Hospital performance measures, such as prices and costs, are associated with hospital-insurer contract types.
Objectives: To describe the association between the form of hospitals’ contracts—either markup from a benchmark or a discount from a list price—and performance: price, charge, cost, and length of stay.
Study Design: Retrospective observational study using administrative claims data matched with hospital characteristics from the American Hospital Association Annual Survey and the Healthcare Cost Reporting Information System. Data include a balanced panel of 1889 general acute care hospitals for the years 2011 to 2015.
Methods: Inpatient hospital commercial claims data from the Health Care Cost Institute were used to classify claims by contract type based on claim-line billed and allowed charges. Hospital-level performance measures—prices, charges, costs, and length of stay—were analyzed using linear regression models to identify the association of each measure with contract types. All measures were risk adjusted to control for differences in hospital case mix, and the regression specifications controlled for numerous hospital characteristics.
Results: Our estimate of the distribution of contract types in the data is similar to estimates using other methods. We find that discounted charges contracts are associated with higher prices and higher costs but not higher charges. Fixed-rate contracts are associated with lower prices as well as lower costs.
Conclusions: Limited research exists on the relationship between contract structure and hospital performance. Our results suggest that hospital performance is related to contract structure, possibly due to factors such as differences in bargaining strategies or ex post incentives.
Am J Manag Care. 2021;27(6):242-248. https://doi.org/10.37765/ajmc.2021.88663
Hospital performance measures, such as prices and costs, are associated with hospital-insurer contract types.
Hospital prices have been rising more rapidly than other health care provider prices in recent years.1 Hospital spending alone accounted for more than 5% of US gross domestic product in 2018.2 Hospitals are also rapidly consolidating and, in some cases, gaining more leverage in negotiations with health payers, leading to price increases without substantial quality improvements.3-6 Historically, hospital market power and its effects on price and quality have been addressed through antitrust enforcement, such as contesting mergers.
However, policy makers, regulators, and payers have begun to address market power, prices, and quality through hospital-insurer contracts. For example, California reached an antitrust settlement with a large hospital system that limits the terms of its contracts with private insurers.7 Large self-insured employers have responded to rising prices by renegotiating the prices and structure of contracts with hospitals, tying their reimbursements to a multiple of Medicare’s Inpatient Prospective Payment System (IPPS).8,9 The state employee health plan in Montana reportedly will save $15.6 million annually using a Medicare baseline instead of a discount off hospital charges (ie, “list prices”).8 The federal government and private payers have also implemented numerous alternative payment programs (eg, accountable care organizations, value-based purchasing) under the presumption that payment structure affects ex post provider incentives to control costs and improve quality.10,11
Understanding the implications of contracting is fundamental to effective policy because contracts create a variety of incentives related to costs and utilization. The incentives generated by hospital-insurer contracts are well understood.12 Specifically, fixed-rate (FR; fixed payment per admission type) and per diem (PD; fixed payment per day) contracts incentivize hospitals to reduce treatment costs but shift risk to the hospital. If a patient requires more resources than anticipated, the hospital is responsible for the extra cost. Conversely, discounted charges (DC; payment of a discounted list price for each service billed) incentivize hospitals to use more services, which may be inefficient. However, the connection between contract structure and hospital performance is complex, likely involving many factors (eg, differences in bargaining ability, risk preferences, ex post incentives to control costs) and correlated with hospital market power and other characteristics. Analysis of the relationship between hospital-insurer contracts and hospital performance has been limited, largely due to the confidentiality of contracts.
Our study seeks to address the question: Is there an association between hospital-insurer contract types (ie, FR, PD, DC) and hospital performance? We measure performance using prices, charges, costs, and length of stay (LOS). We first present a novel contract type identification methodology leveraging detailed claims data. We then examine the relationship between contract types and hospital performance using both between-hospital and within-hospital variation, controlling for patient and hospital characteristics including hospital market power. Our study is the first, to our knowledge, to empirically examine the relationship between hospital-insurer contract types and hospital performance. Our findings have 2 policy-relevant contributions. First, although descriptive, the results document the direction and magnitude of associations between contract type and hospital performance, informing how changing contract structures may have additional, potentially unintended effects. Second, our findings contribute to a better understanding of hospital-insurer bargaining strategies. Research and policy have focused on bargaining independent of contract form. Our findings suggest that the contract type may also be an important element of negotiations.
Our analyses employ employer-sponsored insurance (ESI) claims data from the Health Care Cost Institute (HCCI) from 2011 through 2015. HCCI maintains a deidentified, Health Insurance Portability and Accountability Act–compliant data set of commercial health insurer claims from 3 large insurers, which include self-insured and fully insured health plans. The data account for approximately 25% of the US ESI population with enrollees in all 50 states and the District of Columbia. The claims data include age category, sex, and plan benefit type (eg, health maintenance organization, preferred provider organization).
