Chapter 2 How to use Modeling to Learn
Click on the hyperlinks or video thumbnails to watch the videos. Transcripts of the videos are also provided.
2.1 How can we get more patients better without more hours in the day?
How can we get more Veterans better without adding hours in our day? To get Veterans better, we have to account for the number of hours available in the day. Yet because we don’t have the right information about how we manage our time, we can pursue unrealistic treatment plans or improvement strategies that violate the laws of physics. Modeling to Learn encourages us to account for clinician time differently, to make our plans feasible locally. Most VA data systems offer snapshots of cross-sectional data, but this kind of information doesn’t help us account for the dynamics of evidence-based care over time. While many VA data systems are available, most display either quarterly lags from a high-level view or an in the weeds patient by patient view. Modeling to Learn helps to bridge these two views. Modeling to Learn introduces trends over time based on the clinic selections carefully reviewed by the staff on site. Trends help us zoom out to understand how we’re doing for most of the Veterans we serve. Modeling to Learn also introduces time-based definitions of care in the units of patients per week, appointments per week, or episodes of care per week. We use a week because over the last several years clinicians told us this is how they think in the clinic. There often is no typical day and a lot has happened by the end of a month. Defining care by week is more beneficial. High quality care in the clinic is always constrained by the available staffing capacity and the hours in the day to see patients, the number of weeks between visits, and the weeks of engagement over time. But do you have time to run these equations in your head or calculate them from the other dashboards you use? As an alternative, Modeling to Learn puts these components of care together to understand their interdependence over time. How can Modeling to Learn consultation help? Watch that video to find out.
2.2 How does Modeling to Learn benefit Substance Use Disorder or SUD programs?
How does Modeling to Learn benefit substance use disorder, or SUD programs? When Modeling to Learn began almost 10 years ago, as a partnership among patients, providers and policy makers across VA, we worked iteratively to define the dynamics of common care problems for the primary reasons Veterans seek addiction and mental health care. We found that if we focused on improving timely, high-quality evidence-based psychotherapy and pharmacotherapy for alcohol use disorder, depression, opioid use disorder and PTSD, we could support VA staff in meeting around 80% of Veterans needs for behavioral healthcare. We work with frontline teams and partners from the VA Academic Detailing Program and the Psychotropic Drug Safety Initiative or PDSI and developed the Medication Management Module of Modeling to Learn. This module helps manage the dynamics of evidence-based pharmacotherapies for alcohol use disorder and opioid use disorder based on their effectiveness in reducing craving, relapse, and overdose. You can find the Medication Management module in Modeling to Learn data mtl.how/data and Simulation User Interface mtl.how/sim. The Modeling to Learn Medication Management module supports improvement in the AlcTop or Alcohol Topiramate indicator from the Mental Health Information System or MIS and the SUD 16 indicator from the Strategic Analytics for Improvement in Learning or SAIL. How does Modeling to Learn help? Prescribers describe the need to better visualize how to locally optimize their medication management appointment supply, including top of license care from staff with and without a Drug Enforcement Agency X-waiver for OUD medication. Modeling to Learn helps a team assess this appointment supply, new patient start rate, no show or missed appointment rate, and the clinically appropriate return-to-clinic visit interval needed for a therapeutic response to alcohol use disorder and opioid use disorder medications. The Medication Management module helps teams and sites find the optimal number of patients who can be engaged in AUD and OUD therapies over time and helps them distinguish the flow of their patients who require antidepressants or other medication needs. Given that community needs and team staffing is dynamic and can change over time, these resources can empower teams to find local improvements that otherwise may be hard to find. Another common module for addressing Veterans SUD needs for group and other therapies is the Team Care module, which helps teams assess and optimize their overall mix of local multidisciplinary services as a function of staffing and patient needs. We are evaluating how Modeling to Learn improves Veterans who start and complete a therapeutic course of medication for AUD and OUD with research funding from the VA and the National Institute on Drug Abuse or NIDA. What about other common SUD comorbidities and presenting concerns? For example, how does Modeling to Learn benefit PTSD, clinical teams or PCTs? Watch that video to find out.
