In American health care, there can be considerable differences in the way we treat two patients with the same diagnosis. Although these differences often are justified, this variation is sometimes based not on a patient’s special characteristics, or what the data indicates, or even a physician’s intuition—but rather on our own habits as practitioners.

Scholars of health care have a name for this phenomenon—unwarranted variation—and it is one of the most relentless problems in medicine. While estimates vary, unfounded variation in care adds up to as much as $600 billion in avoidable health care spending per year in the United States. Reducing this clinical variation is not just a cost-control measure; it is a necessary step toward better outcomes and more satisfied patients. Knowing that preventable medical errors are a leading cause of death in this country provides tremendous incentive to follow proven protocols in typical cases.

Here’s an example: Anterior cervical discectomy and fusion, or ACDF, is a type of neck surgery that involves removing a damaged disc to relieve spinal cord or nerve root pressure causing pain or numbness. In the past, each of the various orthopaedic surgeons and neurosurgeons at our five adult hospitals had their own set of orders governing pre- and postsurgical care.

For instance, some surgeons would have the patient go from recovery to the floor, where nurses feed patients and get them out of bed right away. Other surgeons would request a swallowing evaluation first. Some patients would be required to wear a neck collar and others would not. The timing of X-rays varied widely. All of that could tack days onto the recovery, depending on who operated. And, because the process varied by provider, even within the same hospital, residents and nurses had to memorize preferences specific to each surgeon. Moreover, patients would get jumbled signals.

So surgeons in our departments of neurosurgery and orthopaedics decided to see if they could establish a consistent, evidence-based care pathway for ACDF. They involved colleagues from Sibley Memorial Hospital and Suburban Hospital, nursing, occupational therapy, and others who formed a multidisciplinary Spine Clinical Community.

This team came up with a single optimal pathway that takes ACDF patients all the way from the beginning of their check-in through their hospital stay and discharge. Then came the harder part: getting others to buy in.

Initially, some resisted change. But physicians are scientists, and they respond to compelling data. Ultimately, the group persuaded surgeons by showing the benefits of streamlining care—including a one- to two-day reduction in average length of stay. Now, many across the system have adopted this pathway. They see that standardization brings clear expectations and fewer errors. Plus, everyone stays up-to-date on best practices.

Currently, Johns Hopkins Medicine has 21 clinical communities focused on quality-improvement projects in areas ranging from diabetes to clinical disinfection.

In a community of exceptionally gifted and experienced physicians, these efforts could be perceived as attacking their professional autonomy. Instead, we expect these projects to increase their creativity. First, an established care pathway gives us a standard for measuring new innovations and determining whether they actually make a difference in outcomes. Second, when we apply these protocols in straightforward cases, it frees up time and mental energy for physicians to operate more autonomously in complex cases and to focus their creativity on big-picture initiatives aimed at improving health care.

Most importantly, no administrators in central offices are handing down clinical order sets. In fact, you might describe this movement as replacing a physician’s autonomy with physicians’ autonomy—a profound distinction achieved by a shift of the apostrophe. I am confident that when you bring together world-class clinicians to share notes and strategize, their collective wisdom will improve health care in ways we have yet to imagine.