Can We Improve the Detection of Old Patients at Higher Risk for Readmission After An Emergency Department Visit?

Individuals aged 65 and older are highly prevalent in emergency department (ED).1 They spend more time in the ED and frequently have more-severe diseases and present with atypical symptoms than younger adults.2 After an ED visit, older adults are at high risk for adverse outcomes such as early readmission and functional decline.1 Specific geriatric interventions applied to this population can reduce ED readmission.3 Screening tools were developed to target individuals at high risk for these adverse outcomes. The most studied and validated ones are the Identification of Senior at Risk (ISAR)4 and the Triage Risk Stratification Tool (TRST).5 Although these tools were designed for patients aged 65 and older, their specificity decreases over the age of 75,6 leading to almost all individuals being classified as at high risk. Based on patient safety quality improvement program results, it was previously proposed that some items of the ISAR tool be modified to increase its specificity in this population.7 This study aimed at exploring the efficiency of these modifications on the predictive performance of the ISAR. It also compared the modified tool with the TRST and with the predictive value of a multiple regression model. Responses to these questionnaires are dichotomous (yes–no). Readmission was defined as a new visit to the ED or a direct unplanned hospitalization at 1, 3, 6, or 12 months after the initial ED visit. The three tools were used as predictive models. Each score (ISAR, modified ISAR, and TRST) was used to compute a multiple regression model according to the stepwise backward method and including the most statistically relevant (P < .2) items of the three tools and comorbidities. To assess the predictive power of each model, receiver operating curves were calculated and area under the curve (AUC) was reported. Statistical analyses were performed using Stata software version 11 (Stata Corp., College Station, TX). During the study period, 345 individuals aged 75 and older were included (mean age 84; 63% female). Rates of readmission were 25% at 1 month, 38% at 3 months, 49% at 6 months, and 60% at 12 months. The multiple regression model combined the first item from the ISAR (presence of home help), the second item from the modified ISAR (increased dependency), the fifth item from the TRST (professional recommendation), and the presence of vascular disease. All models were equivalent in terms of predictive value at anytime (P > .05). Table 1 shows the AUC for each model for readmission at 1 and 12 months. Results for 3 and 6 months were similar. According to these results, the screening tools studied have low predictive power to assess readmission risk at 1, 3, 6, and 12 months. Modifying or rearranging some items of these scores did not improve their efficiency. This study is in accordance with a recent Dutch prospective study8 and shows that ISAR, TRST, and modified scores are not accurate enough to be used routinely to detect high-risk individuals. As was recently shown, these tools remain clinically useful for avoiding further geriatric intervention in negative-screened individuals because of their high negative predictive value (70–89% for the ISAR and 67–84% for the TRST).6 Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper. Author Contributions: Graf C. E.: Review of the literature. Graf C. E., Zekry D., Herrmann F. R., Chevalley T.: Drafting of the manuscript. Giannelli S. V., Sarasin F. P., Michel J.-P.: Critical revision of the manuscript for important intellectual content. Sponsor's Role: None.