Chapter 6: Emergency Physician Resource Planning
Current health workforce planning is siloed, focused on one profession in a single region. This patchwork also tends to ignore changing professional lifespans, demographics, and population health needs. We need a planning framework that reflects the reality of complex, interdependent health care labour markets. We need to create environments where people want to work and where workloads are manageable. (1)
Canada’s future emergency care systems will need a workforce that is competent, reliable, and adaptive. Integrated health human resource planning (HHRP) is about having the right healthcare providers and teams, in the right place/medium, at the right time, with the right training.
Resource modelling is one of several aspects of this planning that involves calculating how many physicians are needed now and into the future, to meet population needs. The key concept here is to meet population needs. Physician resource planning must follow clinical services planning, which must follow patient/population needs. Again, form must follow function. Chapters 3, 4 and 5 in this section are about just that: how do we set up the system to meet the emergency care needs of the population now, and into the future?
Deciding how many emergency physicians we need as part of the broader health workforce will depend on optimizing the number, distribution, and standards of EDs, matched with the best make up of their teams. Additionally important is how the various emergency departments are connected through transportation, and/or peer-to-peer virtual support.
A cautionary tale recently played out south of the border. (2)(3) Several dynamics in the American healthcare system (or at least in some parts of what is many different systems) have led to an erosion in timely and quality care. For example, many physician groups have been taken over by venture capitalists or profit-oriented management firms. (4) This has resulted in cost-cutting and profits for owners (over value-based care for patients) as the ultimate driver, potentially at the expense of quality. Mid-level providers are hired where emergency physicians used to tread, sometimes with the blurring of lines so the public is left unaware. (4) And rural care deserts have been created without an overarching strategy to maintain and improve emergency care at a regional/systems level. (5)(6)
We cannot allow this to happen in Canada. If our national association does not advocate strongly for timely access to quality emergency care, system decisions around the number, distribution and staffing of EDs will be made for us and may not be in the interests of the patients and populations we serve.
Background and Rationale of Modelling
There are several examples in the literature of Canadian physician resource modelling for specialties such as cardiac surgery, gynecologic oncology, emergency medicine and rural generalists. (7–10) The studies provide the current and forecasted need for these specialties at the national (7,10) and provincial levels. (8,9)
Although this is not meant to be an exhaustive review of physician resource modelling across Canada, many provinces have tried to perform more comprehensive analyses of physician resource modelling that encompass several, and in some cases all, of their medical and surgical disciplines. (11)(12)(13)(14)(15)(16) A variety of modelling approaches were used, with different underlying assumptions and data sources. Most of these past exercises concluded that more emergency physicians would be needed in the future to meet population demand.
Ontario, however, was an outlier. The modelling conducted in 2010 (12) showed there would be enough emergency physicians by 2011, and the surplus would grow to almost 500 physicians by 2022. (12) These projections were clearly incorrect, especially as EDs are now struggling to stay open under current staffing shortages. Why was the projection so flawed? Despite applying a standard modelling approach, the assumptions used to determine and depict the current and future needs, from both supply and demand perspectives, failed to reflect the system’s actual dynamics.
The description of this model provides little data to support health human resource planning (HHRP) but demonstrates that a “one size fits all” approach to different medical specialties does not work. Should we be using the same methods to plan our emergency medicine, primary care, medical and surgical workforces when the practices are so different in terms of population demand, infrastructure, and resource needs? In addition, making sound predictions of future HHR needs must be based on a reasonably precise grasp of the present and anticipated clinical services plan. This needs to be combined with the ability to adjust assumptions in response to known, evolving, and potential challenges in a specific care area of delivering clinical services.
Training Programs and the Challenge of Increasing the Number of Physicians
Up to this point, the discussion has focused on developing a physician resource model for emergency medicine to improve decision-making about the number of physicians necessary to meet the clinical care needs, and their geographical distribution. Within this lies an underlying assumption that if a greater number of emergency physicians are needed, then a mechanism is in place to achieve this goal. Currently, our resident training programs may not be able to adjust their enrollments dynamically based on the forecasted shortfall; instead, they produce a similar number of physicians within each specialty each year. More robust modelling makes a case for more adaptable residency training programs that can turn up or down various specialty outputs, based on more accurate predictions.
The Logic of the Savage Model
One challenge in using any model is the capacity to understand whether the actual system is appropriately represented. The overall goal of physician resource modelling is to find the right balance between physician supply and patient or population demand. (17)
The difference between supply and demand is often termed the gap, or variance. In our healthcare system, the supply or number of practicing physicians is a dynamic process of inflow and outflow. The inflow of physicians in a particular region is related to the number of residents finishing their training and staying to practice. It is also dependent on the transfer or migration of physicians from other regions, both nationally and internationally.
Physician outflow is more complex; physicians in a region can reduce their full-time equivalency (FTE) for several reasons including:
Retirement or death
To focus on academics, administration, other clinical (medical) disciplines, or changes in lifestyle; and
Outmigration to other regions.
Understanding the complexity of the inflow and outflow of physicians at a collective level can be challenging as it can vary by years in practice, the number of opportunities (both clinically and non-clinically) and may be regionally dependent.
