Redefining the core diversion detection workflow

Drug diversion by hospital staff poses a severe threat to patient safety, employee well-being, and hospital finances. A 2023 industry survey revealed that 95% of respondents strongly agreed that drug diversion incidents occur in hospitals, while 78% believed that most occurrences went undetected.

Starting in late 2023, as the lead UX designer on the ControlCheck product team, I was tasked with reimagining the next iteration of the recently improved Individual Risk Identification Score (IRIS) analytics experience to create a more proactive, efficient, and user-friendly system. The existing ad hoc process led to many high-risk users not being reviewed in time, and the changing risk scores made it difficult to maintain an accurate view of high-risk users.

Understanding customer needs

The financial impact of drug diversion on hospitals is staggering, with estimates ranging from $70 to $100 billion annually in costs related to healthcare, lost productivity, and criminal justice. However, the human cost is even more devastating. Drug diversion can lead to patient harm through denied pain relief, exposure to infectious diseases from contaminated needles, and compromised care from impaired healthcare workers. In some cases, diversion has even resulted in patient deaths.

These issues are not just inconveniences; they pose potential risks in healthcare settings where rapid identification and management of potential diversion is crucial. Leveraging ongoing customer feedback, prod support ticket data, and roadmap initiatives, we formulated the following assumptions about the current review process:

In addition to the existing analytics advisory group we established, one-on-one customer interviews revealed some key themes when considering the existing analytics experience — specifically the pain points that users were running into when they were trying to review activity for potential diversion. Our findings revealed that users struggled with:

Based on our research findings and alignment with roadmap goals, we established the following design goals for the Worklist:

The newest iteration

The Worklist was the third iteration and improvement to modernize the IRIS review workflow. This version was the biggest change — introducing a true workflow — and as a result carried the biggest risk, needing to educate and train customers on the new workflow without introducing unnecessary anxiety over large changes.

One size does not fit all

Customers were using the analytics in different ways in order to detect potential diversion. The majority were taking data out of the tool in order to roll their own diversion analysis spreadsheets to compile many data sources in one spot.

Using IRIS 1.0 and 2.0 nurse managers usually log in daily, for example, and see a list of the highest risk providers, often referred to as “red users”. They can see daily changes in scores, but only if they log in every day. If a nurse manager misses a day, many of the high-risk users do not get reviewed in time. By the time they would review someone, the provider’s risk score may have changed, and unless they go back and adjust the date range, they may not have an accurate view of high-risk providers anymore.

No more side by side

The previous version of the IRIS Dashboard that the Worklist replaced, had a long list of providers next to a "detail pane" — both of which were static. The Worklist introduced a much more dynamic, full-width list to help maximize screen real estate. It also introduced a detail panel — only visible when necessary — that would open above the content to ensure focus.

IRIS user card layout

Now, instead of simply a provider name and score, the full-width Worklist cards maximized the screen real estate and the data available providing a glanceable snapshot of critical information. This ensured users could quickly scan or sort the list to determine the highest risk users to review first.

Streamlined review statuses

We created a clear workflow with defined statuses (Needs Review, In Review, Reviewed, Watching, Snoozed) to help manage the review process more effectively. With those new statuses and the automatic flagging, we added distribution cards to the top of the worklist in order to provide clear visibility into what needed done first (e.g. Needs Review) but also a clear mechanism for narrowing the view without lots of filtering.

If you hopped in the way back machine and took a look at other areas within the application, you'd see the same distribution cards. In subsequent redesigns, they were removed, but were always something users referenced anecdotally and expressed a preference for. The Worklist felt like a natural place to resurrect them, and as a result they were reintroduced in other areas of the application as well.

Automated user flagging

The highest risk users were automatically flagged and added to the Worklist based on customizable criteria, which ensured we could stay flexible as a product team and cover a variety of workflows without being too rigid. It also eliminated the need for daily manual checking by nurse managers. Now, instead of scrolling through an entire list of red, yellow, and green users, the automatically flagged users would end up in the "Needs Review" section, with a clear starting point.

One stop historical context shop

We made the decision to only allow investigations after a review, which meant we needed a way to show users a single view of any past or present reviews and investigations. If a provider was constantly under review, or had many investigations, that would obviously increase the potential risk of the current review. We created a consolidated view of a user's IRIS reviews and investigations, providing historical context and improving decision making — all from within the User Detail Panel.

Improved documentation

We designed a dedicated documentation tab where reviewers could add notes, attach files, and tag review outcomes with the specific metrics they reviewed. This eliminated, or at least significantly reduced the need for external documentation. It also created a single "hub" where nurses and nurse managers could quickly see what had happened, any decisions that had been made, and the relevant supporting notes or documents.

On the road to general availability

To validate our design, we planned a 60-day beta test with 17 hospital sites. Our goals for the beta were:

The beta provided us a way to de-risk the introduction of such a large workflow change by:

Beta feedback

Out of our beta feedback calls, we heard two main pieces of feedback that we prioritized and addressed:

Putting Snooze To Sleep

A byproduct of workarounds in IRIS 1.0, Snooze no longer provided value with the introduction of a new workflow, and we got the green light to sunset it.

Not Enough Date... uhh

The flagged date range alone wasn't enough. As much as we wanted to move away from the existing IRIS workflow, users were not ready to give up the ability to see a user's IRIS score on any given day.

The data and performance implications to support this were massive. After a quick sync with the product manager, he and I recommended that we support the "current" IRIS date range — the last 30 days from the current date — in order to satisfy user needs, and limit the performance impact. Users gave us the thumbs up and we got it built.

Redefining workflows

The IRIS Worklist project represents the most recent iteration and most significant step forward for the analytics section. By doing a deep dive into current workflow shortcomings, understanding user needs, existing pain points, and business goals we were able to create a solution that addressed current inefficiencies. It also set the stage for more proactive and effective diversion detection and diversion management.

While the beta phase was wrapping up and we prepared for the general availability (GA) launch, I was unfortunately laid off as the company restructured teams. While I was not able to measure the outcome of the beta or watch the GA launch in the ways I had hoped, I am proud of the shift toward true workflow. While it's an assumption on my part (informed by user research and feedback) I believe the shift will be a bit bumpy as users experience the initial shock of change. After that I expect the new workflow will start saving nurse managers time and produce clearer outcomes when there's potential diversion.

The next phase of features were focused on adding more in app collaboration — allowing users to collaboratively work on the review and investigation process — mirroring how they work today. The goal? Keep customers in the tool, lowering the need to leave the tool.