Prescribing opioids correctly requires close attention to clinical guidelines and Washington state rules in order to maximize patient benefit and minimize harm. Our team at the University of Washington has developed and implemented a new tool that health organizations and practices can use to access the Washington State PDMP database. The COSRI system can either be integrated into the Electronic Health Record (EHR) or run as a freestanding application. COSRI provides easy access to patient prescription histories in the PDMP as well as integrated clinical decision support with Centers for Disease Control guidelines and Washington state rules.

Since September 2021, COSRI has been in use at two non-academic clinical sites. We work closely with these sites to provide continuous, iterative system development based on user experience and to conduct short- and longer-term evaluation of COSRI.

Responsive, effective clinical application development requires a unique set of engagement and resources. The COSRI timeline below both reflects our experiences and planning and serves as an example of the commitment required for the development successful electronic clinical support tools.

Initial Development

Between October 1, 2019 and September 30, 2020 the Clinical Informatics Research Group (CIRG) at the University of Washington and the Washington (WA) State Department of Health (DoH) partnered to develop novel, standards-based informatics tools and methods to facilitate clinician access and use of the Prescription Drug Monitoring Program (PDMP) database. As one objective of this partnership CIRG at the University of Washington (UW) developed a clinical decision support (CDS) tool known as Clinical Opioid Summary with Rx Integration (COSRI). COSRI was built to provide CDS messages, alerts, and data visualization to help providers to more easily follow Washington State and CDC opioid prescribing guidelines.

User engagement and initial participatory design interviews were facilitated by introductions through the UW’s WWAMI region Practice and Research Network (WPRN) to providers across WA state. These interviews indicated that providers had a strong interest in testing of COSRI and a need for this type of clinical support.

Use Case Development

Starting in March 2021 CIRG identified two clinical partners and began detailed user engagement and experience assessments in order to model the initial COSRI application into a testable application that could be evaluated for use across two clinical care settings in WA State. This included:

1. User engagement interviews
2. Iterative user experience focus groups and interviews with respect to application utility and features
3. Development of new technical features including:
   • ability to launch from outside of an EHR
   • real time query integration with the PDMP database
   • inclusion of CQL logic rules for WA State
   • inclusion of CQL logic for morphine equivalent dose (MED)
   • Conversion from NDC codes to RxNORM for reasoning, using NLM RxNAV tool
   • Extension of SMART on FHIR launch security tokens to provide user level authentication between system components
   • User alerts for unreliable and/or unavailable data
   • Tailoring of MED history summative graphs to demonstrate current and historical MED based on dispensed prescription
   • Suppression of summative graphs when MME cannot be calculated
   • secure file upload and integration of clinical data from the EHR
   • sorting and searching for patient data for PDMP database calls
   • classification of types of controlled substance medications to allow for easier searching
   • Provision of all controlled substance prescribing medications with the ability to search, sort and filter in order to help providers explore patterns of controlled substance use in their patients for appropriate discussion, treatment and decision support when prescribing opioids
   • CDS alerts for prescribing risks including age > 65 years and co-prescribing of opioids, sedatives, benzodiazapines, carisoprodol, and barbiturates
   • urine drug screen order dates
   • dates for controlled substance prescribing agreements signed with patients
   • usage and event logging features to ensure data quality and system functionally, and provide high quality user support
   • inclusion of a security model for protecting all services
4. Extensive testing and validation of application components including, but not limited to:
• Development of hundreds of tailored mock patient xml PDMP messages in order to test features and functionality
• Validation of the MED calculator based on 250 mock patient xmls and >4000 opioid medication NDC
• Validation of alerts correctly triggered
• Validation of error messages correctly triggered
• Validation of correct assignment of drug class
• Extensive authentication testing, including introduction of known authentication problems
• Intensive use of the system across many test patients in order to capture unexpected errors
• activity and inactivity testing with time lapse recording to determine if system time outs we operating as expected

Clinical Testing and Evaluation

Beginning on September 14, 2021 we implemented the COSRI application into our first clinical test setting in WA State, followed by implementation at our second site on October 6, 2021. Both sites provide primary care to the local community and care for a large number of patients who receive controlled substances.

Since implementation, all care providers at both sites have been able to access PDMP data through COSRI, with high volume use. Initial feedback, suggests that it is fast and easy to use. Minor authentication problems have been the only reported issues thus far.

We are collecting data for robust evaluation of COSRI including:

1. User experience via a series of qualitative interviews and record of service desk requests and user initiated issues to determine utility, satisfaction and acceptance by clinical care providers.
2. System usage via system audit logs to determine both the frequency and volume of use.
3. System performance via system audit logs to determine number of errors encountered.
4. Value of support materials and other features via system audit logs to determine the utility of specific features included in the system.
5. MME calculator performance via audit logs to determine accuracy of key information provided to the end users.
6. Numbers of alert notifications via system audit logs to determine possible issues averted.
7. Changes in prescribing 6 months pre and 6 months post-implementation via clinic electronic health record data and PDMP returns to determine if prescribing behaviors changed after COSRI implementations.
8. Changes in expediency of urine drug screens and opioid contracts 6 months pre and 6 months post-implementation via electronic health record data presented within the COSRI system.
9. Change in PDMP usage 6 months pre and 6 months post-implementation via audit of PDMP access and self-reported access in qualitative interviews.