Denise Fournier

Application Support Specialist

How often have you heard of data mix-ups between two similarly named members in a software system, or when finding a member in your system, the record is out of date?

Long ago, I remember hearing talk of making sure my dad, Cyril, and his sister, Catherine, didn’t both have accounts at the same store when they were younger and living in the same town, or their purchases would end up getting charged to each other’s account. This would have been the late 40’s to early 50’s, long before the advent of electronic record keeping. Now that everything is computerized and digitized, of course this can’t happen anymore, right?

But, as we all know, data mix-ups do still happen.  And somehow, now they seem even harder to detect and correct.

Currently, my work at TCS Healthcare is to help clients keep their member records accurate and up to date via electronic data loads.  This is still a universal challenge regardless of using a sophisticated software solution, a home grown software solution, or even when keeping track of records via Excel!

So, where do these “challenges” come from? One of the most common problems we run across involves changes to what we refer to as “matching info”. If you’re lucky, you have a single unique identifier that can be used to match up incoming records to previously created ones. In years past that may have been a social security number. But even that had issues – not everybody had one, duplicates did creep into the system, numbers got transposed – and so on.

Currently, the use of social security numbers as unique identifiers is slowly disappearing. So, if there is no other “unique identifier”, you have to rely on information such as name, birth date, gender, etc., all of which can be changed at any time for a variety of reasons. Especially difficult is the common task of recording newborn babies due to changes with the baby’s first name.

Of course, in managed care, there is usually some sort of member ID, which works just fine for most situations. But the same issues can occur that we see with social security numbers – duplicates, transposed numbers, family members with the same base number, etc.

Members with dual coverage, often under different payers, pose yet another dilemma. Detecting dual coverage members adds even more complexity to keeping data records accurate since there is no uniform identifier between different payers. Compounding the problem, data collection practices can vary significantly from payer to payer.

So, what can be done to keep your data as accurate as possible?

The obvious solution is a uniform identifier assigned to each person that NEVER changes regardless of which or how many payers cover that person. On paper, this is a nice idea, but it is probably not realistic.

In truth, there is no “best solution”. However, the following are some practices you can incorporate into your routine which can help to identify potential issues.

Incorporate safeguards into your process to detect and “exception out” bad data before it ends up in your system. For example, finding a member name that was “John Smith” yesterday compared to “Sandra Jones” today, might indicate a transposed member ID at some point in your process.

Monitor your exceptions and make corrections not just to your target but to your source as well. This means if you fix an error on a target system but never circle back and correct the source, that same error can and probably will recur the next time that record comes back through your data feed.

Don’t rely solely on error processing to catch everything. Make sure you also incorporate safeguards during processing to avoid inserting or updating bad data. For example, if a specific value needs to be unique, make sure that an insert is only attempting to add just one record with that value, and that value doesn’t already exist. The more complex things get, the more likely you are to run into unexpected scenarios, and the unexpected can cause a variety of issues like unintended duplicates.

Set up reasonable and strong matching rules to detect your dual covered members. Trying to find dual coverages by using JUST first and last name is NOT a good plan. Include more data items such as birth date, gender, social security number (if you can collect it), address, etc., which can help to detect more dual coverages. There will always be some records that simply don’t line up, so you’ll also want to have a process in place that allows you to mark your dual coverage records whenever they are discovered outside your normal process.

Overall, data maintenance is definitely a challenging business! And for those who have accepted that challenge, keeping that “bad” data at a minimum is an on-going process. However, adding in safeguards and consistent monitoring can help significantly in the “fight” for good member data.