Breaking the COVID-19 Transmission Chain

Focusing on ‘Impactability,’ Not Risk in Behavior Change Programs

Right now the entire world is focused on ameliorating the impacts of the COVID-19pandemic. Health experts clearly understand many of the things that people need to do: wash hands frequently and thoroughly; avoid unnecessary social interactions; avoid touching your face; self-quarantine if you have symptoms, and so on. Some essential public health measures can be taken by law and enforced by police. But most behavior change will result from voluntary compliance.


Convincing people to modify their behavior patterns in ways that are
uncomfortable and unnatural is an enormous challenge. Many public health officials believe that providing fact-based information is the key to voluntary behavior change. This works for some people some of the time, but not everyone in the world thinks like somebody with a doctorate in epidemiology. Accordingly, determining what messages, delivery channels, timing and other aspects of engagement will be most effective for each individual is essential to allocating finite resources to drive the greatest overall behavior change.

Health plans and care providers are well positioned to create this desired behavior change because they typically have the following key ingredients:

  • Detailed member / patient data;
  • Individual engagement channels at scale, including email, texts, contact
    centers and in-person healthcare visits; and,
  • Trust-based relationships with members and patients.

Both health plans and care providers will also need to drive more targeted
behaviors as the pandemic progresses, such as convincing members / patients to use telehealth early in the process, and then proceed to specialized clinics rather than simply arriving at the ED, or encouraging adherence with care management protocols.

There is extreme urgency in executing quickly because successful behavior change now is disproportionately important in terms of flattening the infection curve. But there is also a need to build and sustain this kind of behavior change capability, since current estimates suggest we will be living with this problem for months and years.


In our estimation, the key missing ingredient for both health plans and care
providers to execute these kinds of engagement programs most effectively is the analytical capability to reliably predict member or patient ‘impactability’ (sometimes called ‘response propensity’ or ‘persuadability’).

This can sound surprising, as over many decades health plans have developed advanced risk scoring models of various types. More recently, and mostly in response to their growing exposure to risk under value-based care arrangements, many health providers have built comparable risk stratification capabilities by partnering with plans, using third-party resources, and creating their own risk tools in-house.

But each of these approaches miss a key dimension of the current problem, which is the ability to reliably answer the following question: “Will person X actually change behavior in response to message Y?” For example, in any health plan or care provider adherence program, there are high-risk members / patients who will not change behavior in response to even a well-timed and well-executed call, text, email or counseling session. The resources devoted to these contacts are therefore entirely wasted. More perversely — but in our experience, always true for a nontrivial minority of members / patients for any given intervention — there may be members / patients who would have behaved correctly but fail to adhere because we contacted them. In that case, we actually invest resources to drive down adherence. The instinct to engage with everybody all the time in a crisis is natural, but we have finite resources. And further, some messages to some people can actually be counter-productive. Allocating resources for maximum benefit is even more important in a crisis.

We can therefore represent the possibilities in the following matrix:

Ideally, we would only contact the “Persuadables,” and prioritize within this group using our risk models. The most effective approach to do this is to: (1) estimate for each member / patient the risk (or more formally, the expected value) of nonadherence if not contacted; then (2) estimate the probability of change in adherence if contacted for each member / patient; and then, (3) select the members/ patients to call based on maximum projected change in expected value of non-adherence caused by calling them.


Effectively targeting efforts to drive adherence to COVID-19 behavior change
recommendations demands two important extensions to the historical risk-centric approach of most health plans and provider networks.

  • Predicting ‘impactability’ requires causal analysis of engagement campaigns, e.g., “What was the actual incremental effect of that adherence calling program, and how did it vary by member or patient?” This is the foundation of all reliable propensity prediction. It requires different mathematical methods than risk modeling, many of which have only recently emerged from academic research in the past several years, and that have not been widely deployed in the healthcare sector.
  • As a health plan or care provider attempts variants of a program (for example, calls versus texts, message A vs message B, etc.), the inherent risk profile of each member / patient doesn’t change, but the ‘impactability’ of a given individual, dependent on different program variants, does change, often dramatically so. This implies the need to build many such sophisticated ‘impactability’ models, which in turn implies a need for infrastructure to semi-automate model building at an advanced level that integrates analytical methods with a variety of internal and external data sources.

These extensions are not simple, but new technologies have made them practical for real-world implementation. Doing so now is urgent and will help to build capabilities that will be of growing importance over time, both for management of COVID-19 and for ongoing engagement programs in the increasingly value-based healthcare environment.


Jim Manzi — Partner,

Jim is a Partner and co-founder of He was co-founder, CEO and Chairman of Applied Predictive Technologies, which became the world’s largest cloud-based AI software company and the dominant platform for rapid, iterative randomized trials of business programs. Previously, Jim developed pattern recognition software at AT&T Laboratories, and worked as a corporate strategy consultant.

Jim is the author of several software patents for the automation of randomized experiments, as well as the 2014 Harvard Business Review article “The Discipline of Business Experimentation.” His 2012 book Uncontrolled on the development and application of randomized trials was widely reviewed in the New York Times, Wall Street Journal and other national publications.

Jim received an SB in mathematics from MIT, and was subsequently awarded a Dean’s Fellowship in statistics to the doctoral program at the Wharton School of the University of Pennsylvania. ​ He serves on the Board of Directors of Aledade, an innovative value-based care company backed by Venrock and Google Ventures.

Learn More



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Jim Manzi

Jim Manzi


Partner & co-founder of, MIT grad, author. Former co-founder, CEO and Chairman of Applied Predictive Technologies.