Achieved unified data view of 50+ hospitals and more than 100,000 caregivers
Reduced cost of care by $20M in first year using Tableau to visualise care value
Reduced knee replacement cost by $297 per patient, approximately $3M savings
Providence St Joseph Health is a 51-hospital system with 100,000-plus caregivers who deliver high-quality, cost-effective healthcare to millions of patients annually. Patient data resides in many systems, including electronic medical records (EMRs). As Providence continues to expand, managing and integrating the ever-growing volume of data is a top-of-mind challenge.
Providence needs a holistic understanding of patient health and satisfaction levels, cost of care, patient-reported outcomes and other data to optimise quality of care as well as improve the patient and provider experience. Prior to adopting Tableau, analysts were producing different versions of the same analytics for key performance indicators (KPIs), but the dashboards lacked important qualities: statistical discipline, comparative hospital or regional data, process measures and data on cost.
As a first step, Providence identified the universal problems that could be solved with data, creating unified views to highlight best practices and help reduce waste. Wasteful practices included using unnecessarily costly supplies and medications which can drive up the cost per patient case and make healthcare less affordable.
Using the Tableau platform, Providence built dashboards accessible to the entire hospital system, displaying detailed quality data and cost data. The dashboards allow practitioners and clinicians to see analytics that pertain to every hospital, clinician and individual nursing unit. This data transparency has been associated with substantial improvements in quality measures and large reductions in cost of care.
I think where Tableau has been most helpful with respect to our growth is creating simplicity in scaling. Once data is organised, it’s very easy to add another small multiple in our stack of visualisations related to a specific region.
Value-driven analytics shift analytics perceptions for Providence clinicians
Keeping up with the needs of an expanding healthcare organisation requires analysing complex data (including hundreds of KPIs) and producing actionable insights that support better quality of care. But within Providence’s dispersed network, different performance reports were saved in different places. This made it difficult for providers to see all relevant data, or use the data to collaborate and make better decisions.
Complicating matters, many doctors approach healthcare data with skepticism. They question whether EMR data is adding value to their practice and if the analysts know medicine well enough to make EMR data meaningful. That’s why Providence created the Chief Medical Analytics Officer position and hired Dr Ari Robicsek to oversee clinical data analysis to improve patient care, treatment affordability and the healthcare experience.
The system-level clinical analytics team, led by Ari, builds and implements Tableau dashboards and other analytics tools for system-wide clinician use. The team includes people with working knowledge of the specific needs of clinicians: physicians who oversee cost of care versus outcomes of care measurement and tool development; nurses who translate between the clinical, analytics and technical teams to ensure the analytics work and results are relevant to all clinicians; and biostats staff with deep, healthcare-related data analytics experience. Regional and hospital-specific analytics teams exist in addition to this central analytics team, resolving local problems.
What my team set out to do was build a kind of 'one ring to rule them all', the one central dashboard that included all of the KPIs that were important to all of our hospitals.
Using Tableau, Ari’s team created relevant, contextual dashboards that Providence clinicians and other staff can use to explore and understand quality-of-care metrics. These dashboards replaced reports that weren’t intuitive or adaptable, eliminating the inefficiencies associated with static reports. Users can also subscribe to additional dashboards that influence the work of specific doctors, nurses and hospital units.
Optimising patient data with other care metrics, Tableau helped Providence see what works and what doesn’t across all their hospitals. Now they can identify the biggest opportunities to maximise value and cut out waste using a single, integrated platform. Ari adds, “We’ve seen really meaningful moving of the needle on the difficult-to-improve quality outcomes across the system, and I believe part of that is because we’re all speaking a common language.” Currently, Providence is testing Tableau integration with its EMR environment so clinicians can consume analytics in the same place that they spend most of their day and capture all data from patient charts.
By building one single network of Tableau dashboards that allows a user to see at the system level how we are doing, how all of our hospitals are doing, and to very rapidly drill down to the individual hospital and nursing unit, and down to a list of the affected patients in that unit, has been helpful for communicating and collaborating around improving many of our KPIs.
Real-world results: Tableau uncovers wasteful practices and improves patient care
According to the Centers for Medicare and Medicaid Services, more than $3 trillion a year is spent on healthcare in the US, and approximately one third of that is waste. Providence was determined to reduce its own wasteful practices, and at the same time lower costs and improve patient care using Tableau. But highlighting data isn’t enough to convince physicians to change their behaviour. The analytics team needed an innovative, data-informed approach.
With Tableau, Providence analysed data and revealed wasteful practices to clinicians. Knee surgeons often opted for antibiotic-impregnated bone cement to prevent patient infections, which cost three or four times more per patient. Providence analysed thousands of patients who had antibiotic-impregnated and regular cement to see if materials made a difference with infection rates in first-time knee replacements, and they didn’t. By not using unnecessarily costly material in as many replacements, Providence cut hundreds of thousands of dollars in a year from the 10,000 procedures performed and reduced bone cement cost per case by $76. A similar approach with other items resulted in an overall cost per case reduction by $297 in the first year of this work.
Doctors don’t want to increase costs for their patients unnecessarily, so when the Providence surgeons were shown the data and the financial impact of their choices, they were appreciative and hungry for more information. By collecting data, analysing it and sharing it in a more organised way, Providence now makes it easier for doctors to understand what behaviours positively or negatively impact patient care. It is now bending the cost curve downward or keeping it flat across all patient costs and clinical areas for its 51 hospitals, even in the face of inflationary pressures. Across 10 clinical conditions, Providence has reduced cost of care by $20 million in the first year of these efforts.
What was interesting was the total number of (knee-replacement) patients that we were able to study was much larger than the entire combined world literature on that subject, because there haven’t been great studies in that area.
Leadership supports governed, self-service analytics to encourage transparency
Providence data sources are complex, including Premier, Press-Ganey, Epic and more. Before Tableau, analysts coded their own visualisations using data from these sources, and different dashboards were used by different user groups across the hospital network. These were cut and pasted into lengthy PowerPoint or PDF reports that were saved to local servers, making it difficult for clinicians to share analyses and performance metrics for the benefit of patients, the practice, the hospital and Providence. Likewise, local and regional hospital leadership, corporate executives and the Board of Directors at Providence struggled with getting a consolidated, high-quality view of the organisation’s performance.
By gaining corporate leadership support for Tableau from the beginning, Providence’s local and regional hospital leaders as well as clinicians took cues and quickly saw value in using the platform. Consistency was also required by the Board of Directors; by making an enterprise shift to Tableau, Providence gained one visual data language for leaders and frontline providers to assess quality of care. In fact, many of the metrics used to determine executive bonuses are now based on Tableau dashboards.
Building on success: Providence prioritises visual analytics
To take enterprise transparency and data analytics to the next level, Providence is shifting from data monitoring and reduction of cost per case in certain clinical scenarios (e.g. joint replacement, spinal fusion, coronary bypass surgery) to all conditions cared for across its hospital network.
“I’m excited about expanding some of the value work we’ve been doing, and we’re working hard to build the platform that will allow that to happen,” Ari explains. Analytics is also pre-baking statistics into the ETL process, and Providence has started using Tableau R integration for more flexibility with data parameters and filtering so the statistical significance in the data translates through to all physicians.
It’s somewhat unusual to have a physician full-time in a role directing analytics. And I would argue in healthcare that’s a really useful thing.