How Much More Than Medicare Do Private Insurers Pay? A Review of the Literature

Appendix Table: Summary Of The Data Sources And Methodological Details For The Reviewed Studies
Study Data Year(s) Payments Addressed Data Sources, Methodology, and Factors Influencing Results
American Hospital Association Annual Survey 2018 2010-2017 Hospital Source data: American Hospital Association (AHA) Annual Survey of Hospitals for 2010-2016, including facility-level payment and cost data provided by over 6000 AHA member hospitals nationwide, representing roughly 85% of U.S. hospitals. Respondents are community hospitals, defined by the AHA as all non-federal short-term hospitals, including academic medical centers and teaching hospitals.

 

Methods: Using hospital-reported payment data for hospitals participating in the AHA annual survey, the authors calculate overall payment-to-cost ratios (PTCRs) for each year. This review further calculates ratios of the PTCR for private payers to the Medicare PTCR, resulting in estimates of the private-to-Medicare payment ratio for each year. Methodology is consistent across years, so changes in the Medicare-private payment differential are likely to reflect true changes over time. The analysis does not control for hospital case mix or patient-specific factors such as medical complexity.

 

Factors Influencing Reported Rates:  The hospitals included in the AHA annual survey include academic medical centers, community hospitals, and non-profit hospitals, which may result in more broadly representative estimates of the private-to-Medicare hospital payment ratio than studies that focus on for-profit hospitals. Medicare PTCRs are based on the amounts that the Centers for Medicare and Medicaid Services (CMS) actually paid, including payments for Disproportionate Share Hospital (DSH) status, indirect medical education (IME), geographic adjustments, pass-through payments including those for direct graduate medical education (DGME), etc., which may result in higher estimates of Medicare rates than studies that exclude these payment components. Self-pay and privately insured patients are not differentiated in the reported private PTCRs, which may skew estimates of private insurance payments downward. However, because the private PTCRs are based on payments rather than claims, out-of-network payments are included, which may result in higher estimates of private insurance rates than studies that exclude out-of-network claims.

Baker et. al. 2016 2012 Hospital Source data: Health Care Cost Institute (HCCI) claims data for 2012, which includes claims for three large nationwide private insurers – UnitedHealthcare, Humana, and Aetna, representing hospital service claims for approximately 4 million individuals nationwide. Medicare claims were obtained from the Medicare Provider Analysis and Review (MEDPAR) File for 2012, which includes individual hospital price data for all Medicare FFS enrollees.

 

Methods: Private insurance rates were calculated based on prices for commercial hospital claims for beneficiaries aged 65 and older and their dependents enrolled in HMO, PPO, or POS plans, aggregated to the event level. Medicare rates were calculated based on a 100% sample of Medicare FFS enrollees, aggregated to the event level. Authors compared average commercial and Medicare prices for the top 100 DRGs by volume, at both the CBSA level and nationally. The analysis does not control for patient-specific factors such as medical complexity, though calculating private insurance rates based on claims only for beneficiaries age 65 and older may partially account for differences in medical complexity among privately insured and Medicare beneficiary claims.

 

Factors Influencing Reported Rates: The specific insurers represented in this dataset may not reflect the national distribution of enrollees in private insurance plans, so the estimates of private insurance rates may not reflect rates negotiated by other insurers, particularly smaller local insurers with less negotiating power relative to providers that may have higher provider payments. Medicare payments are based on the amounts that CMS actually paid, including payments for DSH status, IME, geographic adjustments, pass-through payments, etc., which may result in higher estimates of Medicare rates than studies that exclude these payment components. Claims with zero or negative payments and those in the top and bottom 1% of the payment amount distribution by DRG and insurance type were excluded, reducing the effect of extreme payments relative to studies that do not exclude outliers. The data do not contain information on provider payments made outside of the usual claims process. Out-of-network claims are included, which may result in higher estimates of private insurance rates than studies that exclude out-of-network claims.

Biener and Selden 2017 2014-2015 data consolidated Physician Source data: Medical Expenditures Panel Survey (MEPS) Household and Medical Provider components for 2014-2015, which include provider billing records containing full established charges and payments for over 30,000 office-based physician visits, across private insurers, Medicare, and other payers.

 

Methods: The authors analyze payments associated with all office-based physician visits in the dataset for adults aged 18 and older (excluding visits characterized by the reporting household as emergency visits, as well as visits for counselling only, vision exams, or surgical procedures). Visit payments were regressed on coverage type and several patient and visit characteristics, and mean payments were estimated for private payers and Medicare FFS based on the regression model. The MEPS dataset is not large enough to be representative at the state level, so the authors do not address geographic variation, and report national averages only. The authors do not attempt to control for service intensity or patient-specific factors such as medical complexity.

