English
Immune-mediated inflammatory diseases (IMIDs), which include diseases affecting the skin (psoriasis, eczema, and lupus) and joints (rheumatoid arthritis and lupus), lead to a substantial physical, psychosocial, and economic burden for up to 7% of people globally. New immunomodulators, which are medicines that change the immune system, have transformed the lives of people with IMIDs. These are mostly high-cost treatments that target proteins that control inflammation. Patients with IMIDs and their clinicians rely on evidence to guide their decisions to start an immunomodulator. This includes information from randomized controlled trials (RCTs), where an experimental drug is given randomly to a group of participants and another group of participants is given a comparator drug or dummy medicine; and observational studies, where large groups of people are observed in routine settings without active intervention, and these studies have complementary strengths and limitations. Observational studies are representative and large, but can be biased if not conducted robustly, at worst introducing patient harm. Conversely, RCTs are reliable but there is an under-representation of people from minority populations; it is therefore unclear from RCTs whether drugs work the same in these populations and this likely leads to an over-estimation of benefit and under-estimation of risk for the treatment overall. The aim of this fellowship is to use data from both RCTs and observational studies together to look for ways to make future studies better. Firstly, I will use results from RCTs, which often have high reliability, as a benchmark to conduct analyses of observational studies to evaluate treatment effects. I will do this by using data from IMID patient registries, which are studies that follow people with these conditions over time. By mimicking the design of RCTs in this data and using RCTs as a reference, I can understand if observational studies can give reliable estimate of treatment effect and identify situations where they should not be used. Secondly, I will test treatment effects in subgroups based on measured participant characteristics in both trial and observational data, e.g. male and female participants, giving me a much bigger sample than if I relied on one type of study. If I find that the treatment effect is different between the subgroups, and this is consistent in both the trial and observational data, these subgroup factors could then help stratify treatment for personalised clinical decision-making. Thirdly, I will use information about the UK patient population (for example the distribution between male and female patients) and "weight" the trial populations to mimic the UK patient population. This will help identify any differences in our overall understanding of the treatment outcomes if the trial was conducted in the UK patient population instead of the strictly controlled trial population. Fourthly, I will take the learnings from the analyses above and apply them to use registry data to predict the results from ongoing trials and test how closely our results can match the results from the RCT. This study will help us understand whether observational studies can produce reliable treatment effects in IMIDs; compare factors affecting treatment effects across different IMIDs; understand treatment effects in trial-excluded populations; and identify clinical subgroup effects that may help personalise treatment, leading to more efficient and appropriate use of high-cost immunomodulators to treat people suffering from a group of life-ruining conditions.