2.3 How does Modeling to Learn benefit PTSD Clinical Teams or PCTs?
How does Modeling to Learn benefit PTSD clinical teams, or PCTs? Since I’m at the National Center for PTSD, this priority is very close to my heart. When Modeling to Learn began almost 10 years ago as a partnership among patients, providers and policy makers across VA, we worked iteratively to define the dynamics of common care problems for the primary reasons Veterans seek addiction and mental health care. We found that if we focused on improving timely, high-quality evidence-based psychotherapy and pharmacotherapy for alcohol use disorder, depression, opioid use disorder and PTSD, we could support VA staff in meeting around 80% of Veterans needs for behavioral healthcare. Modeling to Learn helps teams better evaluate the local dynamics of balancing the needs of new and existing patients for evidence-based psychotherapies, particularly cognitive processing therapy and prolonged exposure. Teams described the challenges of starting new patients who had never had their needs met without compromising care for Veterans already engaged in therapy. Modeling to Learn enables teams to see where their Veterans get stuck, to decide when to graduate patients, or to recognize when the weeks between return to clinic visits start to get too long. Have you been trying to improve PTSD 56 in the Strategic Analytics for Improvement and Learning, or SAIL? We worked with local PTSD clinical teams from the beginning and have supported many more teams over the years who have engaged with the evidence-based psychotherapy programs or the PTSD mentorship program to meet these needs. The Psychotherapy module of the Modeling to Learn data user interface, and simulation user interface helps teams to see how they’re doing for patients who are starting psychotherapy, how many flow through to complete a therapeutic dose, and whether Veterans are getting near weekly therapy consistent with the PTSD treatment evidence base. There’s also the ability to assess emerging approaches such as masked or intensive outpatient psychotherapy and other scenarios that are commonly considered and implemented in PCTs. You may also be wondering about the flow from primary care or from primary care mental health integration, or PCMHI, teams into general mental health and specialty mental health programs like PCTs. Stepped care up and down the mental health continuum of care can also be optimized to local needs using the Modeling to Learn Team Flow module. Speaking of the care continuum, you may be wondering How does Modeling to Learn benefit Behavioral Health Integration program or BHIP teams? Watch that video to find out.
2.4 Can Modeling to Learn benefit Behavioral Health Integration Program or BHIP teams?
How does modeling to learn benefit Behavioral Health Integration Program or BHIP teams? What are the challenges with providing excellent team care? One challenge is the data sources to support team decision making jump from the individual patient and clinician counter all the way to the entire clinic or facility. It can mean that teams are flying blind to how they’re coordinating care for their Veterans. This can be especially hard when teams are short staffed or when staff, patient needs, and local policies continually change. It’s hard to know what improvements you might be able to find in your own team for coordinating evidence-based psychotherapies and pharmacotherapies over time. One way that Modeling to Learn can complement other BHIP resources is through the Data User Interface available at mtl.how/data. The Clinic Selection and Team Flow Selection tabs empower teams to carefully review the clinics that comprise their data views. Data are never all good or all bad. Modeling to Learn emphasizes transparency. Teams need to know what data they’re getting and its strengths and weaknesses for guiding a particular decision. The Clinic Selection tabs update every day to reflect all the Mental Health 500 series Stop Code, Clinic, or Grid updates that may be going on locally. Teams can save bookmarks that reflect their specific clinic selections, for example to include trainees or to select any of the clinics they have used within the last two years. This can be helpful when new BHIP teams are being formed or remapped. The team Patient Data tabs update daily too, so teams can view the patient diagnosis, encounter, health factor, measurement-based care, and high-risk flag information all in one place, and then the team can zoom out to filter and view these data as team trends. Coordinating care within a team is challenging. No single clinician can offer all the services a Veteran may need, so treatment plans must be coordinated in real time, over time, to support Veteran improvement. Clinicians may not have visibility on exactly how appointment supplies divvied up into intake evaluations, individual and group psychotherapy, medication management, or other adjunctive supportive therapies. What is particularly important about the Modeling to Learn simulation user interface at mtl.how/sim is the ability to assess not only the appointment supply, but even more important, the service proportions of patients that flow through to these services after intake or treatment plan review. One of the most common scenarios teams assess is how to respond to patient needs when they may not have the perfect staffing mix. Even when you know patient flow through care is divvied up into a mix of services, the dynamics over time are too much to understand in your head. The Modeling to Learn simulation user interface enables teams to quickly assess the dynamics of available appointment supply, service proportions, and the other two major factors that govern patient improvement, the return-to-clinic visit interval in weeks and the overall duration in care. We’ve often used Modeling to Learn to help teams find a way to ensure evidence-based care through the decisions that they make all day and find improvements in quality of care and quality of work life that they didn’t think was possible due to limited staffing. You might wonder How does Modeling to Learn support stepped care up to BHIP from primary care or primary care mental health integration, or from general mental health up or down into specialty mental health programs such as substance use disorder programs or PTSD clinical teams? Watch that video to find out.