 

Factors Influencing Reported Rates: MEPS data contains information on physician office visits, including visit type (PCP, specialist, etc.) and content, allowing the authors to control for factors such as practice setting and physician specialty in their regression model. Payments in the top 1% of visits ranked by charges were excluded for all payment types. Because private rates are calculated based on physician office billing data, payments outside of the claims process are included, as are payments for out-of-network claims, which may result in higher estimates of private insurance rates than studies that exclude these payments. Because the authors exclude procedures rendered in hospital outpatient departments (which are often reimbursed at higher rates than identical procedures rendered in physician offices under current Medicare policy), this analysis may produce lower estimates of average Medicare rates that studies that incorporate additional care settings.

Colorado Department of Health Care Policy and Financing 2019 2010-2017 Hospital Source data: Colorado Hospital Association’s “DATABANK” database for 2010-2017, which contains aggregated hospital utilization and payment data reported by 67 participating Colorado hospitals representing over 97% of licensed beds in the state.

 

Methods: The authors calculate aggregate payment-to-cost ratios (PTCRs) for each payer type (e.g. commercial insurance, Medicare) across all hospitals in the dataset. This review further calculates ratios of the PTCR for private payers to the Medicare PTCR, resulting in estimates of the private-to-Medicare payment ratio for each year. Methodology is consistent across years, so changes in the Medicare-private payment differential are likely to reflect true changes over time. The analysis does not control for hospital case mix or patient-specific factors such as medical complexity.

 

Factors Influencing Reported Rates: Data are specific to participating hospitals in Colorado, so are likely biased toward payment rates prevailing in Colorado hospital markets. White and Whaley (see table entry below) find that hospital prices in Colorado are relatively high compared to those in most of the 24 other states included in their 2019 study, suggesting that Colorado hospitals’ high degree of market power relative to insurers may result in higher estimates of private insurance hospital payment rates. Reported Medicare payments are based on the amounts that CMS actually paid, including payments for DSH status, IME, geographic adjustments, pass-through payments, etc., which may result in higher estimates of Medicare rates than studies that exclude these payment components. Separate PTCRs are reported for self-pay and privately insured patients, so estimates of private insurance rates are not affected by differences between self-pay and privately insured payments, in contrast to the AHA analyses. Out-of-network payments are included in the private insurance PTCR, which may result in higher estimates of private insurance rates than studies that exclude out-of-network claims.

Cooper et. al. 2018 2011 Hospital Source data: Health Care Cost Institute (HCCI) claims data for 2011, which includes claims for three large nationwide private insurers – UnitedHealthcare, Humana, and Aetna – representing hospital service claims for approximately 4 million individuals nationwide. Medicare claims data were obtained from the Medicare Provider Analysis and Review (MEDPAR) File for 2011, which includes individual hospital price data for all Medicare FFS enrollees.

 

Methods: The authors calculate mean hospital prices for privately reimbursed and Medicare-reimbursed stays based on approximately 2,400 general medical and surgical hospitals in the dataset for which inpatient stays across both payer types were available. The authors adjust for differences in medical complexity between the privately insured and Medicare populations by risk-adjusting price, for differences in each hospitals’ mix of admission types by adjusting for the relative volume of DRGs, and for differences in care quality by analyzing price variation within each hospital.

 

Factors Influencing Reported Rates: The specific insurers represented in this dataset may not reflect the national distribution of enrollees in private insurance plans, so the estimates of private insurance rates may not reflect rates negotiated by other insurers, particularly smaller local insurers with less negotiating power relative to providers that may have higher provider payments. Medicare payments are based on the amounts that CMS actually paid, including payments for DSH status, IME, geographic adjustments, pass-through payments, etc., which may result in higher estimates of Medicare rates than studies that exclude these payment components. The data do not contain information on provider payments made outside of the usual claims process, which may underestimate privately reimbursed or Medicare payments that have non-claim components. Out-of-network claims are included, which may result in higher estimates of private insurance rates than studies that exclude out-of-network claims.

Ge and Anderson 2018 2016 Hospital Source data: Florida Agency for Health Care Administration hospital financial information records for 2010-2016, which includes audited hospital-reported gross revenues, deductions, and net revenues by type of payer (e.g., commercial insurance, Medicare) for 153 private short-term hospitals in Florida.