We supplement the HCCI data with a rich set of hospital characteristics from the American Hospital Association Annual Survey, including hospital bed size, ownership type, payer mix, urban/rural location, teaching program size, hospital system affiliation, service offerings, and vertical integration with physicians. We also include measures of operating expenses (ie, costs) from CMS’ Hospital Cost Report Information System. Our analysis sample contains inpatient claims from a balanced panel of 1889 general acute care hospitals. Hospital market power, which has been shown to be a significant driver of prices and may affect other performance measures, is estimated using the hospital-specific Herfindahl-Hirschman Index (HHI).13 We also include the hospital county-level percentage of HCCI coverage of the ESI population as a proxy for insurer market power. Additional data set details are provided in the eAppendix (available at ajmc.com).
Contract Type Categorization
Hospital-insurer contracts for inpatient payments come in myriad forms but are commonly categorized as 1 of 3 types.14 FR contracts reimburse a fixed amount for each admission based on the patient’s diagnosis, level of complications and comorbidities, and general course of treatment. The most common FR payment arrangement is Medicare’s IPPS. PD contracts specify a fixed daily reimbursement based on the type of service, type of treatment, or some combination. Alternatively, DC contracts specify a payment rate for each billable service, usually in the form of a discount on the billed charge.
Contracts can include variations of these 3 basic contract types. For instance, some contracts use FR payments for some service lines and DC for others. Alternatively, a FR or PD contract may include a “stop loss” or outlier clause that shifts some or all of the payment to DC for expenses incurred beyond a negotiated threshold defining outlier cases. There are also FR contracts that “carve out” specific services or items such as drugs or medical devices that are paid on a DC basis.
Analysis of hospital-insurer contracts has been limited because of contracts’ confidentiality. Recently, researchers have developed methodologies to infer contract types using detailed administrative claims data. Two previous studies have also used HCCI data. One used a bunching algorithm to categorize contracts and found that DC contracts are more prevalent among hospitals with more market power.14 The other derived an index of commercial prospective payment and found that shares of prospective payment admissions were positively correlated with a hospital’s Medicare and managed care plan patient volumes.15 A third study used the Colorado All-Payer Claims Database and assigned claims to the contract type that minimizes variation within year–insurer–hospital–diagnosis-related group (DRG) categories.16 They report trends in spending by contract type over time and by service. There have also been a small number of physician-insurer contract studies.17,18
The HCCI data are unique in that they include claim line detail. Every inpatient admission is comprised of 1 or more claim lines, with an individual service (eg, laboratory test) or item (eg, room and board) listed on a line. Each claim line includes the “list price” or billed charge for the service or item. However, the allowed payment (ie, actual amount the insurer paid) depends on the contract. Thus, the occurrence of allowed payments varies by contract type. We confirm this empirically as described below. We also confirmed that claim line payment patterns differ by contract type through conversations with the data scientists and actuaries involved in insurers’ claims data processing. We use the number of claim lines with allowed payments relative to total claim lines and LOS to categorize inpatient admissions by contract type. FR contracts include a single payment for the admission, DC contracts pay for each billed line item, and PD contracts pay based on the LOS.
We categorize admissions with a positive allowed payment on only 1 claim line as a FR contract. We find that the payment line corresponds to the claim line item related to the stay, (eg, room and board, an all-inclusive rate, or an intensive care unit bed) for 79.3% of admissions categorized as FR. Admissions with an allowed payment on the number of claim lines equal to the length of stay are categorized as PD. DC admissions are identified as admissions with a payment on at least 80% of the claim lines. We set the threshold below 100% of claim lines to allow for the possibility that payments for some services or items may be bundled. However, we find that of the DC admissions, the mean (SD) percentage of claim lines with a payment is 98.8% (3.1%).
We categorize 78.4% of the 5,298,699 admissions in our data. Table 1 presents the percentages and volumes of admissions and allowed payments by contract type. The percentage of admissions by contract type is consistent over time, with 53.6% of all admissions categorized as FR, 19.8% as DC, and 5.5% as PD. Total allowed payments are also stable over time, with the overall percentages generally reflecting the annual payments by contract type.
The distribution of admissions by contract type is consistent with that seen in a prior study using HCCI data, which found that approximately 23% of the claims were DC and up to 57% were FR.9 However, our method classifies substantially more claims. For example, using a bunching algorithm similar to the previous method, we were able to classify only 49.9% of the claims in our sample. Additional contract classification validations are presented in the eAppendix.
