Why Not the Best (WNTB) http://www.whynotthebest.org
A Commonwealth Funded health care reporting tool, web-based public comparative health care performance report card. Data come from a range of sources: CMS Hospital Compare, State Hospital Discharge data are the largest data sources. Reported measures come for nationally accepted and defined quality indicators aggregated to an entity. Hospitals are the main type of entity, but there are also state, nation, hospital type, HRR, hospital systems, counties entities.
Data for WNTB are stored in the IPRO eServices mysql database dubbed Pellucid. Below is an overview of how the CMS Hospital Compare is prepared for WNTB. Of course, the devil is in the details. Some details can be found at the Pellucid Wiki.
The CMS Hospital Compare data is downloaded as a MS Access database four times a year from the CMS website (shortly after CMS updated it). The data already has defined individual measures (like aspirin on discharge for AMI patients) by hospital, state, and nation. However it does not include measure composites or a range of benchmarks. The measures mostly consist of rates. IPRO eServices extracts the relevant measures for WNTB and creates the composites, benchmarks, roll ups to the range of entities, maps the measures and entities to Pellucid, and formats a “values” table to be loaded to Pellucid.
The screenshot of the website‘s Compare Hospitals tab displays a summary of the measures populated for a report. The measures (columns) and values (cells) are displayed by a range of entities (rows) that the user selects when building a report.
There are several types of reports and graphs than can be constructed. This next image shows you what options are available for building a report in the Compare Hospital tab.
The user can: 1) select the up to 30 hospitals from anywhere in the US then 2) select benchmark(s).
WNTB automatically creates the report. Once the report is created it will display for those entities and the user can select the measure by clicking the drop down button. Some measures are composite measures, where the individual item measures can also be viewed.
All of the data used to populate WhyNotTheBest.org is stored in Pellucid tables. Entity_values is the table that stores the rows of data values by measure_id , entity_id, and start and end dates. The screenshot of the mysql Pellucid shows how the data is stored. Each value is its own row.
Composites and Benchmarks. In addition to reporting performance for each CMS process-of-care measure, IPRO creates and reports on WNTB summary performance scores for each hospital and a range of benchmarks. This work is programmed and completed using in SAS. It involves aggregating data to various levels, ranking, creating percentiles and percentages, and conducting statistical comparisons from entity to benchmark. The measures are also rolled up to define performance for a given geography (zip, county, region, HRR), a hospital type (teaching, rural, government, etc), or hospital system.
Footnotes. Not all data is displayed on WNTB. For example, values are not displayed for hospitals with too cases in the denominator of that measure. This data still gets loaded into Pellucid, but does not get displayed on WNTB. For the composites and benchmarks only data for hospital with sufficient denominator sizes 30 get included in these created values. This is an example of a footnote. Details can be found at the Pellucid Wiki. IPRO uses some of these footnotes and adds others. For each entity_values data load, there must be a data load for the footnote_values table. For each footnote, there is one row of data that contains the measure_id, entity_id and start and end date. These fields are used to link to the entity_values table.
This is only an example of one data source, WNTB does contain many more measures derived from other data sources (not described here).