Frailty and survival of older Chinese adults in urban and rural areas: Results from the Beijing Longitudinal Study of Aging
Article Outline
- Abstract
- 1. Introduction
- 2. Materials and methods
- 3. Results
- 4. Discussion
- 5. Conclusions
- Conflict of interest statement
- Role of the funding source
- Acknowledgements
- References
- Copyright
Abstract
Differences in frailty between rural and urban older adults have been demonstrated in developed countries. It is not understood how the apparently greater differences in living conditions between different types of regions in China may affect health and outcomes of older Chinese adults. Here, a frailty index (FI) based on the accumulation of health deficits was used to investigate health and survival differences in older Chinese men and women. We studied rural (n
=1121) and urban (n=2136) older adults (55–97 years old) in the Beijing Longitudinal Study of Aging (BLSA), of whom 48.9% (rural) and 35.4% (urban) died over 8 years of follow-up. The FI was generated from 35 self-reported health deficits. The mean FI increased exponentially with age (r2=0.87) and was higher in women than in men. The death rate increased significantly with increases in the FI, but women showed a lower death rate than did men. The mean FI in urban older adults (0.12±0.10) was lower than that in their rural counterparts (0.14±0.12, p<0.001). Urban dwellers showed better survival compared with their counterparts in the rural areas. Adjusted by age, sex, and education level, the hazard ratio for death for each increment of the FI was 1.28 for urban people and 1.27 for rural people. Chinese urban dwellers showed better health and survival than rural dwelling older adults. The FI readily summarized health and mortality differences among different geographic regions, reflecting the impact of the environment, socioeconomics, and medical services on deficit accumulation and on survival.
Keywords: Aging, Frailty index, Geographic areas, Rural, Urban, China, Frailty
1. Introduction
The term frailty is used to capture differences in health and survival of people of the same chronological age. This state of increased vulnerability to adverse health outcomes can be operationalized in many ways, including a phenotype approach (Fried et al., 2001) and a deficits accumulation approach, typically referred to as the FI (Mitnitski et al., 2001). The latter is especially useful in secondary data analyses, as it can be operationalized from virtually any health dataset. Using this approach, the level of frailty can be conveniently expressed as the number of deficits present over the total number of deficits considered (Rockwood and Mitnitski, 2011). The deficits measured in the FI construction can be of different types, including symptoms, diagnosis, medical history, disabilities, etc. No specific instrumentation or dataset-specific requirement is required, as long as each of the deficits included in the FI satisfies a few inclusion criteria (i.e., health-related, increasing with age, not saturating too early in the lifespan, <5% missing data; Searle et al., 2008). The FI has been applied in studies from many countries (Goggins et al., 2005, Mitnitski et al., 2005, Kulminski et al., 2007, Kulminski et al., 2008, Dupre et al., 2009, Gu et al., 2009, Armstrong et al., 2010). It has proved to be robust, in that some basic characteristics of the FI have been consistently reported. For example, deficits accumulate exponentially with age, at an average rate of 3% per year. The higher the value of the FI is, the more vulnerable people are to adverse health outcomes, including death, institutionalization and the use of hospital and community service (Mitnitski et al., 2001, Mitnitski et al., 2005, Goggins et al., 2005, Rockwood et al., 2006). Women accumulate more deficits than men on average; but at any level of the FI, men show higher mortality than do women (Rockwood and Mitnitski, 2011). The FI approach has also been used to compare group differences in health and adverse outcomes. For example, in the Canadian Study of Health and Aging, rural older adults were frailer (at all ages had higher mean FI scores) than did urban older adults and had also demonstrated a higher death rate (Song et al., 2007).
China has the world's largest population. In 2009, its population was 1.334 billion, with 712million (53.4%) rural and 622 million (46.6%) urban dwellers (Wikipedia: Urbanization in the People's Republic of China). Considerable differences in socioeconomic status, demographic characteristics, and health care utilization exist between rural and urban populations in China (Ying et al., 2007, Woo et al., 2010). In comparison, rural people have been found to be older, less educated, and less healthy, to have lower incomes and more unemployment than urban people. They also have less accessibility to affordable health care (Chau, 2010).
