An Interactive Interface to Explore Patient Venipunctures at a University Hospital

ANDREW HAMMES
Research Instructor, Colorado School of Public Health

During a hospitalization, patients routinely receive blood draws (i.e. venipunctures) in the course of diagnosis and treatment of their conditions. From the patient perspective, frequent blood draws can be distressing and decrease patient satisfaction with care. From the hospital perspective, blood draws represent an expense to the hospital both in terms of personnel time and material cost. Even though venipunctures impact patient care and hospital expenditures, integrated aggregate data on the patients receiving blood draws, the personnel performing the blood draws and overall trends in volume are not currently available to decision makers at the University of Colorado Hospital. The University of Colorado SOM NavLab seeks to address this issue by developing a clinician-usable interface which will allow clinical leaders to examine the number and timing of venipunctures done to patients.

This study was performed at the University of Colorado Hospital, including inpatient data between March 2019 and April 2020. All inpatient venipunctures were potentially candidates to be included in the interface.

Data was acquired from the Clarity tabulation of the EPIC Electronic Health Record including all tests done on inpatient venipuncture blood draws between March 2019 and April 2020. Consultation with the Ancillary Health Technician leadership informed the aggregation of tests for summarization. Tests were collapsed into individual patient draws in a five-minute rolling window, that is any vials which were recorded as collected within five minutes of the previous vial within a contiguous patient and collector pairing were considered one stick. This was done to allow many potential tests from one draw to be correctly subset into a single draw. Individual patient sticks were then aggregated to yield the number during a calendar day per-patient, as well as totals by the collecting user. All data aggregation was done in R v 3.6.0.

Aggregated data was transferred from R to Microsoft Power BI for visualization and dissemination. Summaries of data including daily tracking of number of tests and sticks, sticks by patient, sticks by user, and comparisons of sticks between groups were built within Power BI. Having these within Power BI also allowed the use of filtering, such as by date, user, or department, to be done by the end-user without additional effort by the team. Updates to the completed Power BI will be automated using stored commands within EPIC and R to load updated data into the report.

Excessive venipunctures can impact patient satisfaction and hospital resources. To address this problem in a data-driven way data must be presented to decision makers. The University of Colorado NavLab produced a Power BI report which allows clinicians and leadership to examine data on their own in a dynamic way with informative visualizations of the data.