Meet the Scientists: Dr Alice Richardson

By Sasha Nimmo

Dr Richardson taught undergraduate statistics for two decades at the University of Canberra, and collaborated on a variety of quantitative research projects. In 2016 she took up the role of biostatistician at the National Centre for Epidemiology & Population Health at the ANU. Her research interests are in linear models; robust statistics; statistical properties of data mining methods; statistical methods applied to large data sets in biomedical sciences; and statistics education.

Dr Richardson recently published ‘Weighting of orthostatic intolerance time measurements with standing difficulty score stratifies ME/CFS symptom severity and analyte detection‘ in the Journal of Translational Medicine, April 2018. Read about it in ‘How much can people with ME stand?‘.

What are you working on at the moment?
I’ll tell you about three projects that capture the breadth of what I do. My ME/CFS work is focused on refining the Weighted Standing Time to improve its performance as a measure of symptom severity. One of my projects with other researchers in Population Health is looking at the effects of traumatic events on the mental health residents of war-torn areas such as Kashmir. And in statistics theory, I’m working on how to fill the gaps in datasets when there are missing values, particularly when the data has a very complex structure.

Why is the role of biostatisticians important?
Biostatisticians have a really important contribution to make to advancing knowledge in all sorts of areas. Our skills in design of research studies, data collection, analysis and visualisation of results make us an asset to a project from the very beginning to the very end.

How is technology impacting your work?
Technology impacts on my work in a range of ways. My favourite computer software, called R, releases an update every 6 months or so, meaning I’m constantly upgrading it. I’m also being introduced to new software packages all the time. Technology also impacts upon the type of data I get to analyse, as gene sequencing and wearable devices generate a deluge of data bigger and faster than even five to ten years ago.

Your colleague, Associate Prof Brett Lidbury described your work as ‘dimension reduction and hypothesis generation’ rather than hypothesis testing. Does this reflect the complexity of ME or the way that biostatisticians work?
It reflects the way medical research goes, I think, as the process from dimension reduction to hypothesis generation then hypothesis testing can apply across the board. The size of the gaps in knowledge around ME mean that hypothesis generation is an important step to help fill those gaps. A biostatistician’s toolkit contains graphical and numerical ways to reduce the complexity of data sets and descriptive ways to identify what hypotheses might be useful to test in the next round of analysis.

How did you become involved in studying ME?
ME came to me rather than the other way round – I was approached a number of years ago by Brett Lidbury to collaborate on the statistical analysis of biomarker data associated with infectious diseases, and that collaboration extended over time from the infectious disease to more complex conditions such as ME.

What appeals to you about studying ME?
I like that ME is a complex condition, and that answers to its diagnosis and treatment are not going to be easy. The sense of satisfaction making an advance in this area is going to be immense!

Your recent paper measures how long and how difficult it is for people with ME to stand. Does your work have an impact right now, to help clinicians measure degree of illness severity and to provide evidence of disability?
Absolutely this measure can (and is) having an impact right now. It’s in use in one of the biggest CFS clinics in Australia already. The paper we published about the weighted standing time measure is getting plenty of hits online, and we’d like to encourage clinicians and researchers who use our measure to let us know, so we can track its usefulness and validate the measure.

The paper also supports earlier research that found Activin B is a potential biomarker for ME. How important are these results, considering that in Australia less than a third of people diagnosed with chronic fatigue syndrome by GPs meet the criteria for ME (Johnston et al 2016) and it can take years for a diagnosis?
Identifying a biomarker in the form of the cytokine activin B cytokine that as having a potential correlation with the presence of CFS symptoms is very important. It provides a starting point for the investigation of pathways from the biomarker to the development of chronic fatigue, and it provides a clearcut direction for GPs to look when faced with attempting to diagnose a patient with suspected CFS. Results like this should therefore take years off the time taken to diagnosis.

Diagnosis is a difficult area in Australia. The 2002 Australian chronic fatigue syndrome guidelines 2002 say it is an ‘arbitary’ diagnosis; ‘physical examination of people with CFS is normal’; there is ‘the most difficult diagnostic uncertainty between CFS and psychological illness’; and it is not a specific delineated entity. This is very different from the definition used in your studies: the Canadian Consensus Criteria, an earlier version of the International Consensus Criteria which ‘removes ME patients from the broader CFS category’ to select a ‘homogenous patient set that can be studied to identify biopathological mechanisms, biomarkers and disease process specific to that patient set’.

How important is criteria to your work (for example, does the data do a large part of the identification for you, or is the CCC still identifying a few subgroups?)
The Canadian Consensus Criteria are an important part of the work I do, because they provide a quantitative way to check whether the suite of symptoms observed in a patient are sufficient for a CFS diagnosis to be made. The ICC represent an important update to the CCC and provide clear decision rules around the number of symptoms observed across different categories that are required for a diagnosis of the more specific condition, ME.

I think the criteria both have the potential to identify further subgroups in the relevant conditions. Whilst my work to date has mostly been using the Weighed Standing Time to identify subgroups, the richness of the CCC (close to 50 questions are included!) and related tools means that they contain information that can be further extracted and analysed to identify interesting subgroups.

You are excited about the ANU’s Grand Challenges winner, personalising medical technologies for diabetes and Multiple Sclerosis. Could this benefit people with ME?
Both diabetes and MS are complex conditions with symptoms that vary from person to person and from day to day. Some of the technologies being explored in the Grand Challenge include wearable devices, apps and other cutting-edge data collection tools that could be adapted for use by people with ME.

You have twenty years’ of teaching experience, what is the most important thing people should know about statistics?
The most important thing people should know about statistics is that it is all about variation while mathematics is all about similarity and patterns. Statistics touches every corner of human existence, wherever data can be collected and wherever people are curious about how a random process behaves when it is repeated. I love the look on students’ faces when they “get” this idea of variation and start to ask really good questions of the world around them as a result.

You recently spoke at a Biostatistics conference about your latest paper on weighted standing in ME, what surprised you about their response?
Not all biostatisticians are aware of the impact of ME and they were really struck by the statistics I presented on the number of Australians with ME, and the impact it has on the economy as well as on lives.

Is the ANU, and Australia, a good place to be a researcher?
I moved to Canberra from Wellington in New Zealand (one capital city to another!) as a post-graduate student nearly 30 years ago, and I’ve been really satisfied with the move. ANU is a great institution in a really liveable city. The access to health policy makers as well as medical and statistical researchers is a huge bonus.

What are you most proud of in your career so far?
Right at the end of my teaching career at the University of Canberra, I presented a short course on introductory statistics to a group of Commonwealth public servants. I put all my experience in teaching statistical concepts to non-statisticians to work in that course. I used a careful focus on the language of statistics, small-group discussions, and real-world data sets to keep everyone interested. I even organised a walking tour to discover statistics in action all around the university campus, and a “Statistics in the Pub” session, with singing, on the last afternoon to debate statistical controversies. I’ve kept in touch with a number of the students and they have all benefitted hugely from the course.

What would you like to achieve in your career?
I’d like to be known as the go-to person for statistical analysis of health data because I’m passionate about the application of the statistical methods I know well to data that can improve people’s health and lives. I’d also like to be known as the collaborator who asks great questions and helps other researchers tease out their questions into problems with elegant statistical solutions.

What do you enjoy outside work?
I love singing in choirs! I get a real thrill out of performing on stage with a big group. I also direct the choir in my workplace (the National Centre for Epidemiology and Population Health) which will be performing at the 30th anniversary of NCEPH in October 2018.

See the full list of Dr Richardson’s publications and blog ‘Alice in Statistics Land‘.