Recently, we used the FI in the BLSA to evaluate aging and health (Shi et al., 2011). Here, we extend the analyses to consider how residential conditions may influence frailty and its relationship to survival of older Chinese adults. In this secondary analysis of the BLSA database, we explore differences in health among people living in urban and rural areas. Our specific objectives were: to compare, in different geographic regions, (1) levels of frailty as a function of age, (2) differences in frailty and mortality, and (3) differences in survival between men and women in relation to frailty and geographical region.
2. Materials and methods
2.1. Data collection and study population
The BLSA is a prospective cohort study of 3257 community dwelling Chinese people aged 55 years and older at baseline. Distributions of gender, age group, and educational categories of the study sample represent those of older population in Beijing, as obtained from the Fourth National Census Data (Jiang et al., 2002). As described elsewhere (Tang et al., 1999, Jiang et al., 2002), the cohort was assembled in 1992, with follow-up every 2–3 years in 1994, 1997 and 2000 (Fig. 1). The survey was based on self-reported information, which covered demographic characteristics, socio-economic status, functional ability, life style, the use of medical services, physical health including diseases, psychological health, and cognitive status (e.g., the Mini Mental State Examination – MMSE). All information was collected at the respondent's home by trained interviewers, mostly nurses or physicians. The presence of disease was based on self-report questionnaires as to whether a subject has a particular disease as diagnosed by a physician.
Fig. 1.
Composition of BLSA cohort.
For the present study, variables from the baseline (1992) survey were used to construct the FI, while the 8-year survival outcomes (i.e., in 2000) were evaluated. Of the 3257 participants sampled at baseline (1992), 1593 (48.9%) were men and 1664 (51.1%) were women, with an average age of 70.1±9.0 years. By the year 2000, 1705 people were still alive, 1155 had died, and 397 people were lost to follow-up. Survival status was determined through interviews with surviving household members and with neighbors when surviving household members were not available and the information was verified in a subset of participants based on household registration records. Geographical regions were defined by socioeconomic, demographic, and geographical characteristics as detailed in previous publications (Tang et al., 1999, Jiang et al., 2002, Kaneda et al., 2005). Briefly, two districts within the Beijing municipality were represented in the data collection. Xuan Wu is an urban district located in the city center of Beijing. Rural regions were represented by Da Xing (rural plains district south of Beijing) and Huai Ruo (rural mountainous district northeast of Beijing).
2.2. Construction of the FI
The 35 variables included in the FI satisfied the criteria of being associated with health status, increasing with age, not saturating too early (i.e., not becoming fully prevalent at some young age), and having >1% prevalence and <5% missing values (Searle et al., 2008). The variables covered a range of health problems, including symptoms (n=7), basic and Instrumental Activities of Daily Living (ADL and IADL) disabilities (n=14), diseases (n=8), psychological problems (n=5), and the MMSE total score (Table 1). The questionnaires assessing health and cognition (e.g., MMSE) have been modified and validated to be applicable to Chinese people. Each variable was either a binary or a 3 scale variable. For the 15 binary variables, “0” was used to indicate the absence of the deficit, and “1” the presence of the deficit; for the remaining 20 3-item variables, an additional value “0.5” was used to represent “sometimes”. The FI was expressed as the proportion of deficits accumulated and was calculated for each individual as the number of the deficit items present over the total number of deficit items considered. For example, if a person had 6 of the 35 deficits considered in this study (e.g., hypertension, arthritis, requiring walking stick to walk, incapable moving 300m or using public transportation, and needing assistance with housework), this person's FI was calculated as 6/35=0.17. For individual cases with missing values, the missing variable was excluded from both the numerator and the denominator – participants had a denominator of 35; the lowest denominator used was 29, in 48 participants.
