Factors Predicting Early Non-Elective Hospital Readmission of Nutritionally Compromised Older Adults

Abstract

With the implementation of Diagnostic Related Groups (DRGs) as the basis of hospital reimbursement, the average length of hospital stay has declined. As a result, elderly patients are being discharged from short-stay hospitals more rapidly than those with similar illnesses in the past. This may result in more frequent need for early non-elective readmission. Predicting which elderly persons may be readmitted will enable preventative interventions to reduce overall health care costs. The purpose of this study was to determine predictors and assess predictive models of early non-elective hospital readmission. The sample consists of 134 male and female Medicare patients who were found to be nutritionally compromised at the time of discharge. The study is a secondary data analysis utilizing a repeated measures design of multiple variables representing demographics, anthropometries, laboratory values, and impairment in Activities of Daily Living (ADLs). Data have been collected at four time points: at hospital discharge, and during dietitian or nurse home visits at one week, one month, and three months post discharge. Bivariate logistic regression analysis reveal that serum albumin and change in weight measured at one month post discharge are highly significant predictors of readmission. There is a 31% decrease in the log odds of readmission associated with every 0.1 unit increase in serum albumin at one month post discharge (Wald χ2 = 8.2748, p = 0.0040), such that an albumin value of 3.3mg/dL relates to a 51% probability of early non-elective readmission compared to an albumin value of 3.7mg/dL with a 20% probability of readmission. Serum albumin at one month post discharge accounts for approximately 28% of the variance in non-elective readmission (McFadden's R2 = .2753). For every kilogram of weight loss, there is a 75% increase in the odds of readmission (Wald χ2 = 5.1757, p = 0.0229). Change in weight accounts for 8% of the variability in non-elective readmission (McFadden's R2 = .0843). It is of interest that serum albumin measured at the time of hospital discharge was not significantly associated with non-elective readmission in this population. This suggests that an evaluation of serum albumin and change in body weight should be routine during early post-hospitalization office visits to identify those at increased risk of early non-elective readmission.