Figure 16. The dynamic modelling of Emergency Physicians in a system (at the departmental, regional, or provincial level) where the supply of EPs minus the demand equals the variance.
The demand side of the equation reflects the population’s need for emergency department services. Forecasting this can be as simple as looking at historic trends: how many EDs does a region have, and how many total hours have to be covered? These predictions can also be more sophisticated by looking at population needs through changing demographics, population health status and the availability of healthcare resources (e.g., primary care). It can also include anticipated changes in clinical services planning for a region: how many EDs are needed, and how will their hours of coverage be determined? As with the supply aspect of the problem, a change in demand due to policy, funding, or more recently a pandemic, requires a re-examination of the underlying assumptions, model structure and available data.
The assumptions and potential variables around inflow, outflow, current supply, and demand for services can be seen in Figure 19. There will always be a trade-off in the number of variables proposed, and the administrative feasibility of their use. Following the Pareto Principle, (18) 80% of the predictive value may come from 20% of the variables. In addition, from a statistical perspective, these parameters are estimated and may have some degree of uncertainty. But the enemy of good enough is perfect.
Figure 17. The important variables aligning with the four components of the model: Supply, Demand, Inflow and Outflow.
Using sensitivity analysis to systematically set the model’s parameter variables allows the modeller to identify which variables may have the greatest effect on the results and determine how much confidence can be placed in them. This process is especially important when parameter values are unknown or estimated from little data (as demonstrated in Figure 18).
Figure 18. In this example region the gap between the broken black line (# of FTEs required) and the starting point of the 3 coloured lines (actual # of FTEs under different scenarios) is the shortage of EPs at year zero, which are then projected forward.
Policy Implications of the Savage Model
Multiple data sources need to be incorporated to create a model for physician resource allocation. This serves multiple and equally important roles. The modelling not only estimates current and future physician needs, but also identifies data gaps. It helps planners and policy makers better understand system dynamics, the current workforce, their practice patterns, the demand for services, and the future workforce required. Various policy interventions can also be tested in a low-risk virtual environment.
Forecasting allows decision-makers to identify the effect of maintaining the status quo but can also provide significant information about the effect of implementing different policy interventions. These models should also include some level of geographic integration to ensure regions that need more attention with regards to hiring and retaining physicians are identified. Developing a model, parameterizing it (expressed in terms of parameters), and planning for the next decade’s physician workforce is not a static process. As George Box famously said: “All models are wrong, but some are more useful than others”. We believe that a population needs-based (12), behaviourally-informed, (19) continually revised and updated model proposed by Savage will be far more useful than those previously based on supply or fee billing models.
Mechanics of the Savage Model
Physician resource models can be formulated in several ways, but one of the most popular would be a systems-dynamics framework. (7) All models must be based on variables that use current information for the specific system being studied. Getting data can be challenging which means relying on assumptions and expert opinion instead.
From a health system planning perspective, the COVID-19 pandemic taught decision-makers that planning needs to be flexible, dynamic, and responsive to an ever-changing environment. For this reason, physician resource models should not be run once and put on the shelf; instead, they need to be updated with new emerging data and shifts in the healthcare environment with repeated, iterative analysis performed at least every 2-4 years to account for evolving changes.
Future Directions
As mentioned, one of the greatest challenges facing physician resource modelling is the lack of necessary data to drive it. This is an issue for both physician supply and demand. To start, many physicians, administrators, and policy leaders would suggest we do not have a robust approach to estimate the existing shortfall in physicians. We also do not have a firm understanding about how their careers progress over time. At what rate do they reduce or increase their clinical full-time equivalent (FTE) (working in the emergency department)?
Each physician will have a personalized ratio of clinical FTE (ED +/- other clinical work), academic FTE, and an administrative FTE, which varies by region, hospital type, and stage of career. At what rate are these physicians migrating from one region to another? When do they retire? From the demand side of the equation, what is the best method for predicting future need? Do population health or demographics affect the number of ED visits? All these questions can be handled if we can agree on some common approaches, definitions, governance structures, and data-collecting accountabilities across the country.
In fact, once we know what baseline data/variables are required, attaining and maintaining up-to-date rosters and shift commitments for example should become an important expectation of ED site chiefs, and emergency care clinical network administrative responsibilities.
Conclusion
Before the pandemic, the media regularly highlighted the challenges in finding full-time work that many new surgeons across several disciplines faced. (20) These surgeons are often under-employed and not working to their full potential or scope of practice. Meanwhile, our emergency departments, primary care system, and addictions and mental health systems have suffered because of a lack of physicians. (21) If we truly want to meet Canada’s healthcare needs, our system of health human resources planning (HHRP) must be dynamic and responsive to the changing requirements of our population and workforce over time. (22,23)
Recommendations for Emergency Physician Resource Planning
Emergency Physician Resource Planning should adopt a needs-based, behaviourally-influenced, iteratively-updated approach (the Savage Model).
ED directors at the site level should understand the logic and variables of the Savage Model so that they can keep the current data points necessary for the model to be accurate.
