By Aiden Flynn, Exploristics CEO
I recently went to see an outstanding and thought-provoking play called Dr Semmelweiss in London’s West End. It is a true story based on the life of Ignaz Philipp Semmelweis, a Hungarian obstetrician and scientist who is now widely regarded as the pioneer of antiseptic procedures. However, at the time, his ideas were rejected by the medical establishment leading to the loss of tens of thousands of lives. Watching the play, I was struck by some of the common challenges he faced that can still impact healthcare research today.
A tale of two clinics
Semmelweiss worked on the maternity clinics at the Vienna General hospital where there were two clinics which had very different mortality rates due to postpartum infection. One clinic had an average maternal mortality rate of around 10% whilst the other was considerably and consistently lower at less than 4%. Pregnant women were aware of the reputation of the first clinic, and they were desperate to avoid giving birth there, crying to be admitted to the second clinic. Indeed, some women preferred to give birth in the streets. The only major difference between the clinics was the individuals who worked there.
The first clinic was run by the physicians who were also involved in conducting autopsies and in training medical students whilst the second clinic was managed by midwives. The established opinion within the medical profession was that the discrepancy could be explained by differences the quality of light and air. However, Semmelweiss was not content to accept this explanation.
Six ways Semmelweiss addressed his clinical challenges and today’s parallels.
1.Accept you have a problem
Semmelweiss himself had run a clinic which had experienced much higher mortality rates than that run by midwives. Despite potential risk to his professional reputation, he realised that there was a problem and was so concerned about the infection rates that he set about trying to understand the difference in mortality.
In pharma today, attrition rates in clinical trials remain stubbornly high. Recent estimates indicate that the level of overall success in getting a candidate therapy to market through all phases of development is around 10% [1, 2]. Moreover, while estimates for late-stage trial success are around 50% (depending on the study), the figure for novel first-in-class agents is considerably lower with some estimates at below 25% [3]. This suggests that there is a problem with clinical development approaches or at the very least room for significant improvement in the way we currently de-risk clinical trials.
2.Test your assumptions
Semmelweiss’ first step was to switch the clinics so that the physicians took over the second clinic and the midwives the first. He showed that the mortality rates also switched between the clinics, thereby discounting the current assumptions that it was due to air and light quality.
Designing a clinical trial also requires the use of assumptions but how do you know which ones are reasonable? Understanding the uncertainties involved in a clinical trial and testing the assumptions you use in designing one with techniques like simulation can be a powerful way of optimising study design. However, a more comprehensive approach to de-risking studies and build in more success through design and analysis strategies is still not well integrated into the study design process.
3.Gain more insights from historical data
Semmelweiss collated the historical data on mortality from all births from both clinics and street births. His analysis uncovered some alarming results- even the street births had lower rates of postpartum infection than those from the first clinic.
Evaluating historical data is still an important way to help make decisions regarding key study factors and get a better overview of the clinical trial landscape. Accessing and collating reliable data you can trust can support this.
4.Create a robust process to improve outcomes
Eventually, Semmelweiss proposed that the only explanation for the different infection rates was that those involved in conducting autopsies carried decaying organic matter on their hands to the patients they examined in the clinic. This explained why the clinic run by midwives, who were not engaged in autopsies and had no contact with corpses, saw a much lower mortality rate. He then introduced a strict handwashing procedure with chlorinated lime solution which drastically reduced the incidence of infection.
Amending entrenched processes is always hard and this also applies to the traditional approach to clinical trial design. Integrating a more robust evidence-based approach into the design process using methods like simulation can feel uncomfortable if you are not familiar with the methodology. However, it has been shown to dramatically improve clinical trial outcomes by ensuring studies are quickly and comprehensively de-risked at the design stage. This means that avoidable reasons for failure like unsuitable design are prevented.
5.Publish your research
Even though Semmelweiss’ empirical evidence led to a procedure that worked in practice, he was unable to offer a theoretical explanation for his findings. He did not publish his results at the time and his ideas conflicted with the established beliefs and so were rejected. Some doctors were even offended at the suggestion that they should wash their hands or that they were implicated in the deaths of many women and infants.
To ensure effective clinical trials and improve success rates, it helps to understand common reasons behind clinical trial failures. However, this is not always straightforward. Failed investigational drug studies are not often published in peer reviewed journals. This means that there is limited systematic data on how frequently or why novel agents fail, even in late-stage development. Consequently, it is hard to fully scrutinise the clinical development process and pinpoint key areas for improvement.
6.Remember a data-driven approach puts patients first
Unfortunately, Semmelweiss became increasingly disillusioned, outspoken and outraged by the resistance to adopt his antiseptic procedures. He was dismissed from the hospital, suffered a nervous breakdown and died in an asylum. Nevertheless, decades after his discovery his findings earned widespread acceptance when Louis Pasteur confirmed germ theory. This resulted in widespread adoption of hygienic methods which has saved many lives.
More effective clinical trial design not only helps make the most of R&D investment, it also can help to reduce patient burden by ensuring that the right patient populations are recruited. It also ensures that patients are not involved in clinical trials that are destined to fail from the outset due to poor design. This is a tangible way of building a more patient-centric approach to clinical trials.
Working in a conservative space
The story of Dr Semmelweiss is a tragic one but the parallels with some aspects of the current clinical development process are striking. For new treatments in clinical development, the rate of attrition is greater than 90% and has remained at this level for decades. Over the same period, the cost to bring a new medicine to the market has ballooned to more than $6Bn. While there is widespread recognition that the status quo is unsustainable, the life sciences sector has been slow to adapt, often sticking to a conventional but inefficient approach to conducting trials rather than adopting new technologies, methods or processes.
The need for disruption
Whilst the conservative nature in a highly regulated, high risk and high-cost environment is understandable- after all no one wants to oversee a failed study following a departure from convention- the industry is in real need of disruptive change which cannot happen without taking some risks. In recent years, there have been numerous initiatives and innovations aimed at improving the likelihood of success of clinical trials including AI, precision medicine, patient centricity, novel study designs, decentralised trials, digital health solutions and many others. While there is no magic bullet for improving an increasing complex space, greater uptake of these disruptive technologies can certainly bring incremental improvements.
Investing more in the design process
As a statistician, there is one area that I know will result in a step change in the success of clinical trials- investing in study design. I am constantly amazed by how many trial sponsors continue with the current convention of spending millions on conducting a study but only a tiny fraction of that on de-risking the investment through data and model driven design. Many costly failures could have been completely avoided by properly evaluating and mitigating the study risks at the design stage. I’m not suggesting that investing in design is the only answer, as clinical trials face numerous challenges, but improving design is a very easy and cost-effective fix that should be part of the disruption the industry needs.
Challenging convention
Dr Semmelweiss was brave and prepared to challenge convention for the sake of improving patient care. Once more, today’s clinical researchers have the chance to be brave and scrutinise conventional development approaches more closely and adopt new technologies and methods that can help transform them. This should be done to reduce wasteful unnecessary failure and improve ROI on R&D. However, most importantly this should be done for the sake of people with unmet clinical need still waiting for healthcare breakthroughs.
References
[2] Wong et al., 2019 Biostatistics 20:2, 273-86
Read more:
Plague, problem and profiting from failure
Shifting the development paradigm with innovative trial design
Designing clinical studies for success and not failure
Statistical Consulting Services
Hear more:
The power of simulations for designing clinical studies and beyond
Simulations your most powerful study design tool
Watch more:
Why do correlations matter in clinical trial design?