Report 2004-123 Summary - March 2005

California Public Employees' Retirement System: It Relied Heavily on Blue Shield of California's Exclusive Provider Network Analysis, an Analysis That Is Reasonable in Approach but Includes Some Questionable Elements and Possibly Overstates Estimated Savings


Our review of the decision by the California Public Employees' Retirement System (CalPERS) board of administration (board) in May 2004 to approve an exclusive provider network for CalPERS members in the Blue Shield of California (Blue Shield) health maintenance organization (HMO) found the following:


In an effort to control health-care costs, the board of administration (board) of the California Public Employees' Retirement System (CalPERS) voted on May 19, 2004, to approve an exclusive provider network1 for the 427,000 CalPERS members2 in the Blue Shield of California (Blue Shield) health maintenance organization (HMO). The approval excluded 38 hospitals. Subsequently, as a result of further negotiations, Blue Shield permitted some of the hospitals to remain in the network, and the Department of Managed Health Care, which is responsible for regulating health-care service plans in the State, denied the exclusion of four others. As of January 1, 2005, 24 hospitals were excluded from the Blue Shield HMO provider network. Blue Shield estimated the hospital savings resulting from this network to be $20.6 million and the savings associated with some medical groups to be $10.8 million, for a total of $31.4 million in savings to CalPERS in 2005.3

In establishing this network, Blue Shield expected to save CalPERS money by excluding high-cost hospitals and medical groups that admit only to those hospitals. An actuary Blue Shield hired to review its models (model-review actuary), at the request of CalPERS, found that its original savings estimate did not incorporate recent financial contract terms with a health system that were expected to produce savings only if CalPERS retained the full provider network. According to the model-review actuary, Blue Shield should have factored these savings into the baseline for the full provider network before projecting the savings CalPERS would realize by switching to the exclusive provider network. The new contract between Blue Shield and the health system contained a clause with certain financial terms that would be available to CalPERS if it did not adopt an exclusive provider network. If the CalPERS board approved the exclusive provider network, the financial terms would not apply in 2005 and 2006. Blue Shield estimated that the financial terms would result in substantial savings to CalPERS in 2005.

In response to the model-review actuary's recommendation, Blue Shield stated that due to the private and confidential nature of its hospital agreements, it was unable to present to the public or in open board or health benefits committee (committee) meetings the hospital's contract savings. However, Blue Shield did present an estimate of the impact of the savings to CalPERS' committee during a closed meeting held on March 16, 2004, and did present to the board the impact of the health system's financial terms during a closed meeting held on May 11, 2004.

During an open session at the board's May 11, 2004, meeting, Blue Shield presented two options. One option was to maintain the full network under the following conditions: accept the health system's financial offer, provide higher-cost hospitals an opportunity to bring their costs closer to the industry average, and maintain the option of adopting an exclusive provider network in the future (for example, January 1, 2006). The other option was to adopt the exclusive provider network effective January 1, 2005. On May 19, 2004, the board approved the latter option, which resulted in the health system's financial terms no longer applying in 2005 and 2006. According to the current deputy executive officer for benefits administration, the board chose to proceed with the exclusive provider network because it sought to generate savings beyond the health system's contract period and to initiate structural reform in the health care industry. This statement is consistent with our review of the transcripts of meetings prior to the board's approval.

However, Blue Shield's estimate of $31.4 million in savings for the exclusive provider network does not take into consideration the impact of members leaving its HMO provider network and joining other health-care plans. A preliminary analysis prepared by Blue Shield, using CalPERS data, estimated that it lost 34,000 members during CalPERS' December 2004 open enrollment period. According to an analysis prepared by CalPERS in February 2005, almost 16,000 Sacramento-area members left the Blue Shield HMO provider network during the open enrollment period.4 There could be a number of reasons why members chose to leave Blue Shield's HMO provider network. However, as a result of this member movement, according to our consultant, Blue Shield's savings estimate of $5.5 million for the Sacramento area could drop to between $1.7 million and $3.5 million.5 CalPERS' analysis did not include similar information for other areas of the State. Therefore, we are unable to quantify the full effect that member movement has had on Blue Shield's savings estimate.

Blue Shield also did not adequately address a recommendation made by the model-review actuary to investigate differences in the emergency room assumptions for one hospital system. According to our consultant, Blue Shield's hospital savings estimate of $20.6 million could drop to only $8.9 million if the model-review actuary's assumptions were used. Given the sensitivity of the hospital savings estimate to differences in emergency room assumptions, Blue Shield should have reconciled the two analyses to ensure that its estimate of emergency room costs was reasonable.

According to Blue Shield, if the savings from the exclusive provider network differ materially from its original estimate, it will adjust CalPERS' future premiums. Specifically, a provision in Blue Shield's contract with CalPERS requires it to compare the actual cost of health care to its projected costs. If the difference falls outside a certain range, Blue Shield will adjust for the difference when calculating the projected health-care costs for the following year, which could potentially increase CalPERS' premiums.

Blue Shield's savings estimate are based on an analysis that consists of three distinct models: the Milliman USA, Inc., RBRVS for Hospitals™ Relative Value Unit Fee Schedule (Milliman model); the cost model; and the savings model. Blue Shield also developed a fourth model to estimate savings from financial terms for calendar years 2004 and 2005 that it negotiated with one health system after its initial analysis was completed. Our consultants found that many components of Blue Shield's analysis appear reasonable but it contains some questionable elements such as using non-CalPERS claim data sources.

