Beginning in 1998, all CoC-accredited cancer programs are required to submit case reports to the NCDB in response to the annual Call for Data. In the NCDB PUF, facilities are assigned a random ID, PUF_FACILITY_ID. This ID is assigned regardless of cancer site, so researchers may identify the same facilities across cancer sites. The number of CoC-accredited cancer programs changes from one diagnosis year to the next. Thus, not all of the hospitals available in the PUF have been accredited for every one of the diagnosis years included in the PUF. If a planned analysis includes the calculation of hospital volume, investigators should recognize that CoC reporting requirements affect the methodological approaches to computing and estimating hospital volume.
Some patients receive treatment in more than one CoC accredited program, and are noted in the PUF Multiple Source item; however only one of the cancer programs where diagnosis/and or treatment was received is included in the PUF, in order to exclude duplicate records for the same patient. Thus calculation of hospital volume will not include hospitals with duplicate records that were excluded from the PUF. Approximately 10% of patients in the PUF data have duplicate records, but this percentage varies by cancer site. In addition, the summary surgical procedure of the primary site variable is based on information from the hospital where the surgery was performed, but this is not necessarily the hospital report that is included in the PUF data. The at this facility surgery variable may be used to account for surgery performed at that particular facility included in the PUF.
Whether an analysis uses total case volume or surgical volume, the easiest way to begin assigning average volume to each hospital appearing in the PUF is to create an aggregated dataset of the number of cases by hospital and diagnosis year. Such an aggregated file can be used to assess a particular cancer program's CoC accreditation history. Hospitals that have remained accredited throughout the years covered by the PUF pose minimal challenges when attributing volume metrics. If there are observed trends (either upward or downward) or spikes in hospital case counts, investigators may deem it more appropriate to calculate an average volume from the most recent years or a select set of years. Where significant shifts in annual caseloads are observed, investigators might consider recalculating their volume metric using a minimum and maximum volume value for each hospital in the aggregated dataset.
Hospitals that have previously discontinued and subsequently re-established their CoC accreditation throughout the span of diagnosis years available in the PUF will display a seemingly inconsistent or incomplete reporting pattern across years. Investigators should be certain to check their aggregated data set to ensure that computed volume metrics appropriately account for these hospitals.
In addition, researchers need to only include surgeries performed at the facility included in the PUF, by using the variable, Surgery at this Facility.