Over the past couple of years, the Bureau of Labor Statistics (BLS) reported that the prices hospitals receive from insurers and patients have grown by less than 3 percent per year after adjusting for improvements in the quality of care (Bureau of Labor Statistics 2015c). The BLS hospital producer price index estimates have been widely reported (Dobson et al. 2014, Health Research Institute 2015). BLS also presents data on a subcategory of payers called “private insurance and all other patients” that had price growth of 3.3 percent in 2011, 3.8 percent in 2012, 5.2 percent in 2013 and 2.6 percent in 2014 (Bureau of Labor Statistics 2015c). While Altarum also reports this data, this series has received less attention (Altarum Institute 2015).
In contrast to the BLS data, HCCI (which obtains data from private insurer claims) shows both inpatient and outpatient hospital prices for those with employer sponsored insurance growing by over 5 percent on average from 2011 through 2013 (Health Care Cost Institute 2014). HCCI does has not yet reported data for 2014 or 2015, but the chief financial officers of Tenet and HCA, two large publicly traded firms stated they expect revenue per commercial discharge to increase by “mid-single-digits” at Tenet and by 4 to 5 percent per adjusted discharge at HCA (The Street 2015a, The Street 2015b). Similarly some insurers have projected price growth in the neighborhood of 5 percent in their rate filings (California Department of Insurance 2014a, California Department of Insurance 2014b). The price growth reported by HCCI and hospitals themselves is consistent with the strong profit growth from 2012 through 2014 at many large hospital systems (e.g., Tenet, HCA, Mayo Clinic, Cleveland Clinic). The strong profit growth is consistent with hospital prices increasing faster than the relatively low rate of input price inflation in the overall economy and the hospital sector in particular.
The purpose of this blog post is to point out some technical caveats in when using BLS hospital price data. These technical issues may explain the divergence between BLS data and the pricing data reported by insurers and hospitals for their privately insured patients. We will discuss technical issues with three different BLS measures of hospital price inflation.
BLS All-payer data.
BLS provides a hospital producer price index for all payers (PCU 622110622110), which reflects a blend of changes in prices for private insurers, fee-for-service (FFS) Medicare, Medicaid, Medicare Advantage and other payers. In recent years, it reflects the slow growth in Medicare rates (including the effect of the sequester) as well as reductions in rates charged to the uninsured in recent years. Therefore, this should not be used to make general statements about how hospital consolidation activity has affected prices paid by private insurers. For example, Dobson and DaVanzo use this index to argue that “hospital consolidation activity has not led to higher hospital price growth” (Dobson et al. 2014). The correct series to look for that type of analysis would be a private-insurer data set, and the better question would be whether price growth continues to be higher than input prices or general wage growth. This would indicate changes in the affordability of hospital services. As we discuss below, the data suggest that private hospital prices continue to grow faster than the general level of inflation in the economy and the average increase in workers’ total compensation (2.2% in 2014) as reported by the BLS in series CIU1010000000000A (Bureau of Labor Statistics 2015a).
BLS “Private insurance and all other” data.
The BLS “A6” series (PCU 62211A62211A6) covers prices paid to hospitals for those with “private insurance and all other patients.” Since the passage of the Patient Protection and Affordable Care Act of 2010 (PPACA) there are several reasons why the data in this series may present slower price growth than a pure private-insurer database.
During 2011 through 2013, the BLS A6 series includes price changes for Medicare Managed Care (MA) patients. We have found that MA plans follow Medicare FFS prices (MedPAC 2015). Because Medicare FFS price growth was lower than privately insured price growth in 2011 through 2013, the BLS data for “private insurance and all other patients” would be expected to present price growth that was lower than the pure rate of commercial price growth among those under 65. This could in part explain the divergence between the A6 series and the HCCI data for 2011 and 2012. Starting midway through 2014, the BLS improved the methodology so that Medicare Advantage prices are now included in the Medicare price levels and not in the “private insurance and all other patients” category.
