Original Article |
2014, Vol.36, No.2, pp. 249-254
Estimation of mortality with missing data using logistic regression
Amornrat Chutinantakul, Marzukee Mayeng, and Phattrawan Tongkumchum
pp. 249 - 254
Abstract
The present study aims to improve estimation of mortality data with unknown demographic factors using logistic regression, based on inflation factors for distributing such cases. The method is illustrated on death proportions based on numbers of deaths reported in 1996-2009 classified by gender, age and province in the death registration (DR) system. The results indicate that cases with unknown province mostly occurred in ages 0-4 years and 15-44 years and cases with unknown age mostly occurred in the central and southern regions. The method is straightforward, provides confidence intervals, and can be generally used for eliminating biases due to cases with unknown values of demographic factors in DR data. The resulting estimated numbers of deaths were used to examine age-specific mortality using cubic spline function. Cubic spline interpolation is effective to smoothly interpolate non-negative mortality data. These methods provide valuable information to the Ministry of Public Health.