Professor Alex Bottle joined the Dr Foster Unit in 2002 following an unusual route into statistics. Having originally studied medicine, a BSc in statistics and epidemiology fuelled his interest in healthcare data. An MSc in Statistics followed, with a year working for a drug company for practical experience. Alex then undertook a PhD in Epidemiology at Imperial College London where he is now a Professor of Medical Statistics. Prior to joining the Dr Foster Unit, he was seconded to the Harold Shipman public inquiry and asked to develop a statistical method for finding GPs with high death rates. He used this knowledge and devised the statistical methodology that underpins Dr Foster’s Healthcare Intelligence Portal. Here Alex explains how his experience is being used by the Dr Foster Unit and what this means for Dr Foster customers.
How does your experience benefit the Dr Foster Unit?
My main area of interest is research to measure and explain variations in the quality of healthcare. A lot of my time is spent with Dr Foster analysts acting as a sounding board, discussing how to tackle problems and the challenges of running various analyses. Providing expertise in statistics and data sets and applying them to health service research helps the Dr Foster analysts define the big questions that need answering.
We work together from idea conception to carrying out analysis and its final presentation at the end. There is a strong focus on helping non-statisticians understand the analysis process and communicating the outputs. The beginning of any project always considers how the end results will be presented and who the audience will be: academic, hospital specialist, NHS manager, GP or the general public.
Has Covid-19 had an impact on the direction of your work?
Recently we worked with hospitals to help them understand their Covid-19 admission and mortality data; for instance, we are looking at the impact of the pandemic and lockdown on A&E attendances in children and adults. Hospitals are also concerned about the impact Covid-19 has on their overall mortality rates, so it is important for them to be able to separate the Covid-19 from the non-Covid-19 data.
At first, results were presented in a certain way, which meant that Covid-19 admissions were being grouped together with other illnesses, like pneumonia. To enable hospitals to look solely at Covid-19 admissions and mortality we had to change the underlying database and how it is processed. Our starting point was to consider how hospitals would use this data and what they would want to see, rather than what was statistically easy to do or what made sense from a modelling point of view. This ensured we created and made available the most useful analysis to hospitals.
How does your work bring academic credibility to Dr Foster?
A good example of this is my work with GPs in Northwest London, using the Discover-NOW database to analyse patients with chronic conditions. Subsequently, I had two papers published on this dataset analysis: one descriptive and one an analysis of healthcare and social care use by patients with heart failure. Dr Foster customers can be assured that when we advise on data, analytics and the output created, the underlying methodology is credible because our work and methods are routinely published in peer reviewed journals and so are easily available and transparent. I have co-authored more than 250 such articles to date. At the Dr Foster Unit we feel this is important, and we know it is highly valued by Dr Foster.
This is a great distinguishing feature for them. Typically, commercial organisations keep their methods proprietary. As these methodology remains unpublished, no one else knows how they processed their data and how they ran their analysis, which can result in a question mark over how good it is.
What role will analytics and data will play in future healthcare?
There have been several changes in healthcare data analysis that are impacting how we work. Covid-19 is the big one as it has provided the catalyst to do things differently. Remote care through telemedicine and telemonitoring, for example, is fairly new, and its continued rapid rise is in part due to organisations’ responses to the pandemic. However, whilst these new options bring another layer of data into play, which could prove very useful in modelling care pathways, personal health data confidentiality remains an ongoing issue for many people, and there are sometimes safety concerns if the doctor only speaks to the patient by phone or via videocall.
I’m particularly interested in chronic diseases and anticipate finding a way to measure outcomes along the whole patient pathway. Another development is the greater availability of linked datasets, (linking GP records to hospital records, death certificates and increasingly social care records), as we have seen in NW London with Discover-NOW. Linked datasets will help to create a way of tracking patients over time to see the whole process or patient journey from first symptoms, to tests, appointments, hospitalisations, community support, follow-up, and longer-term outcomes. In terms of making a big impact on patients by enabling health and social care to work together better, it’s imperative that we are able to pull all those parts of the journey together.
Mapping a patient’s interactions with health services, providing context, and putting it all together statistically is challenging. However, using analytics and data to model the patient journey and identify areas for improvement can produce greater efficiency and improve outcomes for patients and the NHS, which is ultimately the driving force behind what I do.