Hospital Provider Cost Report

Overview

The Hospital Provider Cost Report dataset contains annual financial statements filed by Medicare-certified hospitals through the Healthcare Cost Report Information System (HCRIS). Each cost report is submitted on CMS Form 2552-10 and covers a hospital-defined fiscal year, reporting approximately 300+ financial line items across standardized worksheets (S-2, S-3, G-3, S-10, and others). The data includes revenue, expenses, utilization statistics, balance sheet items, and uncompensated care figures. HCRIS is maintained by the Centers for Medicare & Medicaid Services (CMS) and updated quarterly as hospitals submit or amend their filings.

Cost reports answer questions about hospital financial performance, operating margins, cost per discharge, payer mix, occupancy rates, uncompensated care burden, and departmental cost allocation. Because they capture the full institutional cost structure — not just Medicare claims — they are the primary public source for understanding hospital finances. Researchers, journalists, and policymakers use HCRIS data to compare hospital efficiency, assess financial distress, and evaluate the impact of payment policy changes. CareGraph derives key financial metrics from raw worksheet cells and presents them alongside quality and utilization data on hospital entity pages.

Join Strategy

Each cost report record is keyed by the Provider CCN (CMS Certification Number), a 6-character zero-padded string that uniquely identifies a Medicare-certified facility. CareGraph normalizes the CCN to a consistent 6-digit zero-padded format during ETL, stripping any leading/trailing whitespace and left-padding with zeros where the source data stores it as an integer. The CCN joins directly to hospital entity pages at /hospital/{ccn}. When a hospital files multiple cost reports in a single release (e.g., an initial and an amended report for the same fiscal year), CareGraph retains the most recent filing based on the fiscal year end date and report status indicator, preferring settled reports over as-submitted reports. Hospital-based subprovider units (SNFs, home health agencies) that share the parent hospital's CCN are not separated into distinct entity pages; their cost data remains embedded in the parent hospital's cost report and is presented at the institutional level.

Known Limitations

• Fiscal year misalignment. Hospital fiscal years are institution-defined and do not necessarily align with the calendar year or federal fiscal year (October–September). Comparing two hospitals labeled as the "same year" can mix time periods by up to 11 months. CareGraph displays the fiscal year end date for each cost report to make this transparent, but cross-hospital comparisons on a calendar-year basis remain imprecise.
• Self-reported, partially audited data. Cost reports are self-reported financial statements. CMS audits only a fraction of filings in any given year through its Medicare Administrative Contractors (MACs). Unaudited reports may contain errors, inconsistent cost allocations, or strategic classifications that overstate or understate specific line items.
• Late and amended filings. Approximately 5–10% of hospitals file late or submit amended cost reports after the initial deadline. The HCRIS quarterly release may not reflect the most recent amendment for every hospital, creating a lag between the filed data and the published snapshot.
• Critical access hospital form differences. Critical access hospitals (CAHs) file on CMS Form 2552-96, which has a different worksheet structure than the standard 2552-10. Field mappings between the two forms are not one-to-one, and some derived metrics cannot be computed identically across both populations. CareGraph flags CAH cost reports where worksheet cell references differ.
• Hospital-based subproviders. Hospital-based SNF, home health, and psychiatric units report on the same cost report as the parent hospital. Isolating department-level costs from institutional overhead requires extraction from supplemental worksheets, and these allocations follow hospital-chosen cost allocation methods (step-down or reciprocal), which vary across institutions.
• Uncompensated care reporting variability. Worksheet S-10 (uncompensated care) data has been under increased CMS scrutiny since FY2014 for its role in Medicare DSH payment calculations, but reporting quality and definitional consistency still vary across hospitals, particularly for charity care vs. bad debt classification.

Data Quality Notes

• Derived metrics require cross-worksheet calculation. Key financial indicators — operating margin, cost per discharge, occupancy rate — are not present as fields in the raw data. CareGraph computes them from specific worksheet cells: operating margin from Worksheet G-3 (net revenue and total operating expenses), cost per discharge from Worksheet S-3 (total costs) divided by Worksheet S-3 Part I (discharge counts), and occupancy rate from Worksheet S-3 Part I (inpatient days divided by available bed days). Cell references are version-specific to Form 2552-10.
• Numeric fields stored as strings. Several financial line items in the HCRIS flat files are encoded as character fields with embedded commas, parentheses (for negative values), or trailing spaces. ETL normalizes these to numeric types, converting parenthesized values to negative numbers and stripping non-numeric characters.
• Null and zero ambiguity. Many financial line items contain null values that may indicate either a true zero (the hospital had no activity in that category) or a non-response. CareGraph does not impute missing values; nulls are preserved and distinguished from explicit zeros in the source data. Fields with particularly high null rates include Worksheet S-10 columns for hospitals that filed before FY2014 reporting requirements took effect.
• Duplicate report handling. A single hospital may have multiple cost report records for overlapping fiscal periods due to amended filings, short-period reports (e.g., change of ownership), or corrections. ETL deduplicates by retaining the record with the latest RPT_STUS_CD (report status code), preferring status values indicating a settled or audited report over as-submitted drafts.