Process mining for the analysis of a Rapid Referral Pathway of Colorectal cancer in an integrated healthcare area in Galicia (Spain).

1. Colorectal cancer (CRC) is one of the most common types of cancer and a rapid referral pathway (RRP) has been developed in Galicia in order to diminish the time of its diagnosis. Process Mining (PM) helps examining and analyzing the data logs from the electronic medical records. We used PM to study those patients finally diagnosed with CRC in an Integrated Area in Galicia along a 3-years period of time.

2. Structured variable data (timestamps and coded data for every activity and event) were extracted from healthcare databases. The moment of extraction of the pathological biopsy was taken as a reference for selecting patients. Process mining analysis was performed to compare processes according to whether or not they were included in the rapid diagnostic pathway.

3. From January 2016 to December 2018, we found 444 cases of CRC. 132 (29,7%) were included in the RRP and 312 were not included (70,3%). Referral to RRP came from Primary Care the 38% of events, 10% from the Emergency department and 52% from the Outpatient consultations. The median time from the colonoscopy request was 7 days in RRP vs 34 days in non-RRP cases. We identify two patterns: first, ""from outpatient consultation to colonoscopy"" is repeated 21% in the RRP, compared to 7% of the cases in the non-RRP; another one, “referral from Emergency department“ is the second repeated pattern of the RRP cases (12%), with 20 days of delay until biopsy extraction.

4. Process mining analysis makes it possible to describe the complete care process along the time axis without limitations based on levels of care. We have identified the most frequent usage patterns in both rapid and non-rapid referral pathways. It has also helped to know the real performance of the rapid diagnostic pathways, providing new information with the data registered in the digital electronic health information systems.

5. PM advantages in healthcare management are unbounded and applications are several. Most important one: it leads to better medical treatment, more effectiveness.

6. We easily identify performance patterns in healthcare assistance of CRC patients in an integrated area analysing structured data of Electronic Health Records with Process mining.

7. More long-term studies should be developed to determine the survival and quality of life of patients to determine the impact of PM in RRF in clinical practice and on healthcare related outcomes.

8. Next possible step could be the development of a digital algorithm in Electronic Health Records as a clinical decision support system.