Brief Description of the Surveillance, Epidemiology & End Results (SEER)-Medicare Health Outcomes Survey (MHOS)-Level Analysis File
The following is a brief description about how the analysis file was created, who is included in the file, the structure of the file, and the content of the file. For ease of reference, variable names are included in parentheses.
The SEER-MHOS is a survey-level analysis file organized chronologically, based on the earliest survey administration date. The file includes all Medicare Advantage enrollees from Cohorts 1 to 18 that have completed at least one MHOS. Each cohort consists of a baseline survey and a two-year follow-up survey. Beneficiaries who responded to a baseline survey may or may not have completed a follow up survey. Some beneficiaries were sampled in more than one cohort, resulting in multiple baseline and follow-up surveys per person. The survey records are arranged chronologically starting with the earliest completed survey, and include the survey date (SRVDATE).
The survey-level file includes information for people with and without cancer. The variable (INSEER) identifies MHOS respondents who were found in the SEER file and were diagnosed with cancer between 1973 and 2015. Dates of diagnosis and clinical information, for up to 10 cancer sites, from the SEER file are included in each survey record that belongs to SEER-linked respondents. A number of variables were calculated to describe the timing of the survey relative to cancer diagnosis including the patient's cancer status at the time of the survey (CASTAT). These variables can be used to select a subset of respondents or surveys based on pre/post-cancer diagnosis; for example, selecting number of cancers before survey equal to 0 (NUMCABEF = 0) will keep only surveys prior to any cancer diagnosis. For more examples, refer to Sample SAS Code for Patient and Survey Selection (SAS, 5 KB) . The SEER variables were copied from the SEER-Medicare file (PEDSF) and are described briefly in MHOS-PEDSF of the Data Dictionary. For a detailed description of the SEER variables including data dictionary, refer to Attachment A of the Data Dictionary.
The file also contains demographic information obtained from sources including the Enrollment Database (EDB) maintained by the Centers for Medicare & Medicaid Services (CMS), SEER file, and self-reported information. The census tract and zip code around the time of each cancer diagnosis, and zip code at the time of the survey were encrypted and added to the survey record. Separate files that contain geographically-based (ZIP code and census tract level) socioeconomic information from the 1990 and 2000 Censuses and the 2008 – 2012 American Community Survey are provided and can be matched by the encrypted patient census tract and zip code. SF-36 scales, as well as physical and mental summary scores are provided on the T-score metric; for Cohorts 1-6 Baseline and Follow up as well as Cohorts 7-8 Baseline. Physical and mental summary scores are also available for Cohorts 7-18. For more information, refer to How to Use the SF-36 Measures. For 2006 survey years and later, physical and mental summary scores are provided based on the Veterans Rand (VR-12), but the SF-36 scales and T-scores are no longer available. In addition, VR-12 equivalent scores were calculated for Cohorts 1-6 Baseline and Follow up and Cohorts 7-8 Baseline. The eight scales (physical functioning, role limitation (physical), pain, general health, emotional well-being, role limitation (emotional), social functioning, and energy/fatigue) have all been adjusted by survey mode and language.
For each data request, the file will be subset on respondents with the requested cancer site. Respondents may have more than one cancer site, and information for up to 10 sites is stored in an array for each SEER data item. The variable (CAINDX) indicates the number of the array element (1-10) which corresponds to the requested cancer site. The file(s) extracted for each request are sent in SAS Cport format.