Variation—in clinical workflow efficiencies, processes, and more—is one of the biggest challenges healthcare organizations face in outcomes improvement work. Healthcare isn’t alone, as variation has been recognized across other industries as a common foe. In the book Basic Statistical Tools for Improving Quality, physicist and statistician W. Edwards Deming (1900-1993) is credited as stating, “Uncontrolled variation is the enemy of quality.” Revered for his contributions to the science and process of management, most notably in the automotive industry, Deming proposed that the key to improving quality was reducing statistical variation.
Today, healthcare is increasingly recognizing the relationship between reducing variation and improving outcomes. Interventions to reduce variation for a targeted process can improve care by establishing consistency based on best practices. The industry’s evolving data capabilities, such as Late-Binding™ technology, are expanding health systems’ ability to reduce variation in delivery.
Creating new protocols, however, such as bundles of care and care processes based on best practices, is only one step toward reduced variation. Organizations also need to develop strong analytic solutions to evaluate the effectiveness of these outcomes improvement initiatives and, perhaps just as importantly, create methods to track the decision-making process and rationale in instances when these initiatives are not followed.
The ability to identify the type of variation, as well as when it occurs and why, is fundamental for healthcare improvement. Patients frequently present to clinics or hospitals with varying degrees of complexity and other unique circumstances. For example: Two patients present to the emergency department (ED) with pneumonia. One has a history of renal failure and severe chronic obstructive pulmonary disease (COPD), while the other has no significant comorbidities. The ED physician will likely treat each patient differently. This can include the use of distinct antibiotics, based on types of bacteria associated with severe COPD. Each patient (the one with renal failure and COPD, and the one with no comorbidities) may also receive antibiotic treatment at different dosages and frequencies.
The type of variation described above—in which care is altered to serve the needs of a specific patient—is considered appropriate, or intended, variation. Conversely, if the above patient with renal failure and CPOD were treated in the same way as the patient without these comorbidities (with the same antibiotics at the same dose and frequency), then they might be at risk of another form of variation: unintended, or unwanted, variation. This form of variation occurs because the patient was not treated per their specific needs. As such, unwanted variation is responsible for suboptimal outcomes, including increased morbidity and mortality.
The opportunity to reduce variation in outcomes improvement work lies in unwanted variation: by working towards data-driven best practices that reduce variation, health systems further quality improvement by taking actions that support better care and reduced cost.
The Institute for Healthcare Improvement (IHI) describes variation as a common culprit behind burdens in the healthcare system: “Many quality and cost problems in a process or product are due to variation,” it states. The IHI adds, “The process that produces 95 percent on-time delivery or good product is the same process that produces the other 5 percent late deliveries or bad product.”
To improve quality and lower cost, health systems need to identify the causes of unwarranted variation in outcomes, and develop ways to manage them. A failure to combine analytics with best practices and adoption principles can result in suboptimal outcomes and higher costs—which directly oppose the goals of healthcare improvement.
A natural answer to overcoming variation is standardization. Systems, however, need to be cautious about how they employ standardization. A blanket standardized approach can be too much of a “cookbook”—a collection of instructions and precise measurements that don’t account for differences in patients, facilities, and resources.
Clinicians will encounter cases that initially appear similar, but—after a closer look—have more variables that call for a modification in care. Interventions are often initially developed based on the Pareto principle (also known as the “80/20” rule), which identifies a limited number of input factors that will have the greatest impact on output. The Pareto principle maintains that 80 percent of the output from a situation or system is determined by 20 percent of the input.
few years ago, I worked on a project to develop a protocol to treat hyperglycemia (elevated glucose levels) and hypoglycemia (low blood glucoses levels) in hospitalized patients. We created three different options for ordering insulin that would meet the needs of at least 80 percent of the population. We developed a protocol that created intended variation around a standardized process; realizing that when two patients experienced a similar event (e.g., glucose of 280 mg/dL), one may require different treatment to get back into a normal glucose range.
Healthcare analytics plays a critical role in reducing variation in healthcare by revealing actionable information. This includes indicators, such as if facilities are adhering to best practices, if interventions are effective, and specific details about individual patients. As leaders assess variation in their organization, they may use a control chart, also known as a Shewart chart. These charts, conceived in the 1920s, continue to gain value in outcomes improvement, as they provide a robust ability to determine how closely a process is being followed as it was designed.
