Risk stratification of New Zealand general practice patients for emergency admissions in the next year: adapting the PEONY model for use in New Zealand.

INTRODUCTION Patient-centred case management programmes in general practice are needed for patients at high risk for emergency admissions to hospital. AIM To adapt and assess the Predicting Emergency Admissions Over the Next Year (PEONY) model for use in New Zealand to provide risk stratification of general practice patients aged ≥40 years for emergency hospital admissions in the next year. METHODS A retrospective observational cohort study modelling 2008-2010 hospital utilisation and medicine use was undertaken to estimate for each patient a risk of emergency admissions in 2011. Health care data were integrated from four national data collections relating to general practice patient registers, hospital admissions, pharmacy dispensed medicines, and mortality. Logistic regression was used to estimate coefficients for variables in the model. Model performance was assessed by calculating its positive predictive value (PPV), sensitivity, and specificity at incremental risk thresholds and receiver operating characteristic. RESULTS The patient cohort included 1,409,506 registered patients; 154,892 (11.0%) had an emergency admission in the follow-up year. Patient age, sex, ethnic group, deprivation status, prior emergency admissions and use of medicines for chronic conditions were all strong predictors of admissions in the next year. The model's PPV for the validation dataset was 58.2% for patients with risk ≥ 50%, and the area under its receiver operating curve = 0.72. DISCUSSION The PEONY model provides an effective methodology for stratifying New Zealand general practice patients' risk for future emergency admissions. High-risk patients may benefit from patient-centred case management programs to address risk and reduce unplanned admissions.