 

Methods: The authors calculate net payment ratios (net revenue to gross revenue) for both Medicare and private HMO and PPO plans across all hospitals in the dataset, as well as the ratio of these net payment ratios, yielding an estimated private HMO/PPO-to-Medicare payment ratio, which the authors report for 2016. The authors do not control for hospital case mix or patient-specific factors such as medical complexity.

 

Factors Influencing Reported Rates: Data are specific to participating hospitals in Florida, so are likely biased toward payment rates prevailing in Florida hospital markets. The authors report that the proportion of for-profit hospitals is much higher in Florida than nationwide, which may result in higher estimates of private insurance payment rates. However, White and Whaley (see table entry below) find that hospital prices in Florida are roughly comparable to those in most of the 24 other states included in their 2019 study. Reported private insurance rates are based on payments from HMO and PPO plans only (excluding other plan types such as POS), which may bias reported payments toward HMO and PPO rates. Medicare payments are based on the amounts that CMS actually paid, including payments for DSH status, geographic adjustments, pass-through payments, etc., which may result in higher estimates of Medicare rates than studies that exclude these payment components. Out-of-network payments are included, which may result in higher estimates of private insurance rates than studies that exclude out-of-network claims.

Ginsburg 2010 2010 Physician and Hospital Source data: Physician and hospital payment rates directly provided by four large private insurers – Aetna, Anthem Blue Cross Blue Shield, Cigna, and UnitedHealth – for services reimbursed in 2010 in eight diverse health care markets throughout the U.S. (Cleveland, OH; Indianapolis, IN; Los Angeles, CA; Miami, FL; Milwaukee, WI; Richmond, VA; San Francisco, CA; and rural Wisconsin). Medicare payment rates are also provided by the same insurers, based on simulated Medicare equivalent payments for the provided private claims.

 

Methods: The authors used insurer-provided average private insurance and Medicare payment rates to calculate average private insurance-to-Medicare payment ratios for all inpatient hospital services, all outpatient hospital services, and for physician services across several medical specialties, respectively. Exact calculation methods may have varied for each insurer, though additional information is not available due to the authors’ data-sharing agreement with the participating insurers. Because information on the relative volume of claims in each market is not provided, this brief further calculates unweighted average ratios across all markets.

 

Factors Influencing Reported Rates: Although only eight markets are represented, the authors specifically chose these markets due to their geographic spread and expected variation in payment rates, which may moderate the effects of nationwide variation in the private-to-Medicare payment ratio. The specific insurers represented in this dataset may not reflect the national distribution of enrollees in private insurance plans, so the estimates of private insurance rates may not reflect rates negotiated by other insurers, particularly smaller local insurers with less negotiating power relative to providers that may have higher provider payments.  The insurer-reported “standard” private insurance rates for physician services are derived from a sample in which typically “hospital-based” specialties such as anesthesiologists and radiologists are underrepresented, which may bias private insurance rates downward relative to Medicare levels. The data do not contain information on provider payments made outside of the usual claims process, which may underestimate privately reimbursed or Medicare payments that have non-claim components. The authors requested that participating insurers exclude out-of-network claims from their reported data, which may result in lower estimates of private insurance payments than among studies that include out-of-network claims.

Kronick and Neyaz 2019 2015-2016 data consolidated Hospital Source data: California hospitals’ Annual Financial Disclosure Reports filed with the California Office of Statewide Health Planning and Development (OSHPD) for 2015-2016, which includes payment and cost data by type of payer (e.g., commercial insurance, Medicare) for all California hospitals.

 

Methods: The authors calculate aggregate payment-to-cost ratios (PTCRs) for each payer type (e.g., commercial insurance, Medicare) across all hospitals in the dataset, as well as ratios of the PTCRs for each payer type, yielding an estimated private-to-Medicare charge ratio. Inpatient and outpatient services are aggregated together. The authors limit their analysis to acute care general hospitals, excluding psychiatric, rehabilitation, substance abuse, long-term care, and children’s’ hospitals. The authors do not control for hospital case mix or patient-specific factors such as medical complexity.

 

Factors Influencing Reported Rates: Data are specific to participating hospitals in California, so are likely biased toward payment rates prevailing in California hospital markets. OSHPD reports only include cost data at the level of hospital units rather than for specific procedures or services, though the authors adapt formulas used by OSHPD to account for indirect hospital payments not associated with specific hospital units, preventing the potential underestimations of private insurance and Medicare rates that might have otherwise been caused by excluding these payments. Specifically, Medicare payments are based on the amounts that CMS actually paid, including payments for DSH status, geographic adjustments, pass-through payments, etc., which may result in higher estimates of Medicare rates than studies that exclude these payment components. Out-of-network payments are included in the private insurance PTCR, which may result in higher estimates of private insurance rates than studies that exclude out-of-network claims.