Control charts monitor the extent to which variation is occurring. They also determine if the variation is caused by sources common (stable and predictable) to the process, or if it’s the result of special causes (unpredicted events or processes that are significantly different from usual practices).
For example, a certain hospital is in a region participating in the CMS Comprehensive Care for Joint Replacement payment model. The hospital’s Outcomes Improvement Team is working to improve the quality of care for patients having a total hip or total knee replacement surgery. They are currently focusing on a process to ensure that patients with diabetes have a preoperative hemoglobin A1C result to confirm that their diabetes is being controlled (otherwise, surgery should be postponed).
Recently, a patient who underwent a total hip replacement was slow to recover following hospitalization. The patient was discharged to a skilled nursing facility (SNF), where they had a prolonged stay due to post-operative surgical site infection.
What could have been done to prevent the surgical site infection? Thanks to a strong outcomes improvement structure, the team was able review information in addition to content present within their own EMR. They identified that there had not been a preoperative hemoglobin A1c. In addition, the patient’s primary care provider’s EMR showed a history of poorly controlled diabetes. Thanks to the team’s ability to identify where in the delivery of care the variation occurred, the team could improve their messaging and process to ensure that all patients with diabetes have a hemoglobin A1c prior to surgery.
The outcome improvement process might start with a group representing a portion of a population, and improvement teams could alter that over time as they add data from more of the population. Using risk stratification tools, they can stratify patients to make the best use of scarce resources by targeting high-risk, high-cost patients who need more aggressive care.
Health systems can dramatically impact outcomes by shifting the focus of the entire organization to make improvement a common goal. As teams from across systems and facilities begin to work together, success is often more efficient and robust. Cohesiveness is built when sites and departments share information—particularly successes—around a common improvement goal. This brings teams together from the ground up.
In addition, frontline clinicians must feel supported when they diverge from a certain process. One of the most important things improvement teams can do in intervention design is consider how to capture and identify clinician rationale around intended variation. By incorporating frontline judgement, improvement teams can continue to improve and refine their interventions over time.
Throughout systems with multiple facilities, there can be a dramatic difference in how care is delivered; these systems must remove unintended types of variations so that all facilities can achieve a similar degree of success. The fundamental goal is to work across facilities to enable patients to receive exceptional care anywhere in the system. Open communication and productive relationships among staff and team members across facilities is a critical step in this process. As these individuals come together and form cohesive team structures, they can achieve goals that once might have previously seemed impossible. This is the power and principle of reducing variation to achieve sustained outcomes improvement.
Once health systems have identified and formed outcome improvement teams, they can improve outcomes by taking three steps to reduce unwanted variation:
This principle is particularly important for health systems that have multiple facilities, within which the way care is delivered can vary. There may be some intended variation between facilities in the same system (e.g., location, staffing, and resources), but by reducing variation, they can all adopt the same fundamental strategy. The goal is to work across facilities to enable high-quality care for all patients, no matter where they are treated.
This objective requires open, systemwide communication and productive relationships among frontline staff and members of the outcomes improvement team. It can help teams from different sites and departments to understand that other groups have faced similar struggles. It’s a sense of inclusion that reminds disparate groups that reducing variation in healthcare is not a competition between entities, but rather a unified effort towards improvement.
As health systems look to reduce variation in care, one of the most important things they can do is share success stories. Disclose strides towards improvement—large and small—among staff at the department and facility level, and among the greater healthcare community (between health systems). Make routine announcements—in the form e-mails, newsletters, or meetings—a standard step in the initiative to reduce variation.
Health systems deliver care to patients with intended variation—as determined by comorbidities, personal factors, etc. They might start with an intervention focused on a population or portion of a population and alter the intervention over time to reduce variation, they gain experience and expertise. By decreasing the extent of variation in care delivery, coupled with continued tracking of processes and interventions, teams gain the ability to quickly identify issues and address them before they increase in frequency or severity.
Healthcare systems face many trials as they work towards outcomes improvement—from creating a strong governance and team structure, incorporating best practices, and building a strong analytic system, while ensuring organizational and financial alignment. Each of these challenges makes up the journey of outcomes improvement. Reducing variation persists along the entire journey, as it shows health systems where their opportunity for improvement lies.
Would you like to use or share these concepts? Download this presentation highlighting the key main points.