The client state, via its Department of Safety and Professional Services (DSPS), was developing a new enhanced Prescription Drug Monitoring Program as part of its efforts to reduce opioid addiction by preventing phenomenons like doctor shopping and overprescribing.
The ePDMP allows a healthcare provider to evaluate a patient's complete history of controlled substance prescriptions from any provider before writing a new prescription. It contained more than 45 million prescription records at launch, with millions more added every year.
The ePDMP's effectiveness would rest on its ability to provide complete and accurate controlled substance prescription histories for individuals. However, the data in the ePDMP is submitted by pharmacies and prescribers, so the state does not have the control to enforce quality standards at the source.
The ePDMP's data suffered from typical data quality issues such as missing data, typos, and the evolution of a person's information over time: for example, name changes.
The common issue of a person using different forms of a name at different times was also a concern—particularly because patients with a concerning history of opioid usage may be actively trying to avoid detection by the ePDMP by using such variations frequently.
We created a data quality automation solution that is a critical part of the ePDMP, establishing a gateway data quality threshold and using sophisticated data matching to identify individual patients from non-identical data.
As part of the gateway data quality, the solution created checks the data submitted by pharmacies and prescribers for adherence to the required format, valid data in selected fields, and record completeness. Records that pass the checks are ingested, while records that fail can be corrected online. If a file has too many failing records, the entire file is returned to the sender and must be resubmitted.
Using our expertise in unsupervised machine learning and scenario-based match rules, all records related to the same patient are automatically brought together. The result is truly effective patient matching that uses a relatively small number of rules to address thousands of potential scenarios.
With individual patients identified, our solution uses sophisticated rules to identify concerning histories with controlled substances. It proactively alerts healthcare providers when any of their patients develop a concerning history, enabling intervention as soon as possible.
API hooks we built in allow the client to retrieve everything they need from our solution’s database for the front-end application they built to enable medical professionals, law enforcement, and others to easily interact with the ePDMP.
Finally, controlled substance monitoring is subject to a large amount of governmental oversight and requirements. Our solution generates the complex reports that state and federal agencies require.
With MIO, the state’s ePDMP automatically enforces basic data quality requirements before the data is accepted. As a result, the state’s DSPS does not have the administrative burden of following up with individual submitters regarding unusable data.
MIO’s solution also empowers the ePDMP to locate records about the same individual even when data is inconsistent, incomplete, and changing over time. This allows the ePDMP's users to make data-based decisions based on their patients' complete history of controlled substance usage in the state.
Automated alerts are also sent proactively to relevant parties when a patient’s history exhibits concerning characteristics.
As the ePDMP has been in operational use, DSPS has identified additional needs for the ePDMP's behavior, some in response to new legislative requirements. With its centralized approach to data quality, monitoring, and reporting, MIO experts have been able to respond quickly to modify the solution to meet new needs.