Second, changes in the prices hospitals charge the uninsured will affect the A6 series. Initially hospitals would often report to BLS that the price for the uninsured was “full charges” even if the hospital did not expect to receive anything close to full charges from those patients. These charges are usually over triple Medicare rates (MedPAC 2015). This should have changed for most hospitals after the passage of PPACA. PPACA limits the prices hospitals are allowed to charge to poor uninsured patients. Non-profit hospitals are not allowed to charge these patients more than they charge those with insurance (e.g., Medicare and private insurance) (Internal Revenue Service 2014, Internal Revenue Service 2015). PPACA was in effect in 2011 and could have had a material effect on the 2011 to 2012 rates charged to the uninsured. For example, if a hospital shifted from charging the uninsured full charges to charging them Medicare rates (consistent with the new law) the price charged them and recorded by BLS would be expected to fall by over 70 percent. Therefore, even a small shift in the number of patients being charged full charges could affect the overall rate of cost growth in the A6 series.
The A6 series also includes small numbers of patients with other types of insurance such as Veterans Administration insurance, which also pays Medicare rates. In addition, the BLS data is adjusted for quality, so the actual cost per service is slightly higher than the quality adjusted inflation rate reported by BLS. BLS measures quality using hospital compare process measures for inpatient cases. As hospitals increasingly adhere to guidelines (such as increasing the share of heart failure patients given discharge instruction), the BLS views that as an improvement in the value of the inpatient service which results in a slight reduction in the producer price index (PPI) all else equal. While the MA effect, the uninsured effect, the quality adjustment, and the effect of other payers such as the VA may each be moderate in magnitude, together they may have contributed to pushing down the price growth reported in the BLS data compared to the data from HCCI, which relies purely on claims for those with commercial insurance. In addition, starting in 2014 the numbers would also be affected by entrance of plans on the exchanges that have narrower networks. By averaging in lower prices for new narrow network products with the historic prices for traditional private insurance, the BLS rates of price growth would be expected to be slightly lower than price increases for existing insurance products. For all the reasons above, researchers should be cautious about using the BLS data for 2012 through 2015 as a proxy for estimating rate increases that insurers are negotiating with hospitals for existing insurance products.
BLS Medicare price data.
One additional word of caution is that BLS reported an unusually large drop in Medicare hospital prices in October 2014 (-2.2 percent from September 2014 to October 2014). This reduction in prices is reported in the series on Medicare prices (PCU62211A62211A2) and is factored into the all-payer series. This reduction of -2.2 percent is about 1.5 percent lower than the CMS estimated change in inpatient prices (Centers for Medicare & Medicaid Services 2014). It is also low compared to the revenue per adjusted admission expected by HCA, the largest hospital chain; they expect Medicare revenue per adjusted admissions (which also reflects intensity changes) to increase by roughly 1 percent in 2015 (The Street 2015a). In part the BLS lower level of reported inflation is because they report a quality-adjusted PPI. However, the quality adjustments are not large enough to fully explain the 1.5 percentage point difference in the BLS and CMS estimates. BLS also reports a Medicare inpatient price index (inpatient services only, series WPU5121010111) that showed a 1 percent decline from September to October of 2014 (Bureau of Labor Statistics 2015b). This appears to be roughly correct given the reductions in rates reported by CMS in their final rule and the effect of improved process measures, which would act to decrease the PPI on a quality-adjusted basis. However, if Medicare inpatient prices declined by 1 percent and outpatient prices did not materially change (outpatient prices change in January), then it is hard to see how Medicare hospital prices overall declined by 2.2 percent from September to October. We are not sure what is driving the low BLS estimate, but given that it differs substantially from the CMS, MedPAC, HCA, and BLS inpatient-only estimates, it should be viewed with caution as should the overall movement of the hospital PPI for the last few months.
