Summary of Research Findings

We can not present here the results of all of our research team’s data analyses due to space limitation. Instead, we briefly summarize the major new findings by the core members.

Childhood SES is associated with health and mortality at the oldest-old ages. Based on the CLHLS longitudinal datasets of 1998-2002, our analyses show that receiving adequate medical service during sickness or not (or rarely) suffering from serious sickness in childhood significantly reduces the risk of being ADL impaired, cognitively impaired, and in self-reported poor health by 18 to 33% at the oldest-old ages. Estimates of effects for five other indicators of childhood conditions are similarly positive, but mostly not statistically significant. Multivariate survival analysis shows that better childhood socioeconomic conditions in general tend to reduce the four-year period mortality risk among the oldest-old; but after additionally controlling for fourteen covariates, the effects are not statistically significant. We argue that policies enhancing childhood health care and well-being can have large and long-lasting benefits up to the oldest-old ages (Zeng, Gu, and Land 2007).

Association of late childbearing (early life conditions) and healthy longevity at oldest-old ages. The Cox hazards model and logistic regression analysis based on our large and unique longitudinal data set demonstrate that late childbearing after age 35 or 40 is significantly associated with survival and healthy survival in the two-year follow-up period among oldest-old Chinese women and men. The association is stronger in oldest-old women than in men. All of these statistical model estimates are adjusted for a variety of confounding factors of demographic characteristics, family support, social connections, health practice, and health conditions. The analysis based on the Fixed Attribute Dynamics method (initially proposed by Vaupel, 1992, as discussed in our previous application, and extended by Zeng and Vaupel, 2003a) and our survey data has shown that late childbearing is positively associated with long-term survival and healthy survival from ages 80-85 to 90-95, and 100-105. This association exists among oldest-old women and men, but again, the effects are substantially stronger in women than in men. We also discussed four factors that may explain why late childbearing affects healthy longevity at advanced ages: (1) social factors, (2) biological changes in women caused by late pregnancy and delivery, (3) genetic and/or other biological characteristics, (4) selection (Zeng and Vaupel, 2003a). We suspect that women who had 2 or 3+ births after age 35 or 40 and survived in good health up to ages 80-105 may have specific gene(s) or other biological characteristics in favor of healthy longevity. These women with rich follow-up data and DNA samples may need to be targeted in our future study to integrate non-genetic and genetic factors for discovering genotypes associated with healthy longevity.

Analysis of the cohort data from CLHLS and the censuses indicates that, as compared to no schooling, having at least one-year schooling, which was an indicator of childhood socioeconomic conditions in old China, dramatically increases the probability of survival and healthy survival from ages 84-89 to 92-97, and 100-105 (see Figures 1a and 1b) (Zeng, Gu, and Land 2005, paper presented at the XXV IUSSP conference, Paris, July, 18-23). It is clear that policies promoting childhood health care and education can have large and long-lasting benefits up to the oldest-old ages.



Different profiles of institutionalized elders in China as compared to their West counterparts. Using three waves of the CLHLS datasets collected in 1998, 2000, and 2002, we find that unlike in Western societies where institutionalization is mainly due to poor health, the primary reason for the oldest-old to be institutionalized in China is a lack of caregiving resources. Although the institutionalized elders tend to exhibit slightly poorer health and higher mortality, the differences are largely reduced or eliminated once the sociodemographic conditions and family care for the oldest-old are taken into account (Gu, Dupre, and Liu 2007).

Sociodemographic factors still play some role in determining dynamics of ADL disability at oldest old ages. Based on the CLHLS multi-waves dataset, our results show that each of sociodemographic factors (age, gender, urban/rural residence, ethnicity, education, primary lifetime occupation, primary source of daily expenses, living alone, and marital status) still play some role in the dynamics of both overall ADL functioning and each of the tasks (Gu and Zeng 2004; Gu and Xu 2007).

