LCP Health analytics team share highlights from the R for HTA annual workshop, June 2023 

LCP Health Analytics highlights from the R for HTA annual workshop, June 2023

R for Health Technology Assessment (HTA) is an academic consortium whose aim is to explore and advocate the use of the statistical software R for cost-effectiveness analysis. Its annual workshop, first held in 2018, is an opportunity for representatives from academia, industry, and government institutions to meet to discuss latest developments. Dr Katja Grasic (Health Economist) and Dr Mei Chan (Senior Statistician) represented LCP Health Analytics at the 2023 workshop in June and share their reflections on the event.

Why did you choose to attend R for HTA this year?

Katja: My background is in health policy research, and I recognized the importance of expanding my skillset to include R programming and of staying up to date with the latest trends in HTA. Having completed my PhD in York and having worked for many years at the Centre for Health Economics there, the event also provided an excellent opportunity for me to reconnect with my academic and professional networks.

Mei: I was fortunate to attend and speak at last year’s R for HTA workshop in Oxford. The workshop was a great opportunity for me (with a background in biostatistics and epidemiology) to explore the translation of my disease trajectory modelling experience in R to health economic contexts. In fact, I discussed potential uses of a biomarker-based measure of biological age (that I developed during my PhD at Oxford) in health economic modelling in my talk. This year, I was keen to hear the latest views on the use of R for HTA from stakeholders and disseminate learnings to the team.

What were your main takeaways from the workshop?

Katja: One aspect that stood out for me was the focus on survival analysis. The discussions surrounding this topic were particularly engaging and highlighted the crucial role of survival analysis in understanding long-term outcomes and estimating the potential benefits of different treatments. Overall, the workshop emphasised the shift towards using R for data analysis instead of relying solely on software like Excel. I have personally experienced the limitations of Excel in terms of traceability and potential for errors. The workshop showcased how R can enhance reproducibility, facilitate version control and provide a more efficient platform for conducting complex statistical analyses.

Mei: Aside from survival analysis, strategic discussions around the latest developments in large language models (LLMs), such as ChatGPT, also stood out to me. It was suggested that these tools could make it easier for sponsors to present analyses in the most suitable programming language and for agencies to accept languages outside their core capabilities. ChatGPT is currently unreliable in supporting the development of code-based models, but it has the potential to make development of these models even more efficient relative to Excel models.

How are you hoping to apply what you learned at the R for HTA workshop in your day-to-day work?

Katja: I have already implemented code from the University of Manchester to download, organise and analyse frequently used datasets, such as NHS Reference Costs and ONS demographic data. I also look forward to using a UKHSA tool in development for currency conversion which uses a combination of Purchasing Power Parity (PPP) indices, GDP deflators and official currency exchange rates. Finally, one particular Shiny app that caught my attention was SurvInt, which is a simple tool from the University of Warwick, designed to obtain parametric survival extrapolations from only a couple of data points.

Mei: I have already initiated discussions with the Health Analytics team at LCP on the novel modelling techniques and applications of LLMs that we mentioned. I also plan to spread the word about the R for HTA community’s plans to hold a hackathon later this year, where participants will work together on open-source health economic and decision science projects.

Was there anything missing from the event that you’d like to see included next year?

Katja: Overall, I found the conference to be comprehensive in its coverage. If I were to suggest an addition to next year's event, it would be incorporating sessions on modelling techniques related to generating Real World Evidence (RWE) using R. This could encompass topics such as causal inference and big data.

Mei: In addition to more content related to RWE, there have also been recent advances in modelling techniques that could be captured in future sessions, for example assessing and improving fairness in machine learning. This is of particular relevance given that HTA bodies such as NICE and ICER are increasingly considering health inequalities in their work.

Any final thoughts?

Katja: I’m grateful to have had the opportunity to attend the R for HTA workshop. The conference provided a wealth of knowledge, networking opportunities, and practical insights that will undoubtedly have a positive impact on my work in health policy outcomes and economic evaluation.

Mei: Having attended the R for HTA workshops twice now I’m keen to participate in future events and help advance the field with my ‘outsider’ perspective in biostatistics and actuarial science. The general consensus at the workshop was that advancement in using R to support HTAs has been held back by the slow pace in adoption and framework-setting by regulators, even though R-based analyses may be easier to validate if HTA bodies had the relevant resources. At LCP, we’ve seen first-hand the benefits of modelling in R. It enables us to develop complex analytical solutions and models that are more reflective of real-world health issues and our stakeholders’ analytical needs. I hope to see more dialogue and developments in this area in the near future.