Table 1. Baseline characteristics and the percentage present of the FI-composing deficits in the rural and urban samples.
| Variables | Women | Men | ||
|---|---|---|---|---|
| Rural (n=572) | Urban (n=1092) | Rural (n=549) | Urban (n=1044) | |
| Age, mean (SD) | 70.1 (9.1) | 70.1 (9.4) | 70.4 (8.8) | 70.1 (8.7) |
| Junior high graduates | 0.3 | 19.4 | 4.2 | 45.1 |
| MMSE score, mean (SD) | 19.8 (3.8) | 23.2 (4.2) | 23.2 (3.8) | 25.5 (3.1) |
| Do not have much energy | 68.4 | 62.5 | 58.6 | 51.5 |
| Fell less useful | 84.3 | 63.2 | 69.9 | 47.3 |
| Do not feel a lot of fun in life | 38.2 | 38.8 | 37.3 | 37.4 |
| Do not feel very happy | 26.3 | 22.0 | 23.0 | 17.9 |
| Feel there is nothing to do | 29.0 | 22.8 | 25.2 | 15.8 |
| Hypertension | 12.8 | 26.0 | 9.1 | 24.0 |
| Coronary heart disease | 9.4 | 19.9 | 6.4 | 19.1 |
| Stroke | 4.7 | 4.0 | 4.9 | 8.1 |
| TIA/small stroke | 0.0 | 2.0 | 0.2 | 2.9 |
| Arthritis | 5.2 | 9.0 | 3.1 | 6.7 |
| Thyroid diseases | 0.3 | 2.4 | 0.0 | 1.0 |
| Glaucoma | 2.8 | 2.9 | 1.5 | 2.0 |
| Cataract | 5.2 | 16.0 | 3.5 | 15.6 |
| Urinary incontinence | 36.4 | 23.4 | 13.7 | 9.3 |
| Falls | 18.5 | 11.1 | 9.7 | 7.7 |
| Fracture | 5.2 | 10.8 | 4.9 | 6.0 |
| Tremor | 7.2 | 7.1 | 8.9 | 6.5 |
| Do not hear clearly | 19.1 | 16.8 | 23.1 | 21.3 |
| Wear a hearing aid | 0.5 | 1.3 | 0.2 | 3.0 |
| Use a walking stick | 31.2 | 16.1 | 24.2 | 13.5 |
| Need help with eating | 3.5 | 1.4 | 2.9 | 0.9 |
| Need help with grooming | 3.5 | 2.2 | 3.3 | 1.2 |
| Need help with dressing | 3.3 | 2.3 | 3.1 | 1.9 |
| Need help with getting on/off bed | 4.2 | 2.6 | 3.8 | 1.4 |
| Need help with bathing | 10.5 | 9.4 | 8.0 | 5.7 |
| Need help with moving in house | 4.9 | 3.2 | 4.4 | 1.8 |
| Need help with cooking meals | 22.7 | 14.0 | 26.2 | 14.6 |
| Need help with managing money | 25.0 | 11.9 | 16.0 | 7.6 |
| Need help with taking a bus | 46.2 | 31.2 | 23.1 | 15.7 |
| Need help with shopping | 26.2 | 17.2 | 15.8 | 10.2 |
| Need help with walking 300m | 19.9 | 14.0 | 11.8 | 6.5 |
| Need help with up/down stairs | 26.4 | 15.8 | 16.4 | 8.0 |
| Need help in running housework | 43.5 | 36.9 | 54.5 | 49.1 |
| Need any other personal care | 6.6 | 6.2 | 6.2 | 3.4 |
2.3. Analyses
For the 35 deficits, risk ratios for 8-year mortality were examined in a multivariable logistic regression model. Area-specific frailty distributions were estimated as the logarithm of the FI measured by 5-year aggregated intervals from age 55. The relationship between the FI and 8-year death rate was assessed using a logistic regression model. For each geographical region, the FI as a function of age was evaluated for men and women in each group separately using a logarithm model. In addition, survival probabilities of up to 8 years were assessed using Kaplan–Meier curves stratified by sex and geographical region. Cox proportional hazard model were applied to evaluate the effect of the FI on survival, adjusting for age, sex and education. Comparisons of the means among multiple sample groups were conducted using analysis of variance (ANOVA) and Student–Newman–Keuls’ pair-wise multiple comparisons. The Chi-squared test was used to compare percentages for categorical variables. Data processing and analyses were performed using SPSS version 15.0 and Matlab version v2007.
