DISCUSSION:
A pre-eminent feature of safety critical organisations is the “implementation of highly structured approaches to safety management” such that they are “proactively identifying, assessing, mitigating and monitoring risk”. A variety of human factors methods exist to explore and analyse work systems and processes. Some have been developed in other settings and some adapted or specifically designed for healthcare. This study has used SEIPS to analyse safety risks during the conduct of an experimental medicine study in an academic CRF because it was designed specifically for healthcare contexts and has been adopted for wider use in the NHS.
The analysis of work as a human endeavour has been the subject of studies in social and engineering sciences over the past 70 years . ‘Work as Done’ commonly differs from ‘Work as Imagined’ (see Figure 1) and that discrepancy increases as individuals become more distant from the actual work environment spatially, temporally and experientially. Problems arise when managers or policy makers, or in the case of clinical research, those designing studies, make assumptions about activity and formulate protocols and guidelines which describe how work should be performed, without absolute certainty that what one imagines is achievable will actually be deliverable. This inevitably leads to rule-breaking by the humans undertaking the tasks in order to get the work done.
Whilst the safety of study participants was evidently paramount to the staff of both the facility hosting the observed study (OxCRF) and those who designed and conducted it (COV-CHIM study team), we found that the use of structured observations by individuals trained in human factors methodology recognised latent risks in the protocol as written, the CRF facility itself and the interaction between the two, that had not been identified a priori by standard peer review, institutional, ethical or sponsorship appraisal. In addition, confusion in, or deviations from, expected practice (often unavoidable) and the development of local workarounds was catalogued: behaviour that study and facility leadership were unaware of via conventional pathways. Use of the SEIPS PETT scan aided the design of recommendations to rectify or mitigate these risks by the multidisciplinary team and their prioritisation for implementation based on the established hierarchy of effectiveness of corrective actions in which physical interventions (e.g. pathway or equipment redesign) are considered most effective; procedural interventions (e.g. automation or use of checklists) are considered moderately effective, and person-based interventions (e.g. warnings or training) are considered weak.
Given the single-centre, single-study basis of our work it is inevitable that the specific findings described here will not be wholly generalisable to other facilities and research programmes. However, this was not the intent of the study. Instead we sought to understand whether the use of human factors methods could be extended to early phase and experimental medicine research with meaningful, actionable results to improve participant and staff safety. Our experience supports this assertion but clearly requires both extension and replication. Specific areas that warrant prioritisation due to their likely commonality across study type and relevance to multiple CRFs are discussed below.