Reskilling a workforce quickly by “Rehearsing the Future” using a special simulation method.
This is a story about how we engaged a cross-functional, inter-organizational team to meet specific challenges facing their industry, health care for Medicaid patients, using a simulation exercise. Rather than begin with the “capabilities” we challenged the team to start with the goal and work backwards from that goal to develop what was needed to achieve it. The collective team ended up developing unanticipated capabilities in response to this challenge. What emerged was a targeted method of re-skilling the teams for the challenge before them and at the same time road-testing the relevance of those skills for the problem at hand.
Changes in business have caused major disruptions in how people do what is essentially the same job, often requiring them to develop new competencies and integrate emerging technologies as tools in order to do their work. As a consequence, it’s likely impossible to anticipate ahead of time what skills will be needed for a specific role in an organization or to plan future training in any detail. ACSI Labs’ approach to outcome-oriented simulations may be a way to address the reskilling requirements with the current pace of change. In our simulations the teams are given the goal, and resources to explore solutions, but no “instruction.” Simply put, the exercise — organized by the goals of the organization — also identified which skills needed to be developed, empowered the workforce to figure it out, and then gave them the means to deploy these solutions and skills in their workplace practices.
In this case, the companies were a not-for-profit health care insurance management organization and their “customers,” i.e., clinics in poor communities. The patients were low-income or welfare recipients who were now “members” of the non-profit’s health plan. We will call this non-profit MPH. Simply put, the primary mission of MPH is to eliminate the well-known “outcome disparities” between the privately insured and those who depend on Medicaid or other forms of public assistance. It was well known at the time that poorer patients have worse prognoses for the same conditions; i.e., impoverished or socioeconomically disadvantaged persons tend to get sicker and suffer poorer overall outcomes than privately insured people of the same age or with the same illnesses (Adler &Newman 2002; Braverman, Egerter, and Williams 2011; Woolf & Braverman 2011; WHO report 2008). The purpose of the “membership” in MPH was to provide these patients with an insurance model closer to that enjoyed by private insured patients. We were brought in when MPH’s plan to execute this strategy failed with the clinics.
It turned out competing goals were at play. The staff at the two kinds of organizations also had a different mental model of what would contribute to “wellness.” Although the common aim between MPH and the clinics was to eliminate outcome disparities, the insurer also wished to keep the costs per patient down while the clinics struggled to stay afloat financially. They saw MPH as a provider with unlimited financial resources and assumed the problems of the patients could be solved with more services; a “more=more” assumption. They were also concerned with staying open for these patients. As a result, the clinic employees were focused on cash flow to solve their operating cost problems and had become skilled at targeting patients for whom they could claim the highest reimbursements.
In contrast, the major skill set of the insurance company’s employees was cost auditing and a commitment to a preventive care model. Being insurance specialists, they knew that preventive care could cost less and produce very good patient outcomes. However, their approach assumed that patients could easily make themselves available for screenings and preventive care visits.
That said, MPH’s “membership” model enjoyed considerable success in reducing health outcome disparities among some segments of patients (pregnant women and children) through a commitment to aggressive preventive care and socializing the benefits of wellness care among mothers. The difficulty was getting the same results with other kinds of patient populations (e.g., young men, elderly, and chronically ill but not acutely ill patients). MPH contacted us after reading an article about how we helped for-profit organizations break out of entrenched practices (Bower 2004). They were hoping our methods could apply to a non-profit health care organization as well.
Using interviews, examining key artifacts, and conducting an economic study of the state of healthcare and outcome statistics, we landed on an analysis of the “whole” problem, including the clinic’s/provider’s role in the care of the Medicaid members. The MPH executives indicated a desire to craft a solution that worked around some of the issues with providers. They were committed to eliminating outcome disparities but believed providers (clinics) were contributing to the problems (understaffing, inconsistent records, the high rate of walk-in acute problems, the high rate of uninsured and non-paying patients.) They also believed clinics were struggling financially, which led to providers billing MPH for unneeded services. MPH believed that placing an MPH person in the clinic who could help socialize the preventive care model would fix some problems.