3. Results
Of the 3257 participants in the large Beijing region (the capital city of China), 1121 (34.4%) lived in rural areas and 2136 (65.6%) were urban dwellers. There are no statistically significant differences between respondents to all cycles and those lost to follow-up, with regard to gender, education level and dwelling areas.
There were no age differences between the urban and rural samples, but differences in other demographic features were evident (Table 1). Compared with rural dwellers, urban dwellers had higher MMSE scores and a higher level of education. People in urban areas also tended to have fewer psychological problems (e.g., low energy; feeling less useful; don’t feel very happy). Even so, they reported more diseases (e.g., hypertension; coronary artery disease; arthritis; thyroid disease; glaucoma). For almost all deficits, women showed a higher prevalence than did men in the same region. The prevalence of IADL disabilities was highest in rural women, and lowest in urban men.
In each area, most health deficits which make up the FI were individually not associated with a significantly increased risk of death (Table 2). Combined into the FI, however, the accumulation of the health deficits greatly increased the risk of death; the hazard ratio for death for each increment of the FI was 1.28 for urban people and 1.27 for rural people (Table 3). Being men had a higher risk of death in both areas with the ORs of 0.62 and 0.67, respectively. People of advanced ages had an increased risk of death in both geographical regions, while education did not affect the survival probability (Table 3).
Table 2. The odds ratio for 8-year death for each component variable of the FI, in a multivariable logistic regression model, adjusted for age and sex.
| Multivariable model | ||||
|---|---|---|---|---|
| Rural sample (n=1121) | Urban sample (n=2136) | |||
| Odds ratio | p | Odds ratio | p | |
| Age (year) | 1.10 | 0.00 | 1.14 | 0.00 |
| Gender (female) | 0.51 | 0.00 | 0.52 | 0.00 |
| Do not have much energy | 1.43 | 0.11 | 1.24 | 0.36 |
| Fell less useful | 1.57 | 0.07 | 1.05 | 0.84 |
| Do not feel a lot of fun in life | 0.90 | 0.72 | 1.47 | 0.18 |
| Do not feel very happy | 1.05 | 0.89 | 1.31 | 0.41 |
| Feel nothing to do | 1.27 | 0.46 | 1.63 | 0.15 |
| Hypertension | 1.42 | 0.24 | 1.77 | 0.01 |
| Coronary heart disease | 1.79 | 0.11 | 0.95 | 0.81 |
| Stroke | 1.38 | 0.61 | 1.36 | 0.45 |
| TIA/small stroke | – | – | 1.32 | 0.60 |
| Arthritis | 0.85 | 0.73 | 1.06 | 0.86 |
| Thyroid disease | – | – | 4.64 | 0.01 |
| Glaucoma | 0.36 | 0.19 | 1.61 | 0.39 |
| Cataract | 0.62 | 0.33 | 0.59 | 0.03 |
| Urinary incontinence | 1.16 | 0.52 | 1.35 | 0.20 |
| Falls | 0.75 | 0.33 | 0.67 | 0.21 |
| Fracture | 1.03 | 0.96 | 0.91 | 0.79 |
| Tremor | 1.18 | 0.65 | 0.72 | 0.35 |
| Do not hear clearly | 1.43 | 0.41 | 1.14 | 0.75 |
| Wear a hearing aid | 0.70 | 0.78 | 0.45 | 0.32 |
| Use a walking stick | 2.52 | 0.01 | 1.98 | 0.05 |
| Need help with eating | 0.00 | 1.00 | 0.00 | 1.00 |
| Need help with grooming | 0.61 | 1.00 | 0.00 | 1.00 |
| Need help with dressing | 0.00 | 1.00 | 0.00 | 1.00 |
| Need help with getting on/off bed | 0.00 | 1.00 | 0.00 | 0.99 |
| Need help with bathing | 0.72 | 0.77 | 0.24 | 0.10 |
| Need help with moving in house | 0.00 | 1.00 | 0.00 | 1.00 |
| Need help with cooking meals | 1.26 | 0.64 | 2.30 | 0.20 |
| Need help with managing money | 1.12 | 0.80 | 0.21 | 0.04 |
| Need help with taking a bus | 0.83 | 0.70 | 1.56 | 0.35 |
| Need help with shopping | 1.01 | 0.98 | 3.03 | 0.19 |
| Need help with walking 300m | 0.13 | 0.06 | 0.12 | 0.08 |
| Need help with up/down stairs | 4.62 | 0.13 | 5.03 | 0.12 |
| Need help in running housework | 1.44 | 0.09 | 1.13 | 0.53 |
| Need any other personal care | 0.39 | 0.27 | 0.79 | 0.83 |
| MMSE status | 2.94 | 0.01 | 3.91 | 0.00 |
Table 3. Cox proportional hazard model on the 8-year death rate for the rural and urban populations. The regression parameters were estimated with the covariates of the FI, age, sex, and education level.
