Updated: Mar 24
Potential Duplicate, also known as a task or suspect, is a set of records that a match platform (IBM, Nextgate, Epic, Cerner, Intersystem Health, Informatica, Quadramed) has identified as a “possible” match but lacks the ability to make a definitive decision. The reason is due to the technology that the MDM/EMPI/MPI platform uses which is either a deterministic or probabilistic approach. These older approaches are linear which means that it cannot discern the more difficult match decisions and can only they throw up their hands and leave it to humans to make the final decision.
Are potential duplicates/tasks special? No, they are simply a match decision that couldn’t be made by their platform. But often they are viewed as something different or special which tends to confuse the issue. In addition, the number of potential duplicates/tasks are a byproduct of the platform's match accuracy. The less rigorous the match technology the more potential duplicates/tasks will be raised and vice-versa. But in either case these are simply possible matches that couldn’t render a final decision.
So, do potential duplicates/tasks matter to an organization? Absolutely as their presence directly affects match accuracy. The average MDM/EMPI/MPI platforms generate between 10-20% potential duplicates/tasks and the ONC has mandated that by 2020 healthcare organizations need to achieve a 0.5% duplication rate. This means that tasks do matter and for an organization to achieve the .5% duplication rate would need to be resolve this backlog as well as keep up with the newly generated potential duplicates/tasks.
Ignoring the tasks is a common approach but puts patient safety at risk and various studies such as PEW have found that a duplicate patient record costs $96 on average which is far greater than the cost to resolve the duplicate. This means that only 100 duplicates will cost an organization nearly $10,000. And this cost soars to $1900 for inpatient duplicates.
There is match technology that can resolve these potential duplicates/tasks at equivalent accuracy to a human data steward team and at much greater scale and consistency. This also frees up the team to work on other important work.