Objective The aim of this study was to develop models that predict hospital admission to ED of patients younger and older than 70 and compare their performance.
Methods Prediction models were derived in a retrospective observational study of all patients≥18 years old visiting the ED of a university hospital during the first 6 months of 2012. Patients were stratified into two age groups (<70 years old and ≥70 years old). Multivariable logistic regression analysis was used to identify predictors of hospital admission among factors available immediately after patient arrival to the ED. Validation of the prediction models was performed on patients presenting to the ED during the second half of the year 2012.
Results 10 807 patients were included in the derivation and 10 480 in the validation cohorts. The strongest independent predictors of hospital admission among the 8728 patients <70 years old were age, sex, triage category, mode of arrival, performance of blood tests, chief complaint, ED revisit, type of specialist, phlebotomised blood sample and all vital signs. The area under the curve (AUC) of the validation cohort for those <70 years old was 0.86 (95% CI 0.85 to 0.87). Among the 2079 patients ≥70 years, the same factors were predictive, except for gender, type of specialist and heart rate; the AUC was 0.77 (95% CI 0.75 to 0.79). The prediction models could identify a group of 10% of patients with the highest risk in whom hospital admission was predicted at ED triage, with a positive predictive value (PPV) of 71% (95% CI 68% to 74%) in younger patients and PPV of 87% (95% CI 81% to 92%) in older patients.
Conclusion Demographic and clinical factors readily available early in the ED visit can be useful in identifying patients who are likely to be admitted to the hospital. While the model for the younger patients had a higher AUC, the model for older patients had a higher PPV in identifying the patients at highest risk for admission. Of note, heart rate was not a useful predictor in the older patients.
- emergency department
Statistics from Altmetric.com
Twitter Jacinta A Lucke APOPLeiden
Contributors SPM, GJB, CH, AJF and BdG designed the study. SPM and GJB obtained funding. JAL and JdG collected the data from the electronic patient files and JAL checked them for validity. AJMC provided statistical advice. JAL and FC performed the statistical analysis and drafted the paper. BdG and SPM advised during the drafting process. All authors contributed to its revision and gave approval of the final version of the article.
Funding The Institute for Evidence-Based Medicine in Old Age (IEMO) is funded by the Dutch Ministry of Health and Welfare and supported by ZonMW (project number 62700.3002). The funding organisation had no role in the design or conduct of the study, neither in the data collection and analyses nor in the interpretation of the data.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement For collaboration purposes and data sharing, contact can be made with the corresponding author.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.