Maeda and Nelson, CBO 2017 2013 Hospital Source data: Health Care Cost Institute (HCCI) claims data for 2013, which includes claims for three large nationwide private insurers – UnitedHealthcare, Humana, and Aetna – representing hospital service claims for approximately 4 million individuals.

 

Methods: The authors aggregate inpatient claims to the level of hospital stays (of which they identify over 620,000 in the dataset), then calculate average overall allowed payment amounts per stay by payer type (e.g., private insurance, Medicare). The authors limit their analysis to stays in acute care hospitals in MSAs only, and exclude stays associated with services that uncommon among Medicare patients such as labor and delivery. The authors simulate Medicare payments based on the same claims used in the calculation of private insurance payments, and include simulated payments for Medicare-allowed cost-sharing, DSH status, geographic adjustments, etc., but not pass-through or IME payments.

 

Factors Influencing Reported Rates: The specific insurers represented in this dataset may not reflect the national distribution of enrollees in private insurance plans, so the estimates of private insurance rates may not reflect rates negotiated by other insurers, particularly smaller local insurers with less negotiating power relative to providers that may have higher provider payments. Medicare rates are simulated by the authors, including hospital-specific payment adjustments such as those for DSH status, and geographic adjustments, though not IME and pass-through payments; this may result in higher estimates of Medicare rates compared to studies that do not include any of these payments, but relative underestimates of Medicare rates compared to studies that use the amounts that CMS actually paid. The authors focus on hospitals in Metropolitan Statistical Areas (MSAs) only, which may skew the reported rates toward those prevailing in large urban areas. The data do not contain information on provider payments made outside of the usual claims process, which may underestimate privately reimbursed or Medicare payments that have non-claim components. Out-of-network payments are included, which may result in higher estimates of private insurance rates than studies that exclude out-of-network claims.

MedPAC 2012-2019 2010-2017 Physician Source data: Claims data provided by an undisclosed private insurer that covers a wide geographic area across the U.S., including all paid claims for services covered by the insurer’s PPO plans in 2010-2017, reflecting both the insurer’s allowed amount and patient cost-sharing amounts. Due to MedPAC’s data-sharing agreement with the private insurer, additional information is not available. Medicare beneficiary-level claims data provided directly by CMS for 100% of Medicare beneficiaries in 2010-2017, including Medicare allowed amounts and Medicare’s allowed patient cost-sharing amounts.

 

Methods: The authors do not report detailed methods for their comparison of private insurance (PPO) payment rates and Medicare payment rates. However, the authors state that the methodology has been consistent across years, so changes in the Medicare-private payment differential are likely to reflect true changes over time.

 

Factors Influencing Reported Rates: Although the data represent a wide geographic distribution of claims, the fact that only one private insurer’s PPO plans are represented limits the generalizability of the findings, and may skew estimates toward the generally lower payment rates that large insurers are able to negotiate with providers due to their greater bargaining power. Medicare payments are based on the amounts that CMS actually paid, including Medicare-allowed cost-sharing, payments for DSH status, IME, geographic adjustments, pass-through payments, etc., which may result in higher estimates of Medicare rates than studies that exclude these payment components. The data do not contain information on provider payments made outside of the usual claims process, which may result in lower estimates of private insurance rates than studies that include these payment components. The authors do not report whether out-of-network claims are included in the estimation of private insurance payment rates.

Pelech 2018 2014 Physician Source data: Health Care Cost Institute (HCCI) claims data for 2014, which includes physician service claims for three large nationwide private insurers – UnitedHealthcare, Humana, and Aetna – for approximately 39 million individuals.

 

Methods: The authors aggregate claims for physician services (of which they identify over 230 million in the dataset, representing services provided to over 19 million patients by over 600,000 physicians), then calculate average overall allowed payment amounts across several services by payer type (e.g., private insurance, Medicare). The authors limit their analysis to 12 physician services that are comparably common among both privately insured and Medicare beneficiaries and delivered in MSAs. The authors simulate Medicare payments based on the same claims used in the calculation of private insurance payments, and include simulated payments for cost-sharing and procedure- and provider-specific adjustments.

