New medicines — urgent need to assess outcomes in special groups

Randomised controlled trials (RCTs) provide the most internally valid method of evaluating new pharmaceuticals and, appropriately, provide the core evidence for marketing approval. However, certain patient groups (eg, older people, young people, people with comorbidities, people taking concomitant medications) are typically excluded, and RCTs are often too small and their duration too short to detect uncommon but clinically important adverse drug reactions (ADRs). Consequently, the translation of these internally valid results to the real world of everyday clinical practice is not a trivial task.1

An example of this is highlighted by Baker and colleagues in this issue of the Journal.2 They discuss real-world problems involving serious bleeding events with the new oral anticoagulant dabigatran. These problems are notably in “special” groups excluded from clinical trials of the drug, such as people older than 75 years and people with cancer, liver or renal impairment, pregnancy, or other conditions that alter dabigatran’s pharmacokinetics and pharmacodynamics. Although the reported problems relate to toxicity, equally important is the lack of efficacy data for these groups.

The current difficulties with dabigatran exemplify a recurring problem with the translation of clinical research into practice. We suggest that several strategies, which are neither new nor radical, could feasibly improve this situation:

• Improve access to and linkage of routine data systems

• Increase emphasis in medical education that newly marketed pharmaceuticals require further assessment

• Broaden inclusion criteria for RCTs, so that patients from special groups are included.

First, a low-cost way to obtain useful (albeit incomplete) information on drug safety and effectiveness in the real world is through linkage of routine health service databases (eg, linking prescribing data with hospitalisation data). There have been repeated calls over the years for this to occur.3 Although there are longstanding and successful state-based linkage programs (most notably in Western Australia), little progress has been made at a national level. In contrast, good progress is being made in other countries, such as through the Sentinel Initiative of the United States Food and Drug Administration (FDA)4 and Canada’s Drug Safety and Effectiveness Network.5

Second, recognition of potential ADRs could be improved through education. In our institutions, we have found that many “new” medical diagnoses are actually due to ADRs from existing medications.6 Similarly, education could be used to highlight that drug exposure in RCTs is short compared with exposure in practice, that our ability to detect rare ADRs in RCTs is limited, and that patients in RCTs are very different to patients in clinical practice. Education and counter-detailing (educating prescribing physicians on less expensive equivalent or generic alternatives) from local opinion leaders will also help motivate prescribers to keep up to date with the latest information. The National Prescribing Service’s RADAR bulletins and Australian Prescriber journal serve an important education function by highlighting potential issues with new medications.7

Third, we need broader inclusion criteria for RCTs. Given the benefits associated with randomisation (ie, insurance against confounding by indication or selection bias), a necessary part of the solution is to increase the external validity of RCTs by mandating the inclusion of people with obesity (particularly morbid obesity8), older people, people from different ethnic groups, and those with comorbidities. Although this may seem difficult, it was not that long ago that the FDA ruled that studies of new medications had to include female patients and children (1993 and 1998, respectively).9

There are other issues to consider. Currently, our postmarketing surveillance is reliant on the sponsor providing (often yearly) periodic safety update reports. This is a voluntary reporting scheme, handled by the Therapeutic Goods Administration’s Advisory Committee on the Safety of Medicines and the World Health Organization Programme for International Drug Monitoring. However, as highlighted by Baker et al with regard to dabigatran,2 this postmarketing system has important limitations — most notably, the voluntary nature of reporting, the lack of data linkage, and the small numbers of people in special groups who experience clinically important ADRs.

Lastly, better information on toxicity and efficacy in particular subgroups would allow for more realistic economic analyses to guide the pricing arrangements associated with subsidy of medications included in the Pharmaceutical Benefits Scheme. With the example of dabigatran, the unexpected number of bleeding events and the requirement for monitoring that are becoming apparent will involve unanticipated costs.

Australia’s regulatory authorities are among the best in the world. What we need now is better postmarketing research, more inclusive premarketing research, education on pharmacovigilance, and more integrated, accessible and linked data systems.