Throughout this series on implementation science, we’ve learned that developing and deploying health interventions is anything but easy. From developing adaptations and evaluating economic impacts to integrating user-centered design and operating in virtual environments, we have seen that it takes a great deal of creativity, collaboration, and sustained effort to improve care and boost efficiency.
That’s because the health care system is incredibly complex and interconnected. Every time a patient, provider, executive, or regulator makes a decision, the impacts ripple up and down the care continuum.
Health interventions are intended to make it easy for participants to make the right decisions to improve positive results. But without a clear idea of how each stakeholder’s actions affect the next person in the chain, implementation teams might be flying blind.
To wrap up our recap of the 13th Annual Conference on the Science of Dissemination and Implementation in Health, we will discover how a deeper understanding of participatory system dynamics can clarify the underlying causes of problematic processes and help implementation teams design effective, comprehensive solutions.
Breaking down the basics of participatory system dynamics
System dynamics is the study of how decision-making structures develop, operate, and occasionally break down. Using visual and mathematical modeling with implementers, modeling identifies causal relationships in complex systems in order to improve care decisions.
“Models are designed in the systems dynamics tradition to represent the implementation problem, not the solution,” explained Lindsey Zimmerman, Ph.D., staff investigator at the National Center for PTSD, Dissemination & Training Division. “In participatory system dynamics, we are codifying the problem so that we can use the model to work with stakeholders and together identify the solution that works best for them.”
Zimmerman shared the example of her recent work to understand why well-established, evidence-based psychotherapy treatments were reaching so few patients in the Veterans Affairs (VA) health system.
“We had the infrastructure available, incentivized quality measures, published guidelines, and national training programs that had been running for almost ten years,” she continued. “But we were still only reaching about one in three patients with very strongly supported practices, such as cognitive behavior therapy for depression or medication-assisted therapy for opioid use disorder.”
Zimmerman knew it wasn’t necessarily a problem with the clinical interventions themselves. Patients were starting psychotherapies and pharmacotherapies, but far fewer were engaged sufficiently over time to meet their needs. Roadblocks within the system made it difficult for front line staff to connect with eligible patients while also making it hard for patients to stick with the course of treatment.
“We always start with the notion that none of these stakeholders have an incorrect view of their own priorities and responsibilities,” said Zimmerman, “But, their individual views are likely incomplete. It’s the aggregation of different decisions that can lead to problems with achieving an overarching health care goal. Participatory systems dynamics can help us better understand those chain reactions.”
Engaging health care stakeholders in system dynamics modeling
Zimmerman and her colleagues, including David Lounsbury, Ph.D. and Tom Rust, Ph.D., two simulation model designers and co-facilitators with the VA’s participatory system dynamics team (known as “Team PSD”), started working with frontline VA staff to bring their systemic challenges to light more than six years ago.
“Clinical teams only have a limited amount of time, so the challenge was to identify where that time would be best spent to get people through to the completion of an effective therapeutic program. They’re probably trying to do it all, because they don’t have enough evidence to warrant changing their strategy.”
Team PSD began by talking with patient, provider and policymakers about their stakes in improving care. Then, using scripted, interactive activities to gather key information about the way outpatient addiction and mental health care operated at the local facility level.
“We use scripts that push implementation teams to understand the interconnections between variables and perspectives and how the problem has trended over time,” said Rust.
“We recommend using Scriptapedia, which is a well-used, well-referenced group model building community that offers highly structured, established scripts to help you start seeing how all your pieces and perspectives fit together. These scripts offer a very participatory approach that builds co-ownership of the modeling process.”
Bringing system models to life with real-world simulations
Modeling can be extremely helpful for illuminating hidden pain points, but dynamic simulations take problem solving to the next level. Learning from simulations can build consensus among stakeholders about their decisions and catalyze new solutions without spending too many resources on trial and error.
“At the VA, we don’t want to lose a single veteran to relapse, overdose, or suicide,” said Zimmerman. “Since no veterans are harmed in a simulation, the virtual world offers significant opportunities for staff to learn, iterate, and solve problems. Any time you can dynamically adjust variables and get an immediate result, you’re improving understanding and engagement.”
Rust agreed that simulations are valuable for bringing conceptualized models to life but cautioned that participatory engagement is critical during all phases of modeling, from mapping causal connections, to validating data for model calibration, to selecting solution ideas to test with the simulation model.
“You have to spend a lot of time up front working with people on diagramming and understanding context, because that builds trust that the simulations are based on the right factors and that they will reflect something real and useful,” he said.
The two presenters also suggested that implementation teams should keep simulations grounded in reality.
“We avoid running theoretical simulations about having twice as many staff or three times the budget, because that’s just not very helpful,” said Rust.
Zimmerman concurred. “It’s completely disempowering to simulate something that is out of the team’s control,” she said. “Think very carefully about taking an empowering stance and focus on parameters that can actually change.”
Sharing the results in manageable formats to encourage future problem-solving
The insights from participatory system dynamics models and simulations can provide a platform from which to begin making large-scale changes.
“We know that a fundamental aspect of developing a system dynamics model is that the model is there to represent the problem, not the solution. But we also know that if no one uses the model to test solutions and make new decisions in the real world, then there’s no point in identifying the problem,” said Rust.
“Think about how your local teams are going to interact with the results. What will they get out of it, and how much time will it take them? We don’t want our work to be part of yet another meeting they don’t have time for. We want to empower decision-makers through the transparent review of their own data and processes.”
Rust and Zimmerman suggested trimming down information and workshopping into manageable chunks of time – small enough to fit into a routine daily huddle or existing staff meeting.
“At every stage of the project, we’re tailoring the resources into the type of digestible models that people can learn about in a staff meeting, because most of our stakeholders don’t have time for a five-day retreat,” said Zimmerman. “I would encourage everyone to think about how to break down your insights into small pieces so that learners can absorb them easily and make the most of the learnings we’re generating together.”
Participatory system dynamics modeling is an important complement to the other implementation science strategies we have discussed during this series. By integrating system models and simulations into the process of designing effective interventions, implementation teams can be sure that they are addressing the systemic factors that truly impact outcomes.
To learn even more about implementation science and its role in supporting effective decision making across the health care ecosystem, sign up for the latest updates on the 14th Annual Conference on the Science of Dissemination and Implementation in Health, taking place virtually on December 14-16, 2021.
This blog post highlights quotes and learnings from the panel “Introduction to Participatory Systems Dynamics for Implementation Science” presented at the 13th Annual Conference on the Science of Dissemination and Implementation in Health in December 2020. A full recording of the panel is available here.