| Subjects | Covariate | B | SE | Wald statistic | Hazard ratio (95% CI) |
|---|---|---|---|---|---|
| Rural sample | Age | 0.079 | 0.006 | 173.611 | 1.08 (1.07–1.09)** |
| n=1121 | Sex | −0.477 | 0.091 | 27.291 | 0.62 (0.52–0.74)** |
| Chi2=450.22 | Education | 0.141 | 0.323 | 0.191 | 1.15 (0.61–2.17) |
| p<0.001 | FI | 0.236 | 0.022 | 114.147 | 1.27 (1.21–1.32)** |
| Urban sample | Age | 0.084 | 0.005 | 253.041 | 1.09 (1.08–1.10)** |
| n=2136 | Sex | −0.395 | 0.085 | 21.877 | 0.67 (0.57–0.80)** |
| Chi2=670.44 | Education | −0.264 | 0.103 | 6.529 | 0.77 (0.63–0.94)* |
| p<0.001 | FI | 0.243 | 0.019 | 158.123 | 1.28 (1.23–1.32)** |
*p |
**p |
The mean level of the FI was lower for urban (0.117±0.105) than for rural (0.140±0.116) dwelling people (t=5.53, p<0.001). The FI showed a significant relationship with age (Fig. 2A), which was explained by an exponential function: ln(FI)=A+B×Age, where A=−4.580, B=0.035, r2=0.866, p<0.001. A higher FI was observed in women than in men in both regions (Fig. 2B). The slope of the increase of the FI with age was lower in women than in men in the rural region (B=0.034 vs. B=0.038, t=4.65, p<0.01). In contrast, in the urban region, women tended to accumulate deficits faster than did men (B=0.040 vs. B=0.035, t=3.86, p<0.01).
Fig. 2.
FI as a function of age analyzed for the entire study sample (A) and for rural and urban women and men separately (B). In (A), marks represent the observation data and lines represent exponential fit: FI
=A×eB×age, where A=0.092, B=0.036, r2=0.872.
The 8-year total death rates were 48.9% in rural and 35.4% in urban populations, which increased significantly with the increase of the FI (Fig. 3A). The average survival time over the 8-year survey was 70.7±32.4 months for rural dwellers and 77.4±29.7 for urban dwellers, respectively. Regardless of geographical region, the higher the FI was, the higher the rate of death (Fig. 3B). As expected, people living in the urban areas showed better survival compared with their counterparts from the rural areas (Log rank χ2=55.008, p<0.001). Compared with women, a dose-responsive decline of the survival probability was observed in men in both geographical regions (Fig. 4).
Fig. 3.
Eight-year death rate as a function of the FI for the entire study sample (A) and sub-samples based on sex and geographical region (B). In (A), marks represent the observation data and lines represent logistic regression fit: Death Rate
=A×eB×FI/(1+A×eB×FI), where A=0.259, B=6.865, r2=0.877.
Fig. 4.
The Kaplan–Meier cumulative survival probability for men and women in the rural and urban regions.
4. Discussion
In this study, a FI was calculated based on self-reported health deficit accumulation in the BLSA, and was used to compare the health status of older adults living in two different geographic areas. We found that on average, urban older adults appeared to be healthier than their rural counterparts. Although at all ages, women, on average, had more deficits than men, in each region they were less likely to die.
