Page 83 - IJPS-2-2

Haiyan Zhu, Qiushi Feng, and Danan Gu

was 65 years or older, with a life expectancy of 75.43 (2010–2015); projections to the year 2050 es-
timate that the elderly proportion will increase to 27.6% with a life expectancy of 83.45 (2050–2055).
Chronic diseases are becoming more prevalent among the Chinese elderly population, though the
disability prevalence declined in recent years (Gu, Dupre, Warner et al., 2009; Martin, Feng, Schoeni
et al., 2014). As such, understanding the predictive power of global health measurement such as
SRH and IRH for mortality is particularly important in China. The goal of this paper is to test
whether IRH is an effective complementary measure to SRH to predict mortality across different
subgroups in this important population, specifically gender, age, marital status, rural/urban status,
and socioeconomic status. To ensure statistical power for subgroup analyses, we analyzed a national
survey data with a large sample size.
2. Data and Measures

2.1 Data

The data used for this analysis are from two waves of the Chinese Longitudinal Healthy Longevity
Survey (CLHLS) in 2005 (the fourth wave) and 2008 (the fifth wave). The CLHLS was conducted in
1998 as a baseline for a longitudinal project on health and longevity among the oldest-old population
aged 80+ in China. A multistage, stratified cluster survey design was conducted in 631 randomly
selected counties and cities in 22 out of 31 provinces. In the year 2002 (the third wave), young-old
respondents aged 65–79 were added to the survey sample. The CLHLS attempted to interview
all centenarians in the selected counties/cities. In order to ensure comparable numbers of octogenar-
ians and nonagenarians at each age from 80 to 99, for each centenarian interviewed, one nearby oc-
togenarian and one nearby nonagenarian with pre-designated age and sex were randomly chosen and
interviewed based on a random code assigned to the centenarian. Informed consent was obtained
from each of the respondents. All information was obtained through in-home interviews with profes-
sionally trained and well-educated interviewers. A couple of studies have evaluated the CLHLS data
quality as high, including the accuracy of age-reporting and the validity, reliability, and consistency
of various measures (Gu, 2008; Zeng and Gu, 2008).
In the longitudinal dataset, the 2005 wave interviewed 15,638 individuals aged 65+, with 5,047
young-old respondents aged 65–79 and 10,591 oldest-old respondents aged 80+. Among the re-
spondents aged 80+, there were 3,870 octogenarians, 3,927 nonagenarians, and 2,794 centenarians.
Out of these 15,638 respondents, 5,111 older adults died before the 2008 survey, accounting for
about 33% of the 2005 sample. About 19.5% of the 2005 sample, or 3,055 respondents, were lost to
follow-up and 7,472 respondents (47.8%) were re-interviewed in 2008. Those lost to follow-up were
excluded from analysis because we did not have the information on their survival status and
health conditions in 2008. Therefore, we analyze data from 12,583 respondents who were inter-
viewed in 2005 and had known survival status in 2008.

2.2 Measures
2.2.1 Mortality

The dependent variable is mortality due to all causes, which was estimated by the length of exposure
and survival status during the survey interval from 2005 to 2008. The mortality exposure was meas-
ured in number of days from the interview date in 2005 to either the date of interview in 2008 or the
date of death. For those who died before the 2008 interview, the date of death was collected from
officially issued death certificates whenever available; next-of-kin and local Residential Committees
were consulted in the cases when death certificates were not available (Zhu and Gu, 2010).
2.2.2 Self-Rated Health (SRH)
SRH was measured by the question “How do you rate your overall health?” with five response cate-

International Journal of Population Studies | 2016, Volume 2, Issue 2 77