Linking Professional Fee and Facility Fee Data to Get the Big Picture

A Vermont hospital identified the need for greater analysis across the organization as one way to protect margins and reduce costs.

Like many hospitals, Northwestern Medical Center (NMC), St. Albans, Vt., juggles two separate billing systems for physician professional fees and hospital facility fees. For example, patients who have emergency surgery at NMC and are later admitted will generate facility fees as well as professional fees for surgeons and hospitalists. In the past, these charges would appear as three separate visits in the hospital’s reporting system. But today, leaders at NMC are able to connect data from both billing systems, allowing them to analyze entire encounters in their reporting tools. This is valuable as leaders seek to track individual patients—and the costs and revenue associated with them—across the continuum of care.

NMC has operated in a capped revenue system for several years because Vermont’s Green Mountain Care Board limits hospitals’ patient revenue growth rate at 3 percent annually. As a result, leaders are focused on protecting margins and reducing costs through greater organizational analysis, says Devin Bachelder, decision support and budget manager.

Tracking Outpatient Service Growth

The 50-bed, standalone hospital has seen a significant growth in its outpatient services during the past 10 years because of its acquisition of physician practices. In 2007, NMC added its first employed physician practice and purchased a physician billing system. Today, NMC owns 15 separate practices that account for approximately 20 percent of its gross patient revenue, Bachelder says.

Three years ago, NMC switched billing systems, which prompted Bachelder to think about how his team could link data from the physician billing system and the hospital billing system. “As our physicians’ practices have grown, the need for having connected data has become even greater,” Bachelder says. Unfortunately, the two billing systems were never set up to integrate medical record numbers or account numbers, which presented some issues as NMC was implementing a new decision support system last year.

The challenges of linking data from two different billing systems are often the result of how the systems are built. For example, NMC’s physician billing system cannot automate data reports, which are created in Excel. “We wanted to be able to put our transactional data from the physician system into our decision support system, but to do so, we had to manually export reports in Excel,” he says.

Meanwhile, its hospital billing system creates data reports in structured query language (SQL) format, making it impossible to link the two. Leaders decided they needed to get the physician billing data into SQL as well. Each month, they create a slimmed-down copy of their physician billing database that includes data on transactions, patients, physicians, and insurance in SQL. This allows leaders to integrate the data from both billing systems into their decision support tool.

But Bachelder wanted to go a step further and make the data even more actionable. “With everything in one database, if we were ever going to match records, this was the time to do it,” he says. By matching records between the two billing systems, leaders would have better data to more effectively leverage their decision support tool. The goal was to help leaders access the data they needed in a single report in the decision support tool so they could make better decisions.

Matching at the Patient Level

To match records in the two billing systems, Bachelder sought a unique patient identifier. NMC does not collect patients’ social security numbers, so the organization needed another strategy: matching medical record numbers between the two systems. “This lets us know who’s who all the time,” he says.

To start the matching process, Bachelder and his team focused on identifying patient populations that would generate charges in both the hospital and physician billing system during the same visit. This included new mothers with babies delivered by employed obstetricians. Other examples were patients with lab tests ordered by employed primary care physicians.

From there, the decision support staff could identify common data to match patients in both the hospital billing and physician billing systems. They prioritized patient populations that offered the most value: medical/surgical and ICU, deliveries, orthopedic surgery, general surgery, those with X-rays ordered in the physician office, those with provider-based billing transactions, and those with lab tests sent from physician offices. These patients represented approximately half of the 50,000 accounts in the physician billing database.

Bachelder and his team focused on fields that existed in the hospital billing system and the copy of the physician billing system—specifically, last name, birthdate, service date, and service type.

The first patient population matched was inpatients. Bachelder determined that they needed an exact match on birthdate and last name to establish a match. They also established a range-to-match service date. These efforts resulted in approximately 3,000 patient matches.

Bachelder performed the same exercise to match patients with outpatient labs between the systems. From there, they could move on to more difficult populations to match, such as patients with X-rays taken in physician offices, those with provider-based billing transactions, and those with lab tests sent from physician offices.

Matching at the Visit Level

After creating the patient matches, leaders could then match at the visit, or encounter, level. “It’s a small step from the patient match to the encounter match once you have some good results,” he says.

To do this, leaders matched visits using hospital medical record numbers and physician medical record numbers. They also used more rigid service date parameters to match visits between the hospital and physician billing systems. As a result, leaders achieved 99.9 percent accuracy in matching visits for medical/surgical and ICU, delivery, orthopedic surgery, and general surgery patients.

Getting the medical records to match only took a few weeks, Bachelder says. “The data elements were there, but once I limited the patient populations, I didn’t need many patient data elements in order to match,” he says. “You have to be strict in one area or another, and I found that our data allowed us to be very strict in defining service types, which enabled us to hone in on small groups of patients. In this case, the chances of a last name and birth date incorrectly corresponding within the same service date range were pretty small.”

Visit-level matching allows leaders at NMC to create more meaningful reports in their decision support system. For example, they can now track individual patients throughout the system. Although staff still need to create a copy of the physician billing system each month, they have automated the other matching steps in SQL, saving them time and effort.

Advice for Other Organizations

Bachelder provides the following suggestions for organizations that want to link data from two separate billing systems.

Have a common reporting tool. “This doesn’t necessarily mean common billing systems: It means a single source where the data are held so you can access the data through reporting,” he says.

Employ the right talent to support your efforts. Bachelder manages two analysts. One has SQL-coding skills and focuses on decision support. The other is a financial analyst who oversees cost reporting, budgeting, and monthly financials.

Identify all patient populations that will likely exist in both physician and hospital billing systems. “Make sure when you pick a population you know which data fields you can use to define them,” Bachelder says.

Partner with IT. “This is a very technical process, and it is difficult to be your IT team’s No. 1 priority for the amount of time required to implement and optimize a decision support system,” Bachelder says. “We had to get the skills we needed ourselves.” NMC’s IT team trained Bachelder and his team and provided access to an outside consulting firm for additional training. “Having access to that training helped us move the project to the top of our priority list and get it done,” he says.

Partnering with Physicians

Leaders at NMC are exploring ways they can share this data with physicians, starting with emergency department physicians and hospitalists. Bachelder says one of the hospital’s assistant vice presidents has been working with clinicians to identify ways to reduce variation and, ultimately, costs.

“We need to focus on cost per case and getting as efficient as possible across services and physicians,” he says.

This article is based in part on a presentation at the October 2016 Strata Decision Summit in Chicago.

Interviewed for this article:

Devin Bachelder is decision support and budget manager at Northwestern Medical Center, St. Albans, Vt..