Process mining helps healthcare professionals understand processes within healthcare. While often used in secondary care, there is little work in process mining using primary care data. Serious adverse events that result from hazardous prescribing are common and costly. For example, non-steroidal anti-inflammatory drugs (NSAIDs) and antiplatelets can cause gastro-intestinal bleeds (GiBs). Prescribing typically occurs during primary care; therefore we used this setting to attempt process mining. We extracted events (drug started, drug stopped, GiB) for understanding three prescribing pathways, and applied process mining. We found NSAIDs are often short-term prescriptions whereas antiplatelets are often long-term. This perhaps explains our finding that co-prescription of gastro-protection is more prevalent for antiplatelets than NSAIDs. We identified reasons why primary care data is harder to process mine and proposed solutions. Process mining primary care data is possible and likely useful for improving patient safety and reducing costs.
Keywords: Data mining; Patient safety; Primary care.