Policymakers, healthcare practitioners and researchers are thinking more and more about the real-world impact of new medicines, which means the life sciences industry is starting to ask new questions.
Evidence-based medicine is an approach in healthcare that looks to proven research to direct the best course of action for an illness. Sonal Singh, based at Johns Hopkins’ Center for Public Health and Human Rights, is well versed in evidence-based medicine as an approach to improving public health. Singh’s work looks at how drug safety is evaluated and how these methods can be improved. His recent work focuses on the prevalence of bleeding in a new class of anticoagulants on the market, safety in testosterone products, and the safety of a new class of diabetes drugs.
“We’re seeing an evolution of how drugs are approved. We want to know if these treatments work, but we also want to know how they work in the real world,” Singh said.
Singh notes that while evidence-based medicine can be a useful tool in healthcare decisions, it’s not the only aspect of how medicine is practiced in the real world. “There’s context and patient voice, and there’s cost,” Singh said, adding that the “evidence” in evidence-based medicine needs to evolve to incorporate these different aspects. Rather than be seen as stymying choice, evidence-based medicine has the potential to increase choice in healthcare decisions because it can reveal the variety of options available or physicians, patients and researchers.
Comparative Effectiveness
Comparative effectiveness research evaluates the available treatment choices against each other, and evaluates these therapies in the real world. One type of clinical trial might compare a drug to a placebo, but in the real world the choices are between similar drugs rather than one drug and a placebo.
Comparative effectiveness research can help determine which treatments generally work for everyone, but Singh notes that its true promise lies in individualizing care. “If we can determine that a drug works for me or for people like me, then it’s useful. But if we are just creating general information about an entire population, it may not be useful for ultimately improving care.”
Getting at a certain level of granularity is not out of reach – there are ways to determine how a drug works in different subsets of a population, Singh added. “There is enough data today that - if examined appropriately - we can determine how certain age groups or genders or races react to drugs already on the market. We have the tools, but we just haven’t used them well because we are still swimming on the top, trying to do trials to find drugs that work and then realizing the people included in the trials are not the best group to receive these treatments.”
There is push and pull in measuring both the validity and generalizability of a new drug during clinical research. Randomized controlled trials can provide validity that a drug works in a specific population, but this population may not include the patient that will actually receive the drug in the real world.
On the other hand, once the drug reaches the market and is used as physicians see best, there’s potential to do studies that look at how the treatment works in the real-world population. But without the controlled environment of clinical trials, it can be hard to establish validity in observational research. This conundrum in comparative effectiveness research has no easy answers and is still being examined.
Ask The Right Big Data Questions
Singh welcomes the influx of Big Data and the variety of data being generated through new technology that can inform real-world studies, but he cautions about how researchers approach these large volumes of data. “One of the challenges of Big Data is knowing the right questions to ask. If you don’t ask the right questions, you’re going to sink in the data.”
For a company looking to bring a new drug to market, asking the right questions means setting the appropriate parameters to compare the drug to other drugs on the market, whether it’s quality-of-life issues or safety profiles. “Are you going to make changes based on the information you receive? If not, then it’s redundant data.”
The right question also depends on the actual situation. Singh’s research on atrial fibrillation medication helps provide a comparative analysis of the different options available and could potentially help guide treatment recommendations. But the majority of atrial fibrillation patients don’t receive any medication. For these patients, it doesn’t matter whether they receive the most appropriate atrial fibrillation medication. It’s more important that they receive any medication at all, Singh says.
“Only 40% of atrial fibrillation patients in the United States are treated with any drug, so the drug companies are competing for that 40%. We are fighting a battle between products A, B and C when the question is really about the population’s treatment rate.”
Post-marketing Surveillance and Noise vs. Signal
Post-marketing surveillance, which is used to study a drug once it has entered the market, is an important tool to answering some of these questions, Singh said, but adds that there is a “noise versus signal” problem.
If a patient in a clinical trial dies in a skiing accident, the FDA mandates that it be reported as a potential serious adverse event, despite the extraordinarily small chance that the accident is related to the drug. Most drug safety information in America is obtained from post-marketing surveillance, but without clear regulatory guidance, useful safety signals can get lost in the noise of irrelevant data.
“There’s no reason evidence-based medicine can’t be business friendly. All of us – researchers, companies and regulators – need to get out of the box. We need to get to patients and get their input. The patient voice has real value not only in determining how we treat people, but also how we work in evidence-based medicine to make sure we are asking the right questions.”