The trend of deficits accumulating with increasing age and the correlation of the FI with mortality is similar to findings reported by our group and many others (Rockwood and Mitnitski, 2006, Rockwood and Mitnitski, 2011, Woo et al., 2006, Ying et al., 2007, Kulminski et al., 2008, Dupre et al., 2009, Song et al., 2010, Shi et al., 2011). Also, better health in older Chinese adults living in urban areas is in keeping with work conducted in Canadian data where urban dwellers also fared better (Song et al., 2007).
Even though urban older Chinese were relatively healthier on average compared with their rural counterparts (e.g., fewer deficits, less mortality), urban dwellers in China reported comparatively more diseases. The mean values of diagnosed diseases in urban people were higher than in rural people, possibly reflecting that urban people might have better access to health care services for diagnosis and have better awareness of their health status (Liu et al., 2003). Regional difference in education did not significantly affect survival in a multivariate model (adjusted for age, sex, and health status) even though education has been reported to impact on health and survival (Lange, 2010). On the other hand, survival appeared to be affected significantly by MMSE status. In addition, the mean MMSE scores appeared to be low in this Chinese samples compared to those reported for community samples in West. These differences relating to education and MMSE suggest the inequity of rural and urban Chinese older adults in their access to education across the life course. They also reflect cultural and educational differences between Chinese and Western societies and perhaps the differences in the original and the modified Chinese version of the cognitive assessment tests.
Of note, most deficits did not show association with an increased risk of death when considered individually in a multivariate model. Even so, after they were combined in a FI, the cumulative effect of multiple seemingly insignificant effects was closely associated with mortality. This finding supports the observation that using the FI to aggregate measures of human health can better predict death than do individual health deficits (Song et al., 2007). It also appears to conform to the recent observation that small adverse contributions from allelic variations can sum to have significant impacts (Yashin et al., 2010).
Our data must be interpreted with caution. We focused on the health of the participants at the time of interview; however, we lacked information about the patterns of lifespan, migration, and health transitions among rural and urban older adults. Meanwhile, while the economy is growing rapidly, there still exists a big rural–urban gap in medical condition in China (Jian et al., 2010). For example, there are considerably fewer hospitals, physicians and other health care services in rural communities, and less accessibility and affordability of health care is often limited by transportation and low income in these regions. As the BLSA relied on self-reported information, some variables like disease diagnosis included in the FI may be biased, depending on local norms.
5. Conclusions
This analysis was undertaken as part of the Canada–China Collaboration on Aging and Longevity (Rockwood et al., 2009, Song et al., 2009). In summary, our data suggest that the FI, calculated as deficit accumulation, could be used to compare health and adverse outcomes across different demographic groups. We found that rural dwelling older Chinese adults were frailer than their urban counterparts, and correspondingly also had relatively higher death rates. The rural–urban differences in health and survival may reflect differences in the provision of health care services in different geographic regions in China, which might respond to improved health care accessibility in the less well-served regions.
Conflict of interest statement
Kenneth Rockwood will be applying for finding to commercialize a Clinical version of the FI. No other author of the article has a conflict of interest relating to this work.
Role of the funding source
Data collection was funded by Beijing Geriatric Clinical and Research Center at Beijing Xuanwu Hospital of Capital Medical University. Data analysis was funded by China-Canada Joint Health Research Initiative Program by the Canadian Institutes for Health Research and the National Natural Science Foundation. The funding agents were not involved in the study design, data processing and analysis, result presentation and interpretation, and manuscript preparation and submission.
Acknowledgements
This research was also supported by operating grants from the Canadian Institutes for Health Research and the Fountain Innovation Fund of the Queen Elizabeth II Health Sciences Research Foundation. Kenneth Rockwood receives career support through the Dalhousie Medical Research Foundation as the Kathryn Allen Weldon Professor of Alzheimer Research. The Canada–China Collaboration is funded jointly by the Canadian Institutes for Health Research and the National Natural Science Foundation of China (CIHR CCI-92216: MOP62823 and NSFC30811120439). Collection of data used in this study was funded by Beijing Geriatric Clinical and Research Center at Beijing Xuanwu Hospital of Capital Medical University.
