National Cancer Institute Home at the National Institutes of Health |

Control Characters in the SEER-Medicare Claims Files

It has been determined that some control characters in the SEER-Medicare claims files may cause a problem reading those files with SAS in the Windows operating system.

The Problem

  1. The issue is that control characters in the data files may be incorrectly interpreted as either end of line characters or end of file characters when read in the Windows operating system.
  2. This is NOT a problem when SAS is used in the Unix or Linux operating systems.

How to Tell If You're Affected

  1. The SAS log lists what "records were read" for each input file. These counts must match the number listed in the spreadsheet provided with each request.
  2. In the SAS log the "minimum record length" and "maximum record length" should be the same for each file. The records we provide are all written to a fixed record length. The "lrecl" can be found on the INFILE statement of the Input Statement documentation file.

The Solution

  1. The input statements posted to the SEER-Medicare Web site for the Hospice, HHA and NCH files have been modified to not include "recfm=f" on the infile statement. This extra option was causing some Windows SAS programs to read the files incorrectly.
  2. Add the option "IGNOREDOSEOF" to the INFILE statement when the program is run in the Windows operating system. This option should not be used in the Unix or Linux operating systems as it is invalid.

Last Modified: 18 Oct 2013