of Work: The purpose of this agreement was to support the collection and analysis of data on cause of death and characteristics of the last year of life in the planning of the 1992 Pretest and 1993 Main Survey of the 1993 National Mortality Followback Survey (NMFS), conducted by NCHS, CDC. This survey will supplement information from death certificates in the vital statistics file with information on characteristics of the decedent. The pretest examined approximately 800 deaths of individuals aged 15 years and over who died in 1992. The main survey examined approximately 22,951 deaths of individuals aged 15 years and over who died in 1993. This includes 884 deaths to centenarians. A preliminary death certificate/informant (DC/I) file delivered to the NIA/EDB by the NCHS in 1997 has been extensively examined by EDB staff and the SYTEL computing support contractor. This file contains information keyed from the death certificate file and the informant interviews. Data from the final 1993 mortality file has been merged with the DC/I file and stripped of personal identifiers to form a working file. Post-stratification adjustment factors and non-response adjustments have been developed to form the final sample weights for each file record. A number of anomalies were discovered and reported to the NCHS for production of the final analysis file. A second provisional data file was received from NCHS int he spring of 1998 containing numerous additions to the file and corrections to many of the anomalies that were pointed out earlier. The new file, considered to be a provisional public use data file, has been examined by the EDB data management staff. While some additional problems remain with the revised data set, work has begun on constructing an extract of the file for preliminary analysis work. NCHS has completed linkage operations with the Social Security Administration (SSA) and the Health Care Financing Administration (HCFA) to provide data for the sample of decedents from the administrative records from those two agencies. The SSA has been able to verify (or correct) the social security number for about 98 percent of the decedents. A subset of records from the DC/I file was generated to independently link with SSA and HCFA data. The subset consisted of all DC/I records for which the informant gave permission for a link to be made. At the Kentucky conference on Statistical Methods in Alzheimer's Disease Research, EDB staff held discussions with a neurologist who had completed an analysis of data from the 1986 NMFS on dementia in the last year of life. It is now planned to examine the 1993 data in the same way to explore dementia trends over the 7-year period between the two surveys. Other analytic plans will look at trends in reported lifetime history of other conditions between the 1986 and 1993 NMFS, examination of lifestyle factors related to death at very old ages, hypotheses regarding decreased use of health services in the last years of life among the very old, and possible studies of compression ofmorbidity among the oldest old.
| Foley, Daniel J; Brock, Dwight B; Lanska, Douglas J (2003) Trends in dementia mortality from two National Mortality Followback Surveys. Neurology 60:709-11 |
| Bedford, S; Melzer, D; Guralnik, J (2001) Problem behavior in the last year of life: prevalence, risks, and care receipt in older Americans. J Am Geriatr Soc 49:590-5 |