Hospital Performance Measures
We measure hospitals’ performance using billed charges, allowed payments (ie, prices), costs, and LOS. To assess whether there is a relationship between performance and contract type across hospitals, we severity-adjust each measure at the hospital-year level. Our approach to severity adjustment and use of hospital-level measures are consistent with the hospital competition literature.14,19,20 A description of the severity adjustment calculation is included in the eAppendix. The severity-adjusted measures reflect the average per admission outcome of interest relative to the sample average demographic and comorbidities profile in a year. We use patient-level outcome measures to test for a relationship between performance and contract type within hospitals.
Table 2 (part A and part B) presents the descriptive statistics by quintiles of FR contracts at the patient and hospital levels, respectively. Prices decrease as FR share increases at both levels of aggregation. Patient-level characteristics such as age, plan benefit type, and sex are consistent across quintiles, but admission severity, measured by DRG weight, decreases as the share of FR contracts increases by quintile. A correlation in charges is not apparent from the descriptive statistics. At the hospital level, HHI decreases as the share of FR contracts increases, consistent with previous studies. Most characteristics are generally consistent across quintiles. Additional descriptive statistics are presented in the eAppendix.
We use linear regression to estimate the relationship between contract type and performance measures. In all models, we use the natural logarithm of the dependent variables due to their right-skewed distributions and to allow results to be expressed as percentage changes. We first estimate hospital-year–level regressions that include the shares of admissions with DC, PD, and mixed contract types—FR share is the omitted category. This specification tests for a relationship between contract structure and the performance measure between hospitals. Covariates include hospital characteristics and state and year fixed effects. The dependent variable is the patient severity–adjusted hospital-level performance measure.
We also estimate patient-year–level regressions to test for a relationship between contract structure and performance within hospitals. The dependent variables are unadjusted performance measures. FR is the omitted category again, but this specification includes an indicator variable equal to 1 for DC, PD, or mixed contract types. The model also includes hospital-level controls and year fixed effects, as well as patient-level controls and hospital-DRG fixed effects. The DRG fixed effects control for factors that are constant across admissions for the same DRG within the same hospital. Additional details of the regression specifications, as well as sensitivity analyses employing alternative specifications, are in the eAppendix.
Table 3 presents hospital-level results of interest. Complete results are available in the eAppendix. We find a positive relationship between DC contracts and hospital prices (0.0043; P < .001) and a negative correlation between the PD contract share and prices (–0.0048; P = .020). The mixed share is positively associated with price (0.0068; P < .001) and LOS (0.0011; P = .001), consistent with the mixed category likely including many outlier and carve-out payments for more complicated admissions. PD contract share is negatively correlated with LOS (–0.0030; P < .001), possibly because of the admissions or hospitals where PD contracts are employed. For instance, PD payments are often used for labor and delivery patients, who have shorter LOS than other inpatients. The negative association between PD contract share and LOS may reflect differences in the mix of patients across hospitals.
DC contracts and charges are not significantly related (–0.0002; P = .723), suggesting that hospitals do not necessarily achieve higher prices on DC contracts by simply raising their list prices. Although costs are estimated from charges, the charge-to-cost ratios vary across hospitals. Thus, the relationship between contract types and costs may differ from charges across hospitals. Shares of DC (0.0009; P = .007) and mixed (0.0040; P < .001) contracts are positively associated with costs.
Select patient-level regression results are shown in Table 4, with complete results in the eAppendix. Within hospitals, DC (0.108; P < .001), mixed (0.300; P < .001), and PD (0.138; P < .001) contracts are positively associated with prices after controlling for patient and hospital characteristics, relative to FR contracts. We also find that DC contracts are negatively related to charges (–0.067; P < .001), as with between-hospital variation. However, PD and mixed contracts are both positively related to charges (0.056; P < .001; and 0.244; P < .001, respectively). Only mixed contracts are significantly related to LOS with a positive association (0.220; P < .001). The relationship between contract type and cost is not reported because charge-to-cost ratios are the same for all admissions within a hospital.
We first describe a method of classifying claims by contract type that leverages claim line detail. Our method results in a substantial increase in the overall number of claims classified relative to other approaches (more than 75% of the nearly 5.3 million claims in our data are classified). Moreover, comparison of our approach with a previous approach using the same data source finds the resulting classifications similar for the claims assigned by both methods. The method in this study potentially provides a more efficient technique for assigning admissions to contract types for use in future research.