 

Factors Influencing Reported Rates: The majority of physician services included in the analysis are procedural or imaging studies, which may result in higher estimates of private insurance rates than studies focusing on physician services for specialties such as primary care that have relatively low market power compared to procedural specialists and radiologists. The specific insurers represented in this dataset may not reflect the national distribution of enrollees in private insurance plans, so the estimates of private insurance rates may not reflect rates negotiated by other insurers, particularly smaller local insurers with less negotiating power relative to providers that may have higher provider payments. Medicare rates are simulated by the authors, including Medicare-allowed adjustments for bilateral services, performing multiple same-day procedures , etc.; this may result in higher estimates of Medicare rates compared to studies that do not include any of these adjustments. The analysis is limited to MSAs, which may skew the reported rates toward physician prices prevailing in large urban areas. The data do not contain information on provider payments made outside of the usual claims process, which may underestimate privately reimbursed or Medicare payments that have non-claim components. Out-of-network payments are included in varying proportions by service, which may result in higher estimates of private insurance rates than studies that exclude out-of-network claims.

Selden et. al. 2015 2010-2012 Hospital Source data: Medical Expenditures Panel Survey (MEPS) Household and Medical Provider components for 1996-2012, which include payment data for approximately 39,500 non-maternity hospital stays nationwide, across private insurers, Medicare, and other payers.

 

Methods: The authors analyze payment-to-charge ratios associated with inpatient hospital stays in the dataset for adults aged 19 and older. Hospital stays were regressed on coverage type and several patient and visit characteristics, and mean payments were estimated for private payers and Medicare based on the regression model. Based on these results, the authors calculated mean private-to-Medicare payment ratios for each year for which data was available. The authors do not address geographic variation, and merely report national averages. Methodology is consistent across years, so changes in the Medicare-private payment differential are more likely to reflect true changes over time. The authors do not control for hospital case mix or patient-specific factors such as medical complexity.

 

Factors Influencing Reported Rates: The authors compare average inpatient hospital payment rates based on the “full established charge” for entire hospital stays (weighted by volume) rather than charges for specific procedures, which may be influenced more heavily by differences in patient-specific factors such as medical complexity. Some payments made by supplemental managed care plans (Medicare Advantage) are included in the calculation of Medicare rates, though given that Medicare Advantage payments are generally similar to Medicare rates for inpatient hospital services (see e.g., Trish et. al.), the effect of including these payments is likely minimal.  Reported Medicare rates include payments for DSH status, IME, and outlier payments, but exclude payments not linked to specific events such as pass-through payments; this may result in higher estimates of Medicare rates compared to studies that do not include any of these payments, but relative underestimates of Medicare rates compared to studies that include all Medicare payment components. The authors exclude hospital stays with payments exceeding three times charges, which may disproportionately exclude exceptionally high private insurance payments and result in lower estimates of the private-to-Medicare payment ratio. Because private rates are calculated based on payment data for hospital stays, payments outside of the claims process are included, as are payments for out-of-network claims, which may result in higher estimates of private insurance rates than studies that do not include these payments.

Selden 2020 2013-2016 Hospital Source data: Medical Expenditures Panel Survey (MEPS) Household and Medical Provider components for 2013-2016, which include payment data for approximately 48,000 non-maternity inpatient hospital stays and 236,500 non-emergency outpatient visits nationwide across private insurers and Medicare.

 

Methods: This analysis updates the findings of Selden et. al. 2015 (see table entry above), preserving the same methodology for inpatient services, as well as an additional analysis for outpatient services over the period from 1996-2016. Prices for inpatient stays and outpatient service events were regressed on coverage type and several patient and visit or service characteristics, and mean payments were estimated for private payers and Medicare based on the regression model. Based on these results, the authors calculated mean private-to-Medicare payment ratios for each year. The authors do not address geographic variation, and merely report national averages. Methodology is consistent across years, so changes in the Medicare-private payment differential are more likely to reflect true changes over time. The authors do not control for hospital case mix or patient-specific factors such as medical complexity.

 

Factors Influencing Reported Rates: The authors compare average inpatient hospital payment rates based on the “full established charge” for entire hospital stays (weighted by volume) rather than charges for specific procedures, which may be influenced more heavily by differences in patient-specific factors such as medical complexity. Average outpatient hospital payment rates are calculated based on event-level data (weighted by volume), which may result in lower estimates of private insurance payments for outpatient services than studies that focus on selections of services that disproportionally represent less common but more highly reimbursed services. Some payments made by supplemental managed care plans (Medicare Advantage) are included in the calculation of Medicare rates, though given that Medicare Advantage payments are generally similar to Medicare rates for inpatient hospital services (see e.g. Trish et. al.), the effect of including these payments is likely minimal. Reported Medicare rates include payments for DSH status, IME, and outlier payments, but exclude payments not linked to specific events such as pass-through payments; this may result in higher estimates of Medicare rates compared to studies that do not include any of these payments, but relative underestimates of Medicare rates compared to studies that include all Medicare payment components. The authors exclude hospital stays with payments exceeding three times charges, which may disproportionately exclude exceptionally high private insurance payments and result in lower estimates of the private-to-Medicare payment ratio. Because private rates are calculated based on payment data for hospital stays, payments outside of the claims process are included, as are payments for out-of-network claims, which may result in higher estimates of private insurance rates than studies that do not include these payments.

