Investigating the (mis)match between electronic health records and actual nursing work: An observational study

The process mining research domain uses process execution data to gain insights into work processes and has been applied in a wide variety of domains, including healthcare. The extensive use of electronic health record systems has made the data they capture a common input for process mining, yet data quality issues persist. While these issues are recognized in literature, empirical work examining the (mis)match between nursing interventions and their electronic health record registrations remains scarce. This study addresses this gap through an observational study.

A cross-sectional observational study was carried out between February 23 and April 10, 2024, covering 119.75 h of observation in the urology and neurology wards of a Belgian hospital. Data were collected on both nursing interventions and their electronic health record registrations.

The analysis revealed several mismatches between electronic health record registrations and actual nursing work: (i) 20.34 % of all observed intervention types were never recorded in the EHR, (ii) only 23.32 % of registered interventions were recorded without a time gap between execution and registration, and (iii) there is not always a one-to-one relationship between interventions and registrations.

This study underscores the importance of thorough data quality assessment when using routinely collected data for research or analysis. Beyond assessing the data itself, it highlights the need to understand real-world work processes and how data is recorded in supporting systems. Such insights enable the anticipation and potentially the mitigation of data quality issues prior to their actual use. These efforts are essential to determine how accurately the available data reflects real-world practices and thereby how trustworthy any conclusions based on this data can be.