John Quinnell

Paper 197

Health care insurance companies typically reimburse claims for care provided through managed care networks (e.g., Preferred Provider Organizations and/or Health Maintenance Organizations) and traditional indemnity plans. Managed care plans can be expensive to administer so it is important to verify that the savings from contracting with PPOs are greater than the costs to administer them. Managed care savings can be accurately measured by benchmarking PPO discounts against the cost of indemnity care by using GIS to control the significant influence of regional differences on costs.

The high cost of health care is an important issue to employers and individuals. Employers that choose to provide health care benefits do so to attract and retain good employees, but still must remain cost-competitive with their products; High health care costs impact individuals in employer-sponsored health plans through greater deductibles and/or copayments. Consequently, entities paying for health care—state and federal governments, employers and third-party insurers—place great importance on managing their health care costs.

Preferred Provider Organizations (PPO) are an effective and increasingly popular means to manage health care costs. PPOs can exist in a variety of forms, but common to all is a network of physicians and hospitals that provide services at a discounted rate in return for a predictable, guaranteed block of business. The Principal contracts with numerous PPOs nationwide to provide medical services for about 80% of its enrollees.

Health care savings from PPOs can be significant but difficult to measure. Usually, savings are measured by the size of the PPO discounts; The cost of care after the discount is compare to what the cost of care would have been before the discount to determine "cost savings". The problem with using the cost of care before discount as a benchmark is that there are no restrictions on what the provider can charge. This makes health care charges a "moving target" and discounts—and consequently savings—difficult to value. A 20% discount for the same service will have a different value depending upon the amount charged. For example, a $300 charge discounted 25% ($225) is not as good as a $275 charge discounted only 20% ($220).

The provider’s "moving target" is more complicated because a single provider may have different charges for the same or similar care. The following charges for normal, uncomplicated deliveries are from an actual PPO hospital that occurred last year:

Overview. PPO savings can be measured by comparing the PPO’s discounted charges to care reimbursed through an indemnity plan. Indemnity plans’ charges do not have discounts and provide a more meaningful benchmark. Non-discounted charges are also less subject to inflation by the provider.

GIS can be used to compare the discounted charges to charges without discounts in similar geographical regions. The net difference between the discounted charges and the non-discounted charges in similar geographic regions gives the most precise measure of the real, effective discounts and savings provided by a PPO. Segmenting the health care costs by geographic region minimizes the influence of differences in costs from area to area.

Regions are created from medical claims data based upon the providers’ zip code. Data are imported into the GIS program and the incidence of each health care encounter is plotted. The zip code data are aggregated to regions based upon a review of the data. Each region has non-discounted (indemnity) data and the PPO’s discounted data. The charges without discounts provide the benchmark with which to measure the PPO discounted data within the region.

Creating regions is more of an "art" than "science" and considers four factors:

Cell Size. Each region should contain at least 30 health care encounters within the benchmark group (non-discounted data) and the study group (PPO discounted data) to be statistically meaningful. If there are insufficient data, the zip codes should be aggregated with other zip codes until the minimum cell size threshold is achieved.

Location of Urban and Rural care. Care should be taken not to mix zip codes associated with urban areas with rural zip codes when creating regions. A body of evidence indicates that care provided in metropolitan areas is more expensive to deliver than care provided in rural areas. Tables of Metropolitan Statistical Areas (MSA) identify urban areas by zip code and can be incorporated into the analysis to distinguish urban and rural areas.

Contiguity of zip codes. Regions should be comprised of contiguous zip codes. Contiguity should be observed even if there are no data in zip code areas surrounded by zip code areas with data.

Number of regions. A PPO’s service area may encompass an entire state or even multiple states, or a relatively small area within a corner of a single state. In general, the maximum number of regions supported by the claims data should be created. More regions provide more "granularity" to the analysis and can uncover differences in costs that may be obscured by a higher level of data aggregation. The number of regions created usually range from 3 to 10 regions depending upon the volume of data.

Once the regions have been created from the medical claims data, the "PPO Discount off charges" is calculated. The average PPO Discount off charges can not be accurately directly compared with the average "Non-discounted charge" in each region. This is because no two groups of claims experience are exactly alike, even if they occur in exactly the same geographic area. Each group represents a unique "mix" of care defined by the nature of the illnesses and the frequency of their occurrence. If health care costs were compared without adjusting for this mix of illnesses, differences in cost may simply be due to one group being sicker than the other. Once the data have been "case-mix adjusted", there is greater confidence that any differences in costs can be attributed to differences in the study variable (in this case PPO discounts) than to differences in illness.

There are several methods to case-mix adjust medical claims data. One way is by using "Relative Value Units (RVU)".¹ RVUs are assigned to each service or procedure provided by a physician based upon the time, skill, severity of illness, risk to the patient, and the medico-legal risk to the physician. For example, a procedure with an RVU of "10" is twice as "difficult" as a procedure with an RVU of "5", and its cost should be higher to reflect this greater difficulty. A ratio, or "Relative Value Factor", can be constructed between the cumulative cost and cumulative difficulty of all procedures provided in each group of each region. The RVF can then be used in cost comparisons within a region. The RVF is calculated by:

The formula for the benchmark cohort would not contain the "Discounts" variable.

The Relative Value Factors for the "PPO Discounted charges" and the "non-discounted charges" are then compared in each region. Lower Relative Value Factors indicate better-cost performance because there is more "work" (RVUs) purchased for the amount of dollars paid. The factor representing the "PPO discounted charges" cohorts would expected to be significantly lower than the benchmark’s Relative Value Factors within the same regions.

It’s not uncommon that the Relative Value Factor comparison demonstrates that the PPO’s discounts result in significant savings over non-discounted charges overall, but an individual region(s) is actually more expensive than the non-discounted (i.e., indemnity) care. This observation would be obscured if the power of GIS were not applied to provide a more granular analysis of the PPO’s performance using regions.

The information can be used to direct the use of scarce PPO administrative resources to obtain best discount rates. Negotiation efforts can be directed at selected providers that are not cost-efficient within the PPO.

This analysis can be performed at the desktop with the modest investment of $129.95 for Esri’s BusinessMap Pro 2.0, downloading the RVUs from the Federal Government’s Health Care Financing Administration (AKA Centers for Medicaid and Medicare Services) and the small license cost to use the AMA’s CPT codes.