Workforce burnout from a systems science perspective

ABSTRACT

Workforce burnout is an increasing problem across many industries and professions, with significant impacts on both manufacturing and service sectors. For example, burnout is a major problem for more than 80% of healthcare systems, with the costs of replacing doctors who leave their job and reduced clinical hours related to burnout are estimated at $4.6 billion annually. Productivity is reduced, mental health is affected, family relations are weakened, and a solution to all of this is not clear. More than 10 years ago, the global cost of burnout was estimated to exceed $300 billion annually. Major societal events, such as the recent ongoing COVID-19 pandemic or political unrest, can further exacerbate individual burnout and its impacts. Advancements in burnout research over the past two decades have mostly been to develop methods for measuring and classifying burnout and in lessons learned empirically from various intervention implementations, but with little-to-no analytic modeling research to help inform effective policies. In a recent report in fact, the National Academy of Medicine emphasized the need to develop analytic models that better quantify the extent of the problem in a way that translates into actionable results in addition to approaches for understanding the impact of interventions. The overall aim of our proposed research accordingly is to develop and apply analytic disease progression models to help understand burnout dynamics and evaluate the long-term benefits of interventions prior to wide-scale implementation testing. The proposed dissertation includes three fundamental contributions. First, we develop and introduce two disease progression models of individual and organizational burnout based on Markov chains, parameterized and linked from limited data via optimization and simulation models. We also illustrate the use of the developed models to estimate and compare the relative effectiveness of various strategies and interventions to reduce burnout, with a focus on estimating long-term impacts from limited early testing data, contributing to pre-randomized trial methods. Second, we leverage the models to estimate the effect of COVID-19 on two healthcare professional populations in two case studies. Finally, we propose several potential methodological extensions to disease progression modeling including investigating the effect of higher order nesting, bootstrapping and time non homogeneity. Results indicate that the disease progression models of the proposed type can accurately model individual and institutional burnout progression to help better understand the dynamics of burnout and analyze the effectiveness of potential interventions to make more informed decisions. Sensitivity analysis investigates the impact of data limitations on model accuracy, while sampling provides limits for model results. Model extensions provide empirical approach to the time non-homogenous problem which if approached mathematically requires extensive longitudinal data. (PsycInfo Database Record (c) 2023 APA, all rights reserved)