2.5 How does Modeling to Learn support stepped care?
How does Modeling to Learn support stepped care? We developed the measurement-based stepped care module of Modeling to Learn to help teams assess episodes of care when stepping patients up or down across the continuum of care. General mental health or behavioral health integration program, or BHIP teams, and specialty mental health teams such as substance use disorder, or SUD, programs and PTSD clinical teams, or PCTs, use the team flow selection tab in the Modeling to Learn Data User Interface to define flow from their care up or down to another team. Teams make these team flow selections at mtl.how/data. Based on years of testing the Data User Interface, we found that teams are most accurate in defining who they refer Veterans to after completing an episode of care within their team, rather than where they receive referrals from. And the teams also define the gap between visits that they believe is locally most reasonable for defining a new episode of care in each setting. The primary evaluations in the Team Flow module of the Modeling to Learn Simulation User Interface include assessing the flow through care of Veterans with high-risk flags as well as the proportions of high- and low-symptom patients as they flow through care to recovery and step down or discharge. We know that higher care quality improves recovery among Veterans, but how does that vary as a function of the total patients a team is serving? Team Flow can be used to assess the total manageable patients in a team and the impact of the patient load on care quality. Teams can also account for use of community care, which reduces some of the patients actively managed in the team. But as teams know, it does not reduce this care management to zero as Veterans using community care may still trigger an emergency care response or coordination of other services within a team. Perhaps most helpful is the view at a glance of where all Veterans are accumulating, waiting to step up or down across the continuum of care. This can help leadership and teams understand where there may need to be improved referral flows or service agreements between these settings. Teams can evaluate the flows through care when implementing measurement-based care to reduce the time it takes to detect when Veterans may be getting better or worse, which can also help improve agreements or decisions about when transitions across service settings are warranted. Would you like to know more about using measurement-based care to improve the time to detect patient improvement or risk? Watch that video to find out.
2.6 How does measurement-based care detect patient improvement or risk?
Why use measurement-based care to improve the time to detect patient improvement or risk? To improve care quality in the flow of patients through care to recovery, the major value of measurement-based care is ensuring Veterans are getting appropriate care and are responding to the care they receive. Clinicians are likely familiar with the emphasis on measurement-based care, but from the perspective of care flow, what is powerful is reducing the time it takes for a team to detect whether a patient is getting better or worse. In the Modeling to Learn Data User Interface at mtl.how/data, the dataMeas tab provides tables of all the patient measures recorded at the VA Corporate Data Warehouse, whether they were captured via an evidence-based psychotherapy template, mental health assistant, or other approach. These include measures such as the GAD-7 for anxiety, PHQ-2 or PHQ-9 for depression, PCL-5 for PTSD, and the Brief Addiction Monitor or BAM. As a clinician, I like to use the patient-level views to sort scores from high to low to identify patients who may need to step up to a higher level of care or might benefit from referral to a specialty substance use disorder or PTSD service. The lowest score may indicate Veterans who can graduate from outpatient mental health care and step back down to primary care, which is an amazing day for the Veteran, their loved ones, and their treatment team. The vizMeas tab helps teams see trends in these measures to confirm when their efforts to implement measurement-based care are paying off. Teams can toggle to see what proportion of services include measurement-based care by comparing the vizMeas and vizEnc views. The Modeling to Learn Simulation User Interface at mtl.how/sim also empowers teams to assess the impacts of improved implementation of measurement-based care and how it can drive improvements in care quality. Do you want to learn more about the key drivers of higher care quality to improve recovery and prevent suicide? Watch that video to find out.
