Impact of a Clinical Decision Support Intervention on Albumin Utilization and Appropriateness of Use in an Academic Healthcare System

BackgroundAlbumin is expensive compared to crystalloid intravenous fluids and may be used for inappropriate indications, resulting in low value care. Aim/PurposeTo study the impact of a computerized clinical decision support (CDS) intervention on albumin utilization and appropriateness of use in an academic healthcare system. MethodsA systematic review examining appropriate indications for albumin use in the healthcare setting was used by an interprofessional group of stakeholders locally to develop a CDS intervention to improve the appropriateness of albumin utilization. The order set was implemented across our healthcare system on 4/12/2011, included a list of appropriate indications, and automatically provided albumin concentration, dose and frequency based on the indication selected and patient weight and creatinine. We measured units of albumin ordered across the healthcare system and individually at each of three hospitals in the healthcare system 12 months before and after intervention implementation. An interrupted time series analysis using monthly data examined changes in the level and slope of albumin use during pre-versus post-implementation periods. We also reviewed charts of all adult inpatients receiving albumin in the 3 months prior to and following implementation of the order set at two of the three hospitals within the healthcare system, to determine if "appropriateness" of use had changed, as defined by our consensus criteria. We selected the two hospitals with the most frequent use of albumin in the pre-period. We used chi square tests to compare changes in the proportion of appropriate instances and grams of albumin used. We considered a p-value <0.05 as statistically significant. ResultsThe number of patient encounters analyzed in the 12 months before and after the albumin CDS intervention was 79,108, and 78,240, respectively. There was a statistically significant decrease in mean units of albumin ordered immediately post-intervention across the healthcare system (-4.98 units per 1000 patient days, confidence interval -9.64 to -0.33, p=0.04). At Hospital 1, there were no statistically significant changes in albumin ordering over time. At Hospital 2, albumin ordering significantly increased up to the intervention, but decreased significantly immediately following the intervention and continued to decrease significantly over time following the intervention; the pre and post implementation slopes were significantly different. At Hospital 3, albumin ordering was statistically unchanged up to the intervention, decreased significantly immediately following the intervention, and significantly increased over time following the intervention, but the pre and post slopes were not statistically different. At Hospitals 1 and 3, there was a statistically significant improvement in "appropriateness" of albumin use in the three months following implementation. ConclusionsImplementation of a CDS intervention was associated with an increase in the amount of albumin administered appropriately at two hospitals within an academic healthcare system and an overall decrease in albumin utilization across the healthcare system.