Using our measure of contract type, we test for relationships between contract type and hospital performance. Our analysis includes numerous controls for patient and hospital characteristics, and our results are robust to different identification strategies: between hospital and within hospital. The results are largely consistent with the different ex post incentives of contract types. As expected, DC and mixed contracts are associated with prices 0.43% and 0.68% higher than FR contracts, respectively. For an average per admission price of $11,913, these are relatively modest differences ($51 and $81, respectively). Although this is a minimal amount within 1 claim, the effect can be substantial over millions of claims. For example, $51 for each of the 19.8% of the DC admissions in our data set equates to $53,434,230. DC and mixed contracts are also associated with higher costs per admission on average than FR contracts (0.09% and 0.40%, respectively). We find no association between higher charges and a larger share of charge-based contracts. Counterintuitively, we find that PD contracts are associated with shorter LOS and no difference in LOS between DC and FR contracts.
The evidence of associations between contract type and hospital performance presents important considerations for payers, regulators, and policy makers seeking ways to lower prices and increase quality, even though our results do not provide causal estimates. For example, policies targeting hospital charges may greatly benefit some patients, such as the uninsured, but these polices may have no effect on commercial prices more generally if DC contract prices are unrelated to list prices. Although we find no relationship between LOS and FR or DC contracts, we do find that FR contracts are associated with lower costs. Future research is needed to understand the implications of this on quality.
Hospital-insurer negotiations have been extensively studied by measuring a hospital’s value to the insurer’s network.19,21-23 However, there are many other aspects of these negotiations that relate to contract structure (eg, differences in bargaining ability, risk preferences, ex post incentives to control costs). Early reports suggest that large employers who renegotiated DC contracts with hospitals to FR contracts will achieve substantial savings over the next few years.8 These renegotiations took place without substantial changes to employers’ or hospitals’ market power. Our results are consistent with these reports, in that we find that prices are higher for DC contracts relative to FR contracts, even after controlling for patient, payer, and hospital characteristics.
Our analysis is an empirical analysis of claims data, which may be subject to recording errors. We cannot directly observe contracts, so our contract assignment methodology may be subject to idiosyncratic error. However, as discussed in the eAppendix, our approach is consistent with previous methods, but with fewer data requirements. Our analyses also represent average effects aggregated over multiple commercial insurers. The estimated average effects may not be representative of an individual insurer or of insurers not included in HCCI.
Additionally, our results describe associations between contract type and performance measures but are not causal estimates. Our regression models include a rich set of covariates to control for hospital and patient characteristics, but they do not rule out the possibility that contract types and performance measures are a consequence of unobserved characteristics that vary over time (eg, management changes, bargaining ability, mix of insurers).
Leveraging detailed claims data and a novel classification approach to classify claims by contract type, we find evidence that FR contracts are associated with lower prices, lower costs, and shorter LOS. DC contracts have higher prices and costs, but we do not find evidence of higher charges. Our study demonstrates that variations in hospital prices, charges, costs, and LOS are associated with the structure of hospitals’ contracts with insurers, even after controlling for differences in hospital characteristics, patient mix, and market characteristics.
The authors thank the Health Care Cost Institute and the insurers Aetna, Humana, and UnitedHealthcare for providing the data used in this study. They also thank Bryan Dowd, Stuart Craig, and seminar participants at the Federal Trade Commission who provided feedback on an early version of this project.
Author Affiliations: Medtronic (EB), Washington, DC; Henry W. Bloch School of Management, University of Missouri–Kansas City (CG), Kansas City, MO; Health Care Cost Institute (KK), Washington, DC.
Source of Funding: None.
Author Disclosures: Mr Kennedy is a researcher at the Health Care Cost Institute, whose data are used in this manuscript; however, the Health Care Cost Institute has no financial interest in this manuscript. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.
Authorship Information: Concept and design (EB, CG, KK); acquisition of data (EB, KK); analysis and interpretation of data (EB, CG, KK); drafting of the manuscript (EB, CG); critical revision of the manuscript for important intellectual content (EB, CG, KK); statistical analysis (EB, CG, KK); administrative, technical, or logistic support (KK); and supervision (EB).
Address Correspondence to: Eric Barrette, PhD, Medtronic, 950 F St NW, Ste 500, Washington, DC 20004. Email: email@example.com.