2.7 How can we improve SAIL with Modeling to Learn?
If you’re watching this video, you want to know about strategic analytics improvement in learning, or SAIL metrics, with Modeling to Learn. Modeling to Learn was developed to complement existing VA priorities. One important benchmark is SAIL. Some of the more frustrating aspects of SAIL are the lag in data to support daily decision-making in the clinic and emphasis on high-level facility views and, of course, how challenging it can be to improve sale metrics over time, especially when you’re short-staffed. In fact, when SAIL metrics change over time, either in a positive or negative direction, it is difficult to know what’s driving the change. Facility leadership and clinical teams may not know how to interpret these changes or what improvement strategy to try next. Modeling to Learn supports two primary dimensions of the SAIL mental health quality domain—care coverage and care continuity. But importantly, Modeling to Learn recognizes that these two dimensions may work against each other in a dynamic tension over time. If you’re expanding population coverage measures, such as PSY 32 for depression, PSY 38, or PTSD 56 for PTSD, or SUD 16 for opioid use disorder, then it may be more challenging to ensure continuity of care measures such as MDD43h, MDD47h, and PSY 33 for depression and PSY 39 for PTSD. Of course, the reverse is also true. If you’re doing great with care continuity on those measures, then it may be challenging to expand population coverage at the same time. I’m fond of calling this the physics of our care quality problems. What I mean is that we cannot change the laws of physics, such as the law of conservation, which states that matter and energy in a physical process cannot be created or destroyed, but can only be transformed. In Modeling to Learn, staff time is neither created nor destroyed. This means that when data is exported from the MTL Data User Interface and uploaded to the MTL Simulation User Interface, the scenarios a team or VA explores to find improvements in care quality will always account for or conserve the total local staff time available. This helps to avoid fixes that fail, such as something that improves population coverage in the short term but reduces care continuity in the long term. Have you ever felt like you were playing Whac-A-Mole® with these metrics, where you improve one indicator only to have a problem later somewhere else? Modeling to Learn can help. Let’s start with why Modeling to Learn Red is useful. And why does the Modeling to Learn Data User Interface provide new insights? Watch that video to find out.
2.8 How do five key variables drive care quality over time?
How do five key variables drive care quality? Modeling to Learn emphasizes the dynamics of care over time, which can be accurately simplified to the key time-based variables that drive care quality. But be careful, the important principle is that these variables operate together over time to define an episode of care. That means care quality cannot be improved without understanding how these variables influence one another. If you want to see what Lindsey’s talking about, navigate to the Modeling to Learn Data User Interface at mtl.how/data and review each care problem—care coordination, medication management, psychotherapy, team care, and team flow. Our partners across VA describe how challenging it is to review data in one information system and then in another and end up unsure how to reconcile them, especially when you think that they indicate a different course of action. Folks are extremely busy and any new data resources must really add value to be worth learning. Let’s see if these Modeling to Learn variables meet the commonsense test of value added. Well, what do you think the five time-based variables are that make up an evidence-based episode of care? All outpatient care is defined by whether you can get an appointment when you need help. We focus on this all the time in VA. But then, and this is critical, you must be able to be seen again to complete a therapeutic course of care adequate to meet your need. Clinicians told us that a week was the way they think clinically. So, in Modeling to Learn, teams make their clinic selections to obtain an estimate of their local new patient start rate in patients per week and their appointment supply in appointments per week. Then we define evidence-based engagement as the new patient wait time in weeks, time between visits and weeks, and the engagement duration over time, again in weeks. This is a simplified definition of an evidence-based episode of care that is accurate for time. Why is it wise to focus on the dynamics of care over time? That’s why Modeling to Learn Blue is useful. So watch that video to find out.
