Drug history as a measure of comorbidity and predictor of long term outcome following ICU admission

Abstract

Learning Objectives: Comorbidity in patients in ICU has been shown to have an adverse effect on survival. While many scoring systems exist for assessing disease severity and estimating mortality in critically ill patients, they rarely take into consideration the full burden of comorbidity. Previous scoring systems have been developed for quantifying disease burden, but few have used drug history to directly measure this. This study aims to develop a prognostic tool based solely on patients’ repeat prescriptions, as a method of quantifying disease burden, and assess its ability to predict long term outcomes.
 Methods: The Medication-based Disease Burden Index (2006) was updated and modified. A retrospective search (using CareVue) for patients admitted to Glasgow Royal Infirmary ICU between 10/2007 and 11/2010 was carried out in order to obtain full drug histories from the time of admission. These patients were then individually scored using the modified MDBI. A second search was carried out using Clinical Portal to ascertain long-term survival. Survival analysis using Kaplan-Meier and Cox Proportional Hazards was carried out to illustrate any relationship between total score and survival probability, including correction for APACHE II score
 Results: 562 patients were included in the analysis. Survival probability dropped with increasing score: over 80% survival at 5 yr in those scoring zero, dropping to less than 40% in those with a high score. Log rank test was highly significant (p<0.0001). Hazard ratios for each of the 3 score groups showed an incremental increase in risk when compared to the zero score group, which was significant in each case (low score: HR 2.12(1.40–3.23) p<0.0001, medium score: HR 2.87(1.85–4.45) p<0.0001, high score: HR 5.16(3.08–8.64) p<0.0001). Results remained significant after adjusting for APACHE II score.
 Conclusions: This gives promising, significant evidence of a simple and useful predictive tool for quantifying comorbidity and the effect it has on long term survival following ICU admission. Further work is required to replicate its use in other populations, and in larger samples.