a much higher risk of readmission than were those who maintained or increased their postdischarge weight and had repleted their serum albumin concentrations. More study is warranted to clarify whether routine monitoring of changes in weight and serum albumin after hospitalization is appropriate in older adults.