2.9 How does Modeling to Learn help improve medication management?
How does Modeling to Learn improve medication management? Medication management requires identifying a patient need, starting a medication, and evaluating for a therapeutic response over time. Some of the challenges and effective medication management over time include the realities that not all clinicians can prescribe. And of course, it requires follow up care to ensure that a patient is benefiting from their medication. So, when should the patient be seen again? One thing that can be so frustrating to prescribers is that a scheduler tells them when a patient can be seen again, rather than them telling the scheduler when the patient should be seen based on top of licensed care. The frustration is that medication management return-to-clinic visits may not occur at the appropriate interval to balance patient needs and risks. The VA has guidelines for some patient cohorts such as SAIL, MDD43h, and MDD47h, which define high quality continuity of follow up care as a valuation once every 12 weeks for Veterans diagnosed with depression who are using antidepressants. Other needs may reflect a different follow-up interval. For example, many addiction psychiatrists have described the appropriate medication management return-to-clinic visit interval for buprenorphine or methadone for opioid use disorder as every four weeks. These differences in medication needs, and the evidence-based pharmacotherapy standards that go with them, make it hard for prescribers to evaluate at a patient population level in a real time, especially when you’re completing another medication management encounter every 20 minutes. And of course, depending on the community at any given time, the staffing mix may change and the size or volume of patient needs can change too. Some SUD teams may have a high volume of patients receiving evidence-based pharmacotherapy for alcohol use disorder, whereas the modal presenting concern in a BHIP may be more likely to be depression. And the decisions the prescribers are making using current rules of thumb may be boxing them in to a pattern of care that makes it even harder to be responsive to crises or reduce future wait times. The Medication Management module of Modeling to Learn helps a team evaluate the trade-offs among all these factors as compared to the status quo over the last two years in the clinic. Comparing to the base case of no new decisions, the team can evaluate changes in their X-waivered MOUD appointment supply or evaluate and tailor the return-to-clinic visit interval by patient cohort. The goal is to help prescribers see what daily medication management decisions are adding up to over time and find feasible and effective local rules of thumb for clinically appropriate return-to-clinic orders in weeks that reduce wait times and ensure Veterans are getting better. For coordinating medication management in a multidisciplinary team such as a BHIP or SUD program, the Team Care module of Modeling to Learn can also help. If you want to know how Modeling to Learn helps to improve psychotherapy, watch that video to find out.
2.10 How does Modeling to Learn help improve psychotherapy?
How does Modeling to Learn help improve psychotherapy? VA led the nation in dissemination and training for evidence-based psychotherapy. Many clinicians are ready to offer Veterans the highest quality care available. So, what makes it hard to get Veterans through an evidence-based episode of care when they start? What I’ve learned from working with clinical psychologists like Lindsey and other VA therapists from around the country is that the dynamics of psychotherapy in the clinic are challenging for the same reasons most care problems persist over time. Clinical teams typically do not have good visibility of where all their Veterans are in care. You may know that you plan to do CPT session six today with an individual Veteran, but how many Veterans are currently in the middle of therapy now? How many appointments are on the books? How many weeks are there before they return to clinic for follow-up care and how many graduate when they complete a course of an evidence-based psychotherapy? And how many drop out of therapy after one session or early in care but return to the team later? Evidence-based psychotherapy is typically defined by getting near weekly psychotherapy and having clinically meaningful improvement, which usually starts around session 8 to 12. Many VA data systems provide useful counts and proportions of evidence-based psychotherapy templates and the SAIL continuity of care metrics to find quality based on the number of sessions Veterans get within a specific time period. But the SAIL metric itself is a quarterly snapshot, and we know Veterans don’t all start therapy at the beginning of a new quarter. In the Modeling to Learn Data User Interface, a one-year patient cohort is identified that includes all Veterans seen from 18 months to six months ago based on the clinic selections made with the Data User Interface at mtl.how/data. Then, for the Veterans seen within those 12 months, we look back to see when those individual Veterans began therapy and how they were engaged over time. Once this psychotherapy cohort is defined and uploaded to the Simulation User Interface at mtl.how/sim, we can look at the stocks and see where all the Veterans have accumulated, where they are in care in a typical week, and how many patients per week are progressing through key therapy milestones. Finally, at a local team level, psychotherapists can see how they are doing for most of the Veterans they serve. And even better, they can run experiments to see what the impact will be of making new decisions over time. You might wonder What if we keep making the same care decisions? Will things get better or worse? Watch that video to find out.