Other sources of data
As the general rate of inflation in the economy has remained low, it appears that there has been some moderation in the rate of private insurer price growth. However, it is difficult to determine the magnitude of changes in private insurer prices by only examining the three series of BLS data discussed above. The good news is that there are new sources of data available to create more robust estimates of price growth among private insurers. The first is the new BLS series that has started in the past year that will only use hospital-reported data on those with private insurance (series PCU62211A62211A61). Second, other private sources such as HCCI and Truven use insurer claims data to compute changes in private insurer payment rates to hospitals. Finally, the SEC filings of publicly traded hospital chains often report on revenue growth per adjusted admissions for their private-payer patients. By looking at the improved BLS data in conjunction with other sources of data, estimates of hospital price increases should improve over time.
Implication: The problem of increasing private payer prices has not been solved
In summary, data from private insurers and hospitals both support the finding that the rate of growth in hospital prices has slowed, but the rates paid by private insurers are growing faster than the general rate of inflation and faster than the 2 percent increase in overall employee compensation during the past three years (BLS series CIU10100000000A) (Bureau of Labor Statistics 2015a). This suggests the price problem is not solved. Some may argue that the low rates of growth in Medicare prices forced hospitals to raise prices paid by private insurers by 4 or 5 percent to keep hospitals solvent in 2013 and 2014. That argument does not hold. While restraining Medicare prices can put pressure on hospitals to reduce their cost growth and may help constrain the long-run growth in hospital prices, the full increase in private prices is not caused by Medicare pricing. In 2013, hospitals increased their prices sufficiently to generate the highest overall profit margins in over 20 years. Record profits can be a result of market power, but no industry is forced to raise prices to the point where they have record profits.
Bureau of Labor Statistics. 2015b. Producer Price Index by commodity for health care services: Medicare patients: hospital inpatient care, general and surgical hospitals. http://data.bls.gov/pdq/querytool.jsp?survey=wp.
California Department of Insurance. 2014a. Aetna Life Insurance Company rate filing.https://interactive.web.insurance.ca.gov/apex/f?p=102:9:0::NO::P9_RATE_FILINGS_ID,P9_COMPANY_NAMECalifornia Department of Insurance. 2014b. Blue Shield of California Life & Health Insurance Company rate filing. https://interactive.web.insurance.ca.gov/apex/f?p=102:9:0::NO::P9_RATE_FILINGS_ID
Centers for Medicare & Medicaid Services, Department of Health and Human Services. 2014. Medicare program; inpatient rehabilitation facility prospective payment system for federal fiscal year 2015. Final rule. Federal Register 79, no. 151 (August 6): 45872–45936.
Dobson, A., G. Berger, K. Reuter, et al. 2014. Health care spending slowdown: The consumer paradox. Report prepared by Dobson | DaVanzo for the Federation of American Hospitals. Washington, DC: Federation of American Hospitals. http://fah.org/upload/documents/Dobson-DaVanzo-Federation-Cost-Sharing-Report1.pdf.
Health Research Institute, PricewaterhouseCoopers. 2015. Medical cost trend: Behind the numbers 2016. Washington, DC: HRI. http://altarum.org/sites/default/files/uploaded-publication-files/PwC%20HRI%20Behind%20the%20Numbers%202016_Full%20Report.pdf.
Internal Revenue Service, Department of the Treasury. 2014. Additional requirements for charitable hospitals; community health needs assessments for charitable hospitals; requirement of a Section 4959 excise tax return and time for filing the return. Final rule. Federal Register 79, no. 250 (December 31): 78954-79016.
Internal Revenue Service, Department of the Treasury. 2015. TD 9708: Additional requirements for charitable hospitals; community health needs assessments for charitable hospitals; requirement of a Section 4959 excise tax return and time for filing the return. Internal Revenue Bulletin 2015-5 http://www.irs.gov/irb/2015-5_IRB/ar08.html.
The Street. 2015a. HCA Holdings (HCA) earnings report: Q4 2014 conference call transcript, February 4. http://www.thestreet.com/story/13034409/1/hca-holdings-hca-earnings-report-q4-2014-conference-call-transcript.html.
The Street. 2015b. Tenet Healthcare (THC) earnings report: Q4 2014 conference call transcript, February 25. http://www.thestreet.com/story/13057471/1/tenet-healthcare-thc-earnings-report-q4-2014-conference-call-transcript.html.