RESUMO
AIM: To evaluate the Probability of Repeated Admission (Pra) scale performance in predicting healthcare utilization and death in a community sample of Rio de Janeiro city, Brazil. METHODS: This was a prospective longitudinal study. The sample was composed of community-dwelling older people who participated in the research network Frailty in Brazilian Older People Study, Rio de Janeiro section. It analyzed data from the baseline and the follow up. At the baseline, 764 older adults were stratified by the Pra, and, 3 years later, data on health services utilization and survival status were collected. The Pra accuracy, obtained by the area under the curve, the sensitivity and the specificity to predict hospitalization in 3 years; visits to the emergency department; and death were determined. Cox regression estimated the survival curves, according to the risk strata. RESULTS: The accuracy, the sensitivity and the specificity of Pra were 62% (95% CI 57-68%), 8.8% and 95.4% for hospitalization in 3 years; 59% (95% CI 53-64%), 8.3% and 94.8% for emergency room visits; 66% (95% CI 59-74%) and 68% (95% CI 63-73%), 15.0% and 96.0% for death. Survival curves showed that high-risk individuals were more likely to die compared with the low-risk individuals, adjusted to covariables (P < 0.001). CONCLUSIONS: The Pra presented low accuracy, low sensitivity and high specificity for all outcomes observed, indicating a poor performance to screen vulnerable older adults in Brazil. Therefore, this tool should not be used alone, as a case-finding instrument. Geriatr Gerontol Int 2020; 20: 360-365.