The 17th May 2022 saw the 31st Bradford Hill Memorial Lecture, held annually in memory of Sir Austin Bradford Hill, hosted by LSHTM with the support of the RSS Medical Section. Professor Anne Mills, Deputy Director and Provost of LSHTM, began by reminding us that Sir Bradford Hill initially studied economics but went on to become a leading epidemiologist and write a seminal book on medical statistics and develop the nine criteria of causality in epidemiology.
The many hats of Professor Walker
Professor Mills introduced the speaker Professor Sarah Walker (FMedSci, NIHR Senior Investigator), Professor of Medical Statistics and Epidemiology at the MRC Clinical Trials Unit at University College London and at the Nuffield Department of Medicine at Oxford University. Professor Walker is also the chief investigator and academic lead for the UK’s Covid-19 Infection survey, a partnership between the Office for National Statistics, Oxford University and the Department of Health and Social Care. She is also director of the NIHR Health Protection Research Unit on Antimicrobial Resistance and Healthcare Associated Infections at Oxford.
Beyond ‘Standard-of-Care’
Professor Walker began her talk, 'Beyond "Standard-of-Care": novel trial designs for old problems', by sharing a group photo of LSHTM staff, with Sir Bradford Hill front and centre and emphasising his contribution to the design of clinical trials and their key attribute, to eliminate bias introduced by known and unknown confounders.
Professor Walker introduced the need for innovative design. She did this by describing historical improvements to treatment of tuberculosis, cancer, HIV and other diseases where there are multiple sequential trials evaluating iterative improvements to the standard of care, which can be slow, but reliable. She then introduced platform and multi-arm multi-stage trials. Simplifying a difficult topic, she also summarised basket trials, in which the same interventions are evaluated for their effectiveness against multiple diseases, and umbrella trials which evaluate multiple interventions against one disease.
The tragedy of the commons
Professor Walker explained that the more complex nature of the treatment of bacterial infections with antibiotics led to more challenges. When a patient infected with a true pathogenic bacterium, such as Mycobacterium tuberculosis which causes TB, is given too little antibiotic, then resistance can develop in the infection being treated. In contrast, many infections are caused by the mostly harmless bacteria we live with (commensal bacteria) - these can also develop resistance if the patient receives too much antibiotic. We are looking for a ‘Goldilocks’ zone, where we give enough antibiotic to treat the bacteria causing the infection but not so much that we cause resistance in commensal bacteria. She then reminded us of the important economic principle, ‘the tragedy of the commons’ – to benefit the individual we would always hit hard and fast with the strongest antibiotics we have, but to benefit society we need to start with more targeted regimes and escalate to stronger antibiotics only if needed.
No experiment is ever a complete failure – it can always be used as an example
Clinical trials most often compare novel treatment with standard of care; a key aspect of optimising antibiotic treatment is to optimise duration of treatment. Professor Walker highlighted that current standard of care is not evidence based, as can be seen by the preponderance of treatment regimes focussed on three, seven or 14 days, or, in countries where odd numbers are unlucky, four or eight days. She summarised a published trial looking at optimising treatment regimes which considered days of treatment as the primary outcome, and then went onto describe one of two new trial designs, the 'Durations' design, which uses a multi-arm approach to flexibly model the duration-response curve using fractional polynomials and spline-based methods rather than testing two arbitrary durations of treatment in a standard two-arm trial. This has been used in the PediCAP trial, which is identifying optimal treatment for children with community acquired pneumonia in southern Africa.
Avoiding the worst
The second trial design, the 'PRACTical' (Personalised Randomised Controlled Trial) approach allows trials to move away from the challenge of having to compare the new treatment with the standard of care. The approach is analogous to network meta-analysis, and allows a large number of available treatments to be compared with each other, without needing to specify a single “standard of care” which will not be appropriate for all patients. It is also possible to rank these treatments – she suggested that sometimes we need to at least avoid the worst treatment, even if we cannot identify the very best treatment. This new design will be used in the neoSep trial comparing different antibiotic regimens for neonatal sepsis, in which patients will be switched if not doing well on their first treatment using another type of novel design, a Sequential Multiple Assignment Randomised Trial, 'SMART', approach.
Why walk when you can run?
Professor Walker concluded by challenging us to think: ‘Why walk when you can run?’; challenging the traditional approach of slowly evaluating incremental changes to standard-of-care. She returned to the photo of the department featuring Sir Bradford Hill, and highlighted her mother, Ruth Tall, who worked with Sir Bradford Hill on the early tuberculosis trials.
The audience clearly enjoyed the talk, as was evidenced by the many comments in the ‘chat’. The questions ranged from the role of ethics committees in identifying poor trials and the potential benefits of hindsight, how we should decide precision when studying duration and the role of randomisation.
Professor James Carpenter thanked Professor Walker and her thoughtful comments on innovations, including new, relevant and efficient designs, improved conduct, and the need for timely evidence synthesis. He emphasised the need for methodological work to be supported by funders alongside funding the trials themselves.
A thoroughly enjoyable lecture, with important insights into the role of trial methodology in fighting antimicrobial resistance.
Written by Dr Kate Honeyford, Research Associate, Team Health Informatics, Institute of Cancer Research, on behalf of the RSS Medical Section.