Health data is changing. Now, big data promises to breathe new life into clinical trial design. So, does this mean the end for placebos?
The gold standard for testing interventions has long been randomised placebo-controlled clinical trials. This involves volunteers being randomly assigned to either the test group (receiving the experimental intervention) or the control group (receiving a placebo or standard care). A placebo is a substance or treatment that has no therapeutic value. They are often used in clinical trials to test the effectiveness of treatments. The first known placebo trial took place in 1863. American doctor Austin Flint compared a treatment for rheumatic fever with a placebo derived from a plant extract.
The 1964 version of the Declaration of Helsinki considered it unethical to give placebos if there was an established treatment. However, the revised declaration in 2002 stated that placebo-controlled trials are ethically acceptable even when a proven therapy is available. This is the case when there are compelling and scientifically sound methodological reasons. However, the exact definition of this was not made clear. Currently, most clinical trials, where possible, will involve an ‘active’ control, where the new therapy is compared to the current standard of care available, rather than just a placebo. While cancer trials now avoid placebos, in other areas of medicine, particularly mental health disorders, placebos remain commonly used.
The use of placebos can impose recruitment and retention challenges that often halt the progression of trials. This is because patients are often less willing to participate. The use of active controls was set to address some of these ethical and logistical challenges. However, the field of oncology is rapidly progressing. Therefore, it is not usual for the standard of care treatment to be updated during the course of the trial. This in itself raises ethical questions. Elsewhere, for rare diseases, it can often be difficult to determine what should be used as an active control as there are often no established treatments in these areas. In turn, many clinical trials on rare disease are conducted with very few patients. This translates to insufficient statistical power.
In recent years, a potential solution for these ethical concerns regarding placebos has emerged. Synthetic control methods are statistical methods that can be used to evaluate the comparative effectiveness of an intervention using external control data. This data comes from sources including historical control data and real-world data (e.g. doctors’ notes). The digitisation of health records has meant that real world evidence could be used to create a synthetic control arm to replace clinical trials’ existing control arms. For example, in 2017, the FDA approved cerliponase alfa for a specific form of Batten disease, based on a synthetic control study. The study compared data of 22 patients from a single-arm trial versus independent external control group data with 42 untreated patients.
This shift in paradigm has caused a lot of scepticism amongst the scientific community. Nevertheless, it is likely that a number of clinical trials will start using external data as a synthetic control. To evaluate synthetic control clinical trials, it is critical that researchers assess the external data that is used and also the statistical methods. Therefore, establishing criteria on how they should be assessed is vital to make progress in this space.
It is important to note that these methods do not constitute a complete fix. They do not provide a solution for all the challenges faced by randomised trials and they also do not realise the full promise of real-world evidence in drug development. However, in certain situations, these methods can provide a better option. From an ethical standpoint, this is specifically relevant in cases that would avoid people being placed into a placebo group. In addition, with ongoing improvements in medical record collection and statistical methodologies, it is likely that synthetic control methods will be increasingly referenced and utilised within literature.
Image credit: By freepik – www.freepik.com