Do frailty and comorbidity indices improve risk prediction of 28-day ED reattendance? Reanalysis of an ED discharge nomogram for older people

Abstract

Background: In older people, quantification of risk of reattendance after emergency department (ED) discharge is important to provide adequate post ED discharge care in the community to appropriately targeted patients at risk. Methods: We reanalysed data from a prospective observational study, previously used for derivation of a nomogram for stratifying people aged 65 and older at risk for ED reattendance. We investigated the potential effect of comorbidity load and frailty by adding the Charlson or Elixhauser comorbidity index and a ten-item frailty measure from our data to develop four new nomograms. Model I and model F built on the original nomogram by including the frailty measure with and without the addition of the Charlson comorbidity score; model E adapted for efficiency in the time-constrained environment of ED was without the frailty measure; and model P manually constructed in a purposeful stepwise manner and including only statistically significant variables. Areas under the ROC curve of models were compared. The primary outcome was any ED reattendance within 28 days of discharge. Results: Data from 1357 patients were used. The point estimate of the respective areas under ROC were 0.63 (O), 0.63 (I), 0.68 (E), 0.71 (P) and 0.63 (F). Conclusion: Addition of a comorbidity index to our previous model improves stratifying elderly at risk of ED reattendance. Our frailty measure did not demonstrate any additional predictive benefit.