I too, have ruined healthcare (guest post – Laura Macke MD)

Epic’s In Basket is supposed to be an active workroom — a place where actionable messages live briefly, get handled, and move on. What it had become for many of our users was a hoarding house. Messages from years ago buried under messages from last year buried under messages from last month. Some providers had In Baskets with tens of thousands of messages. That’s not actionable — it’s a psychological weight that makes it nearly impossible to find what actually needs your attention today. The goal was to help people get their heads above water and reset the expectation for what In Basket is for. Not a hoarding house. A workroom.

We ran multiple batch purge jobs covering dozens of message types. We set expiration windows ranging from 90 days to one year depending on message type and deleted old messages that were languishing. It took months of planning, governance approvals, coordination with Epic, and careful batching.

Most users didn’t say anything — which, in healthcare IT, is about as close to a standing ovation as you get.

But we did hear from a few key users, and their complaints were critical to identifying a flaw in our purge. They found that we had eliminated some messages created with a future send date that were scheduled to arrive in a user’s In Basket weeks or months later. Our purge criteria evaluated message age based on when the message was created, not when it was intended to send. 80,000 messages with a future send date were inadvertently deleted.

Our analyst team didn’t panic. They met with affected users to understand the impact and were able to restore all but 147 messages and to help design workflows that would hold up going forward. The restore required its own batch job and careful execution. It was stressful, and it required levelheadedness under pressure.

I remember someone telling me in residency: it’s not the first mistake that kills the patient — it’s the second one, the one you make when you’re scrambling to cover the first. We didn’t make the second mistake. We slowed down, assessed, communicated honestly, and fixed what we could. “Measure twice, cut once” is good advice, but we measured twice and still got burned. That’s sometimes how it goes.

I’m proud of what we accomplished, grateful to the users who flagged the problems, and humbled by how much there still is to learn about operating at this scale. Millions of messages, a handful of critical complaints, and one hard-won restore batch. Our In Basket finally has a chance to work the way it was designed to — and thankfully, because we already set expiration dates on future messages, we’ll never have to do a purge like this again!