Utilising incident data in support of healthcare quality improvement

Abstract

Rapid developments in the scale of healthcare provision have been associated with increases in the amount of healthcare-related harm. Root Cause Analysis (RCA) was developed in high-reliability industries as a method for reducing risk by systematically identifying, analysing, and preventing serious problems. It was subsequently introduced into healthcare organisations in the UK to investigate harmful incidents (referred to as serious incidents) though its overall value and impact have since been debated. In this thesis, RCA is recognised as making a valuable contribution to care quality and safety. However, the disaggregating effect of RCA methodology, which leads to a focus on individual incident review rather than aggregated or case series review, is called into question. In addition, the logic of causality that underpins RCA assumes that incidents can be reductively studied and known and that similarly the state of the system they occur within can be described in sufficient detail for recommendations to be made about how similar incidents might be avoided in the future. This thesis argues that, in healthcare, sector-wide reliance on RCA means that complex patterns of causality that exist within and between case series are not being systematically explored and that methodological innovation is therefore required.

This thesis proposes two methodological strategies for mitigating the disaggregating effect of RCA on serious incident investigations in the hospital setting. In order to develop these strategies, four years of RCA incident data (167 cases relating to care failure and falls resulting in serious harm) from within a single healthcare organisation were analysed through the development and application of a mixed methods research design. This incorporated; a) development of a novel taxonomic framework to code free-text incident data relating to healthcare and, b) development of a novel case-based analytical method formulated through combining principles of Cluster Analysis and Qualitative Comparative Analysis. This mixed method approach was used to aggregate case series data and triangulate causal patterns in order to produce socially useful knowledge for engaging healthcare teams, leaders and safety professionals. Complexity theory has informed the methods developed and been used as a framework that recognises each of the cases studied as mutually constitutive components of a wider system, and that sensitises the research to the limits of modelling in complex systems.

The methods developed have produced findings that are potentially useable by and useful to healthcare organisations. The quantitative aspect of this research identified that three deficits in the function of the system are simultaneously needed for the protective threshold to be breached and for an incident to occur (there were no examples of case clusters where fewer than three deficits were observed). The qualitative aspect identified the near-universal finding that, in the assessment of patients: a) deficits in team function and b) deficits in treatment delivery are the most common combination of conditions in which harm occurs. In the case of falls resulting in serious harm, the environment of care is identified as an additional commonly observed deficit. Overall, the case series clusters demonstrate significant between and within-series cluster heterogeneity, reflecting the inherent complexity of the healthcare system that they represent.

Causality does not emerge from this research in the form that those seeking root causes might anticipate or easily recognise. The pursuit of parsimony that RCA methods promote is important but may lead to over-emphasis of single cases or explanations of causality that do not sufficiently account for the complexity of the system in question. Looking across systems and systemic characteristics through the clustering of cases permits an exploration of complex causality. In this thesis, heuristic causality (HC) is proposed as a complement to the pursuit of parsimonious explanations of serious incidents. HC acknowledges multiple simultaneous interpretations of case series data and seeks to triangulate these using the notion of complex causality and through the application of the quantitative and qualitative methods presented. Importantly, it should be noted that via these methods, heuristic explanations of commonalities between serious incidents, and the data these explanations are founded on, can be made available to healthcare teams, supporting cross-incident enquiry through reflective organisational practice that may have system strengthening effects.

Date of Award	Nov 2023
Original language	English
Awarding Institution	University of Brighton
Supervisor	Phil Haynes (Supervisor) & Mary Darking (Supervisor)