It was clear these MPH “helpers” were not welcome; the clinics quickly figured out that they were there to prevent overbilling or ordering unneeded services. On the other hand, MPH personnel did observe that “social issues” were indeed the cause of many patients not receiving preventative healthcare. This was not factored into MPH’s original model of patient realties.
We started to catalogue the social issues. Most were concrete day to day issues that prevent patients from making and getting to their appointments. These include lack of childcare, lack of transportation, mental illness or physical frailty, responsibilities at home, the economic distress of taking time off work, or an inability to get permission to leave the job for an appointment.
We concluded — after our analysis — that to adequately address improved patient outcomes through preventive care practices, we must take an approach that provides mutual benefits for payers and clinics. Specifically, we felt that a number of things about the way the payers and clinic coordinate their tasks had to change, and this meant new skills on the part of MPH employees and the clinic staff. For example, it became clear that successfully implementing preventive care would at the very least involve a unified medical history for each patient, across all providers. It would also require a unified understanding of patients’ needs and constraints. Further, the financial challenges of practices needed to be addressed. Many were simply poorly run as businesses, finding survival challenging. (See Figure 1 for the “competing forces” that must be resolved for successful implementation).
We decided to help overcome these differences by having staff from MPH and the clinics participate together in a kind of role-playing exercise in which they would solve the collective set of problems together in a simulation. The exercise was a version of a general approach our company uses called the “Strategic Rehearsal”.
The Strategic Rehearsal method is a unique kind of simulation exercise our team has developed over several years with dozens of companies (DiBello, 2001; DiBello Missildine & Struttman 2008; DiBello 2019). This approach allows participants to rehearse, experiment with, fail, and rework novel approaches to organizational problems while being held accountable to non-negotiable outcomes. It involves a reality analogous setting with realistic constraints and opportunities for navigating the options. The ultimate goal of the Strategic Rehearsal method is to help challenge the existing mental models participants use to approach problems by providing a platform of new leading goals and activities to help reorganize entrenched behavioral patterns.
The Strategic Rehearsal Method — like any “wargame” method — is hypothesized to capitalize upon workers’ ability to implicitly learn the temporal correlations and patterns across episodes and events that are typically produced by dynamic situations in the real world (Gureckis & Love, 2019; Franco-Watkins, Rickard, & Pashler, 2019). This ability to implicitly learn dynamic patterns is at the foundation of intuitive decision-making (Evans & Stanovich, 2013; Osman, 2014), which involves decision making by situational pattern recognition and which represents the usual mode of decision-making by individuals with expertise in a wide variety of naturalistic contexts (Klein, Kahneman & Klein, 2019; Patterson et. al 2009) such as in health care delivery systems.
Although implicit learning of temporal patterns and intuitive decision-making is considered to be robust (Stark, Scott, & Todorov, 2018; Newell & Shanks, 2014), it typically operates in situations for which the linkage between events and episodes occurs across short time spans. In a health care delivery system, as well as in other real-world (e.g., business) applications, the linkage among key variables may occur across relatively long time spans for which experiential learning would be difficult
Strategic Rehearsal’s use of highly compressed time periods overcomes this constraint and optimizes this kind of experiential learning, making action and consequence links more obvious. (For more information about the Strategic Rehearsal method and how it’s been used see www.futureviewplatform.com and click the ”research” tab. )
The MPH Strategic Rehearsal: Goals
As indicated, we began our process with the MPH by identifying the core competing forces that drive organizational success and how at the same time those forces can act as barriers to success (See Figure 1 for the competing forces present in the MPH Strategic Rehearsal).
