Context: The landscape of inquiry, insight, and innovation
Finding our way into complexity
A recent news article called attention to a resurgence of interest in paper maps in an age of digital navigation systems. It seems that some travelers are recognizing that the detailed navigation instructions from their phones leaves out the context - the bigger picture of the landmarks, sites of interest, and scenic alternatives we pass during a trip.
The story prompted me look through my camping supplies in search of the topographical maps we use for plotting our backpacking trips. Topographical maps may cover hundreds of square miles and are remarkably detailed, showing the contours of the landscape, mapped trails, rivers and streams, established campsites, and other features of interest.
In fact, the amount of detail can be overwhelming for novice backpackers who might think that the details are unnecessary. They might focus instead on the distance from trailhead to campsite to estimate the time required for the hike. Of course, this estimate could be terribly wrong if the trail passes through steep terrain, offers little shade, and lacks access to water. All the details that make the difference between a strenuous, but enjoyable, hike and a miserable experience.
The need to see, and properly interpret, the bigger picture is important, and not just for camping. The success of applied research, including research in global health and pharmaceutical research and development (Pharma R&D), depends on a proper understanding of context. The depth of our understanding has important implications for the choices we make about the problem we are trying to address, the selection of an intervention, the design of a study, and the interpretation of results.
The challenge with context is that there is so much of it! And, like the novice backpacker, we make choices about which views of context we believe to be most salient for planning and decision-making.
There are two types of context that I consider in this essay - "wild-type" and "tamed"
context.
The wild-type context is the real world. The contextual elements are complex and interrelated, they are dynamic and change over time, and the nature of the context is shaped by individual perspectives and experiences. These characteristics make it hard to accurately define context that is relevant to a particular problem. Wild-type context is frequently encountered in global health research. It is a key reason that a potential innovation that was successfully demonstrated in one context often fails on attempts to scale to larger populations and different contexts.
For example, for nutritional supplementation to be effective, we need to recognize the importance of geography, and the impact of a mismatch between the intervention and differences in the salience of risk factors in different contexts. A comparison of two complementary feeding intervention studies is illustrative.
A study in children in Ecuador showed that eating one egg per day produced a marked improvement in linear growth. Encouraged by these results, a similar study was performed in children in Malawi, but showed no effect. The subsequent assessment of the outcomes revealed important differences in context across the two locations. There was a higher prevalence of linear growth impairment at baseline in the Ecuador study population, while a background diet rich in animal source foods and low prevalence of growth impairment at baseline may have limited the potential benefits of the supplemental protein in the Malawi children.
When we refer to the concept of "global health" we refer to the marked variability in the state of "health" in different contexts across different geographic locations over time. The response to interventions can exhibit marked variability because of the role of risk factors are that salient in each context. One size does not fit all.
The concept of "tamed" contexts refers to the artificially created environments that emerge from the design choices for the randomized, controlled trials performed during the research and development programs for potential new therapeutics.
In a Pharma R&D program, the current understanding of the pathophysiology of the disease process or medical condition of interest and the hypothesized pharmacologic effect of a new medicine is used to frame its potential benefit. This framing reduces the complexity of the biologic system of interest into a manageable, but still complicated, subset. The framing is akin to a topographical map, but with some (potentially important) details removed.
This framing plays a critical role in defining the context for medicine use that will optimize clinical performance. It guides the design characteristics of the clinical study plan, including the specification of inclusion and exclusion criteria to guide subject enrollment, the selection of clinical outcomes, measurement strategy, dosing regimens, and so forth.
The evidence for efficacy and safety that emerges from these trials is inextricably linked to the context circumscribed by the study design choices. Although positive results from clinical trials provide a degree of validation of the frame, even a successful clinical trial may have escaped the effects of unrecognized complexity that will ultimately affect therapeutic outcomes under more complex conditions of use.
Sponsors bear the burden of risk that accompanies the design decisions - a failed trial may lead to a decision to terminate a R&D program. Regulatory agencies, on the other hand, bear the burden of deciding to provide market authorization based on the results of context-specific clinical trials. Consequently, the study designs for large scale clinical trials of a potential new medicine are a compromise between the sponsor’s interest in optimizing the likelihood of a positive outcome and the regulators need to assure valid and robust results regarding safety and efficacy.
The upshot of this "Pharma R&D - Regulatory Review" paradigm is that the wild-type context is always waiting beyond the artificially constructed tamed context. The decisions that are made by regulators regarding market authorization, and subsequent decisions by healthcare providers and patients to use the medicines in clinical practice, requires that these stakeholders put effort into anticipating possible unintended consequences in different contexts.
The anticipatory process can be approached by examining the effects attributable to intrinsic ethnic factors, e.g., genetics, age, gender, and organ function, and extrinsic ethnic factors that are associated with environment and culture (e.g., medical practice, diet, use of alcohol, and concomitant drug use), or interactions among these factors, that might be encountered as a new medicine is introduced to different patient populations.
The challenge to all stakeholders is that the mix of intrinsic and extrinsic risk factors that are present in any given context can vary greatly. This mixture of risk factors, accompanied by the complexity elided by context-specific development plans, can significantly alter the risk to benefit ratio for a medicine.
Recognizing the need for more nuance information about the wild-type context, regulators and other stakeholders turn to a myriad of sources of real-world data to gain insight into the contexts likely to be encountered in clinical practice.
But are these sources of insight enough? Probably not.
W. Ross Ashby was a British cyberneticist and psychologist who, during the 1960s, proposed what is now known as the First Law of Cybernetics which has been described as: In order to deal properly with the diversity of problems the world throws at you, you need to have a repertoire of responses which are (at least) as nuanced as the problems you face.
The challenges we face in global health and Pharma R&D will require a rich variety of sources of insight into wild-type contexts. One source of this insight is the data coming from global health research.
Research programs in global health are conceived and designed by independent academic researchers along different disciplinary lines; the studies consist of different study designs, different target populations, different inclusion and exclusion criteria, different endpoints, and so forth. These varied research programs create a rich ecosystem of hypotheses, results and ideas that provide a glimpse of the many dimensions of wild-type context.
The differences in observed outcomes from similar studies - like the egg studies described above - represent a source of insight and lessons learned that remind us to remain humble in the face of the multitude of factors operant in wild-type contexts. At the same time, the data, ideas, and experiences from non-traditional sources would strengthen the frame we use for decision making and increase our confidence that we can avoid or minimize the cost and harm from unintended consequences.