I’ve been working on a major project for the last couple of months and tomorrow is the go live. Tuesday is the traditional day for new releases at my organization. Although the IT staff likes to do things over the weekend, we know that Monday mornings are the busiest day in most outpatient practices and asking users to accept (let alone successfully adopt) changes on a Monday is just a bad idea.
The project involves a unified laboratory ordering scheme across multiple reference laboratories and hospitals, some of which are competitors. As a regional player, our health system took charge of this project with the goal of allowing physicians to more easily order tests from different facilities based on insurance and patient preference. I’m sure the side benefit of being able to see the ordering behaviors of non-employed physicians so that the hospital-owned labs can lobby for greater market share might have played a role in our leadership as well.
It would have been challenging enough to obtain the historical order data from the hospital-owned labs and create the crosswalk to send the proper codes to the designated facilities. We knew from standardizing the lab orders within our own health system that you can’t always map apples to apples. They could be Golden Delicious, or Granny Smith, or Fuji. Sometimes they are eaten via biting, sometimes sliced with a paring knife, and sometimes with one of those fancy gizmos that never quite fits in your kitchen drawer. They may each be an apple, but in the lab world they are entirely different orders.
Throw in the fact that we had to obtain order data from competing facilities and things started to get interesting. One national reference lab was very cooperative . They have 85 percent market share for some of the physicians and are eager to keep it that way. They provided the data exactly as requested and included all kinds of additional data we didn’t ask for initially, such as reference ranges, order entry questions and their expected responses, and even the type of tube needed for blood draws. They also provided it within one week of the request, which was outstanding.
Another national reference lab was less cooperative. They have a decent market share, but tend to act like they are the only show in town, and their response to our data request was handled accordingly. They initially provided data that lacked vital fields and wasn’t even for the time period we requested. They would send different parameters for different physicians on different spreadsheets. We had to explain to them multiple times that we needed consistent data to keep our analysis functional across all the practices and facilities. It took nearly eight weeks to finally receive the data.
The rest of the facilities fell somewhere between those two on the spectrum. Thank goodness I had a health information management intern to help out. As the data started to come in, we began the analysis. What we found was interesting, namely that physician ordering patterns were all over the place. We knew that we would see a wide variety of ordering behaviors given different specialties and geographies. We didn’t expect to see as wide a distribution within a single specialty, however.
Once we started to see some of the outlier tests that were being ordered, we also asked for data looking at how often the labs were contacting ordering providers for clarifications or substitutions. The preliminary analysis led us to increase scope and add the complicating factor that’s making me the most worried about tomorrow’s go-live: we made it easier to order the right test and a bit more difficult to order the wrong one rather than just mapping everything that had been used in the past. Given the fact that many of the participating providers have at-risk contracts or are part of an Accountable Care Organization, most people were on board with efforts to drive ordering behavior. How users respond to it in a live environment may vary.
Even without that particular challenge, managing the data was going to be difficult. We compared the lab-provided data to order data extracted from some of the provider EHRs and found that quite a few providers had test libraries with incorrect or outdated order numbers. We had to compare the tests they were intending to order with the current order numbers and ensure that we didn’t have duplicates or mismatches.
We had to work closely with a diverse group of resources – physicians, office managers, nurses, laboratory technicians, pathologists, interface specialists, software developers, and more. It was interesting to see each group’s perspective. However, I was surprised at how little some groups knew about the end user experience and what providers need to order labs accurately and efficiently.
Right before testing began, I thought I was losing my mind with collating all the different facility and provider approvals. I’m extremely grateful to a colleague who presented me with a delightful addition to Excel that helped me do the final bit of data cleansing. I don’t know how I lived without it. I am thankful not only for a new tool in my belt, but for someone who cared enough to see a problem and offer to solve it.
I’m sure there will be some unhappy providers who can’t find the tests they’re used to ordering. We’ll have a fully staffed go-live war room with not only directions to find the correct test, but an explanation of why the “old” tests were retired. I’ll be manning the phones as well, not only for escalations, but to see how the process is working overall. Wish me luck!