Song 2019 2016 Physician Source data: Truven Health Analytics MarketScan Commercial Claims and Encounters database 2016 claims data, which contains non-facility physician payment information for approximately 350 private payers throughout the U.S.

 

Methods: The author reports unweighted mean private insurance prices for a selection of 12 physician services for all private insurance plan types other than capitated plans and POS plans with capitation. Medicare prices are simulated based on average non-facility professional fees derived from the 2016 Medicare Physician Fee Schedule for the same set of services, without accounting for geographic adjustments or other payment modifications.

 

Factors Influencing Reported Rates: The MarketScan dataset includes a diverse mix of private insurers, which may result in lower estimates of private insurers’ physician payment rates than studies that examine payments for only a few large nationwide private insurers that likely have stronger negotiating power relative to providers. The simulated Medicare rates do not include payments for DSH status, IME, and outlier payments, etc., which may result in lower estimates of Medicare rates compared to studies that do account for these payments. The data do not contain information on provider payments made outside of the usual claims process, which may underestimate privately reimbursed or Medicare payments that have non-claim components. The author distinguishes between payments for in-network and out-of-network services; for the purposes of this brief, the author’s average reported in-network rates are used in the calculation of the overall cross-study average, which may result in a lower private-to-Medicare payment ratio than studies that include out-of-network payments.

Trish et. al. 2017 2007-2012 data consolidated Physician Source data: Claims data provided directly by an undisclosed private insurer that covers a wide geographic area across the U.S., including over 8.5 million claims for private insurance enrollees from 2007-2012. Due to the authors’ data-sharing agreement with the private insurer, additional information is not available. Medicare claims from a 20% random sample of Medicare beneficiaries for 2007-2012 were provided directly by CMS.

 

Methods: The authors calculate the mean prices for each type of service by plan type (e.g. private insurance, Medicare), care setting (e.g. physician office, hospital outpatient department, etc.), CBSA, and year, then average these relative mean prices across CBSAs for each year, weighting by the private insurer’s enrollment in each CBSA. Finally, the authors calculate mean private-to-Medicare payment ratios using these averages. The authors do not control for patient-specific factors such as medical complexity.

 

Factors Influencing Reported Rates: Although the data represent a wide geographic distribution of claims, the fact that only one private insurer is represented limits the generalizability of the findings, and may skew estimates toward the generally lower payment rates that large insurers are able to negotiate with providers due to their greater bargaining power. The authors incorporate a broad variety of physician specialties, which may result in estimates of the private-to-Medicare physician ratio that fall between those reported by studies in which more procedurally-oriented or more cognitively-oriented specialties are disproportionately represented. Medicare payments are based on the amounts that CMS actually paid, including provider-specific and service-specific payment adjustments, which may result in higher estimates of Medicare rates than studies that exclude these payment components. Claims associated with prices more than four standard deviations above or below the mean for each payer were excluded, reducing the effect of extreme payments relative to studies that do not exclude outliers. The data do not contain information on provider payments made outside of the usual claims process, which may underestimate privately reimbursed or Medicare payments that have non-claim components. Out-of-network payments are included, which may result in higher estimates of private insurance rates than studies that exclude out-of-network claims.

Wallace and Song 2016 2007-2013 data consolidated Physician and Hospital Source data: Truven Health Analytics Medicare and Commercial Claims and Encounters database for 2007-2013, which contains physician service claims for approximately 350 private payers nationwide, as well as for a convenience sample of Medicare patients, together representing approximately 1.4 million individuals that transitioned from private insurance to Medicare between 2007 and 2013.

 

Methods: The authors longitudinally follow a cohort of approximately 200,000 individuals in the dataset who retired from employment and transitioned from private insurance to Medicare at age 65. Mean prices by payer type are estimated by aggregating claims from before and after their transition into separate claims samples for private insurance and Medicare, and using a regression discontinuity model to identify the causal effect of entry into Medicare, controlling for a number of individual-specific and time-specific factors. The authors exclude services that do not have cost-sharing requirements for Medicare beneficiaries such as certain primary care and preventive care office visits, instead focusing on a subset of procedural and imaging services that consistently have patient cost-sharing requirements.

