Feelin’ Lucky With Your Data Migration or Integration Project?

My wife and I recently went to the home of Ernest Hemingway in Key West. Learning about the history of the house and some of the background of the books by Mr. Hemingway, I decided to read some of Mr. Hemingway’s work. I started with The Old Man and The Sea.

For those of you who have not read this wonderful, short, piece of literature, I won’t ruin the story for you. I will just give you a synopsis of the story here so we can get to the point.

The story is located in and off the shores of Cuba (also Mr. Hemingway’s origin). There was an old man. He was a fisherman. He hadn’t had much luck with the fish lately. One day he went out alone on his small fishing boat. There is one quote that I want to focus on for this blog. The old man says it to himself several hours after he set out on the boat.

“Every day is a new day. It is better to be lucky. But I would rather be exact. Then when luck comes you are ready.”

I think this quote especially applies to system testing. In the “Magic!” blog, we discussed the process of integration projects:

Vendors involved in implementation of an interface usually do roundtrip testing (to verify communication) and basic test cases. The responsibility for the more rigorous testing typically falls to the client. This is what is called user testing. The client needs to outline the different business scenarios that the interface was designed to solve and develop test cases that will verify all of the business scenarios. Make sure you devote ample time to testing. It will save you headaches in the long run.

Let’s break this down further.

What is roundtrip testing? For each type of message (demographics/ADT, scheduling/SIU, charge/DFT, etc), the sending side of an interface will send a message. The receiving side will verify receipt and verify the message is formatted as expected (see “Standard is not so Standard” for more discussion of setting expectations). Note that this is more to confirm that the two systems are talking rather than to verify any sort of business logic.

Why do most vendors only do roundtrip testing? Most vendors will not know the workflow intricacies of a practice. The vendor knows what is in the specification document for the integration. They can test what is outlined in the specification, but what is outline in the specification does not necessarily encompass all variations in workflow practice encounters, especially rare scenarios. The vendor relies on the practice to know their workflow and to understand the different scenarios that need to be tested in the integration. This is not always the case. For example, the practice may have a couple patients that come in once or twice a year. These patients are special cases and the practice may need to have a special workflow in their system for these two patients when they visit. An integration built for the standard use case that covers 95% of all encounters may not properly handle the special cases in your practice.

It is the practice’s responsibility to identify these special use cases and test for them. When I discuss testing with a client I ask them to first describe the typical workflow. For example, we have a client that implemented a demographics interface from an associated hospital. The typical workflow for a patient that doesn’t exist in their system is to add the patient, add the guarantor, and add the patient insurance. The typical workflow for a patient that does exist matched on account number, first name, last name, and date of birth is to just update demographic information, leave the guarantor as is, document the patient insurance in the demographic notes, but don’t update the patient insurance on the patient.

Once you have described the typical workflow, start going through the possible variations of the workflow. You need to include knowledgeable people from all departments that are affected by the interface to catch the possible different scenarios. In our example, as we were testing, we found a special scenario where a patient was admitted to the hospital ER, but the new account entered in the hospital system had a temporary patient name. Well, this prompted a message across the interface with the temporary name. This name was entered in the receiving system side as a new patient. When the name was updated, we got a mismatch because the account number matched, but the patient name and date of birth did not. This prompted us to change the scope and logic of the interface on receiving side to discard patients with certain specified names and departments. Once all of the information was filled out, we would then enter the patient on the receiving side.

It is always better to understand these different scenarios when you are scoping out the integration project, but if the practice does not catch the scenario in testing, it can cause bigger data issues in production. In the example, not testing all of the scenarios could have resulted in lots of incorrectly entered and mismatched patients when the interface was put into production.

Another important point with testing is to test everything every time there is a revision. If you caught an error in one of your testing scenarios, make sure to test every step in all of your scenarios again when the error is resolved by the interface vendor. There have been many times when I am testing for a client that when the vendor fixes one thing it breaks something else. The fix had unintended impact on a different part of the integration process. Start from the beginning. Test every scenario thoroughly with every revision. You may even find that other scenarios come up when you find an error that you need to test specifically to make sure the previous error does not rear its ugly head again.

A practice should realize the system is theirs. Accepting what happens to you after the fact is never a good position to take. The practice is the last line of defense before a system or interface is implemented that could benefit or harm the practice. Test it. Understand it. Even if project scope needs to change, it is best to catch an issue up front before your practice is reliant on the system and has to implement work-arounds to accomplish their goal. You may realize that it costs more in the long run to allow things to happen without testing than to put the effort in up front and make the system work as you need it. As the quote said, you might get lucky, but it is better to be exact. It is better to be prepared. Then you will be ready if you are lucky.

Is your practice planning a data migration or integration project, and relying on luck to get you through it? Don’t do it! Luck will only get you so far. It will probably not catch you a fish, and most certainly won’t lead to a successful data migration for your medical practice or physician organization. Migrating data from one electronic health records system to the next takes careful planning and testing. Rely on the data migration gurus at Mi7 to put you on the right path. We’re not lucky, we’re exact. And that’s the way (uh-huh uh-huh) we like it (uh-huh uh-huh).