From the 2021 HVPAA National Conference
Peter McCaffrey (University of Texas Medical Branch), Ragini Desai, Rick Trevino, Todd Masel, Stephen Williams, Gulshan Sharma
Despite the unambiguous value of laboratory testing for diagnosis and patient care, it is estimated that up to 25% of laboratory tests are ordered inappropriately1. Among inappropriately ordered tests, repeat daily labs have been shown to result in sufficient blood loss to cause anemia2, and increased length of hospital stay3. In this study, we focus on reducing unnecessary daily Complete Blood Count (CBC), and daily serum chemistry testing in the form of Basic and Complete Metabolic Panels (BMP, CMP) at the University of Texas Medical Branch, a large, multi-campus academic medical center. As a measurement of repeat testing, we monitor the number of repeat test orders per 100 patient days, with a repeat test defined as a CBC, BMP or CMP ordered within 24 hours of a previous CBC, BMP or CMP having been resulted on the same patient. Moreover, we calculate monthly repeats per 100 patient days by aggregating repeats and patient days across all 47,542 encounters for 34,175 patients according to the month of admission to one of our hospital campuses within a study period spanning from December 1, 2018, until February 29, 2020, as shown in Figures 1 and 2. We exclude repeat tests ordered in ICUs and thus we examine utilization across 82,423 CBCs, and 89,027 BMPs and CMPs with a goal of reducing repeats tests ordered per 100 patient days by 10% compared to pre-intervention baselines.
Previous reviews have described a hierarchy of effectiveness wherein modifications to the order entry workflow (i.e. “forced restraints”) have the most pronounced and durable impact on utilization, followed by alerts, followed lastly by educational initiatives whose effects are among the most transient4. In this study, we implement two consecutive interventions deployed at two consecutive time points over the study period. First, we implemented a Best Practice Advisory (BPA) which would issue an alert when an order was about to be signed that included a repeat test as shown in Figure 3. These alerts were able to be overridden but did require a free text override reason in such circumstances. Since its deployment in October, 93.5% of these alerts have been overridden with 43.7% override reasons consisting of a single keystroke (e.g. “-”, “.”). Moreover, the initial effect of this intervention was a reduction in repeat testing per 100 patient days by –10.4% for CBCs (p=0.03) and –6.9% for BMP/CMPs (p=0.3) with an average month-over-month trend of +3.62%.
Comparatively, forced restraints were implemented in the form of modifications to Computerized Provider Order Entry (CPOE) within our hospital’s Epic Electronic Medical Record System (Verona, WI). These changes eliminated the creation of indefinitely repeating daily labs, resulting in a reduction of –19.9% for CBCs (p=0.05) and –15.9 % for BMP/CMPs (p=0.06). These data illustrate the utility of the hierarchy of effectiveness and demonstrate the enhanced impact of forced restraint compared to alerts and notifications. Similarly, these data illustrate an important point that the impact of alert-based interventions are variable across institutions and are ultimately subject to nullification via override or dismissal.
