Latest posts by Justin Lyon (see all)
Today, it's possible to use an evidence-based dynamic simulation model with a broad national scope to analyse policy proposals.
In the past, when we lacked sufficient computing power and storing data was expensive, we tried to improve a country's health system using models that have only a few variables and overlook certain processes that may delay, dilute, or defeat intervention effects. These simplistic models usually ignored the effects of accumulations, time delays, resource constraints, and behavioral feedback, all of which must be considered to reach correct conclusions about the net impacts of interventions in healthcare systems comprised of many interacting actors, multiple goals, and conflicting interests.
Today, we do things differently!
Simudyne worked on a simulation of the US healthcare system which allows you to experiment with thousands of different ways to drive down costs and improve the US healthcare system. You can use the simulation to generate an infinite number of what-if scenarios. It uses Providence to run on the cloud and pulls national-level data from a variety of sources into a data lake powered by Cloudera's technology.
In one scenario, published in the American Journal of Public Health, healthcare coverage, quality, and capacity interventions were combined with two broad-based interventions for health protection (a concept encompassing both promotion of health and the ensuring of safer, healthier living conditions) to generate a massive $27 trillion in net benefits. The combination of the strategies take time to root, but once they grow they deliver massive improvements over 25 years versus the baseline...
- Reduces the onset of disease and injuries by 13%
- Reduces the number of deaths by 26%
- Reduces the number of unhealthy days by 21%
- Reduces health inequity by 33%
- And, it does all this while reducing healthcare costs by 5%.
Costs went down for 2 reasons.
- First, lower disease and injury prevalence translated directly into lower costs.
- Second, lower prevalence also meant less demand on primary care physicians (PCP), which alleviated PCP shortages and reduced the fraction of visits for acute care going to more expensive hospital emergency departments rather than PCPs.
But, that's not the only way to run a nation's healthcare system!
The simulation allows policymakers to safely experiment with any number of different sets of decisions. There are so many different things that you can do that it becomes a bewildering and complex set of decisions.
- Should we create safer communities or cut reimbursements?
- Should we institute a single-payer system or open it up to market forces?
- Should we invest more in better preventive and chronic care or should we focus on delivering better access to and quality of care for everyone?
- And so on!
So many questions, but no easy way to figure out the possible answers! Right?
Simulations are the only way to explore how a bewildering array of policies can be implemented effectively to drive change for the better in a complex system. There are great data scientists who have cracked this challenge. But, as far as I can tell, they've had minimal impact on the national discussions around healthcare in either the US or the UK.
That needs to change.
Our politicians need to use predictive analytics to generate the insight and foresight required to make great decisions. Better yet, we should let the robots make the decisions.
With the simulation, policymakers can get a firm sense of how much investment will be required, over what time period. And, they can track results over time to ensure they remain on track and adjust course as needed.
Various theories have been offered to explain why the US health system performs so poorly and is so costly.
- Is it the lack of health insurance for millions? Some feel this is the system’s chief problem.
- Do Americans focus too much on disease and not enough on health? Some argue vehemently that the medical industry and the public at large overemphasise disease detection and treatment whilst missing opportunities to reduce preventable risk and protect people’s health.
- Maybe it's greedy doctors? Some people think that perverse incentives and community norms are encouraging physician entrepreneurship and profit making over collaboration, coordination, or conservative practice.
- Or perhaps we just need more and better paid primary care physicians? Some think that the US has too few primary care providers and that the providers they do have are underpaid and unable to offer the highest-quality care.
- Is private insurance the problem? Another favourite villain is private insurance, some of whom pass along high overhead costs to consumers, are unwilling to reimburse adequately for preventive care, and offer a confusing array of coverage plans, creating a substantial administrative burden for providers.
With simulation, you can move conversations into a space where various reform strategies can be tested and refined with no risks to real patients or significant costs to the government.
In the case of this simulation of the US healthcare system, data was pulled from a number of sources. National-level data was used to calibrate the model’s input parameters and to confirm that its output faithfully reproduced key historical metrics. These data included:
- measures of death rates (National Vital Statistics Reports) and their disparity by income level,
- un-healthy days and access to primary care providers by income level (Behavioural Risk Factor Surveillance System),
- rates of health care utilisation (National Ambulatory Medical Care Survey, National Hospital Ambulatory Medical Care Survey, National Nursing Home Survey, National Home Health Care Survey),
- prevalence of disease and asymptomatic disorders (National Health Interview Survey, National Health and Nutrition Examination Survey) and their disparity by income level,
- health care costs by category (National Health Expenditure Accounts).
All of this data is then fed into a robust mathematical model. The model is a complex system dynamics model that we converted from Vensim to Java and loaded into Providence. This model is based on the HealthBound model, produced by the Centers for Disease Control and Prevention, which contains hundreds of equations, constants, and variables that accurately model the behaviors of the U.S. healthcare system.
We're now working with other governments to adapt the model to their unique situation. They do this by pulling their own data into a data lake and then bring the data into a version of the model calibrated to their own situation.
We then wrapped the model with a user interface that makes it fun and interesting to explore and use. We ultimately created a simulated country, which looks much more fascinating than a series of spreadsheets or dashboards with a bunch of charts on it. Here's one of the art assets in development and then placed into the city landscape.
And, here are a few of the various charts and panels that the users can access as they explore the sim...
When folks play the simulation, they get to role-play a number of important groups within the healthcare system . . .
- Providers (healthcare providers, like doctors, who typically work in hospitals and specialised clinics)
- Primary Care Providers (healthcare providers who typically work in clinics and perform general and family practice)
- Employers (those who offer healthcare benefits and enact safe work practices and other programs)
- Government (organizations that pass laws that influence the healthcare system)
- Payers (health insurance providers)
- Advocates (NGOs and non-profits who raise funds and awareness and enact health, safety, and other programs)
Each role contributes to the healthcare system differently. So, they all get to make different decisions. One example of a decision is “Enable Healthy Behaviors,” which is further broken down into “tactics” like “Promote Nutrition Guidance & Education.”
As they play, the users get to fund their decisions, for instance, deciding to spend $250 million from 2015 to 2030 on a number of tactics that drive healthy behaviours in the population. The simulation then evaluates the decisions based on the maths and tells you what will happen with metrics like “Average Healthy Days per Month” and “Healthcare Costs per Capita,” among many others.
It's super cool. It's fun to play. It's rigorous and grounded in the science of peer-reviewed models of the US healthcare system.
It shows, unequivocally, a path forward for saving money and improving the lives of millions of Americans.
I just wish we could get millions of Americans to play it.
I wish we could get the new US President, Donald Trump to play it!!
I'm really looking forward to working with other governments that are interested in bringing big data and computational simulation tools into their healthcare planning toolkit. Here's the sim in action with some US folks.