Predictors of Hospitalized Patients' Access to Postdischarge Home Health Services

Abstract

An ex post facto study was conducted to identify factors associated with diabetic patients' access to home health services following hospital discharge. The Andersen (Aday, Fleming, & Andersen, 1984) model of access served as the conceptual framework for the study which examined the relationship between the predisposing, enabling, and need characteristics of the sample, and the realized access indicators of referral to, and acceptance for post-discharge home health services.

A sample of 300 hospitalized patients was identified at three study sites. One-half of the sample had been referred for home health services and one-half had not been referred for post-discharge home health services. Data regarding acceptance for services were collected on 136 patients. Discriminant analysis was conducted to identify the set of predictors of referral and subsequent acceptance for home health services.

A discriminant function, strongly related to dependency, was found to maximize the differences between groups referred and not referred, and accounted for 33.5% of the variance between groups (p $<$.0001). Length of stay, number of secondary diagnoses, age, and marital status showed moderate association with the function. Stepwise discriminant analysis identified a six variable function, in which dependency was the primary predictor, that significantly (p $<$.0001) discriminated between the groups. No differences were found between the groups accepted and not accepted for home health services following referral.

Nine research hypotheses were tested for significance using chi square and correlations. Patients more likely to be referred for home health service were older, female, lived alone or with friends or family other than a spouse, were covered by Medicare or Medicaid, had a longer hospital stay, had more secondary diagnoses, and were more dependent at discharge. Weaker support was found for persons who had been hospitalized within the prior 24 months. Significant correlations among predictors were found for dependency with age, length of stay, and secondary diagnoses. Additionally, the number of secondary diagnoses were found to correlate significantly with length of stay.

It was concluded that need, primarily dependency, is the best predictor of referral to home health. Continued study, incorporating both population and health system characteristics, is recommended to further examine acceptance for home health services following referral.