This analysis uses a sample of claims obtained from IBM Health Analytics MarketScan Commercial Claims and Encounters Database (IBM Corporation), as well as Medicare payment data obtained from the Medicare Provider Payment and Utilization Data public use files (Centers for Medicare and Medicaid Services). MarketScan is a convenience sample of health care claims provided primarily by large employers and health plans. We used a subset of MarketScan claims provided by large employers, as well as Medicare-reported provider-level payment reports, from 2014 through 2017 to conduct this analysis. The specific Diagnosis-Related Groups (DRGs) examined were selected either due to their relevance to COVID-19-related hospitalizations (DRG 177 – respiratory infections and inflammations with major comorbidities or complications, DRG 207 – respiratory system diagnoses with ventilator support for over 96 hours, and DRG 208 – respiratory system diagnoses with ventilator support for up to 96 hours), or due to their comparably large volume of hospital admissions in both datasets. The latter group includes DRG 470 (major hip and knee replacement without major complications or comorbidities), DRG 247 (percutaneous cardiovascular procedures with drug-eluting stent without major complications or comorbidities), DRG 330 (major procedures of the small and large bowel with complications or comorbidities), DRG 621 (operating room procedures for obesity without complications or comorbidities), DRG 743 (uterine and adnexa procedures for non-malignancy without complications or comorbidities), DRG 392 (esophagitis, gastroenteritis, and miscellaneous disorders without major complications or comorbidities), and DRG 603 (cellulitis without major comorbidities or complications).

To calculate private insurance payment rates, we analyzed a sample of medical claims obtained from the 2017 IBM Health Analytics MarketScan Commercial Claims and Encounters Database. We only included claims for people under the age of 65. This analysis used claims for almost 19 million people representing about 22% of the 86 million people in the large group market in 2017. Weights were applied to match counts in the Current Population Survey for enrollees at firms of a thousand or more workers by sex, age, and state. Weights were trimmed at eight times the interquartile range.  Averages represent the charges paid to the hospitals for an admission.  Across all the DRGs, hospital spending represented about 86% of the total cost of the admission.  Costs include both amounts paid by enrollees in the form of cost-sharing and spending by the plan.  Hospital costs are trimmed to exclude the lowest 1.5% and highest 0.5% of hospital costs within DRG. These data reflect cost sharing incurred under the benefit plan, but do not include balance-billing payments that beneficiaries may make to health care providers for out-of-network services or out-of-pocket payments for non-covered services.

To calculate Medicare payment rates, we analyzed average provider-level total payments (Medicare payments plus enrollee cost-sharing) to hospitals for admissions identified through the diagnostic-related group (DRG) assigned to them in the database, as reported in the Medicare Provider Payment and Utilization Data public use files. These files are prepared by the Centers for Medicare and Medicaid Services (CMS) using Medicare Provider Analysis and Review (MEDPAR) data, employing trimming and weighting methods developed by CMS (see for further information). Provider-level average payments were further weighted by provider-level case volume. This analysis used Medicare payment data representing hospital-specific payments for over 3,000 Medicare-qualified hospitals, encompassing more than 7 million enrollee discharges, or approximately 75% of total Medicare IPPS discharges in 2017. These data reflect Medicare-allowed cost sharing, but do not include balance-billing payments or other payments made outside of the Medicare claims process.

In Medicare, hospital admissions are reimbursed based on DRGs that reflects a patient’s clinical conditions and treatment. In contrast, private insurers pay for hospital admissions using different approaches that may vary with the procedures performed during the stay, including per diem payments, discounted fee-for-service payments, DRGs or other combinations of payments and performance incentives. IBM assigns a DRG to each admission using the Centers for Medicare & Medicaid Services (CMS) Grouper 35. This method selects a DRG for the admission based on the diagnosis and procedures a patient received during the case. The total payments to hospitals in the Marketscan data reflect the payments actually made to the hospitals. Some variation in the payment rates of admissions is accounted for by differences in the intensity or types of services that a patient receives, and not differences in the rates paid for those services. The rates Medicare uses to reimburse DRGs are designed to account for this variation in the intensity of cases and services. This analysis compares the average of DRG payment rates in Medicare to the average amounts paid for admissions by insurers and enrollees in Marketscan.