by Michael Rohwer
In this article, Dr. Michael Rohwer’s fifth in a series that examines healthcare as a Complex Adaptive System, he describes the first of two key innovations necessary to radically improve care while lowering cost.
We will not reinvent the wheel. The core management problems we face happen everywhere people work together to accomplish large objectives. Now a thoughtful look at how the lean startup method has revolutionized product development could prove very helpful to us in our quest to effectively manage healthcare.
Gaining insight from software application development
There was a time when software was created in the same way we construct buildings. A thorough business analysis defines the customer’s need. Then careful and comprehensive documentation specifies exactly what and how engineering creates the new product. When the product is complete, testing (often with a limited release) ensures it satisfies the specifications. Only then is it released to production. This process requires months to years and serious money. Yet in spite of careful planning and a sequence of dedicated processes, many products do not perform as expected—or worse, are useless in that they do not solve the real problem or meet the real need. Moreover, the long process gives time to competitors who have their own “new and improved” products.
But this older “construction” approach goes wrong in several ways. A big flaw is that business analysis is always tricky. Customers often overlook key details. Unrealistic expectations are common. Visualizing something new that has not been used is difficult, but knowing the implications and consequences is impossible. Whatever flaws exist in the specifications are engineered into the product. And regardless of initial perfection, the goal begins to move while the analysis stands still as every customer adapts to their business needs.
- Instead of building a complete product in a single process, lean startup begins with a Minimal Viable Product (MVP) that immediately provides a small benefit to the customer.
- Flaws that are inherent to the comprehensive business analysis are sidestepped using an iterative cyclic of small product enhancements or extensions.
- Each enhancement or extension is tested using a process called Build-Measure-Learn. A small, often unpolished version is created and deployed to the customer. The customer’s response is measured and converted to learning. Learning is the most important product of each cycle because it guides future action.
- Measurement is done with Actionable Metrics. These are measures related to core drivers of the business. For software, the first metric is often whether or not the new product is used (e.g., has value). Once value is demonstrated, the same Build-Measure-Learn determines whether it has enough value to generate revenue (e.g., growth potential).
- Vanity Metrics (e.g., measures of activity) are avoided because they do not drive business decisions.
- This is very efficient. Each iteration of Build-Measure-Learn involves the customer and a real problem in real time, so the product and the customer evolve together. It maximizes value at the same time it conserves resources. Whatever additional time these lean startups consume more than offset development that is useless or error-ridden.
The older method is top-down, like healthcare is today. Lean startup is bottom-up and in direct contact with the customer. It builds strong relationships and in them finds the core drivers of its business because usable value is present from the start. Learning and the creation of new value are intimately connected. It is naturally problem-oriented and is less prone to distraction from extraneous non-value-generating “good ideas.” Its small scale is easier to manage and takes advantage of the natural force we call Complexity.
Applying this process to healthcare as a Complex Adaptive System
While healthcare is a long way from software development, this lean-startup process can revolutionize the way we in healthcare solve our problems by moving us away from a system that is predominantly provider-centered to one focused on the member. I’d like to begin by translating its terminology to healthcare.
Actionable Metrics must connect to the system’s purpose, “to enhance and preserve human health.” This is a measurement of value delivered to the member. For example, for clinical conditions such as diabetes, it might be a measure of control like the A1c. The measure, however, will not always be about a disease. Access to care is measurable and is also an important member value. If the system is about serving members— which it is—then the value delivered to the member is the metric that counts.
Vanity Metrics, on the other hand, are things like some NCQA criteria for the patient-centered medical home. The presence of recommended policies or good communication is valuable, but do not sufficiently ensure the member receives the most important care. Again, the purpose of healthcare is to achieve a beneficial result in the member, not a good process in the provider. Peter Senge, in The Fifth Discipline, his book on systems thinking, observes that “the easy way out usually leads back in.” Output measures are hard to implement, so we have used process measures with the idea they would take us partway there. More often than not, they lead us back into the confounding processes of Complexity. There is no substitute for knowing where we are going.
Build-Measure-Learn cycles are fundamental to knowing what will work. At the heart of Build-Measure-Learn is the idea that we have more to learn and what we learn will ultimately maximize the value we create. Every cycle creates incremental understanding and value; it is both path and destination.
The Community Value System is the first innovation.
I will elaborate on Community Value Systems (CVS) in coming articles, but for now let me say that its purpose is to give each community the tools it needs to develop its capability and to maximize the value delivered to each member. It assumes every community is unique, and composed of individuals having multiple strengths and weaknesses. Rather than suppress variation, it seeks to develop the ones that work better.
It allows the community to define and organize important problems. The definitions are clinical narratives that match how we think as human beings. A community’s system would contain one or more Clinical Programs, each designed around the problem context. That context includes Actionable Metrics, a Build-Measure-Learn mechanism, and performance-based monitoring and payment arrangements. Each program is a direct implementation of the healthcare’s purpose. This will be easier to visualize when the detail of the enabling technology is public.
We have a lot to learn. For years we’ve focused on an inverted order of things. As a result, we have many measures but few are dedicated and tested across multiple local communities. The strength of this lean startup process is that with time it is transforming, like yeast in dough. We start from a position of creating value and build on that; we start with one or two programs and based on that experience, new programs are created to optimize other clinical problems.
In this article I’ve outlined the first innovation I see as necessary to begin the transition to a more Complexity-aware management. It is critical infrastructure for the future, but it is insufficient alone. In a previous article I described how similar processes created value with personal computers and later, with the internet. A CVS System is much like the personal computer of the 1980s. It brings new options and ways to organize that build value and lower cost.
The value created by all systems is through connection and relationship. Each community must connect in a new way. This is the second innovation, a special type of network where performance and operational metadata flow. It forms the framework where cost constraint and the boundaries of scientific knowledge exert control.
Finally, this new infrastructure will produce one more advantage. Natural systems create strength and resiliency through Complex Networks. This can be the case in healthcare too, as we apply what we know to this Complex Adaptive System. Next up is the network. It’s going to be exciting!