Healthcare-Associated Infections

Healthcare-associated infections (HAIs) are infections patients acquire during the course of receiving treatment for other conditions. They are not why the patient came to the hospital — they are a consequence of being there. CMS requires hospitals to report HAI data to the CDC's National Healthcare Safety Network (NHSN) as a condition of Medicare payment.

Unlike most hospital quality measures, which are derived from Medicare claims and apply only to Medicare patients, HAI measures cover all patients treated in acute care hospitals — adults, children, neonates, Medicare and non-Medicare alike. This makes them one of the broadest safety signals in the CMS dataset.

Results are expressed as a Standardized Infection Ratio (SIR) — a single number that compares the actual infections observed at a hospital to the number that would be predicted given that hospital's patient population, care settings, and other risk factors.

CMS tracks five specific infections — chosen because they are common enough to measure reliably, serious enough to matter clinically, and preventable enough to justify holding hospitals accountable for them.

The SIR is calculated by dividing the number of infections a hospital actually observed by the number of infections that would be predicted for a hospital with its specific patient population, care settings, and risk profile. A score of 1.0 means actual infections matched predictions exactly. Below 1.0 is better than expected. Above 1.0 is worse.

The CDC's prediction model accounts for factors the hospital cannot control — the type of care unit where the infection occurred, hospital size and teaching status, patients' existing infection status on admission, and patient age. This risk adjustment is important: a large academic medical center treating sicker patients is expected to have more infections than a small community hospital, and the SIR reflects that.

HAI data flows from hospitals to the CDC's NHSN system — a separate infrastructure from most CMS quality reporting. The CDC calculates the SIRs; CMS publishes them through the Provider Data Catalog. This two-agency pipeline means the data reflects CDC's surveillance definitions and prediction models, not CMS's own calculation methodology.

Collection windows are 12 months. The data is refreshed quarterly — more frequently than most hospital quality programs, which update only once per year. This means HAI scores can shift noticeably between releases, particularly for smaller hospitals with fewer eligible patients.

The data is published in Healthcare_Associated_Infections–Hospital.csv on the Provider Data Catalog. SIR values, observed counts, predicted counts, and comparison categories are all available in the downloadable file.