Screening for risk of rehospitalization from home care: Use of the outcomes assessment information set and the probability of readmission instrument

Abstract

The purpose of this study was to evaluate the Outcomes Assessment Information Set (OASIS) compared with the Probability of Readmission (P ra ) instrument for use in predicting rehospitalization during home care. Using logistic regression and receiver operating characteristic (ROC) curve analysis, the P ra instrument was found to be significantly better at predicting rehospitalization than the OASIS case mix weight, clinical, or service scores. The area under the curve (AUC) for the P ra was .686 compared with .549 for the OASIS case mix weight (p ¼ .010). Similar results were found for the OASIS clinical and service scores. The AUC for the function score of ! 2 (.599) provided the closest approximation to the P ra (.686), and the difference between the two was not statistically significant (p ¼ .120). The OASIS function score could be used to identify at-risk home care patients without having to also use the P ra instrument. ß

The Outcome and Assessment Information Set (OASIS): A Review of Validity and Reliability

“…Four of these five studies evaluated convergent validity (Trochim, 2006; Weiner et al, 2008), using either a gold standard tool like the Center of Epidemiology Studies Depression Scale (Tullai-McGuinness et al, 2009), Structured Clinical Interview Axis I DSM-IV Disorders (Brown et al, 2004), the certified care plan (Kinatukara et al, 2005), or expert-derived “correct” answers (Madigan et al, 2003) as comparison criteria. Bowles and Cater (2003) evaluated the predictive validity of the case mix weight, the clinical, service, and functional domain scores with regard to risk of hospital readmission. In addition, construct validity was investigated by two studies, using statistical procedures to analyze the relationship between sets of items (Madigan & Fortinksy, 2000) or item response categories (Fortinsky et al, 2003).…”

“…Although three studies conducted assessments at two time points, the data were treated as cross-sectional in all studies. Four of the seven studies conducted secondary analyses of existing agency clinical records (Bowles & Cater 2003; Fortinsky et al, 2003; Kinatukara et al, 2005; Madigan & Fortinsky, 2000). Two studies employed prospective designs with data collected by agency staff and research clinicians (Brown et al, 2004; Tullai-McGuinness et al, 2009).…”

“…Some studies used only agency staff for data collection (Fortinsky et al, 2003; Kinatukara et al, 2005; Madigan & Fortinsky, 2000; Madigan et al, 2003) while others used a combination of agency and research staff (Bowles & Cater, 2003; Brown et al, 2004; Tullai-McGuinness et al, 2009). Likewise, some studies collected data during the usual agency assessment process while others employed additional assessment instruments administered at different time points.…”

“…Several studies indicated that they assumed that agency clinical staff were adequately trained in OASIS data collection via routine agency training and provided no additional training for the research data collection (Brown et al, 2004; Madigan & Fortinsky, 2000; Madigan et al, 2003). Two studies provided OASIS training for agency clinical staff (Bowles & Cater, 2003; Kinatukara et al, 2005). Two studies reported achieving adequate interrater reliability on the assessments (Bowles & Cater, 2003; Fortinsky et al, 2003).…”

“…Two studies provided OASIS training for agency clinical staff (Bowles & Cater, 2003; Kinatukara et al, 2005). Two studies reported achieving adequate interrater reliability on the assessments (Bowles & Cater, 2003; Fortinsky et al, 2003). Tullai-McGuinness et al (2009) provided training to the research RNs administering the comparative instruments; however, they did not assess interrater reliability among the research clinicians.…”