Browse > Article

Census Population vs. Registration Population: Which Population Denominator Should be used to Calculate Geographical Mortality  

Hwang, In-A (Department of Preventive Medicine, University of Ulsan College of Medicine)
Yun, Sung-Cheol (Division of Epidemiology and Biostatistics, Clinical Research Center, Asan Medical Center)
Lee, Moo-Song (Department of Preventive Medicine, University of Ulsan College of Medicine)
Lee, Sang-Il (Department of Preventive Medicine, University of Ulsan College of Medicine)
Jo, Min-Woo (Department of Preventive Medicine, University of Ulsan College of Medicine)
Lee, Min-Jung (Department of Preventive Medicine, University of Ulsan College of Medicine)
Khang, Young-Ho (Department of Preventive Medicine, University of Ulsan College of Medicine)
Publication Information
Journal of Preventive Medicine and Public Health / v.38, no.2, 2005 , pp. 147-153 More about this Journal
Abstract
Objectives: Studies on the geographical differences in mortality tend to use a census population, rather than a registration population, as the denominator of mortality rates in South Korea. However, an administratively determined registration population would be the logical denominator, as the geographical areas for death certificates (numerator) have been determined by the administratively registered residence of the deceased, rather than the actual residence at the time of death. The purpose of this study was to examine the differences in the total number of a district population, and the associated district-specific mortality indicators, when two different measures as a population denominator (census and registration) were used. Methods: Population denominators were obtained from census and registration population data, and the numbers of deaths (numerators) were calculated from raw death certificate data. Sex- and 5-year age-specific numbers for the populations and deaths were used to compute sex- and age-standardized mortality rates (by direct standardization methods) and standardized mortality ratios (by indirect standardization methods). Bland-Altman tests were used to compare district populations and district-specific mortality indicators according to the two different population denominators. Results : In 1995, 9 of 232 (3.9%) districts were not included in the 95% confidence interval (CI) of the population differences. A total of 8 (3.4%) among 234 districts had large differences between their census and registration populations in 2000, which exceeded the 95% CI of the population differences. Most districts (13 of 17) exceeding the 95% CI were rural. The results of the sex- and age-standardized mortality rates showed 15 (6.5%) and 16 (6.8%) districts in 1995 and 2000, respectively, were not included in the 95% CI of the differences in their rates. In addition, the differences in the standardized mortality ratios using the two different population denominators were significantly greater among 14 districts in 1995 and 11 districts in 2002 than the 95% CI. Geographical variations in the mortality indicators, using a registration population, were greater than when using a census population. Conclusion: The use of census population denominators may provide biased geographical mortality indicators. The geographical mortality rates when using registration population denominators are logical, but do not necessarily represent the exact mortality rate of a certain district. The removal of districts with large differences between their census and registration populations or associated mortality indicators should be considered to monitor geographical mortality rates in South Korea.
Keywords
Mortality; Geography; Bland-altman test; South Korea;
Citations & Related Records
Times Cited By KSCI : 1  (Citation Analysis)
연도 인용수 순위
1 Park SK, Ha M, Im HJ. Ecological study on residences in the vicinity of AM radio broadcasting towers and cancer death: preliminary observations in Korea. lnt Arch Occup Environ Health 2004; 77: 387-394
2 Chung HK. A Study on the Impact of community-level characteristics on the mortality level [dissertation]. Korea: Seoul National Univ.; 1990 (Korean)
3 Kim J, Sung J. Shin Y. Inequalities of adjusted death rate and years of life lost by region and class in Korea. In: Abstracts of The Korean Society for Preventive Medicine 53th Annual Meeting. 2001. p.257
4 Brand IM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 307-10   PUBMED
5 Son M. The relationship of occupational class. educational level and deprivation with mortality in Korea. Korean J Prev Med 2002; 35(1): 76-82 (Korean)
6 Yoon TH. The relationship between social class distribution and mortality. Korea J Health Policy Admin 2003; 13(4): 99-114   DOI
7 United nations population division. world population prospects: The 2002 revision population database. Available at http://esa.un.org/unpp/.Accessed 18 December 2003
8 Chan YH. Basic statistics for doctors. Biostatistics 104: correlational analysis. Singapore Med J 2003;44(12): 614-619   PUBMED
9 Ro YS. An Analysis of the factors affecting death rate in Korea
10 Jun TY. A Study on the Construction of regional life table in Korea: 1970-1980 [dissertation]. Korea: Seoul National Univ.; 1985 (Korean)