Hospital price transparency
Format errors that break parsers
The file must follow the form and manner CMS prescribes: the CSV template layouts or the JSON schema published in the CMS GitHub repository. A file in another format - a PDF, a spreadsheet screenshot, an ad hoc export - does not satisfy the requirement even if every number in it is right.
Inside the allowed formats, the schema is strict. Data elements have defined names, valid values and structure. Free-text elements that carry several distinct concepts must separate them as the documentation describes. Valid-value lists exist for fields like drug unit of measurement, and inventing new ones breaks parsing.
The failure mode that matters most is silent breakage: the file opens in a spreadsheet and looks fine to a human, while a parser rejects it on the second row. A trailing column, a renamed header, an unescaped quote - each is invisible in casual review and fatal to machine ingestion, which is the entire point of the file.
CMS provides an online validator and the schemas are public. Run the validator on every republication, and then go one step further: parse the file with the same kind of tooling consumers of the data use. The validator checks the schema; it does not prove the data is right. That distinction has its own page in this cluster.