The MPH Strategic Rehearsal was designed so that participants from the practices and payers would have to work out a solution that was jointly beneficial to each set of goals, even if they seemed contradictory. Key to achieving these goals was that the clinics had to refocus the ways of thinking about the problem from the evidence in the patient data, instead of focusing purely on the patients who come to the clinic on a given day. Because of incomplete medical records, however, practices often have trouble identifying at-risk patients. In contrast, the payers often have extensive records and information on patients but are restricted in how they share that information with practices. In addition, the insurance companies had to be assured that identifying at-risk patients would show a tangible financial benefit for them.
No method for resolving the contradictory goals was built into the Strategic Rehearsal; however the resources available to each organization were present. The idea was to find opportunities in these resources for solving the problems.
The MPH Strategic Rehearsal: Design and Execution
In a large event room, we built an environment which included three “practices” of various sizes, represented by Barbie doll scale clinics and a “payer ” company which could pay claims for Medicaid eligible patients. Participants were divided into three teams. The “Practice Team A” ran a “large” clinic. The “Practice Team B” ran two smaller clinics. And the “Payer Team” acted as the payer, modeled heavily after the MPH itself. The Practice Teams were made up of actual doctors and clinic staff that have MPH member patients. The Payer Team was made up of MPH staff. Each Practice Team received a file containing complete charts for all their patients (See Figure 3), as well as financial records and a budget. These records emulated the electronic record keeping system being used at the time, (which also allowed records to be printed out). The patient demographics, histories and current medical conditions were generated by our FutureView Event Generator software, which uses A.I. to generate unfolding future events, data and other patterned data for our Strategic Rehearsals (DiBello 2018). We generated over 3,000 patients, aged 6 months to 90 years with a range of typical conditions, with detailed histories going back as much as 25 years. These records were meant to emulate an electronic chronological patient history and included the results of any diagnostic tests. The demographics of the patient query database were modeled after the known demographics of the Medicaid population. Actual patient data were not used.
The clinics were represented with wooden models outfitted with electronic sensors. Each had miniature examining rooms with slots on the outside for placing charts and claim forms. The forms and charts were actual size for easy reading. The models of the clinics were designed so that when an examining room was being utilized a green light was lit and the lights-on time was recorded by a computer. When the room was in disuse or the patient was not properly prepared (e.g., no chart, no claim form), a red light in the room came on. The red-light time was also recorded. This way the practice could get metrics data on both their utilization and whether or not the patient had been properly attended to.
The Practice Teams did not get credit (reimbursement) for the visit unless the correct chart and correct reimbursement form was in a slot outside the examining room and the patient was inside the room. Our staff worked in these fictional clinics, helping with accurately filling out charts; this ensured the data represented by the practice team (diagnosis, treatment, follow-on appointment) was captured. The Practice Teams got paid by submitting reimbursement forms to the Payer Team. The Payer Team would verify that the patient was in their database, that the services were warranted and reimburse the practice accordingly. Revenue for Practices increased as they filled more of the rooms with patients (more green lights on) and got more patients through by mobilizing adequate resources to get them into the clinic. Revenue per patient was determined by their condition, type of visit, and tests. Costs to the Payor were determined variably by whether or not the patient came in for a regular visit or ended up in the ER.
To create a focus on preventive care, several major chronic health issues were represented, including diabetes, obesity, heart disease, and general indicators of “wellness.” The Payer Team had a database of claims histories (from now on we will call this a “query database”) for every patient in their membership, with the capability to easily make queries of this database for generating reports. Once the database was generated, it could be used to generate various reports that the teams might need by conducting simple queries. However, it was up to the clinic teams to discover this and design meaningful queries.
The “patients” database was programmed with A.I. to make it dynamic . For example, someone at risk for heart disease would get sicker with time if not treated. Conditions such as obesity and a smoking habit could accelerate this path.
