By Lola Butcher

To determine the smartest business approach for providing primary care to the uninsured and underinsured, Baylor is digging into data and developing a detailed financial analysis.

In addition to its 26 hospitals, 174 private practice clinics, and myriad other facilities, Baylor Health Care System has collaborated with community-based, not-for-profit organizations to operate nine medical clinics for low-income patients over the past 10 years. The Baylor Community Care clinics, as they are called, currently serve about 12,000 patients who previously had no regular source of medical care. Most have at least one chronic condition, and the vast majority are uninsured, underinsured, or covered by Medicaid.

Creating medical clinics to serve low-income and uninsured patients is likely to reduce avoidable hospital utilization. Baylor, which collaborates in the operation of nine such clinics, has documented net per-patient savings of about $2,300 a year. However, to determine the health system's future strategy for serving this patient population, Baylor leaders are working to more fully understand the ROI.

This is a sample article from HFMA's CFO Forum, which is an online discussion and networking community for senior finance leaders in hospitals and other healthcare organizations.

Learn more and subscribe to the CFO Forum

More Precision Needed

Although Baylor has a system-level commitment to serving low-income patients, there has not been a formalized system-level strategy for owning low-income clinics. Each of the current Community Care clinics is affiliated with a specific Baylor hospital, and the decision to invest in the clinic was made by hospital leaders, in conjunction with system leadership, who wanted to reduce avoidable emergency department (ED) and hospital use. Patients are referred to the clinic by an ED physician, hospital social worker, or community health worker at the time of hospital discharge.

In the past, creating the clinics has been financially justified by a simple calculation: a comparison of Baylor's cost of care for patients during the 24 months before they were assigned to the clinic against the costs in the 24 months after their assignment. Depending on the clinic, that calculation initially revealed that a Baylor hospital can save between $2,000 and $3,500 in direct care costs per patient per year, says James Walton, DO, MBA, vice president-network performance, Baylor Quality Alliance.

Now, as it transitions to an accountable care organization, Baylor is aggressively moving to a patient-centered medical home (PCMH) model for all its primary care clinics, including those that serve low-income patients who do not have a regular source of health care. To accomplish this, leaders need to determine the following:

  • How many PCMH clinics for low-income, underserved patients should be created?
  • Should Baylor partner with others on the PCMH clinics or own the clinics outright?
  • How much capital should be allocated for clinics dedicated to low-income patients?

To answer these questions, system leaders needed to understand the ROI of clinics that serve low-income patients. "We have a commitment in this space, but we also have a commitment to ensure we are being good stewards of our resources," says Robert Green, senior vice president of strategic finance.

His staff recently completed the first phase of this financial analysis and embarked on a second phase that will be completed later this year. "At that point, we will be able to more definitively determine our strategy for how we roll this out across the Dallas metroplex," he says.

The Financial Analysis

Baylor's primary care clinics only converted to the PCMH model within the past year, so the ROI analysis examines the operation of primary care clinics for low-income patients, not for PCMH clinics.

The first phase of Baylor's detailed analysis was to examine the cost savings attributable to clinics for low-income patients. A single primary care clinic was chosen to study.

Carl MacLachlan, MBA, Baylor's director of advanced analytics and special projects, listed the challenges to this task:

  • Identifying the data source that would reveal which clinic patients are "truly Baylor customers" versus those who are generally served by other providers
  • Overcoming barriers associated with different IT and registration processes at clinics versus those used by Baylor hospitals
  • Analyzing the utilization trends to understand how a patient's clinic use affected ED and hospitalization use

Patient identification. Step 1 was defining the population of patients to study. "We decided to look at patients who had presented themselves prior to being seen at the clinic and also had used the health system after being seen at the clinic," says Maclachlan. "We assumed these were patients that were truly Baylor customers."

With the goal of creating the largest possible sample size and being able to review the data for a consistent timeframe, analysts decided to evaluate patients' clinic use from July 17, 2008 through Jan. 25, 2011, and their hospital discharges between July 1, 2006, and Aug. 31, 2011.

Identifying patients treated at the clinic and the hospital was not always automatic because some clinics did not capture a patient's Baylor identification number and medical record number. An analyst created a program that could match patients by name, telephone number, social security number, and address.

Hospital versus clinic utilization. Analysts then compared patients' hospital utilization before they were assigned to the clinic with their use after they connected with the clinic for three time periods: at six months, 12 months, and 24 months. "At first, it was hard to identify the trends, so we decided to look at the type of services the patient was receiving-inpatient, outpatient, ancillary services, and ED use."

Next, the analysts examined whether greater use of clinic services correlated to reduced use of ED and hospital services. "We wanted to identify patients who were truly engaged at the clinic and not just those who enrolled and then decided it wasn't going to work for them," says Maclachlan. "So we pulled anybody that was seen in the clinic two or more times."

Condition specificity. The final step was eliminating patients whose hospital care was not influenced by care received at the medical clinic. This primarily included trauma patients and newborns and their mothers.

Early Findings

In the end, the analysts were able to review utilization data for patients who represented more than 9,200 clinic visits and 3,800 hospital or ED visits. The analysis revealed the following:

  • About $4,800 in reduced inpatient direct variable costs per patient per year for patients served by the medical clinic
  • Net per-patient savings of about $2,300, after accounting for the higher outpatient costs associated with the clinic
  • Moderate reduction in ED utilization of about $195 per patient per year
  • Substantial reduction in inpatient utilization in two ways: Patients served by the medical clinic had 17 percent fewer inpatient stays, and those who were admitted had lower acuity based on total direct cost and an average length of stay that was 2.31 days shorter
  • Savings accrued in all three time periods analyzed: within six months of being enrolled at the clinic, at 12 months, and at 24 months
  • Greater use of the clinic was correlated with reduced demand for hospital services

"The more times the patient visited the clinic, even though it was a small number, the savings continued to increase," says Maclachlan.

Next Phase of ROI Analysis

The analysis corroborated the health system's assumption that the financial benefits of medical clinics for low-income patients is substantial, as well as Walton's preliminary estimates of the per-patient savings. But more information is needed before Baylor leaders can determine their future investment in opening more owned-clinics, says Green.

The health system recently embarked on the second phase of this study. "We need a randomized, controlled study with a 95 percent confidence level that allows us to extrapolate across our patient base," says Green. "That will allow us to understand what the savings would be against our investment."  

Lola Butcher is a freelance writer and editor based in Missouri.

Interviewed for this article:

Robert T. Green is senior vice president-strategic finance, Baylor Health Care System, Dallas, and a member of HFMA's Lone Star Chapter (

Carl R. MacLachlan, MBA, is director of advanced analytics and special projects, Baylor Health Care System, and a member of HFMA's Lone Star Chapter (

James W. Walton, DO, MBA, is vice president-network performance, Baylor Quality Alliance, Baylor Health Care System (

Discussion Starters

Forum members: Please add your insights, questions, and comments about this article on the CFO Forum's LinkedIn discussion board.

  • What is your health system's strategy for serving low-income patients who have no regular source of medical care? 
  • How do you calculate ROI on services designed to reduce inappropriate ED and hospital use? 


Publication Date: Wednesday, April 11, 2012