 

Factors Influencing Reported Rates: The authors longitudinally observe a consistent cohort from the year before and after their transition from private to Medicare coverage at age 65, which largely controls for medical complexity and other patient-specific features, and allows the authors to account for possible changes in utilization that would otherwise impact reported rates. The authors focus on a subset of mainly procedural and imaging services, which may result in relatively lower estimates of the private-to-Medicare payment ratio than studies that focus on routine evaluation and management services in the hospital setting, or higher estimates than studies focusing on physician services for specialties such as primary care that have relatively low market power compared to procedural specialists and radiologists. The reported private and Medicare rates are based on author-determined “changes attributable to Medicare entry”, which causes estimates of the private-to-Medicare ratio to be lower than if estimates had been based on only the raw reduction in price. The MarketScan dataset includes a diverse mix of private insurers, which may result in lower estimates of private insurers’ physician payment rates than studies that examine payments for only a few large nationwide private insurers that likely have stronger negotiating power relative to providers. Only private payments made by indemnity plans, PPOs, and POS plans were included, which may bias estimates of private insurance payments toward rates paid by these plan types (as opposed to others such as HMOs). The data do not contain information on provider payments made outside of the usual claims process, which may underestimate privately reimbursed or Medicare payments that have non-claim components. Out-of-network payments are included, which may result in higher estimates of private insurance rates than studies that exclude out-of-network claims.

White et. al. 2013 2011 Physician and Hospital Source data: Claims data provided by General Motors, Chrysler, Ford, and the United Auto Workers Retiree Medical Benefits Trust across 5 major metropolitan physician and hospital markets in Michigan, for 2011. Provided claims information included amounts paid to the physicians and hospitals, as well as amounts paid by the insurer and enrollee, for approximately 590,000 active and retired non-elderly autoworkers and their dependents under age 65.

 

Methods: The authors aggregated private insurance claims to the level of DRG codes (for inpatient hospital services) or HCPCS codes (for outpatient hospital services and physician services), and calculated mean provider-level prices for each code. Medicare prices were simulated based on the Medicare IPPS, OPPS, or MPFS, including geographic adjustments, provider-specific adjustments such as IME and DSH payments, and MPFS adjustments. Using these payment estimates, the authors calculated provider-level private-to-Medicare price indices for each hospital (including inpatient and outpatient price indices) and physician practice (for primary care services, medical specialist services, and surgical specialist services). Providers were assigned to markets based on their location, and mean price ratios are reported for each market and overall.

 

Factors Influencing Reported Rates: Although the data represent a wide geographic distribution of claims, all markets were in Michigan, which the authors note may result in relatively low estimates of private insurance payment rates due to the exceptionally strong statewide market power of Blue Cross Blue Shield Michigan relative to providers. Simulated Medicare rates include payments for DSH status, geographic adjustments, IME, and outlier payments, but exclude payments not linked to specific events such as pass-through payments; this may result in higher estimates of Medicare rates compared to studies that do not simulate any of these payments, but relative underestimates of Medicare rates compared to studies that use the amounts that CMS actually paid. Private claims associated with prices outside the range of 20-1000% of the corresponding Medicare payment were excluded, reducing the effect of extreme payments relative to studies that do not exclude outliers. The data do not contain information on provider payments made outside of the usual claims process, which may underestimate privately reimbursed or Medicare payments that have non-claim components. Out-of-network payments are included, which may result in higher estimates of private insurance rates than studies that exclude out-of-network claims.

White 2017 2013-2016 data consolidated Hospital Source data: Claims data provided by the Employers’ Forum of Indiana for 2013-2016 for over a dozen participating employers in Indiana that cover employee health benefits, including prices paid for hospital services for approximately 225,000 covered employees and their dependents under age 65.

 

Methods: The authors aggregated private insurance claims to the level of DRG codes (for inpatient hospital services) or APC codes (for outpatient hospital services), and calculated mean provider-level prices for each code. Medicare prices were simulated based on the Medicare IPPS or OPPS, including geographic adjustments, and hospital-specific adjustments such as IME and DSH payments. Using these payment estimates, the authors calculated provider-level private-to-Medicare relative prices for each hospital (for both inpatient and outpatient services) and overall.