In this toy environment the teams could “wargame” several months in the life of the clinics and their patients, trying various approaches to achieving their goals, with “game periods” of 10 minutes representing a week in the life of a healthcare system. Each team had 8 hours of “gaming” each day and data analysis time to devise and execute their solutions at the beginning and end of each day the exercise was conducted. Each clinic had the medical records in electronic and paper form for all the patients ever seen by each clinic (but not shared between clinics) and the ability to generate “tickler” reports for each game period. These tickler schedules were based on past history only; i.e., patients due to be seen based on the timing and findings of their last visit (e.g., came in for headaches; finding was high blood pressure, 3 months medication prescribed). The “payor” had access to a claims database with the same patients and same history. In addition, their version of the database also had a multi-year history for every patient who was or ever had been a subscriber, across providers (over 3000 patients). The Payer team’s laptops were connected to that comprehensive database. All teams had the ability to run queries on their databases and print various reports.
The last element of the game was the “Patient desk”. Here professional actors had scripts, personal histories and paper dolls representing patients. When a clinic or payor contacted a patient, they went to the patient desk, and an actor played the role of a patient in any interaction regarding getting the patient to visit a doctor, address an issue or even pay a bill.
The Clinics had helped us identify 20 “clinical and social” issues that were common factors preventing patients from complying with health care. The actors played the patient role according to scripts that showed which social issue affected them. For example, an elderly man with a chronically ill wife may not wish to leave her alone to go to a doctor appointment. It was up to the clinics to arrange a solution to his problem and influence the payor to fund it. This resulted in some negotiations between the provider and payor teams to create solutions that were covered. In turn, that could result in the innovation of standard services which could be applied to any patient who had the issue. If a solution was found, the paper doll representing the patient was handed over to the clinic and the payor paid for the care. However, once the “doll” was in the clinic, the right care needed to be identified that would provide the greatest benefit. If the clinics did not administer the right care, the patient would get sicker. As the exercise rolled out, in every game period, patients’ status could change depending on a number of factors, including whether or not they received preventive care for a potential condition. This was managed by A.I. in the patient database.
In summary, participants in the exercise had three key goals
(1) Closing the outcome disparity between privately insured and publicly insured patients, mainly through preventive care practices for patients.
(2) controlling costs to payers, and
(3) increasing revenues for practices.
They had to meet all three goals equally. For example, accomplishing only two of the three was not “winning.” Each team could play through twice on two consecutive days. Below is a table of the timeline for each day. On Day One, the presentation had to include a plan for improving on Day Two.
On both Day One and Day Two, all of the same parameters were present except that new patient charts and claims histories were automatically generated to prevent the teams from memorizing patients and their histories.
The Practice Teams had to run the practices, seeing enough patients per day to generate revenue to cover the costs of operation. The Payers had to figure out how to minimize costs. On day two, at the suggestion from the clinics, the payers used their extensive data to help the practices identify candidates for preventive care using their access to patients’ claims histories. After a while, the doctors from the clinic teams decided to tell the Payers what to look for as pre-morbid indicators, hence helping to design informative queries. As the exercise unfolded, Payers would query the claims history database to identify at risk patients and work with the clinics to overcome obstacles to getting them treated. Some of the innovations that the Payer agreed to fund included: group visits that provided childcare, bussing patients in and giving them lunch or visiting them at home on specific days.
The teams ended each day by preparing a presentation for the whole group after a period of analyzing the results. They had to answer three questions:
1. How did we perform compared to our goals and expectations?
2. What prevented the progress we expected and what contributed to achieving our goals?
3. What will we do tomorrow to meet our goals? They were not allowed to select a spokesperson.
Each team member had to be equally prepared to explain the results and the go-forward plan. This was deliberate; we wanted a shared understanding of the whole problem among all team members. Spokespersons were selected randomly from drawing names out of a bowl.
Results: Improvement between Day 1 and Day 2
As indicated, we conducted this exercise over two days. We warned them that the first “run” is rarely successful but that they would get another try the following day. During the first day of their exercise, no team did well. We got them started with a query on the clinics’ database that produced a list of needed appointments based on past medical history, visit history and age. This generated a set at 360 patients (See Figure 9).
