Efficient RCTs with real world data for benefit assessments of drugs: accelerating evidence-based care

Accelerated drug approvals mean that more and more drugs are reaching the market based on sparse study data. According to a mantra promoted in recent years, further data can be obtained at a later stage during use in routine practice, as so-called real world data (RWD).

The US Food and Drug Administration (FDA) defines RWD as “data relating to patient health status and/or the delivery of health care routinely collected from a variety of sources” such as observational studies, but also from randomised controlled trials (RCTs). However, the European Medicines Agency (EMA) defines RWD differently, as “routinely collected data relating to a patient’s health status or the delivery of health care from a variety of sources other than traditional clinical trials”. This seems to exclude RCTs. RCTs are explicitly not considered in the Data Analysis and Real World Interrogation Network (DARWIN EU), the EMA's main tool for providing RWD.

The authors of the BMJ article argue that while observational data may be useful for other purposes, they are generally inappropriate for examining the effects of drugs in benefit assessments. One can never be sure whether an observed difference in health outcomes between two groups of patients is causally related to treatment or rather to unknown differences in patient characteristics. Even known confounders often cannot be eliminated, because many RWD sources do not include the necessary information.

So why not integrate RWD into the gold standard, RCTs? RCTs have a reputation for being overly complex, burdensome and therefore expensive, and not feasible in practice for many research questions. This has become a self-fulfilling prophecy, especially for questions with small study populations: Researchers who keep hearing that RCTs are only trouble are more likely to set up non-RCTs – and thus further reduce the chances that an RCT on a comparable question will be able to recruit enough participants.

A toolbox for efficient RCTs

The authors therefore present 15 approaches to make RCTs easier, faster and cheaper to conduct, even in small populations. The more standardised patient registries there are and the better the European study infrastructure is developed and standardised, the easier it will be to recruit participants. Multinational studies are useful for recruiting enough patients quickly. Platform studies can be used to compare several new drugs with the same control group. Involving patients in the study design ensures that patient-relevant evidence is generated. To maximise knowledge, individual patient data should always be made available to the EMA and clinical study reports should be published quickly.

Finally, the waste of resources can be avoided by designing clinical trials in such a way that their data can be used both for regulatory approval and for benefit assessments or health economic evaluations. Beate Wieseler, Head of IQWiG's Drug Assessment Department and first author of the article, notes: "Studies that are completely focused on rapid approval are inefficient if the drugs subsequently fail at the fourth hurdle or if completely new studies have to be started for health technology assessment. If you really want to accelerate patient access to evidence-based care, you should start conducting robust 'multi-purpose studies' straight away."

This would require an expansion of the EMA's definition of RWD: "The categorical exclusion of RCTs is outdated. Europe should follow the example of the FDA. Let's hope that DARWIN will be amended accordingly."