Our review found that the CalPERS board, committee, and health benefits branch staff relied primarily on Blue Shield's summary of its analyses and its presentations in deciding to approve the exclusive provider network. A provision of the contract between CalPERS and Blue Shield prohibits Blue Shield from disclosing information that would breach the terms of contracts—including payment rates—with its provider hospitals. As a result, CalPERS did not have access to either hospital rates or Blue Shield's cost model and was therefore unable to verify the model's accuracy. Health benefits branch staff confirmed their reliance on Blue Shield to prepare technical analyses and on an independent actuary to verify the results of Blue Shield's analysis.

Although the model-review actuary concluded that Blue Shield's general method of analysis was reasonable, his report indicates that time constraints prohibited him from fully addressing some important assumptions and elements of the analysis. For example, although the model-review actuary was hired to, among other things, review Blue Shield's cost savings projections for the exclusive provider network, he was unable to express an opinion on Blue Shield's savings estimate of $36.3 million that related to the exclusion of the 38 hospitals. Thus, his report could not provide a credible basis for the CalPERS board to evaluate Blue Shield's savings estimate prior to adopting the exclusive provider network and excluding certain hospitals.

Although the board and committee discussed Blue Shield's savings estimate, our review found that the transcripts and meeting notes for the board's and committee's closed and open meetings held prior to the board's approval of the exclusive provider network did not discuss the actuary's inability to conclude on the savings estimate or all of his remaining findings and recommendations and their impact on CalPERS' decision. CalPERS staff also did not investigate the model-review actuary's other findings and recommendations that Blue Shield did not address fully but may have a significant impact on its analysis. Without addressing the model-review actuary's concerns, CalPERS had no assurance from an independent source that Blue Shield's analysis was accurate.

A review of the Milliman model by our consultant found that it provides a reasonable basis for comparing relative reimbursement levels across hospitals. In addition, our consultant found no evidence of material errors in the claim data underlying the Milliman model. However, Blue Shield's inclusion of non-CalPERS claim data, while arguably necessary to increase its sample size may adversely affect the accuracy of the analysis, due in part, to varying reimbursement rates under different contracts. In deciding which hospitals to include in its network, Blue Shield assigned each hospital a quality-adjusted relative cost factor. Our consultant found that eight hospitals that had cost factors above the exclusion threshold (the cost factor threshold above which hospitals were subject to exclusion from the provider network) when the additional claims were used had costs below the threshold using only HMO claim data.6 Conversely, nine hospitals that had costs below the threshold when the additional claims were used had costs above the threshold using HMO claim data alone. These differences may be important, since these 17 hospitals provided more than 10 percent of hospital services to HMO enrollees as measured by relative value units, a measure of the resources needed by hospitals to provide services. However, our consultant concluded that, in as much as there was no ideal source of claim data to use in its analysis, Blue Shield's decision to include the additional data does not appear unreasonable.

Furthermore, our consultant found that in one instance Blue Shield deviated from its original criteria for excluding hospitals from the network. Specifically, in one geographic area, or cohort, rather than evaluating each hospital in a particular system separately against the cohort average, using HMO and preferred provider organization data, as called for by its rules, Blue Shield evaluated all the hospitals in one system as a group, using a different set of claim data. Since the system as a whole met the threshold, Blue Shield included all hospitals in the system in its exclusive provider network, even though one of them would not have met the threshold if it had been evaluated separately.


The Legislature should consider enacting legislation that would allow CalPERS, during its contract negotiation process, to obtain relevant documentation supporting any analyses it will use to make decisions that materially affect the members of the health benefits program established by the Public Employees' Medical and Hospital Care Act.

To ensure that its decisions are in the best interest of CalPERS members, CalPERS should require its health benefits branch staff to evaluate fully the findings and recommendations of third-party reviews and present their results to the board and committee.


Both CalPERS and Blue Shield disagree with our conclusion that Blue Shield's estimate of cost savings related to the exclusive provider network may be overstated. In addition, both disagree with our conclusion that without addressing the concerns raised in the model-review actuary's report, CalPERS had no assurance that Blue Shield's analysis was accurate. Finally, Blue Shield believes that the elements identified as questionable in our report had no material effect on the exclusive provider network analysis on which CalPERS relied. Our comments follow each response.

1In this report, we refer to Blue Shield's health maintenance organization provider network for CalPERS members as the exclusive provider network.

2In this report, we refer to CalPERS members, retirees, and their survivors and beneficiaries collectively as members.

3As of May 19, 2004, when the board made its decision to exclude 38 hospitals, Blue Shield estimated hospital savings to be $27.7 million and savings associated with some medical groups to be $8.6 million for a total of $36.3 million savings to CalPERS in 2005. For the purposes of our report, unless otherwise stated, the term savings estimate refers to Blue Shield's estimate of $31.4 million that relates to the exclusion of 24 hospitals. In Appendix A we present an overview of Blue Shield's analysis.

4CalPERS includes the counties of Sacramento, Placer, and Yolo in its analysis.

5Information was not available on the cost of providing care to the members who leave Blue Shield and therefore this was not considered in the analysis. If the members moved to health plans that use the high-cost hospitals excluded from the Blue Shield HMO provider network then the lost savings may not be realized by CalPERS. If they move to health plans that do not include these high-cost hospitals or use them less frequently, then CalPERS may realize some of the lost savings. However, without knowledge of the rates these hospitals charge other health plans, it is not possible to further refine the effect that member movement has on Blue Shield's estimate.

6In Appendix A we present an overview of Blue Shield's analysis, including its inclusion threshold.