How we cut clinical admin time, task by task

Marton Bodonyi
Behind the genes at Eugene
7 min readApr 23, 2021

Think back to the last time you visited a healthcare professional, a doctor, dentist, genetic counsellor etc. Did they have a receptionist check you in? Did they organise a followup appointment, book you for some tests, write notes in your file? While you were in the waiting room, were they looking through your file, committing your health record to memory? There’s so much hidden administrative work in caring for people that I didn’t even know existed until I started working at Eugene — and so much of it is ripe for innovation.

In the last 2 years working at Eugene I’ve heard the word ‘fax machine’ more than anybody living in the 2020’s really should.

Clinical admin to some degree is unavoidable, but the processes and technologies that are used to do it have so many opportunities for improvement.

Continuous Deployment at Eugene

One of my favourite parts of working with Eugene is the pace at which we can optimise clinicians day-to-day workflows. Early on we decided to build our own custom patient management system called “Genie” to handle clinical admin and we’ve been in a constant cycle of improvement ever since.

To be honest I’d never seen one of these until I started writing this article. (from Wikimedia)

It turns out the process we use to iterate and innovate Genie fits pretty neatly into the SDLC diagrams you see floating around the web. Where Eugene is different from other places I’ve worked is that we often do the whole cycle in less than a day, sometimes less than an hour, from ideation to deployment. Most of the improvements we ship are small and have modest impacts — but they help us slowly chip away at automating the boring stuff to empower our counsellors to spend more time caring for people and less time filling out forms.

Step 1: Requirement Analysis

Every second Wednesday, 10am, it’s time for the Clinical Ops Efficiency meeting. We use a notion template for the agenda and notes and keep the meeting to a short, sharp 30 minutes. Members of the team are encouraged to drop comments into the notion doc throughout the week which are added to the agenda before the start of the meeting.

An example, cut-down agenda

The first 10 minutes is spent on check-ins where counsellors rate, one by one, how busy they were in the last fortnight. It’s a subjective metric but it helps us facilitate discussions about what tasks make counsellors feel busy. Next, the clinical team briefs us on any internal process changes. Then it’s time to dive into the stats (here’s an article we wrote about how measure ops efficiency https://medium.com/behind-the-genes-at-eugene/improving-clinical-efficiency-at-eugene-aa9f39373f80). The stats are a really great level set that helps pivot the mindset of the team into thinking about and quantifying opportunities.

Accountability

Next we practise some accountability by talking through actions from the previous meeting to see what’s been done and what still needs to be followed up. Sometimes actions get relegated to the too-hard basket where they sit until we have some proper project time to allocate to them. Mostly we try to keep the actions small and easy to achieve.

Discussion and actions

Finally it’s time for discussion and actions. One of the keys to making the meeting run on time is sharing a screen with the meeting agenda and typing out discussion notes in real time as the discussion is happening. It helps confirm that people have been heard, allows us to move on from sticky topics, and gives counsellors the opportunity to validate that we’re interpreting their ideas and feedback correctly.

Someone couldn’t make the meeting? No stress, we can comment and tag them in if we need their input. At 10:30 the meeting is over, although we have some overflow time set in the calendar in case there’s something we want to drill in to — it’s used very rarely.

Step 2: Software Design

Now it’s time to turn those actions from the Clinical Ops Efficiency meeting into features in Genie! This is the part were slack messages start flying around as details get ironed out. Sometimes we’ll shadow a counsellor while they perform a task to double check our assumptions, before we scope out some acceptance criteria and get to the good stuff. This part of the process is usually short and sweet for the tasks that come out of Clinical Ops Efficiency meetings, though this is the point in time where we sometimes discover that an action might be harder than we thought and will need to be scoped into a separate project.

#agile

Step 3: Implementation

This is easily the most fun part of the process if you ask me. It’s time to build! Our Genie stores a lot of information and most of the time we don’t need back-end changes to be able to make new screens and pages thanks to the data being available in GraphQL.

For the lab order screen above, we already had a table component with expanding functionality built into Genie and we re-used the lab-order-number update widget from another screen (the reason why it looks a bit out of place). From there all that was needed was a GraphQL query to list approved requests that were missing a lab order and a new React component to list out all the fields. From start to finish the screen was built in about 2–3 hours including time spent shadowing a counsellor.

GraphQL is the real hero of this story…

From there it gets made into a Pull Request which deploys a preview branch that other members of the team use to demo the feature using development data. Once it’s approved and merged to master it’s automatically deployed to production for counsellors to use.

Step 4: Testing

Besides writing automated unit tests that run in our CI to test React rendering data correctly, we rely on our counsellors to do a lot of the testing for us. But first they need to know that there are new features to test and be taught how to use them.

We use loom videos in a section on Genie called ‘What’s new’ to share updates with counsellors about new features. Bigger features are sandboxed in individual screens under the ‘Tasks’ section so that counsellors can try them out alongside their existing workflows and validate that they work as they should.

We use the Bugherd widget, with screenshots turned off, to allow counsellors to raise bugs and add feedback to any screen at any time from directly within Genie. Minimum friction. And if a bug is urgent then counsellors are encouraged to notify us on Slack as well.

Step 5: Evolution

Most admin task screens added on Genie are done so with some kind of automation in mind for the future. They’re generally single focus, batch-able and optimised for the least amount of clicks possible. In the above screenshot, ‘Post purchase’ Shopify syncing and ‘Pre test’ consult matching tasks are no longer performed by clinicians. The screens for those tasks only exist to monitor the automation, if needed.

The bigger fish

This process works for small, quick wins — some of which are prototypes of bigger more complex projects that get scoped into our quarterly OKR’s. It’s a small part of the work we do at Eugene but it’s the part I enjoy the most. To me the small, quick wins play a big part in connecting the product team and clinical team together and keeping a channel of ideas and innovation flowing freely between the teams. Plus I kind of just like building things sometimes.

Want to work in a place that gives genetic counsellor superpowers, 1 commit at a time? Head over to https://angel.co/company/eugenelabs and get in contact with us.

--

--