Oldest-old mortality trajectory. This project investigated for the first time the mortality trajectories at oldest-old ages 80-105 in a developing country. Zeng and Vaupel (2003) found that a two-parameter logistic model fits Han Chinese death rates at oldest-old ages better than the Gompertz and four other models. Chinese death rates appear to be roughly similar to Swedish and Japanese rates after age 97 for both males and females.  We think that this convergence may be due mainly to mortality selection. We show that in China, as is similar in the U.S. and European countries, the rate of increase in mortality with age decelerates at very old ages. The CLHLS data show that the risk of dying among disabled oldest-old men and women is about twice as high as that of the active oldest-old. The logistic model also fits the Chinese age-sex and ADL status-specific mortality rates at oldest-old ages better than Gompertz and other models (Zeng, Gu, and Land, 2003).

Active life expectancy. The active/disabled life expectancy analysis based on our CLHLS data have shown that the oldest-old living in rural China are more likely to be active in daily living throughout their remaining life than are their urban counterparts. Poor facilities to assist the oldest-old in their daily life in rural areas as compared to urban areas may force rural oldest-old to perform daily activities by themselves; this frequent physical activity may enable them to maintain their capacities for daily life longer than their urban counterparts. A higher percentage of oldest-old men remain active than do women, although women have a longer total life expectancy (Zeng and Vaupel et al., 2001). Our CLHLS data analysis has shown that a man who was active at ages 80, 90 and 100, could expect to spend about 90, 80, and 78 percent of his remaining life in active status, in contrast to 67, 31, and 26 percent for a man who was disabled at ages 80, 90, and 100 (Zeng, Gu, and Land, 2004).

The unique CLHLS data on ADL before dying led to methodological innovation. Based on the unique data on health status before dying collected in our CLHLS and our methodologically innovative investigation, we found that the conventional approach without including data on ADL before dying caused biases in estimating the ADL-status-based active and disabled life expectancies (see Figure 2).  This bias is due to the unreasonable assumption of no changes in ADL status from age x to death if a person dies in the age interval (x, x+n); the biases are mostly sizable and statistically significant. We extended the multi-state life table method for computing active life expectancy with consideration of ADL status before dying and applied it to the new Chinese data to improve the ADL-status-based estimates of active and disabled life expectancies (Zeng, Gu, and Land, 2004).


Oldest-old Chinese women are seriously disadvantaged not only in socioeconomics  but also in health status. A striking finding is that female oldest-old in China are not only seriously disadvantaged in socioeconomic status, but are also in poorer health as measured by activities of daily living, physical performance, cognitive function, and self-reported health, as compared with their male counterparts; these gender differences are more marked with advancing age (Zeng and Vaupel et al., 2002; Zeng, Liu and George, 2003). The large gender differentials among the Chinese oldest-old deserve serious attention.

Extent of disability & suffering before dying at oldest-old ages. Based on 35 variables measuring extent of disability and suffering before dying collected in CLHLS, we found that oldest-old men had a substantially higher chance of experiencing a non-suffering death than did their female counterparts. The age differences in life table proportions of disability and suffering before dying are not substantial. We also found that ADL status reported in the survey is a powerful predictor of the extent of subsequent disability and suffering before dying among the oldest-old (see Figures 3a and 3b)( Zeng, Gu, and Land 2003, presented at the Seminar on Increasing Longevity: Causes, Consequences and Prospects
New York, 20-22 October 2003 ).

Figure 3a. Life-Table Percentages of Slight Morbidity Before Death, by ADL Status at Initial Ages



Figure 3b. Life-Table Percentages of Severe Morbidity Before Death, by ADL Status at Initial Ages



Association of religious participation with health and survival at oldest-old ages. Elderly Chinese religious participants were mostly healthier than non-participants, adjusting for various confounding factors including health status one year ago (see Figure 4a); and the effects among oldest-old are substantially stronger than among younger elders. Risk of dying was 17-31% and 9-17% lower among the frequent and less frequent oldest-old participants than among the non-participants, adjusted for demographics, family/social support and health practices. After health conditions are adjusted, the survival benefits were reduced to 17% (p<0.05) and 4% (not significant) for frequent and less frequent oldest-old participants (Figure 4b). Survival benefits of religious participation among younger elderly were substantially stronger than that of the oldest-old. Oldest-old and younger elderly women are 2.2 and 2.6 times more likely to participate in religious activities than their counterpart men do, but no clear-cut pattern of gender differentials in the impacts of religious participation on health and survival was found (Zeng, Gu, and George 2007).



