Predicting mortality in acutely hospitalised older patients; the APOP study

---

J de Gelder, J.A. Lucke, N. Heim, A.J.M. de Craen, A.J. Fogteloo, G.J. Blauw, S.P. Mooijaart

Voorzitter(s): prof. dr. M.M.H. Kramer, VUmc, Amsterdam & dr. C.G. Vermeij, Deventer Ziekenhuis

Woensdag 22 april 2015

15:00 - 16:00u in Zaal 0.5

Categorieën: parallelsessie (case reports/research)

Parallel sessie: Parallelsessie 4: Case reports/research

---

Introduction:
Acutely admitted older patients are at increased risk for mortality, but at the moment of presentation it is unknown which factors predict mortality. A feasible approach to improve outcomes is needed, by starting to identify the vulnerable patient. In the present retrospective follow-up study we therefore aimed to develop a prediction model on 90-days mortality.

Methods:
A retrospective follow-up study among all patients aged 70 years and over who were admitted to the Acute Medical Unit (AMU) of the Leiden University Medical Centre in 2012 was conducted. Potential early predictors of 90-days mortality were assessed and included vital signs, laboratory results, comorbidity and number of medications used at home.

Results:
In total 517 individual patients were admitted to the AMU in 2012. Ninety days after admission 94 patients (18.2%) had died. The final prediction model consisted of six significant (P<0.025) predictors: oxygen saturation, Charlson Comorbidity Index, thrombocytes, urea, C-reactive protein and non-fasted glucose, with an c-statistic of 0.738 (95% CI: 0.678-0.798). Using as a cut-off the level for the highest quintile of mortality risk (n=106) resulted in an average 44% mortality risk within 90-days.

Conclusion:
With an internally validated prediction model we were able to predict 90-days mortality in acutely hospitalised older patients, using routinely available clinical parameters. Such prediction models may serve at the first step to target high risk patients and design tailored care trajectories. At the moment a large prospective follow-up study is conducted to assess more determinants and endpoints to enhance performance of the model.