1. National health expenditure data: historical. CMS. Updated December 16, 2020. Accessed September 26, 2019. https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/NationalHealthExpendData/NationalHealthAccountsHistorical.html
2. Cooper Z, Craig S, Gaynor M, Harish NJ, Krumholz HM, Van Reenen J. Hospital prices grew substantially faster than physician prices for hospital-based care in 2007-14. Health Aff (Millwood). 2019;38(2):184-189. doi:10.1377/hlthaff.2018.05424
3. Abelson R. When hospitals merge to save money, patients often pay more. New York Times. November 14, 2018. Accessed September 26, 2019. https://www.nytimes.com/2018/11/14/health/hospital-mergers-health-care-spending.html
4. Fulton BD. Health care market concentration trends in the United States: evidence and policy responses. Health Aff (Millwood). 2017;36(9):1530-1538. doi:10.1377/hlthaff.2017.0556
5. Gaynor M, Ho K, Town RJ. The industrial organization of health care markets. J Econ Lit. 2015;53(2):235-284. doi:10.1257/jel.53.2.235
6. Beaulieu ND, Dafny LS, Landon BE, Dalton JB, Kuye I, McWilliams JM. Changes in quality of care after hospital mergers and acquisitions. N Engl J Med. 2020;382(1):51-59. doi:10.1056/NEJMsa1901383
7. Gold J. California AG details ‘historic’ settlement agreement in Sutter Health antitrust case. California Healthline. December 20, 2019. Accessed September 8, 2020. https://californiahealthline.org/news/california-ag-details-historic-settlement-agreement-in-sutter-health-antitrust-case/
8. Appleby J. ‘Holy cow’ moment changes how Montana’s state health plan does business. Kaiser Health News. June 20, 2018. Accessed September 26, 2019. https://khn.org/news/holy-cow-moment-changes-how-montanas-state-health-plan-does-business/
9. Appleby J. Health plans for state employees use Medicare’s hammer on hospital bills. NPR. March 20, 2019. Accessed September 26, 2019. https://www.npr.org/sections/health-shots/2019/03/20/704640661/health-plans-for-state-employees-use-medicares-hammer-on-hospital-bills
10. Accountable care organizations (ACOs): general information. CMS. Updated January 14, 2021. Accessed September 24, 2020. https://innovation.cms.gov/innovation-models/aco
11. The Hospital Value-Based Purchasing (VBP) program. CMS. Updated February 18, 2021. Accessed September 24, 2020. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/Value-Based-Programs/HVBP/Hospital-Value-Based-Purchasing
12. Quinn K. The 8 basic payment methods in health care. Ann Intern Med. 2015;163(4):300-306. doi:10.7326/M14-2784
13. Capps C, Dranove D. Hospital consolidation and negotiated PPO prices. Health Aff (Millwood). 2004;23(2):175-181. doi:10.1377/hlthaff.23.2.175
14. Tompkins CP, Altman SH, Eilat E. The precarious pricing system for hospital services. Health Aff (Millwood). 2006;25(1):45-56. doi:10.1377/hlthaff.25.1.45
14. Cooper Z, Craig SV, Gaynor M, Van Reenen J. The price ain’t right? hospital prices and health spending on the privately insured. Q J Econ. 2019;134(1):51-107. doi:10.1093/qje/qjy020
15. Baker L, Bundorf MK, Devlin A, Kessler DP. Why don’t commercial health plans use prospective payment? Am J Health Econ. 2019;5(4):465-480. doi:10.1162/ajhe_a_00127
15. Weber E, Floyd E, Kim Y, White C. Peering behind the veil: trends in types of contracts between private health plans and hospitals. Med Care Res Rev. Published online July 23, 2019. doi:10.1177/1077558719859724
17. Town R, Feldman R, Kralewski J. Market power and contract form: evidence from physician group practices. Int J Health Care Finance Econ. 2011;11(2):115-132. doi:10.1007/s10754-011-9091-9
18. Clemens J, Gottlieb JD, Molnár TL. Do health insurers innovate? evidence from the anatomy of physician payments. J Health Econ. 2017;55:153-167. doi:10.1016/j.jhealeco.2017.07.001
19. Gowrisankaran G, Nevo A, Town R. Mergers when prices are negotiated: evidence from the hospital industry. Am Econ Rev. 2015;105(1):172-203. doi:10.1257/aer.20130223
20. Roberts ET, Chernew ME, McWilliams JM. Market share matters: evidence of insurer and provider bargaining over prices. Health Aff (Millwood). 2017;36(1):141-148. doi:10.1377/hlthaff.2016.0479
21. Capps C, Dranove D, Satterthwaite M. Competition and market power in option demand markets. Rand J Econ. 2003;34(4):737-763.
22. Town R, Vistnes G. Hospital competition in HMO networks. J Health Econ. 2001;20(5):733-752. doi:10.1016/s0167-6296(01)00096-0
23. Ho K, Lee R. Insurer competition in health care markets. Econometrica. 2017;85(2):379-417. doi:10.3982/ECTA13570