2.11 How does our appointment backlog extend weeks between visits?
How does an appointment backlog extend the weeks between visits? At first, this sounds like a simple system story that many clinicians describe as we work together to co-develop Modeling to Learn. But they also told us when an appointment backlog extends the week between visits, a long interval between visits effects care too. That’s right. As the number of appointments for a specific service grows on the books, then the clinic has no choice but to increase the number of weeks or push out the time between visits to get everyone seen. We call this the return-to-clinic visit interval. This quantifies the return-to-clinic order in a unit of time—one week—which is critical for finding realistic local improvements in our partnerships with sites. Okay, so the balancing feedback principle operating here is that if you have a clinic calendar filled up with appointments, it’s harder for new patients to start care and it becomes harder for all existing patients to be seen with appropriate therapeutic follow-up standards for psychotherapy and pharmacotherapy. Lindsey, you and Debbie also talked about balancing feedback in the video, What if we keep making the same care decisions, will things get better or worse? For an explanation of how start delays affect treatment decisions over time, watch that video. As I said at the beginning of this video, it’s not just that a growing appointment backlog forces teams and clinics to extend the number of weeks between visits to get Veterans seen. It’s that long return-to-clinic visit intervals impact treatment decisions and care quality. None of these effects occur in isolation. All these components of care interact over time. It’s why system scientists will often point out there are no side effects and systems, there are only effects. Some we want and some we don’t. As clinicians identify Veterans who might benefit from an evidence-based course of psychotherapy for depression or PTSD, they may find that many patients are waiting to start, or they know that those 90833 or 90834 psychotherapy encounters are occurring 12 weeks apart—or both. Clinicians recognize that they cannot see patients in a way that is consistent with an evidence-based episode of care and will have to start finding alternatives. Modeling to Learn can help. What a relief that hundreds of clinicians, managers, data experts, quality improvement leads, and patients have already worked with us over the last decade to develop scalable Modeling to Learn resources for clinical and improvement teams to use today. The Modeling to Learn Simulation User Interface accounts for all these feedback effects so the team can upload their data, review what they see, and get immediate feedback about what the most feasible alternatives are to ensure quality of care for patients and quality of work life for providers and local staff. Based on this discussion, you may wonder How can we better balance the needs of new and existing patients? Should we prioritize new patient start rate or weeks between visits for quality? Watch that video to find out.
2.12 How can we better balance needs of new and existing patients?
How can we better balance the needs of new and existing patients? Should we prioritize new patient start rate or weeks between visits? These trade-offs are challenging for clinicians when you know there is a whole community of patients who need help. When teams are struggling with the limits of their available time in the day to see patients, it can feel like a clinical, ethical, and even moral quandary about how to best balance patients’ needs for services with the staff resources available. Of course, staff resources are always changing. We know many teams that are excelling in providing the highest quality addiction and mental health care available. But the behavioral health workforce shortage in the US is much bigger than VA. And VAs and teams need the right tools to make sure Veterans get the care they need for recovery. Challenges balancing the new patient start rate and the weeks between visits for existing patients apply systems thinking insights that we’ve talked about in other MTL videos, such as the physics of conserving staff time in order to ensure a clinically beneficial, realistic approach to care decisions and quality improvement that meets VA quality standards and meets Veterans needs for evidence-based episodes of care. The Modeling to Learn Blue Simulation User Interface available at mtl.how/sim enables a site or team to evaluate these two balancing system stories as a function of their data for the last two years exported from Modeling to Learn Red Data User Interface at mtl.how/data. As we talked about in the How does an appointment backlog extend the weeks between visits? video and What if we keep making the same care decisions, will things get better or worse? video, balancing feedbacks occur in any system that has a goal, including our healthcare and clinical systems. Balancing feedbacks occur in systems that have constraints of resources, staff, and time. And as a result, the trends that occur over time tend to reset to a status quo or oscillate around the status quo. A brief Modeling to Learn consult uses the site or team reviewed local data plus simulation to efficiently find local improvements that account for all these balancing trade-offs. We aim to help clinical and improvement teams find a couple empowering clinical heuristics, or rules of thumb, that are more effective for ensuring evidence-based care to get more Veterans better. When we partner with a VA or a team, we aim to find the lightest lift we can, like increase groups by 10% or adjust the return-to-clinic order for three weeks for these presenting concerns. And we often find something small that has a big payoff for Veterans. Do you want to be empowered to leverage the feedback, flow, and volume of your local care system? Watch that video to find out.