As can be seen, on Day 1, practices fell far short of their goal (seeing 180 patients). Practices had difficulty meeting their break-even revenues and many of the patients due to be seen were not brought in. The preventable claims cost was high; people who got sicker ended up in emergency rooms or hospitals, hitting the payor with significant costs and providing no income for the clinics.
The second day all the teams did well. As we can see from Figure 9, Practice Teams were able to see 400 patients on Day 2, exceeding their target by 40 patients. This represents a 222% increase in the number of patients seen between Day 1 and Day 2, providing a huge incentive for the practices which were looking to increase the number of patients they were able to see. At the same time, total costs were reduced from $81,000 to $48,500 mainly because the right treatments were administered. This represents a total cost savings of 60% between Day 1 and Day 2. The savings were due primarily to the prevention of acute illness.
Furthermore, queries developed within the Strategic Rehearsal by the participants playing the role of practice personnel uncovered additional at-risk patients (due to symptom history and premorbid indicators as shown in claims) that they knew from experience indicated additional preventive screening or treatment. For example, one cluster included: frequent ER visits, a chronic disease, a history of missed preventive visits and poor diagnostic test results. This also added a significant boost to practice revenues while at the same time keeping the cost per patient very low, which was nearly 25% of what it had been per patient on average.
Figures 10 and 11 show the overall revenue performance for Practices and the cost savings for Payers during the exercise. In general, on Day 1, costs were much higher with fewer patients being seen. Most important, on Day 1, the cost for preventable measures for patients covered by other plans was more than 3 times the cost for preventive care for the MPH members alone. On Day 2, costs for preventable claims were negligible for MPH members and for patients covered by other plans. In other words, the preventive measures that the MPH put into place for their patients had a spill over beneficial effect for all patients in the practices, even if they were not MPH members. The difference in income to the practices as a result of stepping up preventive visits was more than 100%.
Most important, Payer and Practice teams worked out better coordination systems for information flow and records sharing, as well as better identification practices for at-risk patients. We believe that this increase in coordination of information and identification is likely the key skill developed among these teams. These solutions should translate into better care for at-risk patients, particularly among low-income populations who often require extensive “social” services to assist them getting their care. Based on these solutions, doctors were able to design better follow up procedures, allocate time with patients according to a more nuanced understanding of their needs, and increase services for those who require special services, such as transportation for visits.
In sum, on Day 2, the teams discovered three things:
- Working with the Payer to analyze claims histories, Practices could identify patients in their practices who were at risk for becoming sicker, based on all medical visits, including emergency room visits and visits to practices other than theirs. The Payer and practices together identified a number of claim-type clusters that could indicate that a patient might be at risk. Using queries, the teams worked to generate a list of the at-risk patients and contacted them. If the patients had “social” or “clinical” issues preventing a visit, the Practice worked with the Payer to overcome these or arrange a home visit. Such increased identification of at-risk patients represents an increase in patient quality care. Participants felt that they learned not only the “skill” to do queries, but also formulating the right questions to ask of the data.
- Following up on at-risk patients and patients identified by the query database filled the examining rooms to capacity and helped the practices meet their revenue goals without ordering unnecessary tests and avoiding most of the reimbursement disputes with the payer. The practices learned that recurring monthly revenue from preventive care was not only lucrative, but predictable. As a result, they focused on devising and implementing innovations to improve attendance.
- Patients needing preventive care the most were now not only identified in a more comprehensive manner, but innovations were mobilized to overcome the traditional barriers to preventive care (e.g., transportation services, home visits, childcare and follow up calls).
Discussion and implications for re-skilling
There are many ways to interpret these results in terms of learning theory. However, the point of this article is to point out the potential for quickly reskilling a workforce to adapt to changes in the workplace, or even changes in the purpose of the work. The exercise developed capabilities in the participants that addressed challenges they were facing in their real work and gave them practice at refining them.
