Predicting mortality in acutely hospitalized older patients: a retrospective cohort study

J Gelder, JA Lucke, N Heim, AJM de Craen, SD Lourens, Ewout Steyerberg, B de Groot, AJ Fogteloo, GJ Blauw, SP Mooijaart

Research output: Contribution to journalArticleAcademicpeer-review

19 Citations (Scopus)
19 Downloads (Pure)


Acutely hospitalized older patients have an increased risk of mortality, but at the moment of presentation this risk is difficult to assess. Early identification of patients at high risk might increase the awareness of the physician, and enable tailored decision-making. Existing screening instruments mainly use either geriatric factors or severity of disease for prognostication. Predictive performance of these instruments is moderate, which hampers successive interventions. We conducted a retrospective cohort study among all patients aged 70 years and over who were acutely hospitalized in the Acute Medical Unit of the Leiden University Medical Center, the Netherlands in 2012. We developed a prediction model for 90-day mortality that combines vital signs and laboratory test results reflecting severity of disease with geriatric factors, represented by comorbidities and number of medications. Among 517 patients, 94 patients (18.2 %) died within 90 days after admission. Six predictors of mortality were included in a model for mortality: oxygen saturation, Charlson comorbidity index, thrombocytes, urea, C-reactive protein and non-fasting glucose. The prediction model performs satisfactorily with an 0.738 (0.667-0.798). Using this model, 53 % of the patients in the highest risk decile (N = 51) were deceased within 90 days. In conclusion, we are able to predict 90-day mortality in acutely hospitalized older patients using a model with directly available clinical data describing disease severity and geriatric factors. After further validation, such a model might be used in clinical decision making in older patients.
Original languageUndefined/Unknown
Pages (from-to)587-594
Number of pages8
JournalInternal and Emergency Medicine
Issue number4
Publication statusPublished - 2016

Research programs

  • EMC NIHES-02-65-01

Cite this