RESEARCH

Effects of Backlog Maintenance on Patient Harm

Commenced in 2021, the research I am undertaking is to investigate the impact that healthcare infrastructure has on patient harm when it fails. During the study I will seek to understand the relationship between the environment and the patient, and will employ secondary data analysis and simulation modelling to understand the potential impacts on patient harm.

The PDF file adjacent contains a summary poster of the research aims and methodolgy

The Literature

In 2000 the Institute of Medicine in the U.S. published To Err Is Human. This seminal book highlighted that as many as 98,000 people in the United States of America die in any given year from unintended harm whilst under the care of the clinical profession. The report spawned thousands of academic papers, conference notes and books on the subject of patient harm and its many subsets. The chart below the World Health Organization classifications and sub-classifications to graphically categorise the papers reviewed during a literature review search on ‘patient harm’ in SCOPUS.

Of 2,708 papers classified, only 0.148% (n=4) of all the papers were related to infrastructure or buildings. This is in comparison to the 5.3% of all National Patient Safety Agency reports citing work and environmental factors as a direct contributing factor to patient incidents. What was common with the four papers classified as infrastructure, and with the majority of other papers on patient harm – they focus on the how, when and where of the primary cause-effect relationship, not why the incident occurred in the first place. It is through understanding why an incident happened, and what influenced the direct causes of the incident in the first place.

The Research

Based on the literature review searches undertaken and the paucity of prior research in the field, the research being undertaken is of mixed method in nature. The study will use both qualitative and quantitative methodologies in which to explore the research subject. The research will draw together three studies to enable triangulation of the findings to strengthen the veracity of the research:

- Secondary data analysis;

- Systems modelling;

- Expert Engagement

More details of the research will follow as the research project unfolds

STUDY 1 - Secondary Data Analysis

The initial phase of research is focused on the utilisation of secondary data drawn from the NPST. To date the NPST have recorded over 20 million incidents reported to the National Learning and Reporting System since its inception in 2003. This study will focus on the analysis of over six million records.

The research will be conducted across NHS England’s acute sector. The first step will require Count, Descriptive and Inferential analysis of the NRLS data. Following the initial review of the whole data set to attain a broad level of count and descriptive statistics from the data (age, gender, location profiles, etc.). The analysis will also use data from the ERIC data collections to understand whether there is any direct correlation between the number and types of incidents and the level of backlog maintenance across NHS England. Steps two to four will undertake qualitative analysis on the NRLS data:

Primary Classification Review. Initially, focusing on the primary incident category – Infrastructure will provide data on active errors where there is believed to be direct cause/effect relationship between the infrastructure and patient harm. The data will be subject to word frequency and thematic analysis to ascertain frequency of common words, sentences and themes used. This data set will lead to the development of a coding structure and created a foundational base for further topic modelling on the wider data set.

Sample Set of Secondary Classification Data. Using R to undertake text mining, a sample set of the whole data set will be taken to analyse the free text data to define a structure of analysis for the remaining free text data held within the contributing factors.

Full analysis of Secondary Classification data. The remaining records across the 10 contributing factors will then be analysed to determine the frequency rate that themes occur as secondary or latent factors to the primary contributing factors.

The results of the qualitative and quantitative analysis of the NRLS reports will be used to guide two further two studies.

STUDY 2 - Systems Modelling

Healthcare systems have been described as a ‘complex adaptive system’ meaning that looking at the individual facets of the system in isolation will not explain the system function. Given the increasing complexity of the healthcare system, the opportunity for error also increases. Furthermore, the systems within the estate infrastructure are intertwined with, affect and are affected by the clinical services provided within the care setting.

To explore how estate infrastructure interacts with the environment, patients and staff of a healthcare facility, Operations Research (OR) methodologies will be applied. Through using the Strategic OR techniques of simulation and modelling it is possible to capture the whole system and look at the interaction of the system over time.

By using simulation modelling it is possible to analyse component parts of large complex systems and understand the emergent behaviour of the system, or put more succinctly “structure determines behaviour” (Forrester, 1968).

Research Parameters

Following the interpretation of the secondary data analysis in phase one, asset failure will be subject to analysis by Causal Loop Diagrams.

The initial CLD will be developed with Vensim software, using several data sources. Through using the approach initially laid down by Gordon, Helmer, and Dalkey in the 1950’s (Woudenberg, 1991), each of the key variables would be subject to the application of the Delphi method. The initial problem will be mapped out through developing the causal loop diagram and then circulated to a panel of experts for comments and modification and, eventually, consensus. Once a level of consensus has been reached on the model, it will be reviewed to assess whether there is sufficient insight into the problem to understand how it impacts NHS policy.

STUDY 3 - Expert Engagement

While it is important to understand the ‘what’ and the ‘how’ estate infrastructure is harming patient outcomes, it is equally important, if not more so, to understand why the current level of harm is not affecting policy maker’s decisions, or at the very least those who influence the policy makers – the directors of estates and facilities. A cross-sectional survey questionnaire will be devised to understand senior management perceptions of backlog maintenance within the NHS. The questionnaire is directed at all directors of estates across the acute sector to triangulate the findings of both the NRLS secondary data results and the findings from the SD modelling. Due to the low numbers of estates leads within the research area, a sample size of 100% is sought.