Predicting the risk of hospital admission in older persons--validation of a brief self-administered questionnaire in three European countries.

Objectives: To validate the Probability of Repeated Admission (Pra) questionnaire, a widely used self-administered tool for predicting future healthcare use in older persons, in three European healthcare systems.
Design: Prospective study with 1-year follow-up.
Setting: Hamburg, Germany; London, United Kingdom; Canton of Solothurn, Switzerland.
Participants: Nine thousand seven hundred thirteen independently living community-dwelling people aged 65 and older.
Measurements: Self-administered eight-item Pra questionnaire at baseline. Self-reported number of hospital admissions and physician visits during 1 year of follow-up.
Results: In the combined sample, areas under the receiver operating characteristic curves (AUCs) were 0.64 (95% confidence interval (CI)=0.62-0.66) for the prediction of one or more hospital admissions and 0.68 (95% CI=0.66-0.69) for the prediction of more than six physician visits during the following year. AUCs were similar between sites. In comparison, prediction models based on a person's age and sex alone exhibited poor predictive validity (AUC <or= 0.57). High-risk individuals (Pra score >or= 0.5) were 2.3 times as likely (95% CI=2.1-2.6) as low-risk individuals to have a hospital admission, and 2.1 times as likely (95% CI=2.0-2.2) to have more than six physician visits.
Conclusion: The Pra instrument exhibits good validity for predicting future health service use on a population level in different healthcare settings. Administrative data have shown similar predictive validity, but in practice, such data are often not available. The Pra is likely of high interest to governments and health insurance companies worldwide as a basis for programs aimed at health risk management in older persons.