National Long-Term Care Survey

Overview of NLTCS Data

Survey Data

The National Long Term Care Survey produces very large data files. The NLTCS is composed of six survey waves. The year of each survey, the number of variables for each wave and the number of records for each wave follows:

1982 1984 1989 1994 1999 2004*
VARIABLES 2,265 2,425 2,781 2,490 4,336 4,986
RECORDS 20,485 25,401 17,565 19,171 19,907 20,474
*These numbers refer to the 2004 Beta-2 data set; they will be updated with subsequent releases.

Variables are described in the context of each wave's questionnaires.  A crosswalk equating each variable across all waves is available from Unicon.

Survey data files are supplemented by an Analytic File. The Analytic File contains 49,242 observations on 424 variables covering all NLTCS participants from 1982 through 2004. The variables were created through analysis of and imputation on the raw data from the five waves of the NLTCS. The variables are described in the Analytic Data Dictionary. Many of the variables were generated for specific papers published by CDS and are limited by the assumptions made for the analysis presented in those publications (e.g., SY89_spec_phone_telephoning - 1989-special telephone (telephoning)). More useful are variables like SEX and RACE, but imputation was employed for some of the missing values.

The current release of the NLTCS Survey Data is 2004 Beta-2 released on November 30, 2006.

Linked Files

Every sample person in the core survey is given an anonymous identifier chosen from the set of positive integers in such a way that it contains no information.  There are several ancillary studies (e.g. caregiver, next-of-kin, and physical specimen) that collect information about the sample person. For a more complete list of these studies, and more detail about them, see Data Collection.  These studies use the same anonymous identifier as the core survey and are therefore linked to it.  In addition to the linked ancillary studies, one of the more powerful features of the NLTCS is that the complete Medicare billing history for all persons in the sample is available.  These billing history records are matched to the core survey records and assigned the same anonymous identifier used in the core survey.  A linked file consisting of the anonymous identifier from the core survey and the Medicare billing history is prepared by CDS and provided to CMS.  Individuals requiring the linked file can obtain the Medicare billing history directly from CMS. Under current rules, it is not available from CDS.  The user must then merge the billing history file with the NLTCS data acquired from CDS.  The advantage of the Medicare billing data is that information on medical conditions is available even on persons who were not given either of the detail survey instruments.  One use of such data is to obtain more precise date-of-onset of medical conditions than can be obtained from the every-five-years wave structure of the NLTCS.  It also makes possible studies of for example, medical expenditures before and after the onset of a major medical event.

Physical Specimen Data

Some of the ancillary surveys collect blood samples or buccal washes from willing sample persons.  These samples are assayed to determine expression levels of certain genes of interest (e.g. APOE and SOD2).  Protein panels are obtained from the blood samples.  Because of the potential adverse consequences to the sample person the possible misuse of genetic data, such misuse must be guarded against carefully.  Procedures for releasing such data are still being formulated and will be announced on the NLTCS Website soon.