Our team have used this method with dozens of companies, but usually to help the leadership wargame strategy for growing their company or overcoming a constraint in the marketplace or changes in access to capital (economic or political trends). In this example, we achieved a fast way to reskill a workforce in two kinds of organizations.
Many organizations are undergoing dramatic changes in technology (such as electronic patient records, in this case). These technologies offer opportunities for being more effective, but also require different skills, not only in the technology itself but also in working effectively with customers, partners, and suppliers. In this case, rehearsing the new reality helped develop staff, but also revealed some clever ways that people make the best use of new tools. For example, we would not have thought of the clinic-designed queries to identify at risk patients based on hands-on experience with patients. The insurance company had not thought of this either but were very pleased with the results.
Finally, we think an important feature of this exercise is that we got agreement from the collective management teams to allow the participants to implement the solution they devised if the exercise results proved it might work in real life. As a result, some form of all the innovations were implemented among real patients. In the end, MPH grew over 3X in subsequent years after this brief period of struggling.
References
Adler, N. E., & Newman, K. (2002). Socioeconomic disparities in health: pathways and policies. Health Affairs, 21(2), 60–76
Bower, Bruce. “Reworking Intuition.” Science News, vol. [Volume number], no. [Issue number], 2004, pp. [page range].
Braveman, P., Egerter, S., & Williams, D. R. (2011). The social determinants of health: coming of age. Annual Review of Public Health, 32, 381–398.
Commission on Social Determinants of Health. (2008). Closing the gap in a generation: Health equity through action on the social determinants of health. World Health Organization.
DiBello L. (2018). Expertise in Business. The Oxford Handbook of Expertise — Oxford University Press.
DiBello, L. (2001). Solving the problem of employee resistance to technology by eframing the problem as one of experts and their tools. In Linking Expertise and
Naturalistic Decision Making. Salas, E. & Klein, G. (Eds.) Lawrence Erlbaum Associates. Mahwah NJ.
DiBello L., Missildine W., & Struttmann M. (2008). Intuitive expertise and empowerment: the long-term impact of simulation training on changing accountabilities in a biotech firm. Mind, Culture & Activity, 16, 11–31.
Evans, J. S. B., & Stanovich, K. E. (2013). Dual-process theories of higher cognition advancing the debate. Perspectives on Psychological Science, 8(3), 223–241
Franco-Watkins, A. M., Rickard, T. C., & Pashler, H. (2019). Taxing multitaskers: memory retrieval during cognitive skill learning impairs subsequent text study. Psychonomic bulletin & review, 26(2), 503–509
Gureckis, T. M., & Love, B. C. (2019). Autonomy and the deep structure of learning in human and machine. Current Opinion in Behavioral Sciences, 29, 115–120
Klein, G. (2008). Naturalistic decision making. Human Factors, 50, 456–460.
Klein, G. (1998). Sources of power: How people make decisions. Cambridge, MA: MIT Press.
Klein, G. (1997). The recognition-primed decision (RPD) model: Looking back, looking forward. In C.E. Zsambok & G. Klein (Eds.), Naturalistic Decision Making. Mahwah, N.J.: Lawrence Erlbaum Ass.
Klein, G., Kahneman, D., & Klein, G. (2019). The making of a decision theorist: A conversation with Daniel Kahneman. Organizational Behavior and Human Decision Processes, 153, A1-A8
Newell, B. R., & Shanks, D. R. (2014). Unconscious influences on decision making: A critical review. Behavioral and Brain Sciences, 37(1), 1–19
Osman, M. (2014). Future-minded: The psychology of agency and control. Springer
Patterson, R., Pierce, B.P., Bell, H., Andrews, D. & Winterbottom, M. (2009). Training robust decision making in immersive environments. Journal of Cognitive Engineering and Decision Making, 3,
Woolf, S. H., & Braveman, P. (2011). Where health disparities begin: The role of social and economic determinants — And why current policies may make matters worse. Health Affairs, 30(10), 1852–1859.