References
- . Examining three frailty conceptualizations in their ability to predict negative outcomes for home-care clients. Age Ageing. 2010;39:755–758
- . Ecological analysis of health care utilisation for China's rural population: association with a rural county's socioeconomic characteristics. BMC Public Health. 2010;10:664
- . Frailty and type of death among older adults in China: prospective cohort study. Br. Med. J. 2009;338:b1175
- . Frailty in older adults: evidence for a phenotype. J. Gerontol. A: Biol. Sci. Med. Sci. 2001;56:M146–M156
- . Frailty index as a measure of biological age in a Chinese population. J. Gerontol. A: Biol. Sci. Med. Sci. 2005;60:1046–1051
- . Frailty and mortality among Chinese at advanced ages. J. Gerontol. B: Psychol. Sci. Soc. Sci. 2009;64:279–289
- . China's rural–urban care gap shrank for chronic disease patients, but inequities persist. Health Affairs (Millwood). 2010;29:2189–2196
- . Demographic determinants for change in activities of daily living: a cohort study of the elderly people in Beijing. J. Epidemiol. 2002;12:280–286
- . Socioeconomic status differentials in life and active life expectancy among older adults in Beijing. Disabil. Rehabil. 2005;27:241–251
- . Cumulative index of health deficiencies as a characteristic of long life. J. Am. Geriatr. Soc. 2007;55:935–940
- . Cumulative deficits better characterize susceptibility to death in elderly people than phenotypic frailty: lessons from the Cardiovascular Health Study. J. Am. Geriatr. Soc. 2008;56:898–903
- . The role of education in complex health decisions: evidence from cancer screening.. J. Health Econ. 2010;16:(Epub ahead of print)
- . Medical expenditure and rural impoverishment in China. J. Health Popul. Nutr. 2003;21:216–222
- . Accumulation of deficits as a proxy measure of aging. Sci. World J. 2001;1:323–336
- . Relative fitness and frailty of elderly men and women in developed countries and their relationship with mortality. J. Am. Geriatr. Soc. 2005;53:2184–2189
- . Limits to deficit accumulation in elderly people. Mech. Ageing. 2006;127:494–496
- . Frailty defined by deficit accumulation and geriatric medicine defined by frailty. Clin. Geriatr. Med. 2011;27:17–26
- . Long-term risks of death and institutionalization of elderly people in relation to deficit accumulation at age 70. J. Am. Geriatr. Soc. 2006;54:975–979
- . Geriatric medicine and the care of frail elderly people. Chin. J. Geriatr. 2009;28:353–365
- . A standard procedure for creating a frailty index. BMC Geriatr. 2008;8:24
- . Analysis of frailty and survival in the Beijing Longitudinal Study of Aging. BMC Geriatr. 2011;20:17
- . Frailty and survival of rural and urban seniors: results from the Canadian Study of Health and Aging. Aging Clin. Exp. Res. 2007;19:145–153
- . Quantification of health and ageing in Canada and China: a Canada–China joint health research initiative project. Chin. J. Geriatr. 2009;28:793–802
- . Prevalence and 10-year outcomes of frailty in older adults in relation to deficit accumulation. J. Am. Geriatr. Soc. 2010;58:681–687
- . The prevalence of functional disability in activities of daily living among elderly Beijing Chinese. Arch. Gerontol. Geriatr. 1999;29:115–125
- Wikipedia: Urbanization in the People's Republic of China. http://en.wikipedia.org/wiki/Urbanization_in_China.
- . Public health significance of the frailty index. Disabil. Rehabil. 2006;28:515–521
- . Relative contributions of geographic, socioeconomic, and lifestyle factors to quality of life, frailty, and mortality in elderly. PLoS One. 2010;5:e8775
- . Joint influence of small-effect genetic variants on human longevity. Aging (Albany, NY). 2010;2:612–620
- . Demand for private health insurance in Chinese urban areas. Health Econ. 2007;16:1041–1050
PII: S0167-4943(11)00104-X
doi:10.1016/j.archger.2011.04.020
© 2011 Elsevier Ireland Ltd. All rights reserved.
