Uncovering the Real Flaws in How Healthcare Work Is Designed

Burnout is not a failure of grit in healthcare and should be attributed to how the work is designed. Placing the blame on the people is an attempt to say a fatigued clinician should be more resilient. This thinking unfortunately allows the system to remain the same. The truth is, it is the opposite. Burnout is the result of a system that is designed to create structural friction and is the expected result of a workforce that is being stretched to its limits in a system that is poorly designed.

The thinking around grit, in addition to being wrong, actually costs a lot of money. When we say that burnout is the result of a personal flaw, we take the financial hit and let the system that created the flaw take the hit as well. The clinician is told to create better boundaries, and the situational boundary constraints remain the same. This is the most convenient thinking for everyone except the clinician, and has prevented us from finding a solution to the actual problem.

The numbers changed in a dramatic way during the pandemic. The rate of burnout for physicians reached the highest levels ever recorded. The rate has come down since, but is still well above the prior levels. It is still easy to identify the causes. Administrative burden and chronic understaffing are issues that were still in place before the pandemic, and remain now. The goal of the system should actually change from maximizing output to providing necessary support.

Redesigning Work with AI

AI usefully gives companies a reason to reconsider how work is done. The easy version of the story is also the wrong one. The focus is not really about speed and efficiency in carrying out a few simple tasks. It is about absorbing the tedious, mindless load, the document management and administrative work, the reading and sorting that takes up hours of the day. In that sense, AI is about bringing the opportunity to ask a different question. The different question, or the more productive question, is not about how to run the old process faster. It is about what the ideal work process should be.

This is an important distinction. Simply applying a tool of any kind, especially a new tool that is designed to simplify or clarify a work process, to a work process that has severe structural issues and is poorly designed will only result in an equally poor work process that is simply more efficient. There is a term for that, which is "paving the cow path". The work process will then be controlled by a technological constraint. The automation of poor work processes will not resolve the issues. It will simply remove the friction that caused people to question the poor work process. The companies that gain the most benefit from AI will be the companies that take the opportunity to reconsider how work is done, to reassign tasks, and to redefine the value of the human contribution to the work process.

New Job Titles Emerging From This Shift

One of the compelling consequences of AI applications is the almost immediate impact on the job market. New roles and even titles emerge as AI applications automate low-complexity and high-repetitive tasks. Roles like a clinical data steward or a workflow designer are hybrid roles, yet their intersection is not around managing technology. It is about understanding data and organizational purpose and improving the quality of human services. This is work that demands a person who understands both of these worlds.

In practice these roles are unremarkable and very specific. A data steward is someone who determines a definition of a data field so a term does not carry three different meanings in three different departments. A workflow designer sits with a person performing the task to rearrange the steps of the task to reflect the actual flow of care rather than the logic imposed by the vendor software. These roles are not part of traditional staffing and both are critical to having AI be a tool that adds value rather than a tool that adds even more unnecessary complexity to an already complex system. This is intentional redesign of a workforce, which requires a serious commitment from an organization. The organizations that are committed to this redesign are investing in upskilling their workforce and adopting more flexible staffing models.

The Messy Middle

A difficult middle section is apparent in all transitions, and it is the section where most workforce redesigns fail. It is the middle section of a transition where the old roles still exist, the new roles are still in the process of being defined, and the employees are unsure of their responsibilities. A task that was once fully the responsibility of a person is now the shared responsibility of a model and a new role, and in the undefined space, responsibilities are ignored.

Organizations that successfully come through this do a single, unremarkable thing really well. They define and communicate the responsibilities during the messy middle, and they do it while the situation is unclear. They do not wait for the situation to clarify itself. The situation will not clarify itself. If the responsibilities are left undefined, the ambiguity becomes the new workarounds, and the organization has simply replaced one system with another poorly designed work system.

Leadership Is the Variable

None of this is dependent on the technology alone. It is dependent on the culture surrounding the technology, and that makes leadership the defining factor. The leaders who successfully manage this pair clear ownership with the psychological safety of a particular work environment, whereby employees feel safe to experiment and fail. This balance fosters an environment where employees feel empowered to learn new responsibilities and work systems, rather than reverting to their old responsibilities.

The test is small and telling. On a team with psychological safety, a nurse will openly criticize a tool that has become a hindrance, and as a result, the tool will either be improved or discarded. A team without it will leave the same nurse to suffer in silence, with leadership never learning of the issues with the tool, and the problem will remain unsolved until adoption metrics are released and no one can explain the poor results. The culture of the rest of the team will systematically define which scenario will unfold. Taking the human side seriously means pairing the technology with the support and infrastructure to improve emotional health and resilience, and not treating it as a nice-to-have soft addition. Trust is the only thing that matters here and is a necessity for any large-scale change to be successfully implemented.

Use Judgement

If a more effective system is the goal, then the criteria for success should also become more effective. AI in healthcare used to be measured by how fast and how much. Those metrics are relatively easy to observe, but they answer the wrong question. The better metrics are the human ones. For example, are clinicians spending more time with patients, is the cognitive load truly diminished, and is the work safe and something that can be done for many years without being diminished?

In addition to watching for late signals, it is also useful to pay attention to the early signals. Employee turnover is an example of a late signal. When your turnover is high, and your best employees are leaving the company, it is a sign that there is a problem, but it is an indication that is too late to be useful. There are early signs. In the world of software, early signs can be shown by the usage of the application and engagement by your employees. Some employees may slowly start to sign out of the application. A serious software system will try to monitor early signs, knowing that they can do something about those signs.

When those become the questions, the tool that increased throughput while leaving people more depleted becomes a failure. In the long run, the only thing that will ever be the most important test of success will be whether the work became more sustainable for the people doing it.

Limitations of AI

The most important thing here is the limitations of AI. AI cannot get rid of employee burnout. Employee burnout is too entwined with systems, policies, and economics to be solved. AI can reduce some of the employee burnout and employee stress by improving systems and reducing the burden of some of the work on the employees.

The true advantage will not come from simply being the first adopter of AI. It will come from the harder, less glamorous ability to create an environment where AI and employees can thrive, and where the focus is on relief and sustainability rather than squeezing every last drop of productivity. This is the flaw and the opportunity that is most compelling.

Christopher Hutchins Founder and CEO, Hutchins Data Strategy Consultants