2.13 How can we leverage the feedback, rates, and volume of our local care system?
Do you want to be empowered to leverage the feedback, flow, and volume of your local care system? In any system, leverage describes the places where a small shift in one thing can produce big changes in everything. For a big benefit for Veterans in their care teams, the primary places to focus are where the greatest volume is, the fastest flower rates are, and of course to account for feedback dynamics over time. When we partner with the VA or team, we aim to find the lightest lift we can, like finding for one BHIP that increasing group psychotherapy by 10% eliminates wait times and gets five times the number of Veterans near weekly therapy. Or, adjusting the return-to-clinic order by three weeks for some presenting concerns can provide the same medication management supply as hiring another prescriber. We can often find small care decisions that when implemented frequently over time, have a big payoff for Veterans because they add to more than the sum of the parts in the interdependent system over time. The last thing we want to do is make big changes that are difficult for Veterans and clinical teams that produce limited benefit over time. In my experience over many years now, finding something clinically powerful that leverages the dynamics of the system can be a huge boost to morale, productivity, and staff empowerment. You might wonder How can you find powerful insights that you are sure will work locally? We do this by reviewing the Modeling to Learn Simulation User Interface stocks to see where all the Veterans are accumulating the most. We review the rates to see where the floodwaters are rising fastest and require action. Or where things are trending in the right direction and our attention could be better focused elsewhere. Modeling to Learn consultation offers clinical and improvement teams real-time efficient support. We can do this quickly because we’ve partnered with patients, providers, policy makers, data experts, evaluators, and quality improvement consultants from the frontline to VA Central Office for nearly 10 years now to define the primary drivers of care problems and care quality using system dynamics. Modeling to Learn consultation involves empowering clinicians, managers, improvers, data leads, and all of VA to leverage the system for getting more Veterans better. How can Modeling to Learn consultation help? Watch that video to find out.
2.14 What key drivers of higher care quality improve recovery and prevent suicide?
What are key drivers of higher care quality to improve recovery and prevent suicide? A central idea of Modeling to Learn resources is that delivering consistent, timely, high-quality care is challenging for teams because we don’t accurately account for the primary dynamics driving care quality over time. You might be thinking Care dynamics, what does that mean? We literally mean that due to time pressures and cognitive limits, we don’t effectively understand relationships among key care decisions that we make every day over time, especially decisions made in the aggregate across a clinical team. For example, we may try to break down problems by looking at referrals or wait times, but we don’t look at how clinical teams’ knowledge of wait times influence referrals to a specific service or team. In the Modeling to Learn Data User Interface, we define care dynamics in terms of rates over time, like new patients per week, so that we can understand the flow of Veterans through care. As another example in terms of care quality, recovery and suicide prevention, the Modeling to Learn Team Flow module looks not only at the number of patients served by a team or site, or the number of patients with a high-risk flag for suicide, but also the time it takes to unflag high-risk patients in a team and the typical time to improve within the team over the last two years. The Modeling to Learn Simulation User Interface uses an interactive structure fed by local team data from the Data User Interface. The Modeling to Learn Simulation User Interface visually depicts accumulation of patients in desirable or undesirable states of care, like waiting to start a new service. These states are defined by their inflows and outflows. A lot of clinical teams and mental health leadership wonder whether they can get any new insights when they are already doing everything that they can to support the Veterans in their care. Would you like to know how to improve SAIL with Modeling to Learn? Watch that video to find out.