Hospital-Acquired Condition (HAC) Reduction Program

Overview

The Hospital-Acquired Condition (HAC) Reduction Program dataset is published by CMS under Section 1886(p) of the Social Security Act. It contains one row per participating hospital with a Total HAC Score, a binary payment reduction flag, and individual measure scores across two domains: the CMS Patient Safety Indicator composite (PSI-90, claims-based) and six CDC National Healthcare Safety Network (NHSN) Healthcare-Associated Infection (HAI) measures (CLABSI, CAUTI, SSI colon, SSI hysterectomy, MRSA bacteremia, and C. difficile infection). Each measure includes a Standardized Infection Ratio (SIR) and a Weighted Z-Score that contributes to the composite. The current file covers FY2026, with the PSI-90 and HAI measurement periods spanning approximately 2-3 years ending in 2022-2023. CMS sources the data from the Provider Data API on data.cms.gov (dataset ID yq43-i98g).

Hospitals scoring in the worst-performing quartile (bottom 25%) on the Total HAC Score receive a 1% reduction to all Medicare IPPS payments for the fiscal year. The penalty is binary — there is no graduated scale. A hospital at the 24th percentile receives the same 1% reduction as one at the 1st percentile. This dataset answers questions such as: which hospitals are penalized under the HAC Reduction Program, what is a hospital's composite patient safety score relative to peers, and which specific measures (infections or patient safety indicators) drive a hospital's Total HAC Score.

Join Strategy

Each row joins to a hospital entity page using the Facility ID field, which contains the CMS Certification Number (CCN) as a 6-digit zero-padded string. During ETL, the _find_column() function matches the CCN column against a candidate list (facility_id, Facility ID, Facility Id, Hospital CCN, CCN, Provider Number) to handle header variation across CMS file releases. The normalize_ccn() function strips whitespace and zero-pads values shorter than 6 characters. The join is one-to-one: each CCN maps to exactly one row. Matched data is written to the hospital's JSON manifest under data.hac_reduction with fields payment_reduction (boolean), total_hac_score (float), and a measures object containing SIR and Weighted Z-Score values for each of the six individual measures (psi_90, clabsi, cauti, ssi, cdi, mrsa). Measurement period start and end dates for the PSI-90 and HAI domains are stored under measurement_period. A provenance record with dataset ID hac-reduction and vintage FY2026 is appended to the manifest. Hospitals without a matching HAC row display missing data indicators rather than being excluded from CareGraph.

Known Limitations

• IPPS hospitals only. The HAC Reduction Program applies only to hospitals paid under the Inpatient Prospective Payment System. Critical Access Hospitals, hospitals with fewer than 200 discharges during the performance period, Maryland waiver hospitals, and specialty hospitals paid under non-IPPS systems (psychiatric, rehabilitation, children's, long-term care, cancer) are excluded entirely. These hospitals have no row in the dataset, so their absence is indistinguishable from a data gap without cross-referencing CMS's exclusion lists.
• Medicare fee-for-service only. The HAI measures are based on CDC NHSN surveillance data (all payers), but the PSI-90 component uses only Medicare FFS claims. Medicare Advantage enrollees, Medicaid-only patients, and commercially insured patients are excluded from the claims-based measure. In markets with high MA penetration, PSI-90 is calculated on a smaller, potentially non-representative subset of discharges.
• Disproportionate impact on safety-net hospitals. The risk adjustment for both the PSI-90 and HAI measures does not fully account for social determinants of health. Safety-net hospitals and facilities serving high-acuity, socioeconomically disadvantaged populations face systematically higher HAC scores. Peer-reviewed research has documented that the bottom quartile penalty disproportionately affects these facilities.
• Overlapping measures across CMS programs. Several measures in the HAC Reduction Program overlap with those used in the Hospital Value-Based Purchasing (HVBP) safety domain and the Hospital Compare HAI measures. The same hospital can be penalized under the HAC program while simultaneously receiving a reward or penalty for overlapping indicators under HVBP, making cross-program comparisons difficult without understanding each program's scoring methodology.
• Domain weighting changes across fiscal years. CMS has changed the relative weighting of the PSI-90 domain versus the HAI domain between fiscal years. Historical Total HAC Scores are not directly comparable across years without adjusting for these weighting changes.
• Measurement lag. The FY2026 scores reflect care delivered approximately 2-3 years prior. Recent improvements or declines in infection prevention or patient safety are not captured in the current data.

Data Quality Notes

• Score fields stored as strings with sentinel values. The source data encodes Total HAC Score, all SIR columns, and all Weighted Z-Score columns as strings. Suppressed or inapplicable values appear as "Not Available" or empty strings rather than numeric zeros. The ETL's _try_float() function converts non-numeric entries to null in the JSON manifest. A null total_hac_score means the hospital lacked sufficient data for a composite score, not that it scored zero.
• Payment reduction flag encoded inconsistently. The source field for payment reduction uses varying string representations (Y, Yes, 1, true) across file vintages. The ETL normalizes these to a boolean payment_reduction field — true means the hospital is in the penalized bottom quartile, false means it is not. Hospitals excluded from the program entirely have no row and therefore no payment_reduction value.
• Individual measure null rates vary by measure. The SSI measures (colon and hysterectomy, combined as ssi) have higher null rates than CLABSI, CAUTI, MRSA, or CDI because fewer hospitals perform qualifying surgical volumes. PSI-90 has a lower null rate since it is claims-based and requires only a minimum discharge count. When analyzing measure-level data, the effective denominator differs by measure.
• Column name variation across vintages. CMS has used different header names and casing between file releases (e.g., Total HAC Score vs. Total_HAC_Score vs. Total Score, CLABSI SIR vs. CLABSI_SIR). The ETL's _find_column() function resolves this via candidate-list matching, but new header names in future releases may require adding candidates.

---