 

Factors Influencing Reported Rates: The data are specific to a relatively small population in Indiana, so likely skews estimates of the private-to-Medicare payment ratio to those prevailing in the state; the author notes that the hospital market in Indiana is dominated by 6 large hospital systems, and so may reflect a greater degree of hospital market power relative to insures than in other states. Simulated Medicare rates include payments for DSH status, geographic adjustments, IME, and outlier payments, but exclude payments not linked to specific events such as pass-through payments; this may result in higher estimates of Medicare rates compared to studies that do not simulate any of these payments, but relative underestimates of Medicare rates compared to studies that use the amounts that CMS actually paid. The data do not contain information on provider payments made outside of the usual claims process, which may underestimate privately reimbursed or Medicare payments that have non-claim components. Out-of-network payments are included, which may result in higher estimates of private insurance rates than studies that exclude out-of-network claims.

White and Whaley 2019 2015-2017 Hospital Source data: Claims data provided by approximately 50 self-insured employers for covered non-elderly employees and their dependents, private claims from statewide all-payer claims databases for New Hampshire and Colorado, and claims from additional undisclosed private health plans. Together, these sources include payment data for about 1,600 hospitals and 4 million patients in 25 states in 2015-2017.

 

Methods: The authors aggregated private insurance claims to the level of DRG codes (for inpatient hospital services) or APC codes (for outpatient hospital services), and calculated mean provider-level prices for each code. Medicare prices were simulated based on the Medicare IPPS or OPPS, including geographic adjustments, and hospital-specific adjustments such as IME and DSH payments, but excluding pass-through payments. Using these payment estimates, the authors calculated provider-level private-to-Medicare relative prices for each hospital, including relative prices for inpatient and outpatient services. Providers were assigned to states based on their location, and mean price ratios are reported for each state and overall. Methodology is consistent across years, so changes in the Medicare-private payment differential are more likely to reflect true changes over time.

 

Factors Influencing Reported Rates: The variety of data sources included in the analysis capture a broad range of private insurers, health care markets, and health insurance markets across 25 states, allowing for estimates of the overall private-to-Medicare payment ratio that are more likely to be nationally representative than studies that examine markets in only one state, or only a handful of private insurers, etc. Simulated Medicare rates include payments for DSH status, geographic adjustments, IME, and outlier payments, but exclude payments not linked to specific events such as pass-through payments; this may result in higher estimates of Medicare rates compared to studies that do not simulate any of these payments, but relative underestimates of Medicare rates compared to studies that use the amounts that CMS actually paid. The authors also exclude some Medicare hospital payments for uncompensated care, but substitute their own formula to partially account for them, which may result in lower estimates of Medicare payment rates for hospitals that provide high volumes of uncompensated care. The data do not contain information on provider payments made outside of the usual claims process, which may underestimate privately reimbursed or Medicare payments that have non-claim components. The data do not contain enough information to differentiate between in-network and out-of-network claims, so the latter likely represented substantially in the calculation of private insurance rates,   which may result in higher estimates of private insurance rates than studies that exclude out-of-network claims.

NOTES: Abbreviations: AHA – American Hospital Association. CAH – critical access hospital. CBSA – core-based statistical area. CMS – Centers for Medicare and Medicaid Services. DGME – direct graduate medical education. DRG – diagnosis-related group. DSH – disproportionate share hospital. ESRD – end-stage renal disease. FFS – fee-for-service. GAF – Medicare geographic adjustment factor. HCCI – Health Care Cost Institute. HCPCS – healthcare common procedure coding system. HMO – health maintenance organization. HRR – hospital referral region. ICU – intensive care unit. IME – indirect medical education. IPPS – Medicare Inpatient Prospective Payment System. MedPAC – Medicare Payment Advisory Commission. MedPAR – Medicare Provider Analysis and Review. MEPS – Medical Expenditures Panel Survey. MPFS – Medicare Physician Fee Schedule. MSA – metropolitan statistical area. MS-DRG – Medicare-severity diagnosis-related group. OPPS – Medicare Outpatient Prospective Payment System. OSHPD – California Office of Statewide Health Planning and Development. PCP – primary care physician. POS – point-of-service. PPO – preferred provider organization. PPS – Medicare Prospective Payment System. PTCR – payment-to-cost ratio.
Issue Brief

The Henry J. Kaiser Family Foundation Headquarters: 185 Berry St., Suite 2000, San Francisco, CA 94107 | Phone 650-854-9400
Washington Offices and Barbara Jordan Conference Center: 1330 G Street, NW, Washington, DC 20005 | Phone 202-347-5270

www.kff.org | Email Alerts: kff.org/email | facebook.com/KaiserFamilyFoundation | twitter.com/kff

Filling the need for trusted information on national health issues, the Kaiser Family Foundation is a nonprofit organization based in San Francisco, California.