Making combinations of medicines safer and better

Drug-drug interactions (DDIs) are where a combination of drugs interacts, which can be harmful or helpful. The harms from DDI are an international problem, and between a third and half of all drugs are not taken as prescribed. In the UK, the Overprescribing Review called for systematic optimisation of medicines, and Liverpool has responded with the Dynamic Prescribing AI project, DynAIRx, building on decades of globally impactful DDI research.

For example, in HIV, Liverpool researchers recognised DDI dangers for those undergoing HIV and hepatitis treatments as well as potential beneficial use of an HIV protease inhibitor (ritonavir) to ‘boost’ concentrations of other drugs. Since 1998, they have produced prescribing tools to optimise medicines for people living with HIV. Between 2012- 2018, the University helped to characterise the frequency and severity of DDIs involving HIV medications. Of approximately 39,000 patients from the UK, Switzerland, Spain and Uganda, significantly harmful DDIs were reported in 18-35% of patients taking antiretrovirals.

Our DDI reference tools now cover HIV infection, hepatitis and liver disease, and COVID-19. Using clinical and scientific expertise and review of scientific literature, we have developed these tools to provide detailed and comprehensive coverage involving over 100,000 DDI-pairs, providing clinical recommendations for each interaction. The team works with patient advocacy groups, such as i-Base, to find the best ways to provide patients and clinicians with high quality, up-to-date drug information. The Liverpool tools are recommended for use in every HIV clinic in the UK, have established users across Western Europe and North America, and are increasingly used in Latin America, Asia and Africa.

Extending beyond DDI reference sets for specific conditions, our DynAIRx project is now addressing the major challenge of optimising medicines for people with multiple long-term conditions (MLTCs). This is a growing problem. By 2035, it is estimated that two thirds of the over 65s will be living with MLTCs. Overall, there are around 14 million UK residents currently living with MLTCs. In deprived areas like Liverpool, these numbers are higher and affect people earlier in life, including their ability to work. Most people with MLTCs take multiple medicines and many suffer DDI harms.

We are developing DynAIRx AI to work with existing NHS systems to help GPs and pharmacists identify which patients are most likely to benefit from a review of their medicines. We use state-of-the-art causal machine learning and modelling to identify the patients most at risk from their current combinations of conditions and medicines. This DynAIRx AI will use routinely collected NHS data to highlight prescribing patterns, support shared decision making, and enable more joined up care. The work is co-produced with public advisors, clinicians, and researchers to ensure we are creating AI that is as useful and fair as it can be.