Provincial ED leaders should understand the logic and variables of the Savage Model so they can influence ministerial and university policy makers around potential leverage points. This will reduce the current and projected FTE gap in ED coverage in Canada.
Health ministry and authority leaders must understand the link between clinical services planning and HHR planning (including impacts provider burnout) in emergency care systems.
Health ministry and authority leaders must be prepared to adequately fund and support a system that meets the current, future, and surge needs of its population.
References
Canadian Medical Association [Internet]. [cited 2024 Mar 1]. Workforce planning. Available from: https://www.cma.ca/our-focus/workforce-planning
Marco CA, Courtney DM, Ling LJ, Salsberg E, Reisdorff EJ, Gallahue FE, et al. The Emergency Medicine Physician Workforce: Projections for 2030. Annals of Emergency Medicine. 2021 Dec 1;78(6):726–37.
foem_people.pdf [Internet]. [cited 2024 Mar 1]. Available from: https://www.acep.org/siteassets/sites/acep/media/administration/foem_people.pdf
Kelman B, Farmer B. ERs staffed by private equity firms aim to cut costs by hiring fewer doctors. NPR [Internet]. 2023 Feb 11 [cited 2024 Mar 1]; Available from: https://www.npr.org/sections/health-shots/2023/02/11/1154962356/ers-hiring-fewer-doctors
Kelly M. The Looming Jobs Crisis for Emergency Physicians. Ann Emerg Med. 2022 Feb;79(2):A23–6.
currentacepframeworkofworkforceconsiderations_combined050521.pdf [Internet]. [cited 2024 Mar 1]. Available from: https://www.acep.org/siteassets/sites/acep/media/workforce/currentacepframeworkofworkforceconsiderations_combined050521.pdf
Vanderby SA, Carter MW, Latham T, Ouzounian M, Hassan A, Tang GH, et al. Modeling the cardiac surgery workforce in Canada. Ann Thorac Surg. 2010 Aug;90(2):467–73.
Savage D, Petrie D. LO77: Assessing the long-term emergency physician resource planning for Nova Scotia, Canada. Canadian Journal of Emergency Medicine. 2019 May;21(S1):S35–6.
Morgan JS, Graber-Naidich A. Small system dynamics model for alleviating the general practitioners rural care gap in Ontario, Canada. Socio-Economic Planning Sciences. 2019 Jun 1;66:10–23.
Chan JR. Modeling the Gynecology Oncology Workforce Using System Dynamics.
AHS Physician Workforce Forecast 2021-22.
Singh D, Hussein L, Karlj B, Newman E, Goodyear J, Hellyer D, et al. Ontario Population Needs-Based Physician Simulation Model. Ontario Ministry of Long Term Care. 2010;
Doctors Today and Tomorrow: Planning British Columbia’s Physician Workforce [Internet]. 2011 [cited 2024 Mar 1]. Available from: https://www.doctorsofbc.ca/policy-papers/health-human-resources/doctors-today-and-tomorrow-planning-british-columbia%E2%80%99s-physician-workforce
Government of Saskatchewan [Internet]. [cited 2024 Mar 1]. Other Ministry Plans and Reports | Saskatchewan Ministry of Health Plans and Reports. Available from: https://www.saskatchewan.ca/government/government-structure/ministries/health/other-reports/other-ministry-plans-and-reports#
Physician Resource Projection Models [Internet]. [cited 2024 Mar 1]. Available from: http://mchp-appserv.cpe.umanitoba.ca/deliverable.php?referencePaperID=76227
21-905-01W_annexe_qualification.pdf [Internet]. [cited 2024 Mar 1]. Available from: https://publications.msss.gouv.qc.ca/msss/fichiers/2021/21-905-01W_annexe_qualification.pdf
RAMLI MR, ABAS ZA, ARIF F, DESA MI. An Analysis Review Approaches Used In Health Human Resources Planning. International Journal of Computer Science and Information Security (IJCSIS). 2016 Aug;14(8):908–35.
Pareto principle. In: Wikipedia [Internet]. 2024 [cited 2024 Mar 1]. Available from: https://en.wikipedia.org/w/index.php?title=Pareto_principle&oldid=1210312828
Jeon SH, Hurley J. Physician resource planning in Canada: the need for a stronger behavioural foundation. Can Public Policy. 2010;36(3):359–75.
Owens B. Unemployed physicians a sign of poor workforce planning. CMAJ. 2019 Jun 10;191(23):E647–8.
Troubles in Canada’s Health Workforce: The Why, the Where, and the Way Out of Shortages pdf [Internet]. [cited 2024 Mar 1]. Available from: https://www.cdhowe.org/sites/default/files/2022-11/Commentary%20630.pdf
Marchildon G, Matteo LD. Physician workforce planning and boom–bust economic cycles: a retrospective on the Barer–Stoddart report. CMAJ. 2023 Jan 30;195(4):E162–5.
Islam R, Kralj B, Sweetman A. Physician workforce planning in Canada: the importance of accounting for population aging and changing physician hours of work. CMAJ. 2023;195(9): E335–40.