Title : Making our hospitals safer for surgical patients; Using improvement science, health services research and machine learning in acute urological surgery
Abstract:
Patient safety is a serious worldwide public health concern. Adverse events in the United Kingdom (UK) is estimated by the Department of Health to occur in 10% of hospital admissions, which is over 850 000 per year. The cost to the health service is estimated at £2 billion per year. In the UK, serious incidents (SIs) and never events include wrong site surgery, retained foreign objects post procedure, wrong implant/prosthesis and wrong route administration of medication. Fortunately, all serious incidents are reported in a systematic detailed manner and stored in the Strategic Executive Information System (StEIS). This repository, managed by National Health Service (NHS) Improvement, holds an enormous amount of data spanning the whole of the UK, over many years. This data is currently not being systematically reviewed to improve patient care, however we will describe how this data can be used to prioritise Health Service Research by using it to identify; patterns, comparative information, trends and hot spots on a national level and we will explain a methodology that can be used in countries with similar types of centralised databases. The aim is to allow changes applicable nationally to be designed and prioritised for the highest impact on patient safety. After implementing change, the longitudinal rate and contributing factors of serious incidents can be used to assess improvement using improvement science research and the StEIS database. We will use a detailed example of this, as proof of concept, where our team in London have used this repository and mined the data to evaluate one of the common acute paediatric emergency surgical conditions. As result over 1000 patients in the UK were discovered with SIs relating to this condition during a 4-5 year period. These SI reports then underwent thematic analysis with the aim of feeding into a ‘logic model’ and pragmatic trial evaluation of the impact quality improvement intervention can make on safer care outcomes in this condition. We will describe the methodology to process a high volume of information within the database using machine learning. Machine learning or artificial intelligence is based on data pattern analysis and is capable of processing overwhelming amounts of complex data. This can be applied to SIs reported in StEIS and be a huge leap forward for improvement research. It will identify common themes and enable us to strategically make changes to enhance patient safety within our nation’s Health Service.
