Industry Voices—How to rethink benchmarking to achieve performance transparency in healthcare

“Comparison is the thief of joy,” Theodore Roosevelt once said.

The former U.S. president was clearly not a healthcare leader. Because when comparative benchmarking is used as a tool in healthcare, decision-makers can uncover, prioritize and enjoy opportunities for performance improvement and strategic insight.

But benchmarking has not seen an innovative improvement in decades. Traditional benchmarking methods compare one provider’s measures against the averages for other providers in the state and all providers nationally. Designed and performed largely manually, traditional benchmarking methods are limited in the type of insights they deliver as well as cumbersome and expensive to perform.

Even more frustrating, the insights can be inconclusive as to whether an organization has a problem, or the problem may not be addressable. Such results hardly point to a clear improvement path or motivate change. As stated in a 2016 National Academy of Medicine article (PDF): “Research has demonstrated that many of the current public reports make it cognitively burdensome for the audience to understand the data.”

RELATED: 4 steps to benchmarking and improving your practice

To keep up with progress in data, benchmarking needs software automation that removes subjective bias, delivers a broad array of insight types in shareable English and uncovers relevant context that can guide evaluation and subsequent decision-making and action. Here's a look at how.

Start with a question instead of selecting metrics

Today’s technological advances such as artificial intelligence (AI) allow healthcare leaders to rethink benchmarking to make it both more useful and easier to implement.

Typically, an investigator selects one or two organizational goals, a few key metrics relevant to these goals and a peer group (notice the limited and subjective nature of this approach). What if healthcare professionals asked questions about what truly matters to them? For instance, they want to know: “How am I doing?” “What are the biggest improvement opportunities?” “What are the major changes over a period of time?” This approach opens up a multitude of possible insights instead of confining them to a limited—state, national or custom—peer group inherent in a manual approach.

Speak plain English

AI technology can explore answers within an exponentially larger space of possibilities than people can. What’s more, it can deliver objective answers in plain English. When insights are delivered by means of language, not diagrams, tables or complex dashboards, they are easily understood and shared by people. A dashboard is great for conveying real-time sensor alerts, but will it motivate you and your staff to improve?

Using language also allows for the provision of relevant context, e.g., facts that help evaluate the significance or scope of the standout behavior or outcome.

For example: Hospital A has the seventh-highest median time patients spent in the emergency department before being seen by a healthcare professional (58 minutes) of the 1,364 hospitals whose emergency department volume is low. Those 58 minutes are 220% higher than the average of 18.1 minutes across the 1,364 hospitals.

RELATED: 5 ways to use benchmarking to increase physician practice efficiency, profitability

Not only does this example show the area of improvement for Hospital A, but it also indicates by how much it would need to improve if the goal is to perform in line with other hospitals of comparable emergency-department volume.

Novel comparisons for extra insight

Automation can also uncover and report comparative performance insights that are invisible to traditional methods. An especially novel type of benchmarking insight involves aligning two numeric measures: One measure expresses the standout behavior, while the second forms the peer group.

In other words, you select a noteworthy peer group (typically good on a certain measure), and then highlight a certain behavior within that group.

For example: In the Southeast with its 1,191 hospitals, Hospital B has the lowest nurse-communication rating (1 star) of the 760 hospitals with as high a doctor-communication rating (3 stars).

This insight indicates that Hospital B might want to look into the reasons for such a uniquely poor nurse-communication rating in light of its peer group which scores anywhere between satisfactory to high performance on a related measure.

An added benefit to automated benchmarking is that healthcare professionals can do as many comparisons with as much data as they have, as often as they need, with no extra effort required.

This enables cost reduction, reliability, repeatability and handling ever-larger and regularly updated data sets, which should also include changes over time, not just the last period’s measurements. These benefits matter, especially if benchmarking is to be repeated periodically.

Raul Valdes-Perez, Ph.D., is co-founder and CEO of OnlyBoth Inc. in Pittsburgh.