Optimism is one of the secrets of longevity. Our CLHLS data show that the percentage of elders who are active in daily living, have good capacity for physical performance, and have normal cognitive function drop dramatically from age 65-69 to 100-105. The percentage of elders reporting good satisfaction in current life, however, remains almost constant from age 65-69 to 80-84 and declines slightly afterwards (see Figure 5). This may suggest that being more positive in one’s outlook on life is one of the secrets of longevity (Zeng, Liu, and George 2003).


Protective effects of marriage on health and survival at oldest-old ages. Based on 20,105 individuals and more than 36,000 observations obtained from four waves in 1998, 2000, 2002, and 2005, we examined the effects of marital status, marriage quality, length of marriage, and widowhood on health and mortality at oldest-old ages. We found that currently married persons were more likely to have lower mortality risk, especially for men. Loss of spouse increases likelihood of mortality in men but not in women; and it increases risk of disability and reduces cognitive functioning for both men and women, even if such a loss occurred a long time ago. A good marriage lowers mortality risk and improves health for both men and women equally. A long-lasting marriage improves SRH for both men and women. Thus, we argue that if one wants to live longer and healthier, it is suggested to get and stay in a happy and long-lasting marriage. (Gu, Zhu, Zeng, and Xie, presented at the 2006 annual meeting of the GSA, Dallas, TX, Nov 16-20, 2006).

Leisure travel improves health, quality of life and mortality at old ages. Using the 2002 and 2005 waves of the Chinese Longitudinal Healthy Longevity Survey datasets, we analyzed factors determining leisure travel (e.g., tourist activities) in the past two years among the Chinese elderly and examined its impacts on health conditions, quality of life, and mortality after controlling for numerous covariates. Results show that those who were younger, healthier, male, living in an urban area, had a higher socioeconomic status, and involved in regular religious activities, exercise, and other leisure activities were more likely to engage in leisure travel. Multivariate logistic regression shows that leisure travel significantly reduces functional disability, cognitive impairment, self-rated poor health and poor life satisfaction by 40-60% after controlling for demographic factors. Although such protective effects are modified down to 25-50%after controlling controlling for family/social supports and connections and other health practice, they are still significant. Survival analyses further show that leisure travel could reduce risk of 3-year mortality by 36% after adjusting for demographic factors, although the effect is reduced to 19% after controlling for the baseline health and various other confounders. This suggests that leisure travel is an important tool for integrating and enhancing the quality of life of older adults, and thus is an effective way to accomplish the goal of healthy aging set by the WHO (Gu and Zeng, presented at the annual meeting of the Southern Demographic Association, Durham, NC, Nov 2-4, 2006).

Longer-lived parents would somewhat contribute to children’s exceptional longevity. This paper aims to investigate the relationship of exceptional longevity (survival to age 80 or older) of offspring to longevity of parents, by using data from the first three waves of the Chinese Longitudinal Healthy Longevity Survey. The Fixed-Attributes Dynamics method and logistic models are applied to the data. Results of both methods show a relatively strong same-sex inheritance of longevity between parents and offspring, that is, a strong association between father’s and son’s longevity or between mother’s and daughter’s longevity, but a weak or non-significant association of longevity between father and daughter or between mother and son. Having both a long-lived mother and father may contribute to children’s exceptional longevity (You, Gu, and Zeng, paper presented at the annual meeting of the Population Association of America, Los Angels, March 29-March 31, 2006).