Background Incident reporting systems in healthcare are historically based on staff descriptions of adverse events. An increasing body of literature suggests patients provide critical insights to risk and error, but their potential has not sufficiently been investigated at the incident level. This study aims to examine to what extent patient complaints and staff incident reports discuss identical incidents, and how their perspectives could be integrated for more comprehensive safety analysis.

Methods Deterministic data linkage was performed on all complaints (n=5,265) and staff incident reports (‘PSIs’) (n=81,077) between April 2014 and March 2019 at a multisite hospital in London. A total of 402 complaints covered at least one incident also identified in the PSIs, and were included in the study. All incidents reported in complaints and staff incident reports were codified based on problem domain; problem severity; stage of care; staff group implicated; reported harm; and descriptive level (eg, description of human factors and root causes); adapted from the Healthcare Complaints Analysis Tool (HCAT) and the National Reporting and Learning System (NRLS). Aggregated coding outputs informed targeted qualitative analysis of free text incident reports for an in-depth exploration of key overlap and discrepancies in patient and staff descriptions of unsafe care.

Results Our preliminary results indicate staff and patients reported similar problem themes for 81.1% of overlapping incidents (of which 66.5% clinical, followed by 27.1% institutional, and 6.4% relational), but commonly differed in their description of contributing factors and root causes (eg, different time points in patient journey). Alongside overlapping incidents, patients reported an average of 1.4 additional incidents in their complaint, of which 23.6% were high severity. Additional patient-reported incidents included blind spot clinical issues (36.7%; eg care continuity; care omissions) or relationship issues pre- or post mutually identified incidents (39.1%; eg failure to listen to patient concerns; breach of candour).

Conclusion Our study suggests that traditional, clinician-based models of safety and risk are likely to omit critical dimensions of root causes to adverse events. Patients and public are able to contribute to safety monitoring and evaluation in two ways: by highlighting overlooked or undervalued aspects of unsafe care (eg failure to listen; absent communication) and by revealing latent causes of incidents across time (eg failed continuity of care; systemic care omissions).