Revisiting My First Healthcare Quality Dashboard "Five Years Later"

Motivation

Five years ago, I accepted my first role as a data analyst supporting the hospital quality and safety department. One of my earliest assignments was to build a dashboard to monitor hospital-acquired conditions for nursing quality indicators.

Recently, I recreated that original dashboard using synthetic data generated with ChatGPT, as a way to reflect on how my thinking has evolved.

Background: How the Dashboard Was Used

Each month, nursing leadership met with nurse managers, educators, and frontline staff to review performance across key measures:

  • CAUTI: Catheter-Associated Urinary Tract Infection
  • CLABSI: Central Line–Associated Bloodstream Infection
  • HAPI: Hospital-Acquired Pressure Injuries
  • Falls with Injury

I attended many of these meetings. Every injury was reviewed individually. Teams discussed what led to the event, what could have prevented it, and what lessons could be applied going forward.

Outside of leadership meetings, the dashboard was printed and posted on unit boards across the hospital. The goal was for nurses to recognize and feel familiar with the dashboard when they encountered it on the unit after the meeting.

Design Choices and Analytical Intent

Leadership was primarily interested in the metrics and how units and the hospital compared to benchmark, while frontline staff focused more on the number of patients affected. Because each hospital-acquired injury is reviewed in detail through a separate event review process, the dashboard only needed to provide a clear summary.

The executive view was designed to give a high-level snapshot of unit performance at a given point in time. The goal was for users to understand performance across all four measures in less than a minute.

The trend view shows how each measure compares to benchmark over time. In the redesign, I also added a Statistical Process Control (SPC) view to help identify anomalies and bring them up for further investigation.

You can view the recreated dashboard on Tableau Public here.

Lessons Learned

Recreating this dashboard showed me that understanding your audience and how the dashboard will be used is the first, and perhaps most important, step in creating a successful dashboard. It is not enough to create a product that shows accurate numbers; you have to create something that is easy to use and effective.

Moving forward, my first question will be how you plan to use this dashboard.